848 files changed, 46830 insertions, 64525 deletions

diff --git a/src/egl/drivers/demo/demo.c b/src/egl/drivers/demo/demo.c
index aea4894448..0933c0bdaa 100644
--- a/src/egl/drivers/demo/demo.c
+++ b/src/egl/drivers/demo/demo.c
@@ -177,7 +177,7 @@ demoCreatePixmapSurface(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf, Nat
       }
    }
 
-   if (conf->Attrib[EGL_SURFACE_TYPE - FIRST_ATTRIB] == 0) {
+   if (GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE) == 0) {
       _eglError(EGL_BAD_MATCH, "eglCreatePixmapSurface");
       return NULL;
    }
diff --git a/src/egl/drivers/glx/egl_glx.c b/src/egl/drivers/glx/egl_glx.c
index 4685f600e2..96292b0e9e 100644
--- a/src/egl/drivers/glx/egl_glx.c
+++ b/src/egl/drivers/glx/egl_glx.c
@@ -33,23 +33,12 @@
  * Authors: Alan Hourihane <alanh@tungstengraphics.com>
  */
 
-/*
- * TODO: 
- *
- * test eglBind/ReleaseTexImage
- */
-
-
-#include <assert.h>
-#include <stdio.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <fcntl.h>
-#include "dlfcn.h"
+#include <stdlib.h>
+#include <string.h>
 #include <X11/Xlib.h>
-#include <GL/gl.h>
-#include "glxclient.h"
+#include <GL/glx.h>
 
+#include "eglconfigutil.h"
 #include "eglconfig.h"
 #include "eglcontext.h"
 #include "egldisplay.h"
@@ -58,46 +47,19 @@
 #include "egllog.h"
 #include "eglsurface.h"
 
-#include <GL/gl.h>
-
 #define CALLOC_STRUCT(T)   (struct T *) calloc(1, sizeof(struct T))
+#define ARRAY_SIZE(a)      (sizeof(a) / sizeof(a[0]))
 
-static const EGLint all_apis = (EGL_OPENGL_ES_BIT
-                                | EGL_OPENGL_ES2_BIT
-                                | EGL_OPENVG_BIT
-                                /* | EGL_OPENGL_BIT */); /* can't do */
+#ifndef GLX_VERSION_1_4
+#error "GL/glx.h must be equal to or greater than GLX 1.4"
+#endif
+
+/*
+ * report OpenGL ES bits because apps usually forget to specify
+ * EGL_RENDERABLE_TYPE when choosing configs
+ */
+#define GLX_EGL_APIS (EGL_OPENGL_BIT | EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT)
 
-struct visual_attribs
-{
-   /* X visual attribs */
-   int id;
-   int klass;
-   int depth;
-   int redMask, greenMask, blueMask;
-   int colormapSize;
-   int bitsPerRGB;
-
-   /* GL visual attribs */
-   int supportsGL;
-   int transparentType;
-   int transparentRedValue;
-   int transparentGreenValue;
-   int transparentBlueValue;
-   int transparentAlphaValue;
-   int transparentIndexValue;
-   int bufferSize;
-   int level;
-   int render_type;
-   int doubleBuffer;
-   int stereo;
-   int auxBuffers;
-   int redSize, greenSize, blueSize, alphaSize;
-   int depthSize;
-   int stencilSize;
-   int accumRedSize, accumGreenSize, accumBlueSize, accumAlphaSize;
-   int numSamples, numMultisample;
-   int visualCaveat;
-};
 
 /** subclass of _EGLDriver */
 struct GLX_egl_driver
@@ -114,6 +76,21 @@ struct GLX_egl_display
    GLXFBConfig *fbconfigs;
 
    int glx_maj, glx_min;
+
+   const char *extensions;
+   EGLBoolean have_1_3;
+   EGLBoolean have_make_current_read;
+   EGLBoolean have_fbconfig;
+   EGLBoolean have_pbuffer;
+
+   /* GLX_SGIX_pbuffer */
+   PFNGLXCREATEGLXPBUFFERSGIXPROC glXCreateGLXPbufferSGIX;
+   PFNGLXDESTROYGLXPBUFFERSGIXPROC glXDestroyGLXPbufferSGIX;
+
+   /* workaround quirks of different GLX implementations */
+   EGLBoolean single_buffered_quirk;
+   EGLBoolean glx_window_quirk;
+
 };
 
 
@@ -131,7 +108,8 @@ struct GLX_egl_surface
 {
    _EGLSurface Base;   /**< base class */
 
-   GLXDrawable drawable;
+   Drawable drawable;
+   GLXDrawable glx_drawable;
 };
 
 
@@ -139,6 +117,7 @@ struct GLX_egl_surface
 struct GLX_egl_config
 {
    _EGLConfig Base;   /**< base class */
+   EGLBoolean double_buffered;
    int index;
 };
 
@@ -173,288 +152,369 @@ GLX_egl_config_index(_EGLConfig *conf)
    return ((struct GLX_egl_config *) conf)->index;
 }
 
-static GLboolean
-get_visual_attribs(Display *dpy, XVisualInfo *vInfo,
-                   struct visual_attribs *attribs)
-{
-   const char *ext = glXQueryExtensionsString(dpy, vInfo->screen);
-   int rgba;
 
-   memset(attribs, 0, sizeof(struct visual_attribs));
+#define MAP_ATTRIB(attr, memb) \
+   { attr, offsetof(__GLcontextModes, memb) }
+
+
+static const struct {
+   int attr;
+   int offset;
+} fbconfig_attributes[] = {
+   /* table 3.1 of GLX 1.4 */
+   MAP_ATTRIB(GLX_FBCONFIG_ID,                  fbconfigID),
+   MAP_ATTRIB(GLX_BUFFER_SIZE,                  rgbBits),
+   MAP_ATTRIB(GLX_LEVEL,                        level),
+   MAP_ATTRIB(GLX_DOUBLEBUFFER,                 doubleBufferMode),
+   MAP_ATTRIB(GLX_STEREO,                       stereoMode),
+   MAP_ATTRIB(GLX_AUX_BUFFERS,                  numAuxBuffers),
+   MAP_ATTRIB(GLX_RED_SIZE,                     redBits),
+   MAP_ATTRIB(GLX_GREEN_SIZE,                   greenBits),
+   MAP_ATTRIB(GLX_BLUE_SIZE,                    blueBits),
+   MAP_ATTRIB(GLX_ALPHA_SIZE,                   alphaBits),
+   MAP_ATTRIB(GLX_DEPTH_SIZE,                   depthBits),
+   MAP_ATTRIB(GLX_STENCIL_SIZE,                 stencilBits),
+   MAP_ATTRIB(GLX_ACCUM_RED_SIZE,               accumRedBits),
+   MAP_ATTRIB(GLX_ACCUM_GREEN_SIZE,             accumGreenBits),
+   MAP_ATTRIB(GLX_ACCUM_BLUE_SIZE,              accumBlueBits),
+   MAP_ATTRIB(GLX_ACCUM_ALPHA_SIZE,             accumAlphaBits),
+   MAP_ATTRIB(GLX_SAMPLE_BUFFERS,               sampleBuffers),
+   MAP_ATTRIB(GLX_SAMPLES,                      samples),
+   MAP_ATTRIB(GLX_RENDER_TYPE,                  renderType),
+   MAP_ATTRIB(GLX_DRAWABLE_TYPE,                drawableType),
+   MAP_ATTRIB(GLX_X_RENDERABLE,                 xRenderable),
+   MAP_ATTRIB(GLX_X_VISUAL_TYPE,                visualType),
+   MAP_ATTRIB(GLX_CONFIG_CAVEAT,                visualRating),
+   MAP_ATTRIB(GLX_TRANSPARENT_TYPE,             transparentPixel),
+   MAP_ATTRIB(GLX_TRANSPARENT_INDEX_VALUE,      transparentIndex),
+   MAP_ATTRIB(GLX_TRANSPARENT_RED_VALUE,        transparentRed),
+   MAP_ATTRIB(GLX_TRANSPARENT_GREEN_VALUE,      transparentGreen),
+   MAP_ATTRIB(GLX_TRANSPARENT_BLUE_VALUE,       transparentBlue),
+   MAP_ATTRIB(GLX_TRANSPARENT_ALPHA_VALUE,      transparentAlpha),
+   MAP_ATTRIB(GLX_MAX_PBUFFER_WIDTH,            maxPbufferWidth),
+   MAP_ATTRIB(GLX_MAX_PBUFFER_HEIGHT,           maxPbufferHeight),
+   MAP_ATTRIB(GLX_MAX_PBUFFER_PIXELS,           maxPbufferPixels),
+   MAP_ATTRIB(GLX_VISUAL_ID,                    visualID),
+};
 
-   attribs->id = vInfo->visualid;
-#if defined(__cplusplus) || defined(c_plusplus)
-   attribs->klass = vInfo->c_class;
-#else
-   attribs->klass = vInfo->class;
-#endif
-   attribs->depth = vInfo->depth;
-   attribs->redMask = vInfo->red_mask;
-   attribs->greenMask = vInfo->green_mask;
-   attribs->blueMask = vInfo->blue_mask;
-   attribs->colormapSize = vInfo->colormap_size;
-   attribs->bitsPerRGB = vInfo->bits_per_rgb;
-
-   if (glXGetConfig(dpy, vInfo, GLX_USE_GL, &attribs->supportsGL) != 0 ||
-       !attribs->supportsGL)
-      return GL_FALSE;
-   glXGetConfig(dpy, vInfo, GLX_BUFFER_SIZE, &attribs->bufferSize);
-   glXGetConfig(dpy, vInfo, GLX_LEVEL, &attribs->level);
-   glXGetConfig(dpy, vInfo, GLX_RGBA, &rgba);
-   if (!rgba)
-      return GL_FALSE;
-   attribs->render_type = GLX_RGBA_BIT;
-   
-   glXGetConfig(dpy, vInfo, GLX_DOUBLEBUFFER, &attribs->doubleBuffer);
-   if (!attribs->doubleBuffer)
-      return GL_FALSE;
-
-   glXGetConfig(dpy, vInfo, GLX_STEREO, &attribs->stereo);
-   glXGetConfig(dpy, vInfo, GLX_AUX_BUFFERS, &attribs->auxBuffers);
-   glXGetConfig(dpy, vInfo, GLX_RED_SIZE, &attribs->redSize);
-   glXGetConfig(dpy, vInfo, GLX_GREEN_SIZE, &attribs->greenSize);
-   glXGetConfig(dpy, vInfo, GLX_BLUE_SIZE, &attribs->blueSize);
-   glXGetConfig(dpy, vInfo, GLX_ALPHA_SIZE, &attribs->alphaSize);
-   glXGetConfig(dpy, vInfo, GLX_DEPTH_SIZE, &attribs->depthSize);
-   glXGetConfig(dpy, vInfo, GLX_STENCIL_SIZE, &attribs->stencilSize);
-   glXGetConfig(dpy, vInfo, GLX_ACCUM_RED_SIZE, &attribs->accumRedSize);
-   glXGetConfig(dpy, vInfo, GLX_ACCUM_GREEN_SIZE, &attribs->accumGreenSize);
-   glXGetConfig(dpy, vInfo, GLX_ACCUM_BLUE_SIZE, &attribs->accumBlueSize);
-   glXGetConfig(dpy, vInfo, GLX_ACCUM_ALPHA_SIZE, &attribs->accumAlphaSize);
-
-   /* get transparent pixel stuff */
-   glXGetConfig(dpy, vInfo,GLX_TRANSPARENT_TYPE, &attribs->transparentType);
-   if (attribs->transparentType == GLX_TRANSPARENT_RGB) {
-     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_RED_VALUE, &attribs->transparentRedValue);
-     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_GREEN_VALUE, &attribs->transparentGreenValue);
-     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_BLUE_VALUE, &attribs->transparentBlueValue);
-     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_ALPHA_VALUE, &attribs->transparentAlphaValue);
-   }
-   else if (attribs->transparentType == GLX_TRANSPARENT_INDEX) {
-     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_INDEX_VALUE, &attribs->transparentIndexValue);
-   }
-
-   /* multisample attribs */
-#ifdef GLX_ARB_multisample
-   if (ext && strstr(ext, "GLX_ARB_multisample")) {
-      glXGetConfig(dpy, vInfo, GLX_SAMPLE_BUFFERS_ARB, &attribs->numMultisample);
-      glXGetConfig(dpy, vInfo, GLX_SAMPLES_ARB, &attribs->numSamples);
-   }
-#endif
-   else {
-      attribs->numSamples = 0;
-      attribs->numMultisample = 0;
-   }
 
-#if defined(GLX_EXT_visual_rating)
-   if (ext && strstr(ext, "GLX_EXT_visual_rating")) {
-      glXGetConfig(dpy, vInfo, GLX_VISUAL_CAVEAT_EXT, &attribs->visualCaveat);
-   }
-   else {
-      attribs->visualCaveat = GLX_NONE_EXT;
+static EGLBoolean
+convert_fbconfig(Display *dpy, GLXFBConfig fbconfig,
+                 struct GLX_egl_config *GLX_conf)
+{
+   __GLcontextModes mode;
+   int err = 0, attr, val, i;
+
+   memset(&mode, 0, sizeof(mode));
+
+   for (i = 0; i < ARRAY_SIZE(fbconfig_attributes); i++) {
+      int offset = fbconfig_attributes[i].offset;
+      attr = fbconfig_attributes[i].attr;
+      err = glXGetFBConfigAttrib(dpy, fbconfig, attr, &val);
+      if (err) {
+         if (err == GLX_BAD_ATTRIBUTE) {
+            err = 0;
+            continue;
+         }
+         break;
+      }
+      *((int *) ((char *) &mode + offset)) = val;
    }
-#else
-   attribs->visualCaveat = 0;
-#endif
-
-   return GL_TRUE;
-}
+   if (err)
+      return EGL_FALSE;
 
-#ifdef GLX_VERSION_1_3
+   /* must have rgba bit */
+   if (!(mode.renderType & GLX_RGBA_BIT))
+      return EGL_FALSE;
 
-static int
-glx_token_to_visual_class(int visual_type)
-{
-   switch (visual_type) {
-   case GLX_TRUE_COLOR:
-      return TrueColor;
-   case GLX_DIRECT_COLOR:
-      return DirectColor;
-   case GLX_PSEUDO_COLOR:
-      return PseudoColor;
-   case GLX_STATIC_COLOR:
-      return StaticColor;
-   case GLX_GRAY_SCALE:
-      return GrayScale;
-   case GLX_STATIC_GRAY:
-      return StaticGray;
-   case GLX_NONE:
-   default:
-      return None;
+   /* pixmap and pbuffer surfaces must be single-buffered in EGL */
+   if (mode.doubleBufferMode) {
+      mode.drawableType &= ~(GLX_PIXMAP_BIT | GLX_PBUFFER_BIT);
+      if (!mode.drawableType)
+         return EGL_FALSE;
    }
+
+   mode.rgbMode = GL_TRUE;
+   mode.haveAccumBuffer = (mode.accumRedBits +
+                           mode.accumGreenBits +
+                           mode.accumBlueBits +
+                           mode.accumAlphaBits > 0);
+   mode.haveDepthBuffer = (mode.depthBits > 0);
+   mode.haveStencilBuffer = (mode.stencilBits > 0);
+
+   GLX_conf->double_buffered = (mode.doubleBufferMode != 0);
+   return _eglConfigFromContextModesRec(&GLX_conf->Base, &mode,
+                                        GLX_EGL_APIS, GLX_EGL_APIS);
 }
 
-static int
-get_fbconfig_attribs(Display *dpy, GLXFBConfig fbconfig,
-		     struct visual_attribs *attribs)
-{
-   int visual_type;
-   int fbconfig_id;
 
-   memset(attribs, 0, sizeof(struct visual_attribs));
+static const struct {
+   int attr;
+   int offset;
+} visual_attributes[] = {
+   /* table 3.7 of GLX 1.4 */
+   /* no GLX_USE_GL */
+   MAP_ATTRIB(GLX_BUFFER_SIZE,         rgbBits),
+   MAP_ATTRIB(GLX_LEVEL,               level),
+   MAP_ATTRIB(GLX_RGBA,                rgbMode),
+   MAP_ATTRIB(GLX_DOUBLEBUFFER,        doubleBufferMode),
+   MAP_ATTRIB(GLX_STEREO,              stereoMode),
+   MAP_ATTRIB(GLX_AUX_BUFFERS,         numAuxBuffers),
+   MAP_ATTRIB(GLX_RED_SIZE,            redBits),
+   MAP_ATTRIB(GLX_GREEN_SIZE,          greenBits),
+   MAP_ATTRIB(GLX_BLUE_SIZE,           blueBits),
+   MAP_ATTRIB(GLX_ALPHA_SIZE,          alphaBits),
+   MAP_ATTRIB(GLX_DEPTH_SIZE,          depthBits),
+   MAP_ATTRIB(GLX_STENCIL_SIZE,        stencilBits),
+   MAP_ATTRIB(GLX_ACCUM_RED_SIZE,      accumRedBits),
+   MAP_ATTRIB(GLX_ACCUM_GREEN_SIZE,    accumGreenBits),
+   MAP_ATTRIB(GLX_ACCUM_BLUE_SIZE,     accumBlueBits),
+   MAP_ATTRIB(GLX_ACCUM_ALPHA_SIZE,    accumAlphaBits),
+   MAP_ATTRIB(GLX_SAMPLE_BUFFERS,      sampleBuffers),
+   MAP_ATTRIB(GLX_SAMPLES,             samples),
+   MAP_ATTRIB(GLX_FBCONFIG_ID,         fbconfigID),
+   /* GLX_EXT_visual_rating */
+   MAP_ATTRIB(GLX_VISUAL_CAVEAT_EXT,   visualRating),
+};
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_FBCONFIG_ID, &fbconfig_id);
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_VISUAL_ID, &attribs->id);
+static int
+get_visual_type(const XVisualInfo *vis)
+{
+   int klass;
 
-#if 0
-   attribs->depth = vInfo->depth;
-   attribs->redMask = vInfo->red_mask;
-   attribs->greenMask = vInfo->green_mask;
-   attribs->blueMask = vInfo->blue_mask;
-   attribs->colormapSize = vInfo->colormap_size;
-   attribs->bitsPerRGB = vInfo->bits_per_rgb;
+#if defined(__cplusplus) || defined(c_plusplus)
+   klass = vis->c_class;
+#else
+   klass = vis->class;
 #endif
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_X_VISUAL_TYPE, &visual_type);
-   attribs->klass = glx_token_to_visual_class(visual_type);
+   switch (klass) {
+   case TrueColor:
+      return GLX_TRUE_COLOR;
+   case DirectColor:
+      return GLX_DIRECT_COLOR;
+   case PseudoColor:
+      return GLX_PSEUDO_COLOR;
+   case StaticColor:
+      return GLX_STATIC_COLOR;
+   case GrayScale:
+      return GLX_GRAY_SCALE;
+   case StaticGray:
+      return GLX_STATIC_GRAY;
+   default:
+      return GLX_NONE;
+   }
+}
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_BUFFER_SIZE, &attribs->bufferSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_LEVEL, &attribs->level);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_RENDER_TYPE, &attribs->render_type);
-   if (!(attribs->render_type & GLX_RGBA_BIT))
-      return 0;
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_DOUBLEBUFFER, &attribs->doubleBuffer);
-   if (!attribs->doubleBuffer)
-      return 0;
+static EGLBoolean
+convert_visual(Display *dpy, XVisualInfo *vinfo,
+               struct GLX_egl_config *GLX_conf)
+{
+   __GLcontextModes mode;
+   int err, attr, val, i;
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_STEREO, &attribs->stereo);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_AUX_BUFFERS, &attribs->auxBuffers);
+   /* the visual must support OpenGL */
+   err = glXGetConfig(dpy, vinfo, GLX_USE_GL, &val);
+   if (err || !val)
+      return EGL_FALSE;
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_RED_SIZE, &attribs->redSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_GREEN_SIZE, &attribs->greenSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_BLUE_SIZE, &attribs->blueSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_ALPHA_SIZE, &attribs->alphaSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_DEPTH_SIZE, &attribs->depthSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_STENCIL_SIZE, &attribs->stencilSize);
+   memset(&mode, 0, sizeof(mode));
+
+   for (i = 0; i < ARRAY_SIZE(visual_attributes); i++) {
+      int offset = visual_attributes[i].offset;
+      attr = visual_attributes[i].attr;
+      err = glXGetConfig(dpy, vinfo, attr, &val);
+      if (err) {
+         if (err == GLX_BAD_ATTRIBUTE) {
+            err = 0;
+            continue;
+         }
+         break;
+      }
+      *((int *) ((char *) &mode + offset)) = val;
+   }
+   if (err)
+      return EGL_FALSE;
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_ACCUM_RED_SIZE, &attribs->accumRedSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_ACCUM_GREEN_SIZE, &attribs->accumGreenSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_ACCUM_BLUE_SIZE, &attribs->accumBlueSize);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_ACCUM_ALPHA_SIZE, &attribs->accumAlphaSize);
+   /* must be RGB mode */
+   if (!mode.rgbMode)
+      return EGL_FALSE;
 
-   /* get transparent pixel stuff */
-   glXGetFBConfigAttrib(dpy, fbconfig,GLX_TRANSPARENT_TYPE, &attribs->transparentType);
-   if (attribs->transparentType == GLX_TRANSPARENT_RGB) {
-     glXGetFBConfigAttrib(dpy, fbconfig, GLX_TRANSPARENT_RED_VALUE, &attribs->transparentRedValue);
-     glXGetFBConfigAttrib(dpy, fbconfig, GLX_TRANSPARENT_GREEN_VALUE, &attribs->transparentGreenValue);
-     glXGetFBConfigAttrib(dpy, fbconfig, GLX_TRANSPARENT_BLUE_VALUE, &attribs->transparentBlueValue);
-     glXGetFBConfigAttrib(dpy, fbconfig, GLX_TRANSPARENT_ALPHA_VALUE, &attribs->transparentAlphaValue);
-   }
-   else if (attribs->transparentType == GLX_TRANSPARENT_INDEX) {
-     glXGetFBConfigAttrib(dpy, fbconfig, GLX_TRANSPARENT_INDEX_VALUE, &attribs->transparentIndexValue);
-   }
+   mode.visualID = vinfo->visualid;
+   mode.visualType = get_visual_type(vinfo);
+   mode.redMask = vinfo->red_mask;
+   mode.greenMask = vinfo->green_mask;
+   mode.blueMask = vinfo->blue_mask;
+
+   mode.drawableType = GLX_WINDOW_BIT;
+   /* pixmap surfaces must be single-buffered in EGL */
+   if (!mode.doubleBufferMode)
+      mode.drawableType |= GLX_PIXMAP_BIT;
+
+   mode.renderType = GLX_RGBA_BIT;
+   mode.xRenderable = GL_TRUE;
+   mode.haveAccumBuffer = (mode.accumRedBits +
+                           mode.accumGreenBits +
+                           mode.accumBlueBits +
+                           mode.accumAlphaBits > 0);
+   mode.haveDepthBuffer = (mode.depthBits > 0);
+   mode.haveStencilBuffer = (mode.stencilBits > 0);
+
+   GLX_conf->double_buffered = (mode.doubleBufferMode != 0);
+   return _eglConfigFromContextModesRec(&GLX_conf->Base, &mode,
+                                        GLX_EGL_APIS, GLX_EGL_APIS);
+}
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_SAMPLE_BUFFERS, &attribs->numMultisample);
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_SAMPLES, &attribs->numSamples);
 
-   glXGetFBConfigAttrib(dpy, fbconfig, GLX_CONFIG_CAVEAT, &attribs->visualCaveat);
+static void
+fix_config(struct GLX_egl_display *GLX_dpy, struct GLX_egl_config *GLX_conf)
+{
+   _EGLConfig *conf = &GLX_conf->Base;
+   EGLint surface_type, r, g, b, a;
+
+   surface_type = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE);
+   if (!GLX_conf->double_buffered && GLX_dpy->single_buffered_quirk) {
+      /* some GLX impls do not like single-buffered window surface */
+      surface_type &= ~EGL_WINDOW_BIT;
+      /* pbuffer bit is usually not set */
+      if (GLX_dpy->have_pbuffer)
+         surface_type |= EGL_PBUFFER_BIT;
+      SET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE, surface_type);
+   }
 
-   if (attribs->id == 0) {
-      attribs->id = fbconfig_id;
-      return EGL_PBUFFER_BIT | EGL_PIXMAP_BIT;
+   /* no visual attribs unless window bit is set */
+   if (!(surface_type & EGL_WINDOW_BIT)) {
+      SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_ID, 0);
+      SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE, EGL_NONE);
    }
 
-   return EGL_WINDOW_BIT;
+   /* make sure buffer size is set correctly */
+   r = GET_CONFIG_ATTRIB(conf, EGL_RED_SIZE);
+   g = GET_CONFIG_ATTRIB(conf, EGL_GREEN_SIZE);
+   b = GET_CONFIG_ATTRIB(conf, EGL_BLUE_SIZE);
+   a = GET_CONFIG_ATTRIB(conf, EGL_ALPHA_SIZE);
+   SET_CONFIG_ATTRIB(conf, EGL_BUFFER_SIZE, r + g + b + a);
 }
 
-#endif
 
 static EGLBoolean
-create_configs(_EGLDisplay *disp, struct GLX_egl_display *GLX_dpy)
+create_configs(_EGLDisplay *dpy, struct GLX_egl_display *GLX_dpy,
+               EGLint screen)
 {
-   XVisualInfo theTemplate;
-   int numVisuals;
-   long mask;
-   int i;
-   struct visual_attribs attribs;
+   EGLint num_configs = 0, i;
+   EGLint id = 1;
 
-   GLX_dpy->fbconfigs = NULL;
-
-#ifdef GLX_VERSION_1_3
-   /* get list of all fbconfigs on this screen */
-   GLX_dpy->fbconfigs = glXGetFBConfigs(GLX_dpy->dpy, DefaultScreen(GLX_dpy->dpy), &numVisuals);
+   if (GLX_dpy->have_fbconfig) {
+      GLX_dpy->fbconfigs = glXGetFBConfigs(GLX_dpy->dpy, screen, &num_configs);
+   }
+   else {
+      XVisualInfo vinfo_template;
+      long mask;
 
-   if (numVisuals == 0) {
-      GLX_dpy->fbconfigs = NULL;
-      goto xvisual;
+      vinfo_template.screen = screen;
+      mask = VisualScreenMask;
+      GLX_dpy->visuals = XGetVisualInfo(GLX_dpy->dpy, mask, &vinfo_template,
+                                        &num_configs);
    }
 
-   for (i = 0; i < numVisuals; i++) {
-      struct GLX_egl_config *config;
-      int bit;
+   if (!num_configs)
+      return EGL_FALSE;
 
-      bit = get_fbconfig_attribs(GLX_dpy->dpy, GLX_dpy->fbconfigs[i], &attribs);
-      if (!bit)
+   for (i = 0; i < num_configs; i++) {
+      struct GLX_egl_config *GLX_conf, template;
+      EGLBoolean ok;
+
+      memset(&template, 0, sizeof(template));
+      _eglInitConfig(&template.Base, id);
+      if (GLX_dpy->have_fbconfig)
+         ok = convert_fbconfig(GLX_dpy->dpy, GLX_dpy->fbconfigs[i], &template);
+      else
+         ok = convert_visual(GLX_dpy->dpy, &GLX_dpy->visuals[i], &template);
+      if (!ok)
+        continue;
+
+      fix_config(GLX_dpy, &template);
+      if (!_eglValidateConfig(&template.Base, EGL_FALSE)) {
+         _eglLog(_EGL_DEBUG, "GLX: failed to validate config %d", i);
          continue;
+      }
 
-      config = CALLOC_STRUCT(GLX_egl_config);
-
-      config->index = i;
-      _eglInitConfig(&config->Base, (i+1));
-      SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_ID, attribs.id);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_BUFFER_SIZE, attribs.bufferSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_RED_SIZE, attribs.redSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_GREEN_SIZE, attribs.greenSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_BLUE_SIZE, attribs.blueSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_ALPHA_SIZE, attribs.alphaSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_DEPTH_SIZE, attribs.depthSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_STENCIL_SIZE, attribs.stencilSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLES, attribs.numSamples);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLE_BUFFERS, attribs.numMultisample);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_CONFORMANT, all_apis);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_RENDERABLE_TYPE, all_apis);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_SURFACE_TYPE, bit);
-
-      /* XXX possibly other things to init... */
+      GLX_conf = CALLOC_STRUCT(GLX_egl_config);
+      if (GLX_conf) {
+         memcpy(GLX_conf, &template, sizeof(template));
+         GLX_conf->index = i;
 
-      _eglAddConfig(disp, &config->Base);
+         _eglAddConfig(dpy, &GLX_conf->Base);
+         id++;
+      }
    }
 
-   goto end;
-#endif
-
-xvisual:
-   /* get list of all visuals on this screen */
-   theTemplate.screen = DefaultScreen(GLX_dpy->dpy);
-   mask = VisualScreenMask;
-   GLX_dpy->visuals = XGetVisualInfo(GLX_dpy->dpy, mask, &theTemplate, &numVisuals);
+   return EGL_TRUE;
+}
 
-   for (i = 0; i < numVisuals; i++) {
-      struct GLX_egl_config *config;
 
-      if (!get_visual_attribs(GLX_dpy->dpy, &GLX_dpy->visuals[i], &attribs))
-	 continue;
+static void
+check_extensions(struct GLX_egl_display *GLX_dpy, EGLint screen)
+{
+   GLX_dpy->extensions =
+      glXQueryExtensionsString(GLX_dpy->dpy, screen);
+   if (GLX_dpy->extensions) {
+      /* glXGetProcAddress is assumed */
+
+      if (strstr(GLX_dpy->extensions, "GLX_SGI_make_current_read")) {
+         /* GLX 1.3 entry points are used */
+         GLX_dpy->have_make_current_read = EGL_TRUE;
+      }
 
-      config = CALLOC_STRUCT(GLX_egl_config);
+      if (strstr(GLX_dpy->extensions, "GLX_SGIX_fbconfig")) {
+         /* GLX 1.3 entry points are used */
+         GLX_dpy->have_fbconfig = EGL_TRUE;
+      }
 
-      config->index = i;
-      _eglInitConfig(&config->Base, (i+1));
-      SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_ID, attribs.id);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_BUFFER_SIZE, attribs.bufferSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_RED_SIZE, attribs.redSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_GREEN_SIZE, attribs.greenSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_BLUE_SIZE, attribs.blueSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_ALPHA_SIZE, attribs.alphaSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_DEPTH_SIZE, attribs.depthSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_STENCIL_SIZE, attribs.stencilSize);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLES, attribs.numSamples);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLE_BUFFERS, attribs.numMultisample);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_CONFORMANT, all_apis);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_RENDERABLE_TYPE, all_apis);
-      SET_CONFIG_ATTRIB(&config->Base, EGL_SURFACE_TYPE,
-                        (EGL_WINDOW_BIT /*| EGL_PBUFFER_BIT | EGL_PIXMAP_BIT*/));
+      if (strstr(GLX_dpy->extensions, "GLX_SGIX_pbuffer")) {
+         GLX_dpy->glXCreateGLXPbufferSGIX = (PFNGLXCREATEGLXPBUFFERSGIXPROC)
+            glXGetProcAddress((const GLubyte *) "glXCreateGLXPbufferSGIX");
+         GLX_dpy->glXDestroyGLXPbufferSGIX = (PFNGLXDESTROYGLXPBUFFERSGIXPROC)
+            glXGetProcAddress((const GLubyte *) "glXDestroyGLXPbufferSGIX");
 
-      /* XXX possibly other things to init... */
+         if (GLX_dpy->glXCreateGLXPbufferSGIX &&
+             GLX_dpy->glXDestroyGLXPbufferSGIX &&
+             GLX_dpy->have_fbconfig)
+            GLX_dpy->have_pbuffer = EGL_TRUE;
+      }
+   }
 
-      _eglAddConfig(disp, &config->Base);
+   if (GLX_dpy->glx_maj == 1 && GLX_dpy->glx_min >= 3) {
+      GLX_dpy->have_1_3 = EGL_TRUE;
+      GLX_dpy->have_make_current_read = EGL_TRUE;
+      GLX_dpy->have_fbconfig = EGL_TRUE;
+      GLX_dpy->have_pbuffer = EGL_TRUE;
    }
+}
 
-end:
-   return EGL_TRUE;
+
+static void
+check_quirks(struct GLX_egl_display *GLX_dpy, EGLint screen)
+{
+   const char *vendor;
+
+   GLX_dpy->single_buffered_quirk = EGL_TRUE;
+   GLX_dpy->glx_window_quirk = EGL_TRUE;
+
+   vendor = glXGetClientString(GLX_dpy->dpy, GLX_VENDOR);
+   if (vendor && strstr(vendor, "NVIDIA")) {
+      vendor = glXQueryServerString(GLX_dpy->dpy, screen, GLX_VENDOR);
+      if (vendor && strstr(vendor, "NVIDIA")) {
+         _eglLog(_EGL_DEBUG, "disable quirks");
+         GLX_dpy->single_buffered_quirk = EGL_FALSE;
+         GLX_dpy->glx_window_quirk = EGL_FALSE;
+      }
+   }
 }
 
+
 /**
  * Called via eglInitialize(), GLX_drv->API.Initialize().
  */
@@ -478,17 +538,33 @@ GLX_eglInitialize(_EGLDriver *drv, _EGLDisplay *disp,
       }
    }
 
-   disp->DriverData = (void *) GLX_dpy;
-   disp->ClientAPIsMask = all_apis;
+   if (!glXQueryVersion(GLX_dpy->dpy, &GLX_dpy->glx_maj, &GLX_dpy->glx_min)) {
+      _eglLog(_EGL_WARNING, "GLX: glXQueryVersion failed");
+      if (!disp->NativeDisplay)
+         XCloseDisplay(GLX_dpy->dpy);
+      free(GLX_dpy);
+      return EGL_FALSE;
+   }
 
-   glXQueryVersion(GLX_dpy->dpy, &GLX_dpy->glx_maj, &GLX_dpy->glx_min);
+   check_extensions(GLX_dpy, DefaultScreen(GLX_dpy->dpy));
+   check_quirks(GLX_dpy, DefaultScreen(GLX_dpy->dpy));
+
+   create_configs(disp, GLX_dpy, DefaultScreen(GLX_dpy->dpy));
+   if (!disp->NumConfigs) {
+      _eglLog(_EGL_WARNING, "GLX: failed to create any config");
+      if (!disp->NativeDisplay)
+         XCloseDisplay(GLX_dpy->dpy);
+      free(GLX_dpy);
+      return EGL_FALSE;
+   }
+
+   disp->DriverData = (void *) GLX_dpy;
+   disp->ClientAPIsMask = GLX_EGL_APIS;
 
    /* we're supporting EGL 1.4 */
    *major = 1;
    *minor = 4;
 
-   create_configs(disp, GLX_dpy);
-
    return EGL_TRUE;
 }
 
@@ -539,8 +615,7 @@ GLX_eglCreateContext(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
       return NULL;
    }
 
-#ifdef GLX_VERSION_1_3
-   if (GLX_dpy->fbconfigs)
+   if (GLX_dpy->have_fbconfig)
       GLX_ctx->context =
          glXCreateNewContext(GLX_dpy->dpy,
                              GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
@@ -548,7 +623,6 @@ GLX_eglCreateContext(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
                              GLX_ctx_shared ? GLX_ctx_shared->context : NULL,
                              GL_TRUE);
    else
-#endif
       GLX_ctx->context =
          glXCreateContext(GLX_dpy->dpy,
                           &GLX_dpy->visuals[GLX_egl_config_index(conf)],
@@ -559,15 +633,6 @@ GLX_eglCreateContext(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
       return NULL;
    }
 
-#if 1
-   /* (maybe?) need to have a direct rendering context */
-   if (!glXIsDirect(GLX_dpy->dpy, GLX_ctx->context)) {
-      glXDestroyContext(GLX_dpy->dpy, GLX_ctx->context);
-      free(GLX_ctx);
-      return NULL;
-   }
-#endif
-
    return &GLX_ctx->Base;
 }
 
@@ -589,17 +654,14 @@ GLX_eglMakeCurrent(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *dsurf,
    if (!_eglMakeCurrent(drv, disp, dsurf, rsurf, ctx))
       return EGL_FALSE;
 
-   ddraw = (GLX_dsurf) ? GLX_dsurf->drawable : None;
-   rdraw = (GLX_rsurf) ? GLX_rsurf->drawable : None;
+   ddraw = (GLX_dsurf) ? GLX_dsurf->glx_drawable : None;
+   rdraw = (GLX_rsurf) ? GLX_rsurf->glx_drawable : None;
    cctx = (GLX_ctx) ? GLX_ctx->context : NULL;
 
-#ifdef GLX_VERSION_1_3
-   if (glXMakeContextCurrent(GLX_dpy->dpy, ddraw, rdraw, cctx))
-      return EGL_TRUE;
-#endif
-
-   if (ddraw == rdraw && glXMakeCurrent(GLX_dpy->dpy, ddraw, cctx))
-      return EGL_TRUE;
+   if (GLX_dpy->have_make_current_read)
+      return glXMakeContextCurrent(GLX_dpy->dpy, ddraw, rdraw, cctx);
+   else if (ddraw == rdraw)
+      return glXMakeCurrent(GLX_dpy->dpy, ddraw, cctx);
 
    return EGL_FALSE;
 }
@@ -642,6 +704,20 @@ GLX_eglCreateWindowSurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
    }
 
    GLX_surf->drawable = window;
+
+   if (GLX_dpy->have_1_3 && !GLX_dpy->glx_window_quirk)
+      GLX_surf->glx_drawable =
+         glXCreateWindow(GLX_dpy->dpy,
+                         GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
+                         GLX_surf->drawable, NULL);
+   else
+      GLX_surf->glx_drawable = GLX_surf->drawable;
+
+   if (!GLX_surf->glx_drawable) {
+      free(GLX_surf);
+      return NULL;
+   }
+
    get_drawable_size(GLX_dpy->dpy, window, &width, &height);
    GLX_surf->Base.Width = width;
    GLX_surf->Base.Height = height;
@@ -649,18 +725,13 @@ GLX_eglCreateWindowSurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
    return &GLX_surf->Base;
 }
 
-#ifdef GLX_VERSION_1_3
 static _EGLSurface *
 GLX_eglCreatePixmapSurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
 			   NativePixmapType pixmap, const EGLint *attrib_list)
 {
    struct GLX_egl_display *GLX_dpy = GLX_egl_display(disp);
    struct GLX_egl_surface *GLX_surf;
-   int i;
-
-   /* GLX must >= 1.3 */
-   if (!(GLX_dpy->glx_maj == 1 && GLX_dpy->glx_min >= 3))
-      return NULL;
+   uint width, height;
 
    GLX_surf = CALLOC_STRUCT(GLX_egl_surface);
    if (!GLX_surf) {
@@ -674,25 +745,39 @@ GLX_eglCreatePixmapSurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLConfig *conf,
       return NULL;
    }
 
-   for (i = 0; attrib_list && attrib_list[i] != EGL_NONE; i++) {
-      switch (attrib_list[i]) {
-         /* no attribs at this time */
-      default:
-         _eglError(EGL_BAD_ATTRIBUTE, "eglCreatePixmapSurface");
-         free(GLX_surf);
-         return NULL;
+   GLX_surf->drawable = pixmap;
+
+   if (GLX_dpy->have_1_3) {
+      GLX_surf->glx_drawable =
+         glXCreatePixmap(GLX_dpy->dpy,
+                         GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
+                         GLX_surf->drawable, NULL);
+   }
+   else if (GLX_dpy->have_fbconfig) {
+      GLXFBConfig fbconfig = GLX_dpy->fbconfigs[GLX_egl_config_index(conf)];
+      XVisualInfo *vinfo = glXGetVisualFromFBConfig(GLX_dpy->dpy, fbconfig);
+      if (vinfo) {
+         GLX_surf->glx_drawable =
+            glXCreateGLXPixmap(GLX_dpy->dpy, vinfo, GLX_surf->drawable);
+         XFree(vinfo);
       }
    }
+   else {
+      GLX_surf->glx_drawable =
+         glXCreateGLXPixmap(GLX_dpy->dpy,
+                            &GLX_dpy->visuals[GLX_egl_config_index(conf)],
+                            GLX_surf->drawable);
+   }
 
-   GLX_surf->drawable =
-      glXCreatePixmap(GLX_dpy->dpy,
-                      GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
-                      pixmap, NULL);
-   if (!GLX_surf->drawable) {
+   if (!GLX_surf->glx_drawable) {
       free(GLX_surf);
       return NULL;
    }
 
+   get_drawable_size(GLX_dpy->dpy, pixmap, &width, &height);
+   GLX_surf->Base.Width = width;
+   GLX_surf->Base.Height = height;
+
    return &GLX_surf->Base;
 }
 
@@ -703,11 +788,7 @@ GLX_eglCreatePbufferSurface(_EGLDriver *drv, _EGLDisplay *disp,
    struct GLX_egl_display *GLX_dpy = GLX_egl_display(disp);
    struct GLX_egl_surface *GLX_surf;
    int attribs[5];
-   int i = 0, j = 0;
-
-   /* GLX must >= 1.3 */
-   if (!(GLX_dpy->glx_maj == 1 && GLX_dpy->glx_min >= 3))
-      return NULL;
+   int i;
 
    GLX_surf = CALLOC_STRUCT(GLX_egl_surface);
    if (!GLX_surf) {
@@ -721,33 +802,44 @@ GLX_eglCreatePbufferSurface(_EGLDriver *drv, _EGLDisplay *disp,
       return NULL;
    }
 
-   while(attrib_list[i] != EGL_NONE) {
-      switch (attrib_list[i]) {
-         case EGL_WIDTH:
-	    attribs[j++] = GLX_PBUFFER_WIDTH;
-	    attribs[j++] = attrib_list[i+1];
-	    break;
-	 case EGL_HEIGHT:
-	    attribs[j++] = GLX_PBUFFER_HEIGHT;
-	    attribs[j++] = attrib_list[i+1];
-	    break;
+   i = 0;
+   attribs[i] = None;
+
+   GLX_surf->drawable = None;
+
+   if (GLX_dpy->have_1_3) {
+      /* put geometry in attribs */
+      if (GLX_surf->Base.Width) {
+         attribs[i++] = GLX_PBUFFER_WIDTH;
+         attribs[i++] = GLX_surf->Base.Width;
+      }
+      if (GLX_surf->Base.Height) {
+         attribs[i++] = GLX_PBUFFER_HEIGHT;
+         attribs[i++] = GLX_surf->Base.Height;
       }
-      i++;
+      attribs[i] = None;
+
+      GLX_surf->glx_drawable =
+         glXCreatePbuffer(GLX_dpy->dpy,
+                          GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
+                          attribs);
+   }
+   else if (GLX_dpy->have_pbuffer) {
+      GLX_surf->glx_drawable = GLX_dpy->glXCreateGLXPbufferSGIX(
+            GLX_dpy->dpy,
+            GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
+            GLX_surf->Base.Width,
+            GLX_surf->Base.Height,
+            attribs);
    }
-   attribs[j++] = 0;
 
-   GLX_surf->drawable =
-      glXCreatePbuffer(GLX_dpy->dpy,
-                       GLX_dpy->fbconfigs[GLX_egl_config_index(conf)],
-                       attribs);
-   if (!GLX_surf->drawable) {
+   if (!GLX_surf->glx_drawable) {
       free(GLX_surf);
       return NULL;
    }
 
    return &GLX_surf->Base;
 }
-#endif
 
 static EGLBoolean
 GLX_eglDestroySurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *surf)
@@ -755,15 +847,35 @@ GLX_eglDestroySurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *surf)
    struct GLX_egl_display *GLX_dpy = GLX_egl_display(disp);
    if (!_eglIsSurfaceBound(surf)) {
       struct GLX_egl_surface *GLX_surf = GLX_egl_surface(surf);
-      switch (surf->Type) {
-      case EGL_PBUFFER_BIT:
-         glXDestroyPbuffer(GLX_dpy->dpy, GLX_surf->drawable);
-         break;
-      case EGL_PIXMAP_BIT:
-         glXDestroyPixmap(GLX_dpy->dpy, GLX_surf->drawable);
-         break;
-      default:
-         break;
+
+      if (GLX_dpy->have_1_3) {
+         switch (surf->Type) {
+         case EGL_WINDOW_BIT:
+            if (!GLX_dpy->glx_window_quirk)
+               glXDestroyWindow(GLX_dpy->dpy, GLX_surf->glx_drawable);
+            break;
+         case EGL_PBUFFER_BIT:
+            glXDestroyPbuffer(GLX_dpy->dpy, GLX_surf->glx_drawable);
+            break;
+         case EGL_PIXMAP_BIT:
+            glXDestroyPixmap(GLX_dpy->dpy, GLX_surf->glx_drawable);
+            break;
+         default:
+            break;
+         }
+      }
+      else {
+         switch (surf->Type) {
+         case EGL_PBUFFER_BIT:
+            GLX_dpy->glXDestroyGLXPbufferSGIX(GLX_dpy->dpy,
+                                              GLX_surf->glx_drawable);
+            break;
+         case EGL_PIXMAP_BIT:
+            glXDestroyGLXPixmap(GLX_dpy->dpy, GLX_surf->glx_drawable);
+            break;
+         default:
+            break;
+         }
       }
       free(surf);
    }
@@ -773,44 +885,12 @@ GLX_eglDestroySurface(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *surf)
 
 
 static EGLBoolean
-GLX_eglBindTexImage(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *surf,
-                     EGLint buffer)
-{
-   struct GLX_egl_display *GLX_dpy = GLX_egl_display(disp);
-   struct GLX_egl_surface *GLX_surf = GLX_egl_surface(surf);
-
-   /* buffer ?? */
-   glXBindTexImageEXT(GLX_dpy->dpy, GLX_surf->drawable,
-                      GLX_FRONT_LEFT_EXT, NULL);
-
-   return EGL_TRUE;
-}
-
-
-static EGLBoolean
-GLX_eglReleaseTexImage(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *surf,
-                        EGLint buffer)
-{
-   struct GLX_egl_display *GLX_dpy = GLX_egl_display(disp);
-   struct GLX_egl_surface *GLX_surf = GLX_egl_surface(surf);
-
-   /* buffer ?? */
-   glXReleaseTexImageEXT(GLX_dpy->dpy, GLX_surf->drawable,
-                         GLX_FRONT_LEFT_EXT);
-
-   return EGL_TRUE;
-}
-
-
-static EGLBoolean
 GLX_eglSwapBuffers(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *draw)
 {
    struct GLX_egl_display *GLX_dpy = GLX_egl_display(disp);
    struct GLX_egl_surface *GLX_surf = GLX_egl_surface(draw);
 
-   _eglLog(_EGL_DEBUG, "GLX: EGL SwapBuffers 0x%x",draw);
-
-   glXSwapBuffers(GLX_dpy->dpy, GLX_surf->drawable);
+   glXSwapBuffers(GLX_dpy->dpy, GLX_surf->glx_drawable);
 
    return EGL_TRUE;
 }
@@ -821,23 +901,24 @@ GLX_eglSwapBuffers(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *draw)
 static _EGLProc
 GLX_eglGetProcAddress(const char *procname)
 {
-   /* This is a bit of a hack to get at the gallium/Mesa state tracker
-    * function st_get_proc_address().  This will probably change at
-    * some point.
-    */
-   _EGLProc (*get_proc_addr)(const char *procname);
-   _EGLProc proc_addr;
-   get_proc_addr = dlsym(NULL, "st_get_proc_address");
-   if (get_proc_addr)
-      return get_proc_addr(procname);
-
-   proc_addr = glXGetProcAddress((const GLubyte *)procname);
-   if (proc_addr)
-      return proc_addr;
-
-   return (_EGLProc)dlsym(NULL, procname);
+   return (_EGLProc) glXGetProcAddress((const GLubyte *) procname);
 }
 
+static EGLBoolean
+GLX_eglWaitClient(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx)
+{
+   glXWaitGL();
+   return EGL_TRUE;
+}
+
+static EGLBoolean
+GLX_eglWaitNative(_EGLDriver *drv, _EGLDisplay *dpy, EGLint engine)
+{
+   if (engine != EGL_CORE_NATIVE_ENGINE)
+      return _eglError(EGL_BAD_PARAMETER, "eglWaitNative");
+   glXWaitX();
+   return EGL_TRUE;
+}
 
 static void
 GLX_Unload(_EGLDriver *drv)
@@ -865,15 +946,13 @@ _eglMain(const char *args)
    GLX_drv->Base.API.CreateContext = GLX_eglCreateContext;
    GLX_drv->Base.API.MakeCurrent = GLX_eglMakeCurrent;
    GLX_drv->Base.API.CreateWindowSurface = GLX_eglCreateWindowSurface;
-#ifdef GLX_VERSION_1_3
    GLX_drv->Base.API.CreatePixmapSurface = GLX_eglCreatePixmapSurface;
    GLX_drv->Base.API.CreatePbufferSurface = GLX_eglCreatePbufferSurface;
-#endif
    GLX_drv->Base.API.DestroySurface = GLX_eglDestroySurface;
-   GLX_drv->Base.API.BindTexImage = GLX_eglBindTexImage;
-   GLX_drv->Base.API.ReleaseTexImage = GLX_eglReleaseTexImage;
    GLX_drv->Base.API.SwapBuffers = GLX_eglSwapBuffers;
    GLX_drv->Base.API.GetProcAddress = GLX_eglGetProcAddress;
+   GLX_drv->Base.API.WaitClient = GLX_eglWaitClient;
+   GLX_drv->Base.API.WaitNative = GLX_eglWaitNative;
 
    GLX_drv->Base.Name = "GLX";
    GLX_drv->Base.Unload = GLX_Unload;
diff --git a/src/egl/drivers/xdri/egl_xdri.c b/src/egl/drivers/xdri/egl_xdri.c
index 518091a2d1..d2affc66dd 100644
--- a/src/egl/drivers/xdri/egl_xdri.c
+++ b/src/egl/drivers/xdri/egl_xdri.c
@@ -48,6 +48,7 @@
 #include "glapi/glapi.h" /* for glapi functions */
 
 #include "eglconfig.h"
+#include "eglconfigutil.h"
 #include "eglcontext.h"
 #include "egldisplay.h"
 #include "egldriver.h"
@@ -104,6 +105,7 @@ struct xdri_egl_config
    _EGLConfig Base;   /**< base class */
 
    const __GLcontextModes *mode;  /**< corresponding GLX mode */
+   EGLint window_render_buffer;
 };
 
 
@@ -162,46 +164,76 @@ get_drawable_size(Display *dpy, Drawable d, uint *width, uint *height)
 }
 
 
+static EGLBoolean
+convert_config(_EGLConfig *conf, EGLint id, const __GLcontextModes *m)
+{
+   static const EGLint all_apis = (EGL_OPENGL_ES_BIT |
+                                   EGL_OPENGL_ES2_BIT |
+                                   EGL_OPENVG_BIT |
+                                   EGL_OPENGL_BIT);
+   EGLint val;
+
+   _eglInitConfig(conf, id);
+   if (!_eglConfigFromContextModesRec(conf, m, all_apis, all_apis))
+      return EGL_FALSE;
+
+   if (m->doubleBufferMode) {
+      /* pixmap and pbuffer surfaces are always single-buffered */
+      val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE);
+      val &= ~(EGL_PIXMAP_BIT | EGL_PBUFFER_BIT);
+      SET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE, val);
+   }
+   else {
+      /* EGL requires OpenGL ES context to be double-buffered */
+      val = GET_CONFIG_ATTRIB(conf, EGL_RENDERABLE_TYPE);
+      val &= ~(EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT);
+      SET_CONFIG_ATTRIB(conf, EGL_RENDERABLE_TYPE, val);
+   }
+   /* skip "empty" config */
+   if (!val)
+      return EGL_FALSE;
+
+   val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE);
+   if (!(val & EGL_PBUFFER_BIT)) {
+      /* bind-to-texture cannot be EGL_TRUE without pbuffer bit */
+      SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGB, EGL_FALSE);
+      SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGBA, EGL_FALSE);
+   }
+
+   /* EGL_NATIVE_RENDERABLE is a boolean */
+   val = GET_CONFIG_ATTRIB(conf, EGL_NATIVE_RENDERABLE);
+   if (val != EGL_TRUE)
+      SET_CONFIG_ATTRIB(conf, EGL_NATIVE_RENDERABLE, EGL_FALSE);
+
+   return _eglValidateConfig(conf, EGL_FALSE);
+}
+
+
 /**
  * Produce a set of EGL configs.
  */
 static EGLint
 create_configs(_EGLDisplay *disp, const __GLcontextModes *m, EGLint first_id)
 {
-   static const EGLint all_apis = (EGL_OPENGL_ES_BIT |
-                                   EGL_OPENGL_ES2_BIT |
-                                   EGL_OPENVG_BIT |
-                                   EGL_OPENGL_BIT);
    int id = first_id;
 
    for (; m; m = m->next) {
-      /* add double buffered visual */
-      if (m->doubleBufferMode) {
-         struct xdri_egl_config *config = CALLOC_STRUCT(xdri_egl_config);
-
-         _eglInitConfig(&config->Base, id++);
-
-         SET_CONFIG_ATTRIB(&config->Base, EGL_BUFFER_SIZE, m->rgbBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_RED_SIZE, m->redBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_GREEN_SIZE, m->greenBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_BLUE_SIZE, m->blueBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_ALPHA_SIZE, m->alphaBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_DEPTH_SIZE, m->depthBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_STENCIL_SIZE, m->stencilBits);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLES, m->samples);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLE_BUFFERS, m->sampleBuffers);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_ID, m->visualID);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_TYPE, m->visualType);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_CONFORMANT, all_apis);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_RENDERABLE_TYPE, all_apis);
-         SET_CONFIG_ATTRIB(&config->Base, EGL_SURFACE_TYPE, EGL_WINDOW_BIT);
-
-         /* XXX possibly other things to init... */
-
-         /* Ptr from EGL config to GLcontextMode.  Used in CreateContext(). */
-         config->mode = m;
-
-         _eglAddConfig(disp, &config->Base);
+      struct xdri_egl_config *xdri_conf;
+      _EGLConfig conf;
+      EGLint rb;
+
+      if (!convert_config(&conf, id, m))
+         continue;
+
+      rb = (m->doubleBufferMode) ? EGL_BACK_BUFFER : EGL_SINGLE_BUFFER;
+
+      xdri_conf = CALLOC_STRUCT(xdri_egl_config);
+      if (xdri_conf) {
+         memcpy(&xdri_conf->Base, &conf, sizeof(conf));
+         xdri_conf->mode = m;
+         xdri_conf->window_render_buffer = rb;
+         _eglAddConfig(disp, &xdri_conf->Base);
+         id++;
       }
    }
 
@@ -363,6 +395,9 @@ xdri_eglCreateContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf,
       return NULL;
    }
 
+   /* the config decides the render buffer for the context */
+   xdri_ctx->Base.WindowRenderBuffer = xdri_config->window_render_buffer;
+
    xdri_ctx->driContext =
       psc->driScreen->createContext(psc,
                                     xdri_config->mode,
diff --git a/src/egl/main/eglapi.c b/src/egl/main/eglapi.c
index 29617b7aff..14cc5fa613 100644
--- a/src/egl/main/eglapi.c
+++ b/src/egl/main/eglapi.c
@@ -9,12 +9,36 @@
  * heterogeneous hardware devices in the future.
  *
  * The EGLDisplay, EGLConfig, EGLContext and EGLSurface types are
- * opaque handles implemented with 32-bit unsigned integers.
- * It's up to the driver function or fallback function to look up the
- * handle and get an object.
- * By using opaque handles, we leave open the possibility of having
- * indirect rendering in the future, like GLX.
+ * opaque handles. Internal objects are linked to a display to
+ * create the handles.
  *
+ * For each public API entry point, the opaque handles are looked up
+ * before being dispatched to the drivers.  When it fails to look up
+ * a handle, one of
+ *
+ * EGL_BAD_DISPLAY
+ * EGL_BAD_CONFIG
+ * EGL_BAD_CONTEXT
+ * EGL_BAD_SURFACE
+ * EGL_BAD_SCREEN_MESA
+ * EGL_BAD_MODE_MESA
+ *
+ * is generated and the driver function is not called. An
+ * uninitialized EGLDisplay has no driver associated with it. When
+ * such display is detected,
+ *
+ * EGL_NOT_INITIALIZED
+ *
+ * is generated.
+ *
+ * Some of the entry points use current display, context, or surface
+ * implicitly.  For such entry points, the implicit objects are also
+ * checked before calling the driver function.  Other than the
+ * errors listed above,
+ *
+ * EGL_BAD_CURRENT_SURFACE
+ *
+ * may also be generated.
  *
  * Notes on naming conventions:
  *
@@ -92,8 +116,8 @@ eglInitialize(EGLDisplay dpy, EGLint *major, EGLint *minor)
       snprintf(disp->Version, sizeof(disp->Version),
                "%d.%d (%s)", major_int, minor_int, drv->Name);
 
-      /* update the global notion of supported APIs */
-      _eglGlobal.ClientAPIsMask |= disp->ClientAPIsMask;
+      /* limit to APIs supported by core */
+      disp->ClientAPIsMask &= _EGL_API_ALL_BITS;
 
       disp->Driver = drv;
    } else {
@@ -496,15 +520,31 @@ eglReleaseTexImage(EGLDisplay dpy, EGLSurface surface, EGLint buffer)
 EGLBoolean EGLAPIENTRY
 eglSwapInterval(EGLDisplay dpy, EGLint interval)
 {
+   _EGLContext *ctx = _eglGetCurrentContext();
+   _EGLSurface *surf;
    _EGL_DECLARE_DD(dpy);
-   return drv->API.SwapInterval(drv, disp, interval);
+
+   if (!ctx || !_eglIsContextLinked(ctx) || ctx->Display != disp)
+      return _eglError(EGL_BAD_CONTEXT, __FUNCTION__);
+
+   surf = ctx->DrawSurface;
+   if (!_eglIsSurfaceLinked(surf))
+      return _eglError(EGL_BAD_SURFACE, __FUNCTION__);
+
+   return drv->API.SwapInterval(drv, disp, surf, interval);
 }
 
 
 EGLBoolean EGLAPIENTRY
 eglSwapBuffers(EGLDisplay dpy, EGLSurface surface)
 {
+   _EGLContext *ctx = _eglGetCurrentContext();
    _EGL_DECLARE_DD_AND_SURFACE(dpy, surface);
+
+   /* surface must be bound to current context in EGL 1.4 */
+   if (!ctx || !_eglIsContextLinked(ctx) || surf != ctx->DrawSurface)
+      return _eglError(EGL_BAD_SURFACE, __FUNCTION__);
+
    return drv->API.SwapBuffers(drv, disp, surf);
 }
 
@@ -518,32 +558,66 @@ eglCopyBuffers(EGLDisplay dpy, EGLSurface surface, NativePixmapType target)
 
 
 EGLBoolean EGLAPIENTRY
-eglWaitGL(void)
+eglWaitClient(void)
 {
-   _EGLDisplay *disp = _eglGetCurrentDisplay();
+   _EGLContext *ctx = _eglGetCurrentContext();
+   _EGLDisplay *disp;
    _EGLDriver *drv;
 
-   if (!disp)
+   if (!ctx)
       return EGL_TRUE;
+   /* let bad current context imply bad current surface */
+   if (!_eglIsContextLinked(ctx) || !_eglIsSurfaceLinked(ctx->DrawSurface))
+      return _eglError(EGL_BAD_CURRENT_SURFACE, __FUNCTION__);
 
-   /* a current display is always initialized */
+   /* a valid current context implies an initialized current display */
+   disp = ctx->Display;
    drv = disp->Driver;
+   assert(drv);
 
-   return drv->API.WaitGL(drv, disp);
+   return drv->API.WaitClient(drv, disp, ctx);
+}
+
+
+EGLBoolean EGLAPIENTRY
+eglWaitGL(void)
+{
+#ifdef EGL_VERSION_1_2
+   _EGLThreadInfo *t = _eglGetCurrentThread();
+   EGLint api_index = t->CurrentAPIIndex;
+   EGLint es_index = _eglConvertApiToIndex(EGL_OPENGL_ES_API);
+   EGLBoolean ret;
+
+   if (api_index != es_index && _eglIsCurrentThreadDummy())
+      return _eglError(EGL_BAD_ALLOC, "eglWaitGL");
+
+   t->CurrentAPIIndex = es_index;
+   ret = eglWaitClient();
+   t->CurrentAPIIndex = api_index;
+   return ret;
+#else
+   return eglWaitClient();
+#endif
 }
 
 
 EGLBoolean EGLAPIENTRY
 eglWaitNative(EGLint engine)
 {
-   _EGLDisplay *disp = _eglGetCurrentDisplay();
+   _EGLContext *ctx = _eglGetCurrentContext();
+   _EGLDisplay *disp;
    _EGLDriver *drv;
 
-   if (!disp)
+   if (!ctx)
       return EGL_TRUE;
+   /* let bad current context imply bad current surface */
+   if (!_eglIsContextLinked(ctx) || !_eglIsSurfaceLinked(ctx->DrawSurface))
+      return _eglError(EGL_BAD_CURRENT_SURFACE, __FUNCTION__);
 
-   /* a current display is always initialized */
+   /* a valid current context implies an initialized current display */
+   disp = ctx->Display;
    drv = disp->Driver;
+   assert(drv);
 
    return drv->API.WaitNative(drv, disp, engine);
 }
@@ -568,8 +642,26 @@ eglGetCurrentContext(void)
 EGLSurface EGLAPIENTRY
 eglGetCurrentSurface(EGLint readdraw)
 {
-   _EGLSurface *s = _eglGetCurrentSurface(readdraw);
-   return _eglGetSurfaceHandle(s);
+   _EGLContext *ctx = _eglGetCurrentContext();
+   _EGLSurface *surf;
+
+   if (!ctx)
+      return EGL_NO_SURFACE;
+
+   switch (readdraw) {
+   case EGL_DRAW:
+      surf = ctx->DrawSurface;
+      break;
+   case EGL_READ:
+      surf = ctx->ReadSurface;
+      break;
+   default:
+      _eglError(EGL_BAD_PARAMETER, __FUNCTION__);
+      surf = NULL;
+      break;
+   }
+
+   return _eglGetSurfaceHandle(surf);
 }
 
 
@@ -586,43 +678,11 @@ eglGetError(void)
 
 void (* EGLAPIENTRY eglGetProcAddress(const char *procname))()
 {
-   typedef void (*genericFunc)();
-   struct name_function {
+   static const struct {
       const char *name;
       _EGLProc function;
-   };
-   static struct name_function egl_functions[] = {
-      /* alphabetical order */
-      { "eglBindTexImage", (_EGLProc) eglBindTexImage },
-      { "eglChooseConfig", (_EGLProc) eglChooseConfig },
-      { "eglCopyBuffers", (_EGLProc) eglCopyBuffers },
-      { "eglCreateContext", (_EGLProc) eglCreateContext },
-      { "eglCreatePbufferSurface", (_EGLProc) eglCreatePbufferSurface },
-      { "eglCreatePixmapSurface", (_EGLProc) eglCreatePixmapSurface },
-      { "eglCreateWindowSurface", (_EGLProc) eglCreateWindowSurface },
-      { "eglDestroyContext", (_EGLProc) eglDestroyContext },
-      { "eglDestroySurface", (_EGLProc) eglDestroySurface },
-      { "eglGetConfigAttrib", (_EGLProc) eglGetConfigAttrib },
-      { "eglGetConfigs", (_EGLProc) eglGetConfigs },
-      { "eglGetCurrentContext", (_EGLProc) eglGetCurrentContext },
-      { "eglGetCurrentDisplay", (_EGLProc) eglGetCurrentDisplay },
-      { "eglGetCurrentSurface", (_EGLProc) eglGetCurrentSurface },
-      { "eglGetDisplay", (_EGLProc) eglGetDisplay },
-      { "eglGetError", (_EGLProc) eglGetError },
-      { "eglGetProcAddress", (_EGLProc) eglGetProcAddress },
-      { "eglInitialize", (_EGLProc) eglInitialize },
-      { "eglMakeCurrent", (_EGLProc) eglMakeCurrent },
-      { "eglQueryContext", (_EGLProc) eglQueryContext },
-      { "eglQueryString", (_EGLProc) eglQueryString },
-      { "eglQuerySurface", (_EGLProc) eglQuerySurface },
-      { "eglReleaseTexImage", (_EGLProc) eglReleaseTexImage },
-      { "eglSurfaceAttrib", (_EGLProc) eglSurfaceAttrib },
-      { "eglSwapBuffers", (_EGLProc) eglSwapBuffers },
-      { "eglSwapInterval", (_EGLProc) eglSwapInterval },
-      { "eglTerminate", (_EGLProc) eglTerminate },
-      { "eglWaitGL", (_EGLProc) eglWaitGL },
-      { "eglWaitNative", (_EGLProc) eglWaitNative },
-      /* Extensions */
+   } egl_functions[] = {
+      /* extensions only */
 #ifdef EGL_MESA_screen_surface
       { "eglChooseModeMESA", (_EGLProc) eglChooseModeMESA },
       { "eglGetModesMESA", (_EGLProc) eglGetModesMESA },
@@ -637,22 +697,23 @@ void (* EGLAPIENTRY eglGetProcAddress(const char *procname))()
       { "eglQueryScreenModeMESA", (_EGLProc) eglQueryScreenModeMESA },
       { "eglQueryModeStringMESA", (_EGLProc) eglQueryModeStringMESA },
 #endif /* EGL_MESA_screen_surface */
-#ifdef EGL_VERSION_1_2
-      { "eglBindAPI", (_EGLProc) eglBindAPI },
-      { "eglCreatePbufferFromClientBuffer", (_EGLProc) eglCreatePbufferFromClientBuffer },
-      { "eglQueryAPI", (_EGLProc) eglQueryAPI },
-      { "eglReleaseThread", (_EGLProc) eglReleaseThread },
-      { "eglWaitClient", (_EGLProc) eglWaitClient },
-#endif /* EGL_VERSION_1_2 */
       { NULL, NULL }
    };
    EGLint i;
-   for (i = 0; egl_functions[i].name; i++) {
-      if (strcmp(egl_functions[i].name, procname) == 0) {
-         return (genericFunc) egl_functions[i].function;
+
+   if (!procname)
+      return NULL;
+   if (strncmp(procname, "egl", 3) == 0) {
+      for (i = 0; egl_functions[i].name; i++) {
+         if (strcmp(egl_functions[i].name, procname) == 0)
+            return egl_functions[i].function;
       }
    }
 
+   /* preload a driver if there isn't one */
+   if (!_eglGlobal.NumDrivers)
+      _eglPreloadDriver(NULL);
+
    /* now loop over drivers to query their procs */
    for (i = 0; i < _eglGlobal.NumDrivers; i++) {
       _EGLProc p = _eglGlobal.Drivers[i]->API.GetProcAddress(procname);
@@ -664,6 +725,9 @@ void (* EGLAPIENTRY eglGetProcAddress(const char *procname))()
 }
 
 
+#ifdef EGL_MESA_screen_surface
+
+
 /*
  * EGL_MESA_screen extension
  */
@@ -838,6 +902,9 @@ eglQueryModeStringMESA(EGLDisplay dpy, EGLModeMESA mode)
 }
 
 
+#endif /* EGL_MESA_screen_surface */
+
+
 /**
  ** EGL 1.2
  **/
@@ -867,33 +934,7 @@ eglBindAPI(EGLenum api)
    if (!_eglIsApiValid(api))
       return _eglError(EGL_BAD_PARAMETER, "eglBindAPI");
 
-   switch (api) {
-#ifdef EGL_VERSION_1_4
-   case EGL_OPENGL_API:
-      if (_eglGlobal.ClientAPIsMask & EGL_OPENGL_BIT) {
-         t->CurrentAPIIndex = _eglConvertApiToIndex(api);
-         return EGL_TRUE;
-      }
-      _eglError(EGL_BAD_PARAMETER, "eglBindAPI");
-      return EGL_FALSE;
-#endif
-   case EGL_OPENGL_ES_API:
-      if (_eglGlobal.ClientAPIsMask & (EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT)) {
-         t->CurrentAPIIndex = _eglConvertApiToIndex(api);
-         return EGL_TRUE;
-      }
-      _eglError(EGL_BAD_PARAMETER, "eglBindAPI");
-      return EGL_FALSE;
-   case EGL_OPENVG_API:
-      if (_eglGlobal.ClientAPIsMask & EGL_OPENVG_BIT) {
-         t->CurrentAPIIndex = _eglConvertApiToIndex(api);
-         return EGL_TRUE;
-      }
-      _eglError(EGL_BAD_PARAMETER, "eglBindAPI");
-      return EGL_FALSE;
-   default:
-      return EGL_FALSE;
-   }
+   t->CurrentAPIIndex = _eglConvertApiToIndex(api);
    return EGL_TRUE;
 }
 
@@ -951,20 +992,4 @@ eglReleaseThread(void)
 }
 
 
-EGLBoolean
-eglWaitClient(void)
-{
-   _EGLDisplay *disp = _eglGetCurrentDisplay();
-   _EGLDriver *drv;
-
-   if (!disp)
-      return EGL_TRUE;
-
-   /* a current display is always initialized */
-   drv = disp->Driver;
-
-   return drv->API.WaitClient(drv, disp);
-}
-
-
 #endif /* EGL_VERSION_1_2 */
diff --git a/src/egl/main/eglapi.h b/src/egl/main/eglapi.h
index 6081e58892..aa0abe3eb6 100644
--- a/src/egl/main/eglapi.h
+++ b/src/egl/main/eglapi.h
@@ -35,13 +35,13 @@ typedef EGLBoolean (*QuerySurface_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurf
 typedef EGLBoolean (*SurfaceAttrib_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surface, EGLint attribute, EGLint value);
 typedef EGLBoolean (*BindTexImage_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surface, EGLint buffer);
 typedef EGLBoolean (*ReleaseTexImage_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surface, EGLint buffer);
-typedef EGLBoolean (*SwapInterval_t)(_EGLDriver *drv, _EGLDisplay *dpy, EGLint interval);
+typedef EGLBoolean (*SwapInterval_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf, EGLint interval);
 typedef EGLBoolean (*SwapBuffers_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *draw);
 typedef EGLBoolean (*CopyBuffers_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surface, NativePixmapType target);
 
 /* misc funcs */
 typedef const char *(*QueryString_t)(_EGLDriver *drv, _EGLDisplay *dpy, EGLint name);
-typedef EGLBoolean (*WaitGL_t)(_EGLDriver *drv, _EGLDisplay *dpy);
+typedef EGLBoolean (*WaitClient_t)(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx);
 typedef EGLBoolean (*WaitNative_t)(_EGLDriver *drv, _EGLDisplay *dpy, EGLint engine);
 
 typedef _EGLProc (*GetProcAddress_t)(const char *procname);
@@ -65,7 +65,6 @@ typedef const char * (*QueryModeStringMESA_t)(_EGLDriver *drv, _EGLDisplay *dpy,
 
 
 #ifdef EGL_VERSION_1_2
-typedef EGLBoolean (*WaitClient_t)(_EGLDriver *drv, _EGLDisplay *dpy);
 typedef _EGLSurface *(*CreatePbufferFromClientBuffer_t)(_EGLDriver *drv, _EGLDisplay *dpy, EGLenum buftype, EGLClientBuffer buffer, _EGLConfig *config, const EGLint *attrib_list);
 #endif /* EGL_VERSION_1_2 */
 
@@ -101,7 +100,7 @@ struct _egl_api
    CopyBuffers_t CopyBuffers;
 
    QueryString_t QueryString;
-   WaitGL_t WaitGL;
+   WaitClient_t WaitClient;
    WaitNative_t WaitNative;
    GetProcAddress_t GetProcAddress;
 
@@ -120,7 +119,6 @@ struct _egl_api
    QueryModeStringMESA_t QueryModeStringMESA;
 
 #ifdef EGL_VERSION_1_2
-   WaitClient_t WaitClient;
    CreatePbufferFromClientBuffer_t CreatePbufferFromClientBuffer;
 #endif
 };
diff --git a/src/egl/main/eglconfig.c b/src/egl/main/eglconfig.c
index d47b99eed4..31d69a7708 100644
--- a/src/egl/main/eglconfig.c
+++ b/src/egl/main/eglconfig.c
@@ -15,302 +15,705 @@
 
 
 #define MIN2(A, B)  (((A) < (B)) ? (A) : (B))
-
-
-void
-_eglSetConfigAttrib(_EGLConfig *config, EGLint attr, EGLint val)
-{
-   assert(attr >= FIRST_ATTRIB);
-   assert(attr < FIRST_ATTRIB + MAX_ATTRIBS);
-   config->Attrib[attr - FIRST_ATTRIB] = val;
-}
+#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
 
 
 /**
  * Init the given _EGLconfig to default values.
  * \param id  the configuration's ID.
+ *
+ * Note that id must be positive for the config to be valid.
+ * It is also recommended that when there are N configs, their
+ * IDs are from 1 to N respectively.
  */
 void
 _eglInitConfig(_EGLConfig *config, EGLint id)
 {
    memset(config, 0, sizeof(*config));
-   config->Handle = (EGLConfig) _eglUIntToPointer((unsigned int) id);
-   _eglSetConfigAttrib(config, EGL_CONFIG_ID,               id);
-   _eglSetConfigAttrib(config, EGL_BIND_TO_TEXTURE_RGB,     EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_BIND_TO_TEXTURE_RGBA,    EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_CONFIG_CAVEAT,           EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_NATIVE_RENDERABLE,       EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_NATIVE_VISUAL_TYPE,      EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_MIN_SWAP_INTERVAL,       EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_MAX_SWAP_INTERVAL,       EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_SURFACE_TYPE,            EGL_WINDOW_BIT);
-   _eglSetConfigAttrib(config, EGL_TRANSPARENT_TYPE,        EGL_NONE);
-   _eglSetConfigAttrib(config, EGL_TRANSPARENT_RED_VALUE,   EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_TRANSPARENT_GREEN_VALUE, EGL_DONT_CARE);
-   _eglSetConfigAttrib(config, EGL_TRANSPARENT_BLUE_VALUE,  EGL_DONT_CARE);
+
+   /* some attributes take non-zero default values */
+   SET_CONFIG_ATTRIB(config, EGL_CONFIG_ID,               id);
+   SET_CONFIG_ATTRIB(config, EGL_CONFIG_CAVEAT,           EGL_NONE);
+   SET_CONFIG_ATTRIB(config, EGL_TRANSPARENT_TYPE,        EGL_NONE);
+   SET_CONFIG_ATTRIB(config, EGL_NATIVE_VISUAL_TYPE,      EGL_NONE);
 #ifdef EGL_VERSION_1_2
-   _eglSetConfigAttrib(config, EGL_COLOR_BUFFER_TYPE,       EGL_RGB_BUFFER);
-   _eglSetConfigAttrib(config, EGL_RENDERABLE_TYPE,         EGL_OPENGL_ES_BIT);
+   SET_CONFIG_ATTRIB(config, EGL_COLOR_BUFFER_TYPE,       EGL_RGB_BUFFER);
 #endif /* EGL_VERSION_1_2 */
 }
 
 
 /**
- * Return the public handle for an internal _EGLConfig.
- * This is the inverse of _eglLookupConfig().
+ * Link a config to a display and return the handle of the link.
+ * The handle can be passed to client directly.
+ *
+ * Note that we just save the ptr to the config (we don't copy the config).
  */
 EGLConfig
-_eglGetConfigHandle(_EGLConfig *config)
+_eglAddConfig(_EGLDisplay *dpy, _EGLConfig *conf)
 {
-   return config ? config->Handle : 0;
+   _EGLConfig **configs;
+
+   /* sanity check */
+   assert(GET_CONFIG_ATTRIB(conf, EGL_CONFIG_ID) > 0);
+
+   configs = dpy->Configs;
+   if (dpy->NumConfigs >= dpy->MaxConfigs) {
+      EGLint new_size = dpy->MaxConfigs + 16;
+      assert(dpy->NumConfigs < new_size);
+
+      configs = realloc(dpy->Configs, new_size * sizeof(dpy->Configs[0]));
+      if (!configs)
+         return (EGLConfig) NULL;
+
+      dpy->Configs = configs;
+      dpy->MaxConfigs = new_size;
+   }
+
+   conf->Display = dpy;
+   dpy->Configs[dpy->NumConfigs++] = conf;
+
+   return (EGLConfig) conf;
 }
 
 
-/**
- * Given an EGLConfig handle, return the corresponding _EGLConfig object.
- * This is the inverse of _eglGetConfigHandle().
- */
-_EGLConfig *
-_eglLookupConfig(EGLConfig config, _EGLDisplay *disp)
+#ifndef _EGL_SKIP_HANDLE_CHECK
+
+
+EGLBoolean
+_eglCheckConfigHandle(EGLConfig config, _EGLDisplay *dpy)
 {
+   EGLint num_configs = (dpy) ? dpy->NumConfigs : 0;
    EGLint i;
-   for (i = 0; i < disp->NumConfigs; i++) {
-      if (disp->Configs[i]->Handle == config) {
-          return disp->Configs[i];
+
+   for (i = 0; i < num_configs; i++) {
+      _EGLConfig *conf = dpy->Configs[i];
+      if (conf == (_EGLConfig *) config) {
+         assert(conf->Display == dpy);
+         break;
       }
    }
-   return NULL;
+   return (i < num_configs);
 }
 
 
-/**
- * Add the given _EGLConfig to the given display.
- * Note that we just save the ptr to the config (we don't copy the config).
- */
-_EGLConfig *
-_eglAddConfig(_EGLDisplay *display, _EGLConfig *config)
+#endif /* _EGL_SKIP_HANDLE_CHECK */
+
+
+enum {
+   /* types */
+   ATTRIB_TYPE_INTEGER,
+   ATTRIB_TYPE_BOOLEAN,
+   ATTRIB_TYPE_BITMASK,
+   ATTRIB_TYPE_ENUM,
+   ATTRIB_TYPE_PSEUDO, /* non-queryable */
+   ATTRIB_TYPE_PLATFORM, /* platform-dependent */
+   /* criteria */
+   ATTRIB_CRITERION_EXACT,
+   ATTRIB_CRITERION_ATLEAST,
+   ATTRIB_CRITERION_MASK,
+   ATTRIB_CRITERION_SPECIAL,
+   ATTRIB_CRITERION_IGNORE,
+};
+
+
+/* EGL spec Table 3.1 and 3.4 */
+static const struct {
+   EGLint attr;
+   EGLint type;
+   EGLint criterion;
+   EGLint default_value;
+} _eglValidationTable[] =
 {
-   _EGLConfig **newConfigs;
-   EGLint n;
-
-   /* do some sanity checks on the config's attribs */
-   assert(GET_CONFIG_ATTRIB(config, EGL_CONFIG_ID) > 0);
-   assert(GET_CONFIG_ATTRIB(config, EGL_RENDERABLE_TYPE) != 0x0);
-   assert(GET_CONFIG_ATTRIB(config, EGL_SURFACE_TYPE) != 0x0);
-   assert(GET_CONFIG_ATTRIB(config, EGL_RED_SIZE) > 0);
-   assert(GET_CONFIG_ATTRIB(config, EGL_GREEN_SIZE) > 0);
-   assert(GET_CONFIG_ATTRIB(config, EGL_BLUE_SIZE) > 0);
-
-   n = display->NumConfigs;
-
-   /* realloc array of ptrs */
-   newConfigs = (_EGLConfig **) realloc(display->Configs,
-                                        (n + 1) * sizeof(_EGLConfig *));
-   if (newConfigs) {
-      display->Configs = newConfigs;
-      display->Configs[n] = config;
-      display->NumConfigs++;
-      return config;
-   }
-   else {
-      return NULL;
-   }
-}
+   { EGL_BUFFER_SIZE,               ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_RED_SIZE,                  ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_GREEN_SIZE,                ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_BLUE_SIZE,                 ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_LUMINANCE_SIZE,            ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_ALPHA_SIZE,                ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_ALPHA_MASK_SIZE,           ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_BIND_TO_TEXTURE_RGB,       ATTRIB_TYPE_BOOLEAN,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_BIND_TO_TEXTURE_RGBA,      ATTRIB_TYPE_BOOLEAN,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_COLOR_BUFFER_TYPE,         ATTRIB_TYPE_ENUM,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_RGB_BUFFER },
+   { EGL_CONFIG_CAVEAT,             ATTRIB_TYPE_ENUM,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_CONFIG_ID,                 ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_CONFORMANT,                ATTRIB_TYPE_BITMASK,
+                                    ATTRIB_CRITERION_MASK,
+                                    0 },
+   { EGL_DEPTH_SIZE,                ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_LEVEL,                     ATTRIB_TYPE_PLATFORM,
+                                    ATTRIB_CRITERION_EXACT,
+                                    0 },
+   { EGL_MAX_PBUFFER_WIDTH,         ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_IGNORE,
+                                    0 },
+   { EGL_MAX_PBUFFER_HEIGHT,        ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_IGNORE,
+                                    0 },
+   { EGL_MAX_PBUFFER_PIXELS,        ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_IGNORE,
+                                    0 },
+   { EGL_MAX_SWAP_INTERVAL,         ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_MIN_SWAP_INTERVAL,         ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_NATIVE_RENDERABLE,         ATTRIB_TYPE_BOOLEAN,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_NATIVE_VISUAL_ID,          ATTRIB_TYPE_PLATFORM,
+                                    ATTRIB_CRITERION_IGNORE,
+                                    0 },
+   { EGL_NATIVE_VISUAL_TYPE,        ATTRIB_TYPE_PLATFORM,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_RENDERABLE_TYPE,           ATTRIB_TYPE_BITMASK,
+                                    ATTRIB_CRITERION_MASK,
+                                    EGL_OPENGL_ES_BIT },
+   { EGL_SAMPLE_BUFFERS,            ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_SAMPLES,                   ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_STENCIL_SIZE,              ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_ATLEAST,
+                                    0 },
+   { EGL_SURFACE_TYPE,              ATTRIB_TYPE_BITMASK,
+                                    ATTRIB_CRITERION_MASK,
+                                    EGL_WINDOW_BIT },
+   { EGL_TRANSPARENT_TYPE,          ATTRIB_TYPE_ENUM,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_NONE },
+   { EGL_TRANSPARENT_RED_VALUE,     ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_TRANSPARENT_GREEN_VALUE,   ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   { EGL_TRANSPARENT_BLUE_VALUE,    ATTRIB_TYPE_INTEGER,
+                                    ATTRIB_CRITERION_EXACT,
+                                    EGL_DONT_CARE },
+   /* these are not real attributes */
+   { EGL_MATCH_NATIVE_PIXMAP,       ATTRIB_TYPE_PSEUDO,
+                                    ATTRIB_CRITERION_SPECIAL,
+                                    EGL_NONE },
+   { EGL_PRESERVED_RESOURCES,       ATTRIB_TYPE_PSEUDO,
+                                    ATTRIB_CRITERION_IGNORE,
+                                    0 },
+   { EGL_NONE,                      ATTRIB_TYPE_PSEUDO,
+                                    ATTRIB_CRITERION_IGNORE,
+                                    0 }
+};
 
 
 /**
- * Parse the attrib_list to fill in the fields of the given _eglConfig
- * Return EGL_FALSE if any errors, EGL_TRUE otherwise.
+ * Return true if a config is valid.  When for_matching is true,
+ * EGL_DONT_CARE is accepted as a valid attribute value, and checks
+ * for conflicting attribute values are skipped.
+ *
+ * Note that some attributes are platform-dependent and are not
+ * checked.
  */
 EGLBoolean
-_eglParseConfigAttribs(_EGLConfig *config, const EGLint *attrib_list)
+_eglValidateConfig(const _EGLConfig *conf, EGLBoolean for_matching)
 {
-   EGLint i;
-
-   /* set all config attribs to EGL_DONT_CARE */
-   for (i = 0; i < MAX_ATTRIBS; i++) {
-      config->Attrib[i] = EGL_DONT_CARE;
-   }
-
-   /* by default choose windows unless otherwise specified */
-   config->Attrib[EGL_SURFACE_TYPE - FIRST_ATTRIB] = EGL_WINDOW_BIT;
-
-   for (i = 0; attrib_list && attrib_list[i] != EGL_NONE; i++) {
-      const EGLint attr = attrib_list[i];
-      if (attr >= EGL_BUFFER_SIZE &&
-          attr <= EGL_MAX_SWAP_INTERVAL) {
-         EGLint k = attr - FIRST_ATTRIB;
-         assert(k >= 0);
-         assert(k < MAX_ATTRIBS);
-         config->Attrib[k] = attrib_list[++i];
-      }
-#ifdef EGL_VERSION_1_2
-      else if (attr == EGL_COLOR_BUFFER_TYPE) {
-         EGLint bufType = attrib_list[++i];
-         if (bufType != EGL_RGB_BUFFER && bufType != EGL_LUMINANCE_BUFFER) {
-            _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig");
-            return EGL_FALSE;
+   EGLint i, attr, val;
+   EGLBoolean valid = EGL_TRUE;
+   EGLint red_size = 0, green_size = 0, blue_size = 0, luminance_size = 0;
+   EGLint alpha_size = 0, buffer_size = 0;
+
+   /* all attributes should have been listed */
+   assert(ARRAY_SIZE(_eglValidationTable) == _EGL_CONFIG_NUM_ATTRIBS);
+
+   /* check attributes by their types */
+   for (i = 0; i < ARRAY_SIZE(_eglValidationTable); i++) {
+      EGLint mask;
+
+      attr = _eglValidationTable[i].attr;
+      val = GET_CONFIG_ATTRIB(conf, attr);
+
+      switch (_eglValidationTable[i].type) {
+      case ATTRIB_TYPE_INTEGER:
+         switch (attr) {
+         case EGL_CONFIG_ID:
+            /* config id must be positive */
+            if (val <= 0)
+               valid = EGL_FALSE;
+            break;
+         case EGL_SAMPLE_BUFFERS:
+            /* there can be at most 1 sample buffer */
+            if (val > 1)
+               valid = EGL_FALSE;
+            break;
+         case EGL_RED_SIZE:
+            red_size = val;
+            break;
+         case EGL_GREEN_SIZE:
+            green_size = val;
+            break;
+         case EGL_BLUE_SIZE:
+            blue_size = val;
+            break;
+         case EGL_LUMINANCE_SIZE:
+            luminance_size = val;
+            break;
+         case EGL_ALPHA_SIZE:
+            alpha_size = val;
+            break;
+         case EGL_BUFFER_SIZE:
+            buffer_size = val;
+            break;
          }
-         _eglSetConfigAttrib(config, EGL_COLOR_BUFFER_TYPE, bufType);
-      }
-      else if (attr == EGL_RENDERABLE_TYPE) {
-         EGLint renType = attrib_list[++i];
-         if (renType & ~(EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT | EGL_OPENVG_BIT)) {
-            _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig");
-            return EGL_FALSE;
+         if (val < 0)
+            valid = EGL_FALSE;
+         break;
+      case ATTRIB_TYPE_BOOLEAN:
+         if (val != EGL_TRUE && val != EGL_FALSE)
+            valid = EGL_FALSE;
+         break;
+      case ATTRIB_TYPE_ENUM:
+         switch (attr) {
+         case EGL_CONFIG_CAVEAT:
+            if (val != EGL_NONE && val != EGL_SLOW_CONFIG &&
+                val != EGL_NON_CONFORMANT_CONFIG)
+               valid = EGL_FALSE;
+            break;
+         case EGL_TRANSPARENT_TYPE:
+            if (val != EGL_NONE && val != EGL_TRANSPARENT_RGB)
+               valid = EGL_FALSE;
+            break;
+         case EGL_COLOR_BUFFER_TYPE:
+            if (val != EGL_RGB_BUFFER && val != EGL_LUMINANCE_BUFFER)
+               valid = EGL_FALSE;
+            break;
+         default:
+            assert(0);
+            break;
          }
-         _eglSetConfigAttrib(config, EGL_RENDERABLE_TYPE, renType);
-      }
-      else if (attr == EGL_ALPHA_MASK_SIZE ||
-               attr == EGL_LUMINANCE_SIZE) {
-         EGLint value = attrib_list[++i];
-         _eglSetConfigAttrib(config, attr, value);
+         break;
+      case ATTRIB_TYPE_BITMASK:
+         switch (attr) {
+         case EGL_SURFACE_TYPE:
+            mask = EGL_PBUFFER_BIT |
+                   EGL_PIXMAP_BIT |
+                   EGL_WINDOW_BIT |
+                   EGL_VG_COLORSPACE_LINEAR_BIT |
+                   EGL_VG_ALPHA_FORMAT_PRE_BIT |
+                   EGL_MULTISAMPLE_RESOLVE_BOX_BIT |
+                   EGL_SWAP_BEHAVIOR_PRESERVED_BIT;
+            break;
+         case EGL_RENDERABLE_TYPE:
+         case EGL_CONFORMANT:
+            mask = EGL_OPENGL_ES_BIT |
+                   EGL_OPENVG_BIT |
+                   EGL_OPENGL_ES2_BIT |
+                   EGL_OPENGL_BIT;
+            break;
+         default:
+            assert(0);
+            break;
+         }
+         if (val & ~mask)
+            valid = EGL_FALSE;
+         break;
+      case ATTRIB_TYPE_PLATFORM:
+         /* unable to check platform-dependent attributes here */
+         break;
+      case ATTRIB_TYPE_PSEUDO:
+         /* pseudo attributes should not be set */
+         if (val != 0)
+            valid = EGL_FALSE;
+         break;
+      default:
+         assert(0);
+         break;
       }
-#endif /* EGL_VERSION_1_2 */
-      else {
-         _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig");
-         return EGL_FALSE;
+
+      if (!valid && for_matching) {
+         /* accept EGL_DONT_CARE as a valid value */
+         if (val == EGL_DONT_CARE)
+            valid = EGL_TRUE;
+         if (_eglValidationTable[i].criterion == ATTRIB_CRITERION_SPECIAL)
+            valid = EGL_TRUE;
       }
+      if (!valid)
+         break;
    }
-   return EGL_TRUE;
-}
 
+   /* any invalid attribute value should have been catched */
+   if (!valid || for_matching)
+      return valid;
+
+   /* now check for conflicting attribute values */
+
+   switch (GET_CONFIG_ATTRIB(conf, EGL_COLOR_BUFFER_TYPE)) {
+   case EGL_RGB_BUFFER:
+      if (luminance_size)
+         valid = EGL_FALSE;
+      if (red_size + green_size + blue_size + alpha_size != buffer_size)
+         valid = EGL_FALSE;
+      break;
+   case EGL_LUMINANCE_BUFFER:
+      if (red_size || green_size || blue_size)
+         valid = EGL_FALSE;
+      if (luminance_size + alpha_size != buffer_size)
+         valid = EGL_FALSE;
+      break;
+   }
 
-#define EXACT 1
-#define ATLEAST 2
-#define MASK 3
-#define SMALLER 4
-#define SPECIAL 5
-#define NONE 6
+   val = GET_CONFIG_ATTRIB(conf, EGL_SAMPLE_BUFFERS);
+   if (!val && GET_CONFIG_ATTRIB(conf, EGL_SAMPLES))
+      valid = EGL_FALSE;
+
+   val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE);
+   if (!(val & EGL_WINDOW_BIT)) {
+      if (GET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_ID) != 0 ||
+          GET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE) != EGL_NONE)
+         valid = EGL_FALSE;
+   }
+   if (!(val & EGL_PBUFFER_BIT)) {
+      if (GET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGB) ||
+          GET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGBA))
+         valid = EGL_FALSE;
+   }
+
+   return valid;
+}
 
-struct sort_info {
-   EGLint Attribute;
-   EGLint MatchCriteria;
-   EGLint SortOrder;
-};
 
-/* This encodes the info from Table 3.5 of the EGL spec, ordered by
- * Sort Priority.
+/**
+ * Return true if a config matches the criteria.  This and
+ * _eglParseConfigAttribList together implement the algorithm
+ * described in "Selection of EGLConfigs".
  *
- * XXX To do: EGL 1.2 attribs
+ * Note that attributes that are special (currently, only
+ * EGL_MATCH_NATIVE_PIXMAP) are ignored.
  */
-static struct sort_info SortInfo[] = {
-   { EGL_CONFIG_CAVEAT,           EXACT,   SPECIAL },
-   { EGL_RED_SIZE,                ATLEAST, SPECIAL },
-   { EGL_GREEN_SIZE,              ATLEAST, SPECIAL },
-   { EGL_BLUE_SIZE,               ATLEAST, SPECIAL },
-   { EGL_ALPHA_SIZE,              ATLEAST, SPECIAL },
-   { EGL_BUFFER_SIZE,             ATLEAST, SMALLER },
-   { EGL_SAMPLE_BUFFERS,          ATLEAST, SMALLER },
-   { EGL_SAMPLES,                 ATLEAST, SMALLER },
-   { EGL_DEPTH_SIZE,              ATLEAST, SMALLER },
-   { EGL_STENCIL_SIZE,            ATLEAST, SMALLER },
-   { EGL_NATIVE_VISUAL_TYPE,      EXACT,   SPECIAL },
-   { EGL_CONFIG_ID,               EXACT,   SMALLER },
-   { EGL_BIND_TO_TEXTURE_RGB,     EXACT,   NONE    },
-   { EGL_BIND_TO_TEXTURE_RGBA,    EXACT,   NONE    },
-   { EGL_LEVEL,                   EXACT,   NONE    },
-   { EGL_NATIVE_RENDERABLE,       EXACT,   NONE    },
-   { EGL_MAX_SWAP_INTERVAL,       EXACT,   NONE    },
-   { EGL_MIN_SWAP_INTERVAL,       EXACT,   NONE    },
-   { EGL_SURFACE_TYPE,            MASK,    NONE    },
-   { EGL_TRANSPARENT_TYPE,        EXACT,   NONE    },
-   { EGL_TRANSPARENT_RED_VALUE,   EXACT,   NONE    },
-   { EGL_TRANSPARENT_GREEN_VALUE, EXACT,   NONE    },
-   { EGL_TRANSPARENT_BLUE_VALUE,  EXACT,   NONE    },
-   { 0, 0, 0 }
-};
+EGLBoolean
+_eglMatchConfig(const _EGLConfig *conf, const _EGLConfig *criteria)
+{
+   EGLint attr, val, i;
+   EGLBoolean matched = EGL_TRUE;
+
+   for (i = 0; i < ARRAY_SIZE(_eglValidationTable); i++) {
+      EGLint cmp;
+      if (_eglValidationTable[i].criterion == ATTRIB_CRITERION_IGNORE)
+         continue;
+
+      attr = _eglValidationTable[i].attr;
+      cmp = GET_CONFIG_ATTRIB(criteria, attr);
+      if (cmp == EGL_DONT_CARE)
+         continue;
+
+      val = GET_CONFIG_ATTRIB(conf, attr);
+      switch (_eglValidationTable[i].criterion) {
+      case ATTRIB_CRITERION_EXACT:
+         if (val != cmp)
+            matched = EGL_FALSE;
+         break;
+      case ATTRIB_CRITERION_ATLEAST:
+         if (val < cmp)
+            matched = EGL_FALSE;
+         break;
+      case ATTRIB_CRITERION_MASK:
+         if ((val & cmp) != cmp)
+            matched = EGL_FALSE;
+         break;
+      case ATTRIB_CRITERION_SPECIAL:
+         /* ignored here */
+         break;
+      default:
+         assert(0);
+         break;
+      }
+
+      if (!matched)
+         break;
+   }
+
+   return matched;
+}
 
 
 /**
- * Return EGL_TRUE if the attributes of c meet or exceed the minimums
- * specified by min.
+ * Initialize a criteria config from the given attribute list.
+ * Return EGL_FALSE if any of the attribute is invalid.
  */
-static EGLBoolean
-_eglConfigQualifies(const _EGLConfig *c, const _EGLConfig *min)
+EGLBoolean
+_eglParseConfigAttribList(_EGLConfig *conf, const EGLint *attrib_list)
 {
-   EGLint i;
-   for (i = 0; SortInfo[i].Attribute != 0; i++) {
-      const EGLint mv = GET_CONFIG_ATTRIB(min, SortInfo[i].Attribute);
-      if (mv != EGL_DONT_CARE) {
-         const EGLint cv = GET_CONFIG_ATTRIB(c, SortInfo[i].Attribute);
-         if (SortInfo[i].MatchCriteria == EXACT) {
-            if (cv != mv) {
-               return EGL_FALSE;
-            }
-         }
-         else if (SortInfo[i].MatchCriteria == ATLEAST) {
-            if (cv < mv) {
-               return EGL_FALSE;
-            }
-         }
-         else {
-            assert(SortInfo[i].MatchCriteria == MASK);
-            if ((mv & cv) != mv) {
-               return EGL_FALSE;
-            }
+   EGLint attr, val, i;
+   EGLint config_id = 0, level = 0;
+   EGLBoolean has_native_visual_type = EGL_FALSE;
+   EGLBoolean has_transparent_color = EGL_FALSE;
+
+   /* reset to default values */
+   for (i = 0; i < ARRAY_SIZE(_eglValidationTable); i++) {
+      attr = _eglValidationTable[i].attr;
+      val = _eglValidationTable[i].default_value;
+      SET_CONFIG_ATTRIB(conf, attr, val);
+   }
+
+   /* parse the list */
+   for (i = 0; attrib_list && attrib_list[i] != EGL_NONE; i += 2) {
+      EGLint idx;
+
+      attr = attrib_list[i];
+      val = attrib_list[i + 1];
+
+      idx = _eglIndexConfig(conf, attr);
+      if (idx < 0)
+         return EGL_FALSE;
+      conf->Storage[idx] = val;
+
+      /* rememeber some attributes for post-processing */
+      switch (attr) {
+      case EGL_CONFIG_ID:
+         config_id = val;
+         break;
+      case EGL_LEVEL:
+         level = val;
+         break;
+      case EGL_NATIVE_VISUAL_TYPE:
+         has_native_visual_type = EGL_TRUE;
+         break;
+      case EGL_TRANSPARENT_RED_VALUE:
+      case EGL_TRANSPARENT_GREEN_VALUE:
+      case EGL_TRANSPARENT_BLUE_VALUE:
+         has_transparent_color = EGL_TRUE;
+         break;
+      default:
+         break;
+      }
+   }
+
+   if (!_eglValidateConfig(conf, EGL_TRUE))
+      return EGL_FALSE;
+
+   /* the spec says that EGL_LEVEL cannot be EGL_DONT_CARE */
+   if (level == EGL_DONT_CARE)
+      return EGL_FALSE;
+
+   /* ignore other attributes when EGL_CONFIG_ID is given */
+   if (config_id > 0) {
+      _eglResetConfigKeys(conf, EGL_DONT_CARE);
+      SET_CONFIG_ATTRIB(conf, EGL_CONFIG_ID, config_id);
+   }
+   else {
+      if (has_native_visual_type) {
+         val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE);
+         if (!(val & EGL_WINDOW_BIT))
+            SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE, EGL_DONT_CARE);
+      }
+
+      if (has_transparent_color) {
+         val = GET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_TYPE);
+         if (val == EGL_NONE) {
+            SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_RED_VALUE,
+                              EGL_DONT_CARE);
+            SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_GREEN_VALUE,
+                              EGL_DONT_CARE);
+            SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_BLUE_VALUE,
+                              EGL_DONT_CARE);
          }
       }
    }
+
    return EGL_TRUE;
 }
 
 
 /**
- * Compare configs 'a' and 'b' and return -1 if a belongs before b,
- * 1 if a belongs after b, or 0 if they're equal.
- * Used by qsort().
+ * Decide the ordering of conf1 and conf2, under the given criteria.
+ * When compare_id is true, this implements the algorithm described
+ * in "Sorting of EGLConfigs".  When compare_id is false,
+ * EGL_CONFIG_ID is not compared.
+ *
+ * It returns a negative integer if conf1 is considered to come
+ * before conf2;  a positive integer if conf2 is considered to come
+ * before conf1;  zero if the ordering cannot be decided.
+ *
+ * Note that EGL_NATIVE_VISUAL_TYPE is platform-dependent and is
+ * ignored here.
  */
-static int
-_eglCompareConfigs(const void *a, const void *b)
+EGLint
+_eglCompareConfigs(const _EGLConfig *conf1, const _EGLConfig *conf2,
+                   const _EGLConfig *criteria, EGLBoolean compare_id)
 {
-   const _EGLConfig *aConfig = (const _EGLConfig *) a;
-   const _EGLConfig *bConfig = (const _EGLConfig *) b;
+   const EGLint compare_attribs[] = {
+      EGL_BUFFER_SIZE,
+      EGL_SAMPLE_BUFFERS,
+      EGL_SAMPLES,
+      EGL_DEPTH_SIZE,
+      EGL_STENCIL_SIZE,
+      EGL_ALPHA_MASK_SIZE,
+   };
+   EGLint val1, val2;
+   EGLBoolean rgb_buffer;
    EGLint i;
 
-   for (i = 0; SortInfo[i].Attribute != 0; i++) {
-      const EGLint aVal = GET_CONFIG_ATTRIB(aConfig, SortInfo[i].Attribute);
-      const EGLint bVal = GET_CONFIG_ATTRIB(bConfig, SortInfo[i].Attribute);
-      if (SortInfo[i].SortOrder == SMALLER) {
-         if (aVal < bVal)
-            return -1;
-         else if (aVal > bVal)
-            return 1;
-         /* else, continue examining attribute values */
-      }
-      else if (SortInfo[i].SortOrder == SPECIAL) {
-         if (SortInfo[i].Attribute == EGL_CONFIG_CAVEAT) {
-            /* values are EGL_NONE, SLOW_CONFIG, or NON_CONFORMANT_CONFIG */
-            if (aVal < bVal)
-               return -1;
-            else if (aVal > bVal)
-               return 1;
+   if (conf1 == conf2)
+      return 0;
+
+   /* the enum values have the desired ordering */
+   assert(EGL_NONE < EGL_SLOW_CONFIG);
+   assert(EGL_SLOW_CONFIG < EGL_NON_CONFORMANT_CONFIG);
+   val1 = GET_CONFIG_ATTRIB(conf1, EGL_CONFIG_CAVEAT);
+   val2 = GET_CONFIG_ATTRIB(conf2, EGL_CONFIG_CAVEAT);
+   if (val1 != val2)
+      return (val1 - val2);
+
+   /* the enum values have the desired ordering */
+   assert(EGL_RGB_BUFFER < EGL_LUMINANCE_BUFFER);
+   val1 = GET_CONFIG_ATTRIB(conf1, EGL_COLOR_BUFFER_TYPE);
+   val2 = GET_CONFIG_ATTRIB(conf2, EGL_COLOR_BUFFER_TYPE);
+   if (val1 != val2)
+      return (val1 - val2);
+   rgb_buffer = (val1 == EGL_RGB_BUFFER);
+
+   if (criteria) {
+      val1 = val2 = 0;
+      if (rgb_buffer) {
+         if (GET_CONFIG_ATTRIB(criteria, EGL_RED_SIZE) > 0) {
+            val1 += GET_CONFIG_ATTRIB(conf1, EGL_RED_SIZE);
+            val2 += GET_CONFIG_ATTRIB(conf2, EGL_RED_SIZE);
          }
-         else if (SortInfo[i].Attribute == EGL_RED_SIZE ||
-                  SortInfo[i].Attribute == EGL_GREEN_SIZE ||
-                  SortInfo[i].Attribute == EGL_BLUE_SIZE ||
-                  SortInfo[i].Attribute == EGL_ALPHA_SIZE) {
-            if (aVal > bVal)
-               return -1;
-            else if (aVal < bVal)
-               return 1;
+         if (GET_CONFIG_ATTRIB(criteria, EGL_GREEN_SIZE) > 0) {
+            val1 += GET_CONFIG_ATTRIB(conf1, EGL_GREEN_SIZE);
+            val2 += GET_CONFIG_ATTRIB(conf2, EGL_GREEN_SIZE);
          }
-         else {
-            assert(SortInfo[i].Attribute == EGL_NATIVE_VISUAL_TYPE);
-            if (aVal < bVal)
-               return -1;
-            else if (aVal > bVal)
-               return 1;
+         if (GET_CONFIG_ATTRIB(criteria, EGL_BLUE_SIZE) > 0) {
+            val1 += GET_CONFIG_ATTRIB(conf1, EGL_BLUE_SIZE);
+            val2 += GET_CONFIG_ATTRIB(conf2, EGL_BLUE_SIZE);
          }
       }
       else {
-         assert(SortInfo[i].SortOrder == NONE);
-         /* continue examining attribute values */
+         if (GET_CONFIG_ATTRIB(criteria, EGL_LUMINANCE_SIZE) > 0) {
+            val1 += GET_CONFIG_ATTRIB(conf1, EGL_LUMINANCE_SIZE);
+            val2 += GET_CONFIG_ATTRIB(conf2, EGL_LUMINANCE_SIZE);
+         }
+      }
+      if (GET_CONFIG_ATTRIB(criteria, EGL_ALPHA_SIZE) > 0) {
+         val1 += GET_CONFIG_ATTRIB(conf1, EGL_ALPHA_SIZE);
+         val2 += GET_CONFIG_ATTRIB(conf2, EGL_ALPHA_SIZE);
       }
    }
+   else {
+      /* assume the default criteria, which gives no specific ordering */
+      val1 = val2 = 0;
+   }
+
+   /* for color bits, larger one is preferred */
+   if (val1 != val2)
+      return (val2 - val1);
+
+   for (i = 0; i < ARRAY_SIZE(compare_attribs); i++) {
+      val1 = GET_CONFIG_ATTRIB(conf1, compare_attribs[i]);
+      val2 = GET_CONFIG_ATTRIB(conf2, compare_attribs[i]);
+      if (val1 != val2)
+         return (val1 - val2);
+   }
+
+   /* EGL_NATIVE_VISUAL_TYPE cannot be compared here */
+
+   if (compare_id) {
+      val1 = GET_CONFIG_ATTRIB(conf1, EGL_CONFIG_ID);
+      val2 = GET_CONFIG_ATTRIB(conf2, EGL_CONFIG_ID);
+      assert(val1 != val2);
+   }
+   else {
+      val1 = val2 = 0;
+   }
+
+   return (val1 - val2);
+}
+
+
+static INLINE
+void _eglSwapConfigs(const _EGLConfig **conf1, const _EGLConfig **conf2)
+{
+   const _EGLConfig *tmp = *conf1;
+   *conf1 = *conf2;
+   *conf2 = tmp;
+}
+
+
+/**
+ * Quick sort an array of configs.  This differs from the standard
+ * qsort() in that the compare function accepts an additional
+ * argument.
+ */
+void
+_eglSortConfigs(const _EGLConfig **configs, EGLint count,
+                EGLint (*compare)(const _EGLConfig *, const _EGLConfig *,
+                                  void *),
+                void *priv_data)
+{
+   const EGLint pivot = 0;
+   EGLint i, j;
+
+   if (count <= 1)
+      return;
+
+   _eglSwapConfigs(&configs[pivot], &configs[count / 2]);
+   i = 1;
+   j = count - 1;
+   do {
+      while (i < count && compare(configs[i], configs[pivot], priv_data) < 0)
+         i++;
+      while (compare(configs[j], configs[pivot], priv_data) > 0)
+         j--;
+      if (i < j) {
+         _eglSwapConfigs(&configs[i], &configs[j]);
+         i++;
+         j--;
+      }
+      else if (i == j) {
+         i++;
+         j--;
+         break;
+      }
+   } while (i <= j);
+   _eglSwapConfigs(&configs[pivot], &configs[j]);
 
-   /* all attributes identical */
-   return 0;
+   _eglSortConfigs(configs, j, compare, priv_data);
+   _eglSortConfigs(configs + i, count - i, compare, priv_data);
+}
+
+
+static int
+_eglFallbackCompare(const _EGLConfig *conf1, const _EGLConfig *conf2,
+                   void *priv_data)
+{
+   const _EGLConfig *criteria = (const _EGLConfig *) priv_data;
+   return _eglCompareConfigs(conf1, conf2, criteria, EGL_TRUE);
 }
 
 
@@ -324,33 +727,33 @@ _eglChooseConfig(_EGLDriver *drv, _EGLDisplay *disp, const EGLint *attrib_list,
    _EGLConfig **configList, criteria;
    EGLint i, count;
 
-   /* parse the attrib_list to initialize criteria */
-   if (!_eglParseConfigAttribs(&criteria, attrib_list)) {
-      return EGL_FALSE;
-   }
+   if (!num_configs)
+      return _eglError(EGL_BAD_PARAMETER, "eglChooseConfigs");
+
+   _eglInitConfig(&criteria, 0);
+   if (!_eglParseConfigAttribList(&criteria, attrib_list))
+      return _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig");
 
    /* allocate array of config pointers */
-   configList = (_EGLConfig **) malloc(config_size * sizeof(_EGLConfig *));
-   if (!configList) {
-      _eglError(EGL_BAD_CONFIG, "eglChooseConfig(out of memory)");
-      return EGL_FALSE;
-   }
+   configList = (_EGLConfig **)
+      malloc(disp->NumConfigs * sizeof(_EGLConfig *));
+   if (!configList)
+      return _eglError(EGL_BAD_ALLOC, "eglChooseConfig(out of memory)");
 
-   /* make array of pointers to qualifying configs */
-   for (i = count = 0; i < disp->NumConfigs && count < config_size; i++) {
-      if (_eglConfigQualifies(disp->Configs[i], &criteria)) {
+   /* perform selection of configs */
+   count = 0;
+   for (i = 0; i < disp->NumConfigs; i++) {
+      if (_eglMatchConfig(disp->Configs[i], &criteria))
          configList[count++] = disp->Configs[i];
-      }
    }
 
-   /* sort array of pointers */
-   qsort(configList, count, sizeof(_EGLConfig *), _eglCompareConfigs);
-
-   /* copy config handles to output array */
-   if (configs) {
-      for (i = 0; i < count; i++) {
-         configs[i] = configList[i]->Handle;
-      }
+   /* perform sorting of configs */
+   if (configs && count) {
+      _eglSortConfigs((const _EGLConfig **) configList, count,
+                      _eglFallbackCompare, (void *) &criteria);
+      count = MIN2(count, config_size);
+      for (i = 0; i < count; i++)
+         configs[i] = _eglGetConfigHandle(configList[i]);
    }
 
    free(configList);
@@ -361,6 +764,28 @@ _eglChooseConfig(_EGLDriver *drv, _EGLDisplay *disp, const EGLint *attrib_list,
 }
 
 
+static INLINE EGLBoolean
+_eglIsConfigAttribValid(_EGLConfig *conf, EGLint attr)
+{
+   if (_eglIndexConfig(conf, attr) < 0)
+      return EGL_FALSE;
+
+   /* there are some holes in the range */
+   switch (attr) {
+   case EGL_PRESERVED_RESOURCES:
+   case EGL_NONE:
+#ifdef EGL_VERSION_1_4
+   case EGL_MATCH_NATIVE_PIXMAP:
+#endif
+      return EGL_FALSE;
+   default:
+      break;
+   }
+
+   return EGL_TRUE;
+}
+
+
 /**
  * Fallback for eglGetConfigAttrib.
  */
@@ -368,15 +793,13 @@ EGLBoolean
 _eglGetConfigAttrib(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf,
                     EGLint attribute, EGLint *value)
 {
-   const EGLint k = attribute - FIRST_ATTRIB;
-   if (k >= 0 && k < MAX_ATTRIBS) {
-      *value = conf->Attrib[k];
-      return EGL_TRUE;
-   }
-   else {
-      _eglError(EGL_BAD_ATTRIBUTE, "eglGetConfigAttrib");
-      return EGL_FALSE;
-   }
+   if (!_eglIsConfigAttribValid(conf, attribute))
+      return _eglError(EGL_BAD_ATTRIBUTE, "eglGetConfigAttrib");
+   if (!value)
+      return _eglError(EGL_BAD_PARAMETER, "eglGetConfigAttrib");
+
+   *value = GET_CONFIG_ATTRIB(conf, attribute);
+   return EGL_TRUE;
 }
 
 
@@ -387,12 +810,14 @@ EGLBoolean
 _eglGetConfigs(_EGLDriver *drv, _EGLDisplay *disp, EGLConfig *configs,
                EGLint config_size, EGLint *num_config)
 {
+   if (!num_config)
+      return _eglError(EGL_BAD_PARAMETER, "eglGetConfigs");
+
    if (configs) {
       EGLint i;
       *num_config = MIN2(disp->NumConfigs, config_size);
-      for (i = 0; i < *num_config; i++) {
-         configs[i] = disp->Configs[i]->Handle;
-      }
+      for (i = 0; i < *num_config; i++)
+         configs[i] = _eglGetConfigHandle(disp->Configs[i]);
    }
    else {
       /* just return total number of supported configs */
diff --git a/src/egl/main/eglconfig.h b/src/egl/main/eglconfig.h
index 36ed96ae95..6b8a259984 100644
--- a/src/egl/main/eglconfig.h
+++ b/src/egl/main/eglconfig.h
@@ -2,27 +2,93 @@
 #define EGLCONFIG_INCLUDED
 
 
+#include <assert.h>
 #include "egltypedefs.h"
-#include <GLES/gl.h>
 
 
-#define MAX_ATTRIBS 128
-#define FIRST_ATTRIB EGL_BUFFER_SIZE
+#define _EGL_CONFIG_FIRST_ATTRIB EGL_BUFFER_SIZE
+#define _EGL_CONFIG_LAST_ATTRIB EGL_CONFORMANT
+#define _EGL_CONFIG_NUM_ATTRIBS \
+   (_EGL_CONFIG_LAST_ATTRIB - _EGL_CONFIG_FIRST_ATTRIB + 1)
+
+#define _EGL_CONFIG_STORAGE_SIZE _EGL_CONFIG_NUM_ATTRIBS
 
 
 struct _egl_config
 {
-   EGLConfig Handle;   /* the public/opaque handle which names this config */
-   EGLint Attrib[MAX_ATTRIBS];
+   _EGLDisplay *Display;
+   EGLint Storage[_EGL_CONFIG_STORAGE_SIZE];
 };
 
 
-#define SET_CONFIG_ATTRIB(CONF, ATTR, VAL) \
-   assert((ATTR) - FIRST_ATTRIB < MAX_ATTRIBS); \
-   ((CONF)->Attrib[(ATTR) - FIRST_ATTRIB] = VAL)
+#define SET_CONFIG_ATTRIB(CONF, ATTR, VAL) _eglSetConfigKey(CONF, ATTR, VAL)
+#define GET_CONFIG_ATTRIB(CONF, ATTR) _eglGetConfigKey(CONF, ATTR)
+
+
+/**
+ * Given a key, return an index into the storage of the config.
+ * Return -1 if the key is invalid.
+ */
+static INLINE EGLint
+_eglIndexConfig(const _EGLConfig *conf, EGLint key)
+{
+   (void) conf;
+   if (key >= _EGL_CONFIG_FIRST_ATTRIB &&
+       key < _EGL_CONFIG_FIRST_ATTRIB + _EGL_CONFIG_NUM_ATTRIBS)
+      return key - _EGL_CONFIG_FIRST_ATTRIB;
+   else
+      return -1;
+}
+
+
+/**
+ * Reset all keys in the config to a given value.
+ */
+static INLINE void
+_eglResetConfigKeys(_EGLConfig *conf, EGLint val)
+{
+   EGLint i;
+   for (i = 0; i < _EGL_CONFIG_NUM_ATTRIBS; i++)
+      conf->Storage[i] = val;
+}
 
 
-#define GET_CONFIG_ATTRIB(CONF, ATTR) ((CONF)->Attrib[(ATTR) - FIRST_ATTRIB])
+/**
+ * Update a config for a given key.
+ */
+static INLINE void
+_eglSetConfigKey(_EGLConfig *conf, EGLint key, EGLint val)
+{
+   EGLint idx = _eglIndexConfig(conf, key);
+   assert(idx >= 0);
+   conf->Storage[idx] = val;
+}
+
+
+/**
+ * Return the value for a given key.
+ */
+static INLINE EGLint
+_eglGetConfigKey(const _EGLConfig *conf, EGLint key)
+{
+   EGLint idx = _eglIndexConfig(conf, key);
+   assert(idx >= 0);
+   return conf->Storage[idx];
+}
+
+
+/**
+ * Set a given attribute.
+ *
+ * Because _eglGetConfigAttrib is already used as a fallback driver
+ * function, this function is not considered to have a good name.
+ * SET_CONFIG_ATTRIB is preferred over this function.
+ */
+static INLINE void
+_eglSetConfigAttrib(_EGLConfig *conf, EGLint attr, EGLint val)
+{
+   SET_CONFIG_ATTRIB(conf, attr, val);
+}
 
 
 extern void
@@ -30,19 +96,76 @@ _eglInitConfig(_EGLConfig *config, EGLint id);
 
 
 extern EGLConfig
-_eglGetConfigHandle(_EGLConfig *config);
+_eglAddConfig(_EGLDisplay *dpy, _EGLConfig *conf);
+
+
+#ifndef _EGL_SKIP_HANDLE_CHECK
+
+
+extern EGLBoolean
+_eglCheckConfigHandle(EGLConfig config, _EGLDisplay *dpy);
+
+
+#else
+
 
+static INLINE EGLBoolean
+_eglCheckConfigHandle(EGLConfig config, _EGLDisplay *dpy)
+{
+   _EGLConfig *conf = (_EGLConfig *) config;
+   return (dpy && conf && conf->Display == dpy);
+}
+
+
+#endif /* _EGL_SKIP_HANDLE_CHECK */
 
-extern _EGLConfig *
-_eglLookupConfig(EGLConfig config, _EGLDisplay *dpy);
+
+/**
+ * Lookup a handle to find the linked config.
+ * Return NULL if the handle has no corresponding linked config.
+ */
+static INLINE _EGLConfig *
+_eglLookupConfig(EGLConfig config, _EGLDisplay *dpy)
+{
+   _EGLConfig *conf = (_EGLConfig *) config;
+   if (!_eglCheckConfigHandle(config, dpy))
+      conf = NULL;
+   return conf;
+}
+
+
+/**
+ * Return the handle of a linked config, or NULL.
+ */
+static INLINE EGLConfig
+_eglGetConfigHandle(_EGLConfig *conf)
+{
+   return (EGLConfig) ((conf && conf->Display) ? conf : NULL);
+}
+
+
+extern EGLBoolean
+_eglValidateConfig(const _EGLConfig *conf, EGLBoolean for_matching);
 
 
-extern _EGLConfig *
-_eglAddConfig(_EGLDisplay *display, _EGLConfig *config);
+extern EGLBoolean
+_eglMatchConfig(const _EGLConfig *conf, const _EGLConfig *criteria);
 
 
 extern EGLBoolean
-_eglParseConfigAttribs(_EGLConfig *config, const EGLint *attrib_list);
+_eglParseConfigAttribList(_EGLConfig *conf, const EGLint *attrib_list);
+
+
+extern EGLint
+_eglCompareConfigs(const _EGLConfig *conf1, const _EGLConfig *conf2,
+                   const _EGLConfig *criteria, EGLBoolean compare_id);
+
+
+extern void
+_eglSortConfigs(const _EGLConfig **configs, EGLint count,
+                EGLint (*compare)(const _EGLConfig *, const _EGLConfig *,
+                                  void *),
+                void *priv_data);
 
 
 extern EGLBoolean
@@ -57,8 +180,4 @@ extern EGLBoolean
 _eglGetConfigs(_EGLDriver *drv, _EGLDisplay *dpy, EGLConfig *configs, EGLint config_size, EGLint *num_config);
 
 
-extern void
-_eglSetConfigAttrib(_EGLConfig *config, EGLint attr, EGLint val);
-
-
 #endif /* EGLCONFIG_INCLUDED */
diff --git a/src/egl/main/eglconfigutil.c b/src/egl/main/eglconfigutil.c
index c9d00e7982..36e94f0d2d 100644
--- a/src/egl/main/eglconfigutil.c
+++ b/src/egl/main/eglconfigutil.c
@@ -52,6 +52,83 @@ _eglConfigToContextModesRec(const _EGLConfig *config, __GLcontextModes *mode)
 }
 
 
+/**
+ * Convert a __GLcontextModes object to an _EGLConfig.
+ */
+EGLBoolean
+_eglConfigFromContextModesRec(_EGLConfig *conf, const __GLcontextModes *m,
+                              EGLint conformant, EGLint renderable_type)
+{
+   EGLint config_caveat, surface_type;
+
+   /* must be RGBA */
+   if (!m->rgbMode || !(m->renderType & GLX_RGBA_BIT))
+      return EGL_FALSE;
+
+   config_caveat = EGL_NONE;
+   if (m->visualRating == GLX_SLOW_CONFIG)
+      config_caveat = EGL_SLOW_CONFIG;
+
+   if (m->visualRating == GLX_NON_CONFORMANT_CONFIG)
+      conformant &= ~EGL_OPENGL_BIT;
+   if (!(conformant & EGL_OPENGL_ES_BIT))
+      config_caveat = EGL_NON_CONFORMANT_CONFIG;
+
+   surface_type = 0;
+   if (m->drawableType & GLX_WINDOW_BIT)
+      surface_type |= EGL_WINDOW_BIT;
+   if (m->drawableType & GLX_PIXMAP_BIT)
+      surface_type |= EGL_PIXMAP_BIT;
+   if (m->drawableType & GLX_PBUFFER_BIT)
+      surface_type |= EGL_PBUFFER_BIT;
+
+   SET_CONFIG_ATTRIB(conf, EGL_BUFFER_SIZE, m->rgbBits);
+   SET_CONFIG_ATTRIB(conf, EGL_RED_SIZE, m->redBits);
+   SET_CONFIG_ATTRIB(conf, EGL_GREEN_SIZE, m->greenBits);
+   SET_CONFIG_ATTRIB(conf, EGL_BLUE_SIZE, m->blueBits);
+   SET_CONFIG_ATTRIB(conf, EGL_ALPHA_SIZE, m->alphaBits);
+
+   SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGB, m->bindToTextureRgb);
+   SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGBA, m->bindToTextureRgba);
+   SET_CONFIG_ATTRIB(conf, EGL_COLOR_BUFFER_TYPE, EGL_RGB_BUFFER);
+   SET_CONFIG_ATTRIB(conf, EGL_CONFIG_CAVEAT, config_caveat);
+
+   SET_CONFIG_ATTRIB(conf, EGL_CONFORMANT, conformant);
+   SET_CONFIG_ATTRIB(conf, EGL_DEPTH_SIZE, m->depthBits);
+   SET_CONFIG_ATTRIB(conf, EGL_LEVEL, m->level);
+   SET_CONFIG_ATTRIB(conf, EGL_MAX_PBUFFER_WIDTH, m->maxPbufferWidth);
+   SET_CONFIG_ATTRIB(conf, EGL_MAX_PBUFFER_HEIGHT, m->maxPbufferHeight);
+   SET_CONFIG_ATTRIB(conf, EGL_MAX_PBUFFER_PIXELS, m->maxPbufferPixels);
+
+   SET_CONFIG_ATTRIB(conf, EGL_NATIVE_RENDERABLE, m->xRenderable);
+   SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_ID, m->visualID);
+
+   if (m->visualType != GLX_NONE)
+      SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE, m->visualType);
+   else
+      SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE, EGL_NONE);
+
+   SET_CONFIG_ATTRIB(conf, EGL_RENDERABLE_TYPE, renderable_type);
+   SET_CONFIG_ATTRIB(conf, EGL_SAMPLE_BUFFERS, m->sampleBuffers);
+   SET_CONFIG_ATTRIB(conf, EGL_SAMPLES, m->samples);
+   SET_CONFIG_ATTRIB(conf, EGL_STENCIL_SIZE, m->stencilBits);
+
+   SET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE, surface_type);
+
+   /* what to do with GLX_TRANSPARENT_INDEX? */
+   if (m->transparentPixel == GLX_TRANSPARENT_RGB) {
+      SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_TYPE, EGL_TRANSPARENT_RGB);
+      SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_RED_VALUE, m->transparentRed);
+      SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_GREEN_VALUE, m->transparentGreen);
+      SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_BLUE_VALUE, m->transparentBlue);
+   }
+   else {
+      SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_TYPE, EGL_NONE);
+   }
+
+   return EGL_TRUE;
+}
+
 
 /**
  * Creates a set of \c _EGLConfigs that a driver will expose.
diff --git a/src/egl/main/eglconfigutil.h b/src/egl/main/eglconfigutil.h
index c477b94737..8c923ee206 100644
--- a/src/egl/main/eglconfigutil.h
+++ b/src/egl/main/eglconfigutil.h
@@ -1,12 +1,10 @@
-
 #ifndef EGLCONFIGUTIL_INCLUDED
 #define EGLCONFIGUTIL_INCLUDED
 
-#include "eglconfig.h"
+
+#include "GL/gl.h"
 #include "GL/internal/glcore.h"
-#if (!defined(WIN32) && !defined(_WIN32_WCE))
-#include "stdint.h"
-#endif
+#include "eglconfig.h"
 
 
 extern void
@@ -14,6 +12,11 @@ _eglConfigToContextModesRec(const _EGLConfig *config, __GLcontextModes *mode);
 
 
 extern EGLBoolean
+_eglConfigFromContextModesRec(_EGLConfig *conf, const __GLcontextModes *m,
+                              EGLint conformant, EGLint renderable_type);
+
+
+extern EGLBoolean
 _eglFillInConfigs( _EGLConfig *configs,
                    EGLenum fb_format, EGLenum fb_type,
                    const uint8_t * depth_bits, const uint8_t * stencil_bits,
diff --git a/src/egl/main/eglcontext.c b/src/egl/main/eglcontext.c
index b094f49bfc..ee4b1b59f5 100644
--- a/src/egl/main/eglcontext.c
+++ b/src/egl/main/eglcontext.c
@@ -45,6 +45,7 @@ _eglInitContext(_EGLDriver *drv, _EGLContext *ctx,
    ctx->DrawSurface = EGL_NO_SURFACE;
    ctx->ReadSurface = EGL_NO_SURFACE;
    ctx->ClientAPI = api;
+   ctx->WindowRenderBuffer = EGL_NONE;
 
    return EGL_TRUE;
 }
@@ -87,6 +88,24 @@ _eglDestroyContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx)
 }
 
 
+#ifdef EGL_VERSION_1_2
+static EGLint
+_eglQueryContextRenderBuffer(_EGLContext *ctx)
+{
+   _EGLSurface *surf = ctx->DrawSurface;
+   EGLint rb;
+
+   if (!surf)
+      return EGL_NONE;
+   if (surf->Type == EGL_WINDOW_BIT && ctx->WindowRenderBuffer != EGL_NONE)
+      rb = ctx->WindowRenderBuffer;
+   else
+      rb = surf->RenderBuffer;
+   return rb;
+}
+#endif /* EGL_VERSION_1_2 */
+
+
 EGLBoolean
 _eglQueryContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *c,
                  EGLint attribute, EGLint *value)
@@ -94,22 +113,29 @@ _eglQueryContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *c,
    (void) drv;
    (void) dpy;
 
+   if (!value)
+      return _eglError(EGL_BAD_PARAMETER, "eglQueryContext");
+
    switch (attribute) {
    case EGL_CONFIG_ID:
       *value = GET_CONFIG_ATTRIB(c->Config, EGL_CONFIG_ID);
-      return EGL_TRUE;
+      break;
+   case EGL_CONTEXT_CLIENT_VERSION:
+      *value = c->ClientVersion;
+      break;
 #ifdef EGL_VERSION_1_2
    case EGL_CONTEXT_CLIENT_TYPE:
       *value = c->ClientAPI;
-      return EGL_TRUE;
+      break;
+   case EGL_RENDER_BUFFER:
+      *value = _eglQueryContextRenderBuffer(c);
+      break;
 #endif /* EGL_VERSION_1_2 */
-   case EGL_CONTEXT_CLIENT_VERSION:
-      *value = c->ClientVersion;
-      return EGL_TRUE;
    default:
-      _eglError(EGL_BAD_ATTRIBUTE, "eglQueryContext");
-      return EGL_FALSE;
+      return _eglError(EGL_BAD_ATTRIBUTE, "eglQueryContext");
    }
+
+   return EGL_TRUE;
 }
 
 
diff --git a/src/egl/main/eglcontext.h b/src/egl/main/eglcontext.h
index 647f24488f..45c7b4717b 100644
--- a/src/egl/main/eglcontext.h
+++ b/src/egl/main/eglcontext.h
@@ -24,6 +24,9 @@ struct _egl_context
 
    EGLint ClientAPI; /**< EGL_OPENGL_ES_API, EGL_OPENGL_API, EGL_OPENVG_API */
    EGLint ClientVersion; /**< 1 = OpenGLES 1.x, 2 = OpenGLES 2.x */
+
+   /* The real render buffer when a window surface is bound */
+   EGLint WindowRenderBuffer;
 };
 
 
diff --git a/src/egl/main/eglcurrent.c b/src/egl/main/eglcurrent.c
index 4431f964f6..df506151b5 100644
--- a/src/egl/main/eglcurrent.c
+++ b/src/egl/main/eglcurrent.c
@@ -9,14 +9,14 @@
 
 /* This should be kept in sync with _eglInitThreadInfo() */
 #define _EGL_THREAD_INFO_INITIALIZER \
-   { EGL_SUCCESS, { NULL }, 1 }
+   { EGL_SUCCESS, { NULL }, 0 }
 
 /* a fallback thread info to guarantee that every thread always has one */
 static _EGLThreadInfo dummy_thread = _EGL_THREAD_INFO_INITIALIZER;
 
 
 #ifdef GLX_USE_TLS
-static __thread const _EGLThreadInfo *_egl_TSD;
+static __thread const _EGLThreadInfo *_egl_TSD
    __attribute__ ((tls_model("initial-exec")));
 
 static INLINE void _eglSetTSD(const _EGLThreadInfo *t)
diff --git a/src/egl/main/eglcurrent.h b/src/egl/main/eglcurrent.h
index 8eb241029e..9503e0aba0 100644
--- a/src/egl/main/eglcurrent.h
+++ b/src/egl/main/eglcurrent.h
@@ -4,8 +4,16 @@
 #include "egltypedefs.h"
 
 
-#define _EGL_API_NUM_INDICES \
-   (EGL_OPENGL_API - EGL_OPENGL_ES_API + 2) /* idx 0 is for EGL_NONE */
+#define _EGL_API_ALL_BITS \
+   (EGL_OPENGL_ES_BIT   | \
+    EGL_OPENVG_BIT      | \
+    EGL_OPENGL_ES2_BIT  | \
+    EGL_OPENGL_BIT)
+
+
+#define _EGL_API_FIRST_API EGL_OPENGL_ES_API
+#define _EGL_API_LAST_API EGL_OPENGL_API
+#define _EGL_API_NUM_APIS (_EGL_API_LAST_API - _EGL_API_FIRST_API + 1)
 
 
 /**
@@ -14,20 +22,19 @@
 struct _egl_thread_info
 {
    EGLint LastError;
-   _EGLContext *CurrentContexts[_EGL_API_NUM_INDICES];
+   _EGLContext *CurrentContexts[_EGL_API_NUM_APIS];
    /* use index for fast access to current context */
    EGLint CurrentAPIIndex;
 };
 
 
 /**
- * Return true if a client API enum can be converted to an index.
+ * Return true if a client API enum is recognized.
  */
 static INLINE EGLBoolean
 _eglIsApiValid(EGLenum api)
 {
-   return ((api >= EGL_OPENGL_ES_API && api <= EGL_OPENGL_API) ||
-           api == EGL_NONE);
+   return (api >= _EGL_API_FIRST_API && api <= _EGL_API_LAST_API);
 }
 
 
@@ -38,7 +45,7 @@ _eglIsApiValid(EGLenum api)
 static INLINE EGLint
 _eglConvertApiToIndex(EGLenum api)
 {
-   return (api != EGL_NONE) ? api - EGL_OPENGL_ES_API + 1 : 0;
+   return api - _EGL_API_FIRST_API;
 }
 
 
@@ -49,7 +56,7 @@ _eglConvertApiToIndex(EGLenum api)
 static INLINE EGLenum
 _eglConvertApiFromIndex(EGLint idx)
 {
-   return (idx) ? EGL_OPENGL_ES_API + idx - 1 : EGL_NONE;
+   return _EGL_API_FIRST_API + idx;
 }
 
 
diff --git a/src/egl/main/egldisplay.h b/src/egl/main/egldisplay.h
index 6575fdf198..ea4e35a8b3 100644
--- a/src/egl/main/egldisplay.h
+++ b/src/egl/main/egldisplay.h
@@ -44,6 +44,7 @@ struct _egl_display
    EGLint NumScreens;
    _EGLScreen **Screens;  /* array [NumScreens] */
 
+   EGLint MaxConfigs;
    EGLint NumConfigs;
    _EGLConfig **Configs;  /* array [NumConfigs] of ptr to _EGLConfig */
 
diff --git a/src/egl/main/egldriver.c b/src/egl/main/egldriver.c
index 87786e36bb..018b06d3be 100644
--- a/src/egl/main/egldriver.c
+++ b/src/egl/main/egldriver.c
@@ -22,59 +22,70 @@
 
 #if defined(_EGL_PLATFORM_X)
 #include <dlfcn.h>
-#elif defined(_EGL_PLATFORM_WINDOWS)
-/* Use static linking on Windows for now */
-#define WINDOWS_STATIC_LINK
 #endif
 
+
 /**
  * Wrappers for dlopen/dlclose()
  */
 #if defined(_EGL_PLATFORM_WINDOWS)
-#ifdef WINDOWS_STATIC_LINK
-   static const char *DefaultDriverName = "Windows EGL Static Library";
-#else
-   /* XXX Need to decide how to do dynamic name lookup on Windows */
-   static const char *DefaultDriverName = "TBD";
-#endif
-   typedef HMODULE lib_handle;
-
-   static HMODULE
-   open_library(const char *filename)
-   {
-#ifdef WINDOWS_STATIC_LINK
-      return 0;
-#else
-      return LoadLibrary(filename);
-#endif
-   }
 
-   static void
-   close_library(HMODULE lib)
-   {
-#ifdef WINDOWS_STATIC_LINK
-#else
-      FreeLibrary(lib);
-#endif
-   }
+
+/* XXX Need to decide how to do dynamic name lookup on Windows */
+static const char DefaultDriverName[] = "TBD";
+
+typedef HMODULE lib_handle;
+
+static HMODULE
+open_library(const char *filename)
+{
+   return LoadLibrary(filename);
+}
+
+static void
+close_library(HMODULE lib)
+{
+   FreeLibrary(lib);
+}
+
 
 #elif defined(_EGL_PLATFORM_X)
-   static const char *DefaultDriverName = "egl_softpipe";
 
-   typedef void * lib_handle;
 
-   static void *
-   open_library(const char *filename)
-   {
-      return dlopen(filename, RTLD_LAZY);
-   }
+static const char DefaultDriverName[] = "egl_softpipe";
+
+typedef void * lib_handle;
+
+static void *
+open_library(const char *filename)
+{
+   return dlopen(filename, RTLD_LAZY);
+}
+
+static void
+close_library(void *lib)
+{
+   dlclose(lib);
+}
+
+#else /* _EGL_PLATFORM_NO_OS */
+
+static const char DefaultDriverName[] = "builtin";
+
+typedef void *lib_handle;
+
+static INLINE void *
+open_library(const char *filename)
+{
+   return (void *) filename;
+}
+
+static INLINE void
+close_library(void *lib)
+{
+}
+
 
-   static void
-   close_library(void *lib)
-   {
-      dlclose(lib);
-   }
-   
 #endif
 
 
@@ -95,14 +106,22 @@ _eglChooseDriver(_EGLDisplay *dpy, char **argsRet)
       path = _eglstrdup(path);
 
 #if defined(_EGL_PLATFORM_X)
-   if (!path && dpy->NativeDisplay) {
+   if (!path && dpy && dpy->NativeDisplay) {
       /* assume (wrongly!) that the native display is a display string */
       path = _eglSplitDisplayString((const char *) dpy->NativeDisplay, &args);
    }
    suffix = "so";
 #elif defined(_EGL_PLATFORM_WINDOWS)
    suffix = "dll";
-#endif /* _EGL_PLATFORM_X */
+#else /* _EGL_PLATFORM_NO_OS */
+   if (path) {
+      /* force the use of the default driver */
+      _eglLog(_EGL_DEBUG, "ignore EGL_DRIVER");
+      free(path);
+      path = NULL;
+   }
+   suffix = NULL;
+#endif
 
    if (!path)
       path = _eglstrdup(DefaultDriverName);
@@ -136,43 +155,48 @@ _eglChooseDriver(_EGLDisplay *dpy, char **argsRet)
 static _EGLMain_t
 _eglOpenLibrary(const char *driverPath, lib_handle *handle)
 {
-   _EGLMain_t mainFunc;
    lib_handle lib;
+   _EGLMain_t mainFunc = NULL;
+   const char *error = "unknown error";
 
    assert(driverPath);
 
-#if defined(_EGL_PLATFORM_WINDOWS)
-/* Use static linking on Windows for now */
-#ifdef WINDOWS_STATIC_LINK
-   lib = 0;
-   mainFunc = (_EGLMain_t)_eglMain;
-#else
-   /* XXX untested */
    _eglLog(_EGL_DEBUG, "dlopen(%s)", driverPath);
    lib = open_library(driverPath);
-   if (!lib) {
-      _eglLog(_EGL_WARNING, "Could not open %s",
-              driverPath);
-      return NULL;
+
+#if defined(_EGL_PLATFORM_WINDOWS)
+   /* XXX untested */
+   if (lib)
+      mainFunc = (_EGLMain_t) GetProcAddress(lib, "_eglMain");
+#elif defined(_EGL_PLATFORM_X)
+   if (lib) {
+      mainFunc = (_EGLMain_t) dlsym(lib, "_eglMain");
+      if (!mainFunc)
+         error = dlerror();
+   }
+   else {
+      error = dlerror();
    }
-   mainFunc = (_EGLMain_t) GetProcAddress(lib, "_eglMain");
+#else /* _EGL_PLATFORM_NO_OS */
+   /* must be the default driver name */
+   if (strcmp(driverPath, DefaultDriverName) == 0)
+      mainFunc = (_EGLMain_t) _eglMain;
+   else
+      error = "not builtin driver";
 #endif
-#elif defined(_EGL_PLATFORM_X)
-   _eglLog(_EGL_DEBUG, "dlopen(%s)", driverPath);
-   lib = open_library(driverPath);
+
    if (!lib) {
-      _eglLog(_EGL_WARNING, "Could not open %s (%s)",
-              driverPath, dlerror());
+      _eglLog(_EGL_WARNING, "Could not open driver %s (%s)",
+              driverPath, error);
       if (!getenv("EGL_DRIVER"))
          _eglLog(_EGL_WARNING,
                  "The driver can be overridden by setting EGL_DRIVER");
       return NULL;
    }
-   mainFunc = (_EGLMain_t) dlsym(lib, "_eglMain");
-#endif
 
    if (!mainFunc) {
-      _eglLog(_EGL_WARNING, "_eglMain not found in %s", driverPath);
+      _eglLog(_EGL_WARNING, "_eglMain not found in %s (%s)",
+              driverPath, error);
       if (lib)
          close_library(lib);
       return NULL;
@@ -384,7 +408,7 @@ _eglInitDriverFallbacks(_EGLDriver *drv)
    drv->API.CopyBuffers = _eglCopyBuffers;
 
    drv->API.QueryString = _eglQueryString;
-   drv->API.WaitGL = _eglWaitGL;
+   drv->API.WaitClient = _eglWaitClient;
    drv->API.WaitNative = _eglWaitNative;
 
 #ifdef EGL_MESA_screen_surface
@@ -428,6 +452,11 @@ _eglFindAPIs(void)
    const char *es2_libname = "libGLESv2.so";
    const char *gl_libname = "libGL.so";
    const char *vg_libname = "libOpenVG.so";
+#else /* _EGL_PLATFORM_NO_OS */
+   const char *es1_libname = NULL;
+   const char *es2_libname = NULL;
+   const char *gl_libname = NULL;
+   const char *vg_libname = NULL;
 #endif
 
    if ((lib = open_library(es1_libname))) {
diff --git a/src/egl/main/eglglobals.c b/src/egl/main/eglglobals.c
index 3ae4c1ad3a..443d0f072c 100644
--- a/src/egl/main/eglglobals.c
+++ b/src/egl/main/eglglobals.c
@@ -15,13 +15,13 @@ struct _egl_global _eglGlobal =
    &_eglGlobalMutex,       /* Mutex */
    NULL,                   /* DisplayList */
    1,                      /* FreeScreenHandle */
-   0x0,                    /* ClientAPIsMask */
    0,                      /* NumDrivers */
    { NULL },               /* Drivers */
    2,                      /* NumAtExitCalls */
-   {                       /* AtExitCalls */
-      _eglFiniDisplay,
-      _eglUnloadDrivers
+   {
+      /* default AtExitCalls, called in reverse order */
+      _eglUnloadDrivers, /* always called last */
+      _eglFiniDisplay
    },
 };
 
diff --git a/src/egl/main/eglglobals.h b/src/egl/main/eglglobals.h
index 58511076d4..5ebb914ca7 100644
--- a/src/egl/main/eglglobals.h
+++ b/src/egl/main/eglglobals.h
@@ -19,9 +19,6 @@ struct _egl_global
 
    EGLScreenMESA FreeScreenHandle;
 
-   /* bitmaks of supported APIs (supported by _some_ driver) */
-   EGLint ClientAPIsMask;
-
    EGLint NumDrivers;
    _EGLDriver *Drivers[10];
 
diff --git a/src/egl/main/egllog.c b/src/egl/main/egllog.c
index 1d7a0a388c..11a9bf7275 100644
--- a/src/egl/main/egllog.c
+++ b/src/egl/main/egllog.c
@@ -9,51 +9,140 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+
 #include "egllog.h"
+#include "eglmutex.h"
 
 #define MAXSTRING 1000
-#define FALLBACK_LOG_LEVEL      _EGL_WARNING
-#define FALLBACK_LOG_LEVEL_STR  "warning"
+#define FALLBACK_LOG_LEVEL _EGL_WARNING
 
-static EGLint ReportingLevel = -1;
 
+static struct {
+   _EGLMutex mutex;
 
-static void
-log_level_initialize(void)
+   EGLBoolean initialized;
+   EGLint level;
+   _EGLLogProc logger;
+   EGLint num_messages;
+} logging = {
+   _EGL_MUTEX_INITIALIZER,
+   EGL_FALSE,
+   FALLBACK_LOG_LEVEL,
+   NULL,
+   0
+};
+
+static const char *level_strings[] = {
+   /* the order is important */
+   "fatal",
+   "warning",
+   "info",
+   "debug",
+   NULL
+};
+
+
+/**
+ * Set the function to be called when there is a message to log.
+ * Note that the function will be called with an internal lock held.
+ * Recursive logging is not allowed.
+ */
+void
+_eglSetLogProc(_EGLLogProc logger)
 {
-#if defined(_EGL_PLATFORM_X)  
-   char *log_env = getenv("EGL_LOG_LEVEL");
-#else
-   char *log_env = NULL;
-#endif
-
-   if (log_env == NULL) {
-      ReportingLevel = FALLBACK_LOG_LEVEL;
-   }
-   else if (strcasecmp(log_env, "fatal") == 0) {
-      ReportingLevel = _EGL_FATAL;
-   }
-   else if (strcasecmp(log_env, "warning") == 0) {
-      ReportingLevel = _EGL_WARNING;
+   EGLint num_messages = 0;
+
+   _eglLockMutex(&logging.mutex);
+
+   if (logging.logger != logger) {
+      logging.logger = logger;
+
+      num_messages = logging.num_messages;
+      logging.num_messages = 0;
    }
-   else if (strcasecmp(log_env, "info") == 0) {
-      ReportingLevel = _EGL_INFO;
+
+   _eglUnlockMutex(&logging.mutex);
+
+   if (num_messages)
+      _eglLog(_EGL_DEBUG,
+              "New logger installed. "
+              "Messages before the new logger might not be available.");
+}
+
+
+/**
+ * Set the log reporting level.
+ */
+void
+_eglSetLogLevel(EGLint level)
+{
+   switch (level) {
+   case _EGL_FATAL:
+   case _EGL_WARNING:
+   case _EGL_INFO:
+   case _EGL_DEBUG:
+      _eglLockMutex(&logging.mutex);
+      logging.level = level;
+      _eglUnlockMutex(&logging.mutex);
+      break;
+   default:
+      break;
    }
-   else if (strcasecmp(log_env, "debug") == 0) {
-      ReportingLevel = _EGL_DEBUG;
+}
+
+
+/**
+ * The default logger.  It prints the message to stderr.
+ */
+static void
+_eglDefaultLogger(EGLint level, const char *msg)
+{
+   fprintf(stderr, "libEGL %s: %s\n", level_strings[level], msg);
+}
+
+
+/**
+ * Initialize the logging facility.
+ */
+static void
+_eglInitLogger(void)
+{
+   const char *log_env;
+   EGLint i, level = -1;
+
+   if (logging.initialized)
+      return;
+
+   log_env = getenv("EGL_LOG_LEVEL");
+   if (log_env) {
+      for (i = 0; level_strings[i]; i++) {
+         if (strcasecmp(log_env, level_strings[i]) == 0) {
+            level = i;
+            break;
+         }
+      }
    }
    else {
-      fprintf(stderr, "Unrecognized EGL_LOG_LEVEL environment variable value. "
+      level = FALLBACK_LOG_LEVEL;
+   }
+
+   logging.logger = _eglDefaultLogger;
+   logging.level = (level >= 0) ? level : FALLBACK_LOG_LEVEL;
+   logging.initialized = EGL_TRUE;
+
+   /* it is fine to call _eglLog now */
+   if (log_env && level < 0) {
+      _eglLog(_EGL_WARNING,
+              "Unrecognized EGL_LOG_LEVEL environment variable value. "
               "Expected one of \"fatal\", \"warning\", \"info\", \"debug\". "
-              "Got \"%s\". Falling back to \"%s\".\n",
-              log_env, FALLBACK_LOG_LEVEL_STR);
-      ReportingLevel = FALLBACK_LOG_LEVEL;
+              "Got \"%s\". Falling back to \"%s\".",
+              log_env, level_strings[FALLBACK_LOG_LEVEL]);
    }
 }
 
 
 /**
- * Log a message to stderr.
+ * Log a message with message logger.
  * \param level one of _EGL_FATAL, _EGL_WARNING, _EGL_INFO, _EGL_DEBUG.
  */
 void
@@ -61,40 +150,26 @@ _eglLog(EGLint level, const char *fmtStr, ...)
 {
    va_list args;
    char msg[MAXSTRING];
-   const char *levelStr;
-   static int log_level_initialized = 0;
 
-   if (!log_level_initialized) {
-      log_level_initialize();
-      log_level_initialized = 1;
-   }
+   /* one-time initialization; a little race here is fine */
+   if (!logging.initialized)
+      _eglInitLogger();
+   if (level > logging.level || level < 0)
+      return;
 
-   if (level <= ReportingLevel) {
-      switch (level) {
-      case _EGL_FATAL:
-         levelStr = "Fatal";
-         break;
-      case _EGL_WARNING:
-         levelStr = "Warning";
-         break;
-      case _EGL_INFO:
-         levelStr = "Info";
-         break;
-      case _EGL_DEBUG:
-         levelStr = "Debug";
-         break;
-      default:
-         levelStr = "";
-      }
+   _eglLockMutex(&logging.mutex);
 
+   if (logging.logger) {
       va_start(args, fmtStr);
       vsnprintf(msg, MAXSTRING, fmtStr, args);
       va_end(args);
 
-      fprintf(stderr, "libEGL %s: %s\n", levelStr, msg);
-
-      if (level == _EGL_FATAL) {
-         exit(1); /* or abort()? */
-      }
+      logging.logger(level, msg);
+      logging.num_messages++;
    }
+
+   _eglUnlockMutex(&logging.mutex);
+
+   if (level == _EGL_FATAL)
+      exit(1); /* or abort()? */
 }
diff --git a/src/egl/main/egllog.h b/src/egl/main/egllog.h
index 2fa352f155..83c8bb72a6 100644
--- a/src/egl/main/egllog.h
+++ b/src/egl/main/egllog.h
@@ -9,6 +9,17 @@
 #define _EGL_DEBUG   3   /* useful info for debugging */
 
 
+typedef void (*_EGLLogProc)(EGLint level, const char *msg);
+
+
+extern void
+_eglSetLogProc(_EGLLogProc logger);
+
+
+extern void
+_eglSetLogLevel(EGLint level);
+
+
 extern void
 _eglLog(EGLint level, const char *fmtStr, ...);
 
diff --git a/src/egl/main/eglmisc.c b/src/egl/main/eglmisc.c
index b37213faf1..e66913320b 100644
--- a/src/egl/main/eglmisc.c
+++ b/src/egl/main/eglmisc.c
@@ -108,11 +108,12 @@ _eglQueryString(_EGLDriver *drv, _EGLDisplay *dpy, EGLint name)
 
 
 EGLBoolean
-_eglWaitGL(_EGLDriver *drv, _EGLDisplay *dpy)
+_eglWaitClient(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx)
 {
    /* just a placeholder */
    (void) drv;
    (void) dpy;
+   (void) ctx;
    return EGL_TRUE;
 }
 
diff --git a/src/egl/main/eglmisc.h b/src/egl/main/eglmisc.h
index a15c839be2..829d4cde79 100644
--- a/src/egl/main/eglmisc.h
+++ b/src/egl/main/eglmisc.h
@@ -37,7 +37,7 @@ _eglQueryString(_EGLDriver *drv, _EGLDisplay *dpy, EGLint name);
 
 
 extern EGLBoolean
-_eglWaitGL(_EGLDriver *drv, _EGLDisplay *dpy);
+_eglWaitClient(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx);
 
 
 extern EGLBoolean
diff --git a/src/egl/main/eglsurface.c b/src/egl/main/eglsurface.c
index e7a1a8329e..940a1b760c 100644
--- a/src/egl/main/eglsurface.c
+++ b/src/egl/main/eglsurface.c
@@ -15,6 +15,22 @@
 #include "eglsurface.h"
 
 
+static void
+_eglClampSwapInterval(_EGLSurface *surf, EGLint interval)
+{
+   EGLint bound = GET_CONFIG_ATTRIB(surf->Config, EGL_MAX_SWAP_INTERVAL);
+   if (interval >= bound) {
+      interval = bound;
+   }
+   else {
+      bound = GET_CONFIG_ATTRIB(surf->Config, EGL_MIN_SWAP_INTERVAL);
+      if (interval < bound)
+         interval = bound;
+   }
+   surf->SwapInterval = interval;
+}
+
+
 /**
  * Do error check on parameters and initialize the given _EGLSurface object.
  * \return EGL_TRUE if no errors, EGL_FALSE otherwise.
@@ -194,7 +210,9 @@ _eglInitSurface(_EGLDriver *drv, _EGLSurface *surf, EGLint type,
    surf->TextureTarget = texTarget;
    surf->MipmapTexture = mipmapTex;
    surf->MipmapLevel = 0;
-   surf->SwapInterval = 0;
+   /* the default swap interval is 1 */
+   _eglClampSwapInterval(surf, 1);
+
 #ifdef EGL_VERSION_1_2
    surf->SwapBehavior = EGL_BUFFER_DESTROYED; /* XXX ok? */
    surf->HorizontalResolution = EGL_UNKNOWN; /* set by caller */
@@ -466,11 +484,10 @@ _eglReleaseTexImage(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surface,
 
 
 EGLBoolean
-_eglSwapInterval(_EGLDriver *drv, _EGLDisplay *dpy, EGLint interval)
+_eglSwapInterval(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf,
+                 EGLint interval)
 {
-   _EGLSurface *surf = _eglGetCurrentSurface(EGL_DRAW);
-   if (surf)
-      surf->SwapInterval = interval;
+   _eglClampSwapInterval(surf, interval);
    return EGL_TRUE;
 }
 
diff --git a/src/egl/main/eglsurface.h b/src/egl/main/eglsurface.h
index f6d44b5922..b75fa9c368 100644
--- a/src/egl/main/eglsurface.h
+++ b/src/egl/main/eglsurface.h
@@ -86,7 +86,7 @@ _eglReleaseTexImage(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf, EGLint
 
 
 extern EGLBoolean
-_eglSwapInterval(_EGLDriver *drv, _EGLDisplay *dpy, EGLint interval);
+_eglSwapInterval(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf, EGLint interval);
 
 
 #ifdef EGL_VERSION_1_2
diff --git a/src/gallium/SConscript b/src/gallium/SConscript
index 89c69d7205..8be84cddbe 100644
--- a/src/gallium/SConscript
+++ b/src/gallium/SConscript
@@ -23,6 +23,7 @@ SConscript([
 	'auxiliary/pipebuffer/SConscript',
 	'auxiliary/indices/SConscript',
 	'auxiliary/rbug/SConscript',
+	'auxiliary/vl/SConscript',
 ])
 
 for driver in env['drivers']:
diff --git a/src/gallium/auxiliary/draw/draw_pt_post_vs.c b/src/gallium/auxiliary/draw/draw_pt_post_vs.c
index 78953bccfc..6c1cb48e8b 100644
--- a/src/gallium/auxiliary/draw/draw_pt_post_vs.c
+++ b/src/gallium/auxiliary/draw/draw_pt_post_vs.c
@@ -104,7 +104,7 @@ static boolean post_vs_cliptest_viewport_gl( struct pt_post_vs *pvs,
    unsigned clipped = 0;
    unsigned j;
 
-   if (0) debug_printf("%s\n");
+   if (0) debug_printf("%s\n", __FUNCTION__);
 
    for (j = 0; j < count; j++) {
       float *position = out->data[pos];
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
index 0d30363484..d31476228c 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
@@ -542,7 +542,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list)
       debug_printf("%10p %7u %7u\n",
                    fenced_buf,
                    fenced_buf->base.base.size,
-                   fenced_buf->base.base.reference.count);
+                   p_atomic_read(&fenced_buf->base.base.reference.count));
       curr = next; 
       next = curr->next;
    }
@@ -556,7 +556,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list)
       debug_printf("%10p %7u %7u %10p %s\n",
                    fenced_buf,
                    fenced_buf->base.base.size,
-                   fenced_buf->base.base.reference.count,
+                   p_atomic_read(&fenced_buf->base.base.reference.count),
                    fenced_buf->fence,
                    signaled == 0 ? "y" : "n");
       curr = next; 
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c
index 1b4df28c70..6e3214ca9c 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c
@@ -350,7 +350,7 @@ pb_debug_manager_dump(struct pb_debug_manager *mgr)
       buf = LIST_ENTRY(struct pb_debug_buffer, curr, head);
 
       debug_printf("buffer = %p\n", buf);
-      debug_printf("    .size = %p\n", buf->base.base.size);
+      debug_printf("    .size = 0x%x\n", buf->base.base.size);
       debug_backtrace_dump(buf->create_backtrace, PB_DEBUG_CREATE_BACKTRACE);
       
       curr = next; 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
index 8a13885da9..53e13b30e6 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
@@ -358,7 +358,7 @@ epilog(
 
 boolean
 tgsi_sanity_check(
-   struct tgsi_token *tokens )
+   const struct tgsi_token *tokens )
 {
    struct sanity_check_ctx ctx;
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.h b/src/gallium/auxiliary/tgsi/tgsi_sanity.h
index ca45e94c7a..52263ff883 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.h
@@ -40,7 +40,7 @@ extern "C" {
  */
 boolean
 tgsi_sanity_check(
-   struct tgsi_token *tokens );
+   const struct tgsi_token *tokens );
 
 #if defined __cplusplus
 }
diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.c b/src/gallium/auxiliary/tgsi/tgsi_scan.c
index c535788819..0db4481a3d 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_scan.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_scan.c
@@ -132,6 +132,7 @@ tgsi_scan_shader(const struct tgsi_token *tokens,
                if (file == TGSI_FILE_INPUT) {
                   info->input_semantic_name[reg] = (ubyte)fulldecl->Semantic.SemanticName;
                   info->input_semantic_index[reg] = (ubyte)fulldecl->Semantic.SemanticIndex;
+                  info->input_interpolate[reg] = (ubyte)fulldecl->Declaration.Interpolate;
                   info->num_inputs++;
                }
                else if (file == TGSI_FILE_OUTPUT) {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.h b/src/gallium/auxiliary/tgsi/tgsi_scan.h
index 2c1a75bc81..8a7ee0c7e4 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_scan.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_scan.h
@@ -45,6 +45,7 @@ struct tgsi_shader_info
    ubyte num_outputs;
    ubyte input_semantic_name[PIPE_MAX_SHADER_INPUTS]; /**< TGSI_SEMANTIC_x */
    ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS];
+   ubyte input_interpolate[PIPE_MAX_SHADER_INPUTS];
    ubyte output_semantic_name[PIPE_MAX_SHADER_OUTPUTS]; /**< TGSI_SEMANTIC_x */
    ubyte output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 1e719940ec..5f6b83b236 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -39,8 +39,9 @@
 #include "tgsi/tgsi_info.h"
 #include "tgsi/tgsi_parse.h"
 #include "tgsi/tgsi_util.h"
-#include "tgsi_exec.h"
-#include "tgsi_sse2.h"
+#include "tgsi/tgsi_dump.h"
+#include "tgsi/tgsi_exec.h"
+#include "tgsi/tgsi_sse2.h"
 
 #include "rtasm/rtasm_x86sse.h"
 
@@ -1360,6 +1361,32 @@ emit_store(
    const struct tgsi_full_instruction *inst,
    unsigned chan_index )
 {
+   switch( inst->Instruction.Saturate ) {
+   case TGSI_SAT_NONE:
+      break;
+
+   case TGSI_SAT_ZERO_ONE:
+      sse_maxps(
+         func,
+         make_xmm( xmm ),
+         get_temp(
+            TGSI_EXEC_TEMP_00000000_I,
+            TGSI_EXEC_TEMP_00000000_C ) );
+
+      sse_minps(
+         func,
+         make_xmm( xmm ),
+         get_temp(
+            TGSI_EXEC_TEMP_ONE_I,
+            TGSI_EXEC_TEMP_ONE_C ) );
+      break;
+
+   case TGSI_SAT_MINUS_PLUS_ONE:
+      assert( 0 );
+      break;
+   }
+
+
    switch( reg->DstRegister.File ) {
    case TGSI_FILE_OUTPUT:
       emit_output(
@@ -1388,19 +1415,6 @@ emit_store(
    default:
       assert( 0 );
    }
-
-   switch( inst->Instruction.Saturate ) {
-   case TGSI_SAT_NONE:
-      break;
-
-   case TGSI_SAT_ZERO_ONE:
-      /* assert( 0 ); */
-      break;
-
-   case TGSI_SAT_MINUS_PLUS_ONE:
-      assert( 0 );
-      break;
-   }
 }
 
 #define STORE( FUNC, INST, XMM, INDEX, CHAN )\
@@ -1747,14 +1761,6 @@ emit_instruction(
    if (indirect_temp_reference(inst))
       return FALSE;
 
-   /* we don't handle saturation/clamping yet */
-   if (inst->Instruction.Saturate != TGSI_SAT_NONE)
-      return FALSE;
-
-   /* need to use extra temps to fix SOA dependencies : */
-   if (tgsi_check_soa_dependencies(inst))
-      return FALSE;
-
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_ARL:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -1768,8 +1774,10 @@ emit_instruction(
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_SWZ:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
-         FETCH( func, *inst, 0, 0, chan_index );
-         STORE( func, *inst, 0, 0, chan_index );
+         FETCH( func, *inst, 4 + chan_index, 0, chan_index );
+      }
+      FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
+         STORE( func, *inst, 4 + chan_index, 0, chan_index );
       }
       break;
 
@@ -1847,7 +1855,6 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_RCP:
-   /* TGSI_OPCODE_RECIP */
       FETCH( func, *inst, 0, 0, CHAN_X );
       emit_rcp( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -1856,7 +1863,6 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_RSQ:
-   /* TGSI_OPCODE_RECIPSQRT */
       FETCH( func, *inst, 0, 0, CHAN_X );
       emit_abs( func, 0 );
       emit_rsqrt( func, 1, 0 );
@@ -1954,7 +1960,6 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_DP3:
-   /* TGSI_OPCODE_DOT3 */
       FETCH( func, *inst, 0, 0, CHAN_X );
       FETCH( func, *inst, 1, 1, CHAN_X );
       emit_mul( func, 0, 1 );
@@ -1972,7 +1977,6 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_DP4:
-   /* TGSI_OPCODE_DOT4 */
       FETCH( func, *inst, 0, 0, CHAN_X );
       FETCH( func, *inst, 1, 1, CHAN_X );
       emit_mul( func, 0, 1 );
@@ -2043,17 +2047,14 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_SLT:
-   /* TGSI_OPCODE_SETLT */
       emit_setcc( func, inst, cc_LessThan );
       break;
 
    case TGSI_OPCODE_SGE:
-   /* TGSI_OPCODE_SETGE */
       emit_setcc( func, inst, cc_NotLessThan );
       break;
 
    case TGSI_OPCODE_MAD:
-   /* TGSI_OPCODE_MADD */
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
          FETCH( func, *inst, 1, 1, chan_index );
@@ -2283,7 +2284,7 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_SEQ:
-      return 0;
+      emit_setcc( func, inst, cc_Equal );
       break;
 
    case TGSI_OPCODE_SFL:
@@ -2291,7 +2292,7 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_SGT:
-      return 0;
+      emit_setcc( func, inst, cc_NotLessThanEqual );
       break;
 
    case TGSI_OPCODE_SIN:
@@ -2303,11 +2304,11 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_SLE:
-      return 0;
+      emit_setcc( func, inst, cc_LessThanEqual );
       break;
 
    case TGSI_OPCODE_SNE:
-      return 0;
+      emit_setcc( func, inst, cc_NotEqual );
       break;
 
    case TGSI_OPCODE_STR:
@@ -2371,7 +2372,6 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_SSG:
-   /* TGSI_OPCODE_SGN */
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
          emit_sgn( func, 0, 0 );
@@ -2929,6 +2929,22 @@ tgsi_emit_sse2(
                          parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX ?
                          "vertex shader" : "fragment shader");
 	 }
+
+         if (tgsi_check_soa_dependencies(&parse.FullToken.FullInstruction)) {
+            uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
+
+            /* XXX: we only handle src/dst aliasing in a few opcodes
+             * currently.  Need to use an additional temporay to hold
+             * the result in the cases where the code is too opaque to
+             * fix.
+             */
+            if (opcode != TGSI_OPCODE_MOV &&
+                opcode != TGSI_OPCODE_SWZ) {
+               debug_printf("Warning: src/dst aliasing in instruction"
+                            " is not handled:\n");
+               tgsi_dump_instruction(&parse.FullToken.FullInstruction, 1);
+            }
+         }
          break;
 
       case TGSI_TOKEN_TYPE_IMMEDIATE:
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
index f7096bd8e2..654426a903 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
@@ -31,6 +31,7 @@
 #include "tgsi/tgsi_ureg.h"
 #include "tgsi/tgsi_info.h"
 #include "tgsi/tgsi_dump.h"
+#include "tgsi/tgsi_sanity.h"
 #include "util/u_memory.h"
 #include "util/u_math.h"
 
@@ -70,6 +71,7 @@ struct ureg_tokens {
 
 #define UREG_MAX_INPUT PIPE_MAX_ATTRIBS
 #define UREG_MAX_OUTPUT PIPE_MAX_ATTRIBS
+#define UREG_MAX_CONSTANT_RANGE 32
 #define UREG_MAX_IMMEDIATE 32
 #define UREG_MAX_TEMP 256
 #define UREG_MAX_ADDR 2
@@ -86,8 +88,10 @@ struct ureg_program
       unsigned semantic_name;
       unsigned semantic_index;
       unsigned interp;
-   } input[UREG_MAX_INPUT];
-   unsigned nr_inputs;
+   } fs_input[UREG_MAX_INPUT];
+   unsigned nr_fs_inputs;
+
+   unsigned vs_inputs[UREG_MAX_INPUT/32];
 
    struct {
       unsigned semantic_name;
@@ -107,9 +111,13 @@ struct ureg_program
    unsigned temps_active[UREG_MAX_TEMP / 32];
    unsigned nr_temps;
 
-   unsigned nr_addrs;
+   struct {
+      unsigned first;
+      unsigned last;
+   } constant_range[UREG_MAX_CONSTANT_RANGE];
+   unsigned nr_constant_ranges;
 
-   unsigned nr_constants;
+   unsigned nr_addrs;
    unsigned nr_instructions;
 
    struct ureg_tokens domain[2];
@@ -119,6 +127,9 @@ static union tgsi_any_token error_tokens[32];
 
 static void tokens_error( struct ureg_tokens *tokens )
 {
+   if (tokens->tokens && tokens->tokens != error_tokens)
+      FREE(tokens->tokens);
+
    tokens->tokens = error_tokens;
    tokens->size = Elements(error_tokens);
    tokens->count = 0;
@@ -228,25 +239,25 @@ ureg_src_register( unsigned file,
 
 
 
-static struct ureg_src 
-ureg_DECL_input( struct ureg_program *ureg,
-                 unsigned name,
-                 unsigned index,
-                 unsigned interp_mode )
+struct ureg_src 
+ureg_DECL_fs_input( struct ureg_program *ureg,
+                    unsigned name,
+                    unsigned index,
+                    unsigned interp_mode )
 {
    unsigned i;
 
-   for (i = 0; i < ureg->nr_inputs; i++) {
-      if (ureg->input[i].semantic_name == name &&
-          ureg->input[i].semantic_index == index) 
+   for (i = 0; i < ureg->nr_fs_inputs; i++) {
+      if (ureg->fs_input[i].semantic_name == name &&
+          ureg->fs_input[i].semantic_index == index) 
          goto out;
    }
 
-   if (ureg->nr_inputs < UREG_MAX_INPUT) {
-      ureg->input[i].semantic_name = name;
-      ureg->input[i].semantic_index = index;
-      ureg->input[i].interp = interp_mode;
-      ureg->nr_inputs++;
+   if (ureg->nr_fs_inputs < UREG_MAX_INPUT) {
+      ureg->fs_input[i].semantic_name = name;
+      ureg->fs_input[i].semantic_index = index;
+      ureg->fs_input[i].interp = interp_mode;
+      ureg->nr_fs_inputs++;
    }
    else {
       set_bad( ureg );
@@ -257,25 +268,14 @@ out:
 }
 
 
-
-struct ureg_src 
-ureg_DECL_fs_input( struct ureg_program *ureg,
-                    unsigned name,
-                    unsigned index,
-                    unsigned interp )
-{
-   assert(ureg->processor == TGSI_PROCESSOR_FRAGMENT);
-   return ureg_DECL_input( ureg, name, index, interp );
-}
-
-
 struct ureg_src 
 ureg_DECL_vs_input( struct ureg_program *ureg,
-                    unsigned name,
                     unsigned index )
 {
    assert(ureg->processor == TGSI_PROCESSOR_VERTEX);
-   return ureg_DECL_input( ureg, name, index, TGSI_INTERPOLATE_CONSTANT );
+   
+   ureg->vs_inputs[index/32] |= 1 << (index % 32);
+   return ureg_src_register( TGSI_FILE_INPUT, index );
 }
 
 
@@ -313,9 +313,57 @@ out:
  * value or manage any constant_buffer contents -- that's the
  * resposibility of the calling code.
  */
-struct ureg_src ureg_DECL_constant(struct ureg_program *ureg )
+struct ureg_src ureg_DECL_constant(struct ureg_program *ureg, 
+                                   unsigned index )
 {
-   return ureg_src_register( TGSI_FILE_CONSTANT, ureg->nr_constants++ );
+   unsigned minconst = index, maxconst = index;
+   unsigned i;
+
+   /* Inside existing range?
+    */
+   for (i = 0; i < ureg->nr_constant_ranges; i++) {
+      if (ureg->constant_range[i].first <= index &&
+          ureg->constant_range[i].last >= index)
+         goto out;
+   }
+
+   /* Extend existing range?
+    */
+   for (i = 0; i < ureg->nr_constant_ranges; i++) {
+      if (ureg->constant_range[i].last == index - 1) {
+         ureg->constant_range[i].last = index;
+         goto out;
+      }
+
+      if (ureg->constant_range[i].first == index + 1) {
+         ureg->constant_range[i].first = index;
+         goto out;
+      }
+
+      minconst = MIN2(minconst, ureg->constant_range[i].first);
+      maxconst = MAX2(maxconst, ureg->constant_range[i].last);
+   }
+
+   /* Create new range?
+    */
+   if (ureg->nr_constant_ranges < UREG_MAX_CONSTANT_RANGE) {
+      i = ureg->nr_constant_ranges++;
+      ureg->constant_range[i].first = index;
+      ureg->constant_range[i].last = index;
+   }
+
+   /* Collapse all ranges down to one:
+    */
+   i = 0;
+   ureg->constant_range[0].first = minconst;
+   ureg->constant_range[0].last = maxconst;
+   ureg->nr_constant_ranges = 1;
+
+out:
+   assert(i < ureg->nr_constant_ranges);
+   assert(ureg->constant_range[i].first <= index);
+   assert(ureg->constant_range[i].last >= index);
+   return ureg_src_register( TGSI_FILE_CONSTANT, index );
 }
 
 
@@ -566,6 +614,19 @@ ureg_emit_dst( struct ureg_program *ureg,
 }
 
 
+static void validate( unsigned opcode,
+                      unsigned nr_dst,
+                      unsigned nr_src )
+{
+#ifdef DEBUG
+   const struct tgsi_opcode_info *info = tgsi_get_opcode_info( opcode );
+   assert(info);
+   if(info) {
+      assert(nr_dst == info->num_dst);
+      assert(nr_src == info->num_src);
+   }
+#endif
+}
 
 unsigned
 ureg_emit_insn(struct ureg_program *ureg,
@@ -576,6 +637,8 @@ ureg_emit_insn(struct ureg_program *ureg,
 {
    union tgsi_any_token *out;
 
+   validate( opcode, num_dst, num_src );
+   
    out = get_tokens( ureg, DOMAIN_INSN, 1 );
    out[0].value = 0;
    out[0].insn.Type = TGSI_TOKEN_TYPE_INSTRUCTION;
@@ -678,17 +741,6 @@ ureg_insn(struct ureg_program *ureg,
    unsigned insn, i;
    boolean saturate;
 
-#ifdef DEBUG
-   {
-      const struct tgsi_opcode_info *info = tgsi_get_opcode_info( opcode );
-      assert(info);
-      if(info) {
-         assert(nr_dst == info->num_dst);
-         assert(nr_src == info->num_src);
-      }
-   }
-#endif
-   
    saturate = nr_dst ? dst[0].Saturate : FALSE;
 
    insn = ureg_emit_insn( ureg, opcode, saturate, nr_dst, nr_src );
@@ -702,6 +754,53 @@ ureg_insn(struct ureg_program *ureg,
    ureg_fixup_insn_size( ureg, insn );
 }
 
+void
+ureg_tex_insn(struct ureg_program *ureg,
+              unsigned opcode,
+              const struct ureg_dst *dst,
+              unsigned nr_dst,
+              unsigned target,
+              const struct ureg_src *src,
+              unsigned nr_src )
+{
+   unsigned insn, i;
+   boolean saturate;
+
+   saturate = nr_dst ? dst[0].Saturate : FALSE;
+
+   insn = ureg_emit_insn( ureg, opcode, saturate, nr_dst, nr_src );
+
+   ureg_emit_texture( ureg, insn, target );                             \
+
+   for (i = 0; i < nr_dst; i++)
+      ureg_emit_dst( ureg, dst[i] );
+
+   for (i = 0; i < nr_src; i++)
+      ureg_emit_src( ureg, src[i] );
+
+   ureg_fixup_insn_size( ureg, insn );
+}
+
+
+void
+ureg_label_insn(struct ureg_program *ureg,
+                unsigned opcode,
+                const struct ureg_src *src,
+                unsigned nr_src,
+                unsigned *label_token )
+{
+   unsigned insn, i;
+
+   insn = ureg_emit_insn( ureg, opcode, FALSE, 0, nr_src );
+
+   ureg_emit_label( ureg, insn, label_token );                  \
+
+   for (i = 0; i < nr_src; i++)
+      ureg_emit_src( ureg, src[i] );
+
+   ureg_fixup_insn_size( ureg, insn );
+}
+
 
 
 static void emit_decl( struct ureg_program *ureg,
@@ -777,13 +876,22 @@ static void emit_decls( struct ureg_program *ureg )
 {
    unsigned i;
 
-   for (i = 0; i < ureg->nr_inputs; i++) {
-      emit_decl( ureg, 
-                 TGSI_FILE_INPUT, 
-                 i,
-                 ureg->input[i].semantic_name,
-                 ureg->input[i].semantic_index,
-                 ureg->input[i].interp );
+   if (ureg->processor == TGSI_PROCESSOR_VERTEX) {
+      for (i = 0; i < UREG_MAX_INPUT; i++) {
+         if (ureg->vs_inputs[i/32] & (1 << (i%32))) {
+            emit_decl_range( ureg, TGSI_FILE_INPUT, i, 1 );
+         }
+      }
+   }
+   else {
+      for (i = 0; i < ureg->nr_fs_inputs; i++) {
+         emit_decl( ureg, 
+                    TGSI_FILE_INPUT, 
+                    i,
+                    ureg->fs_input[i].semantic_name,
+                    ureg->fs_input[i].semantic_index,
+                    ureg->fs_input[i].interp );
+      }
    }
 
    for (i = 0; i < ureg->nr_outputs; i++) {
@@ -801,10 +909,13 @@ static void emit_decls( struct ureg_program *ureg )
                        ureg->sampler[i].Index, 1 );
    }
 
-   if (ureg->nr_constants) {
-      emit_decl_range( ureg,
-                       TGSI_FILE_CONSTANT,
-                       0, ureg->nr_constants );
+   if (ureg->nr_constant_ranges) {
+      for (i = 0; i < ureg->nr_constant_ranges; i++)
+         emit_decl_range( ureg,
+                          TGSI_FILE_CONSTANT,
+                          ureg->constant_range[i].first, 
+                          (ureg->constant_range[i].last + 1 -
+                           ureg->constant_range[i].first) );
    }
 
    if (ureg->nr_temps) {
@@ -890,6 +1001,15 @@ const struct tgsi_token *ureg_finalize( struct ureg_program *ureg )
                    ureg->domain[DOMAIN_DECL].count);
       tgsi_dump( tokens, 0 );
    }
+
+#if DEBUG
+   if (tokens && !tgsi_sanity_check(tokens)) {
+      debug_printf("tgsi_ureg.c, sanity check failed on generated tokens:\n");
+      tgsi_dump(tokens, 0);
+      assert(0);
+   }
+#endif
+
    
    return tokens;
 }
@@ -911,6 +1031,25 @@ void *ureg_create_shader( struct ureg_program *ureg,
 }
 
 
+const struct tgsi_token *ureg_get_tokens( struct ureg_program *ureg,
+                                          unsigned *nr_tokens )
+{
+   const struct tgsi_token *tokens;
+
+   ureg_finalize(ureg);
+
+   tokens = &ureg->domain[DOMAIN_DECL].tokens[0].token;
+
+   if (nr_tokens) 
+      *nr_tokens = ureg->domain[DOMAIN_DECL].size;
+
+   ureg->domain[DOMAIN_DECL].tokens = 0;
+   ureg->domain[DOMAIN_DECL].size = 0;
+   ureg->domain[DOMAIN_DECL].order = 0;
+   ureg->domain[DOMAIN_DECL].count = 0;
+
+   return tokens;
+}
 
 
 struct ureg_program *ureg_create( unsigned processor )
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.h b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
index acbca59040..f04f443b9e 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
@@ -82,10 +82,21 @@ ureg_create( unsigned processor );
 const struct tgsi_token *
 ureg_finalize( struct ureg_program * );
 
+/* Create and return a shader:
+ */
 void *
 ureg_create_shader( struct ureg_program *,
                     struct pipe_context *pipe );
 
+
+/* Alternately, return the built token stream and hand ownership of
+ * that memory to the caller:
+ */
+const struct tgsi_token *
+ureg_get_tokens( struct ureg_program *ureg,
+                 unsigned *nr_tokens );
+
+
 void 
 ureg_destroy( struct ureg_program * );
 
@@ -116,8 +127,7 @@ ureg_DECL_fs_input( struct ureg_program *,
 
 struct ureg_src
 ureg_DECL_vs_input( struct ureg_program *,
-                    unsigned semantic_name,
-                    unsigned semantic_index );
+                    unsigned index );
 
 struct ureg_dst
 ureg_DECL_output( struct ureg_program *,
@@ -130,7 +140,8 @@ ureg_DECL_immediate( struct ureg_program *,
                      unsigned nr );
 
 struct ureg_src
-ureg_DECL_constant( struct ureg_program * );
+ureg_DECL_constant( struct ureg_program *,
+                    unsigned index );
 
 struct ureg_dst
 ureg_DECL_temporary( struct ureg_program * );
@@ -233,6 +244,24 @@ ureg_insn(struct ureg_program *ureg,
           unsigned nr_src );
 
 
+void
+ureg_tex_insn(struct ureg_program *ureg,
+              unsigned opcode,
+              const struct ureg_dst *dst,
+              unsigned nr_dst,
+              unsigned target,
+              const struct ureg_src *src,
+              unsigned nr_src );
+
+
+void
+ureg_label_insn(struct ureg_program *ureg,
+                unsigned opcode,
+                const struct ureg_src *src,
+                unsigned nr_src,
+                unsigned *label);
+
+
 /***********************************************************************
  * Internal instruction helpers, don't call these directly:
  */
diff --git a/src/gallium/auxiliary/util/Makefile b/src/gallium/auxiliary/util/Makefile
index ae8d330a78..1d8bb55bbd 100644
--- a/src/gallium/auxiliary/util/Makefile
+++ b/src/gallium/auxiliary/util/Makefile
@@ -10,6 +10,7 @@ C_SOURCES = \
 	u_debug_stack.c \
 	u_blit.c \
 	u_cache.c \
+	u_cpu_detect.c \
 	u_draw_quad.c \
 	u_format.c \
 	u_format_access.c \
diff --git a/src/gallium/auxiliary/util/SConscript b/src/gallium/auxiliary/util/SConscript
index 28a5ab4256..2187935fa4 100644
--- a/src/gallium/auxiliary/util/SConscript
+++ b/src/gallium/auxiliary/util/SConscript
@@ -24,6 +24,7 @@ util = env.ConvenienceLibrary(
 		'u_bitmask.c',
 		'u_blit.c',
 		'u_cache.c',
+		'u_cpu_detect.c',
 		'u_debug.c',
 		'u_debug_dump.c',
 		'u_debug_memory.c',
diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c
index c516317d70..5038642599 100644
--- a/src/gallium/auxiliary/util/u_blit.c
+++ b/src/gallium/auxiliary/util/u_blit.c
@@ -46,6 +46,7 @@
 #include "util/u_memory.h"
 #include "util/u_simple_shaders.h"
 #include "util/u_surface.h"
+#include "util/u_rect.h"
 
 #include "cso_cache/cso_context.h"
 
@@ -182,47 +183,7 @@ get_next_slot( struct blit_state *ctx )
 }
                                
 
-/**
- * Setup vertex data for the textured quad we'll draw.
- * Note: y=0=top
- */
-static unsigned
-setup_vertex_data(struct blit_state *ctx,
-                  float x0, float y0, float x1, float y1, float z)
-{
-   unsigned offset;
-
-   ctx->vertices[0][0][0] = x0;
-   ctx->vertices[0][0][1] = y0;
-   ctx->vertices[0][0][2] = z;
-   ctx->vertices[0][1][0] = 0.0f; /*s*/
-   ctx->vertices[0][1][1] = 0.0f; /*t*/
-
-   ctx->vertices[1][0][0] = x1;
-   ctx->vertices[1][0][1] = y0;
-   ctx->vertices[1][0][2] = z;
-   ctx->vertices[1][1][0] = 1.0f; /*s*/
-   ctx->vertices[1][1][1] = 0.0f; /*t*/
-
-   ctx->vertices[2][0][0] = x1;
-   ctx->vertices[2][0][1] = y1;
-   ctx->vertices[2][0][2] = z;
-   ctx->vertices[2][1][0] = 1.0f;
-   ctx->vertices[2][1][1] = 1.0f;
 
-   ctx->vertices[3][0][0] = x0;
-   ctx->vertices[3][0][1] = y1;
-   ctx->vertices[3][0][2] = z;
-   ctx->vertices[3][1][0] = 0.0f;
-   ctx->vertices[3][1][1] = 1.0f;
-
-   offset = get_next_slot( ctx );
-
-   pipe_buffer_write(ctx->pipe->screen, ctx->vbuf,
-                     offset, sizeof(ctx->vertices), ctx->vertices);
-
-   return offset;
-}
 
 
 /**
@@ -315,15 +276,13 @@ util_blit_pixels_writemask(struct blit_state *ctx,
 {
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_screen *screen = pipe->screen;
-   struct pipe_texture texTemp, *tex;
-   struct pipe_surface *texSurf;
+   struct pipe_texture *tex = NULL;
    struct pipe_framebuffer_state fb;
    const int srcW = abs(srcX1 - srcX0);
    const int srcH = abs(srcY1 - srcY0);
-   const int srcLeft = MIN2(srcX0, srcX1);
-   const int srcTop = MIN2(srcY0, srcY1);
    unsigned offset;
    boolean overlap;
+   float s0, t0, s1, t1;
 
    assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
           filter == PIPE_TEX_MIPFILTER_LINEAR);
@@ -343,7 +302,8 @@ util_blit_pixels_writemask(struct blit_state *ctx,
     * no overlapping.
     * Filter mode should not matter since there's no stretching.
     */
-   if (dst->format == src->format &&
+   if (pipe->surface_copy &&
+       dst->format == src->format &&
        srcX0 < srcX1 &&
        dstX0 < dstX1 &&
        srcY0 < srcY1 &&
@@ -358,54 +318,83 @@ util_blit_pixels_writemask(struct blit_state *ctx,
       return;
    }
    
-   if (srcLeft != srcX0) {
-      /* left-right flip */
-      int tmp = dstX0;
-      dstX0 = dstX1;
-      dstX1 = tmp;
-   }
-
-   if (srcTop != srcY0) {
-      /* up-down flip */
-      int tmp = dstY0;
-      dstY0 = dstY1;
-      dstY1 = tmp;
-   }
-
    assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
                                       PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
 
-   /*
-    * XXX for now we're always creating a temporary texture.
-    * Strictly speaking that's not always needed.
+   /* Create a temporary texture when src and dest alias or when src
+    * is anything other than a single-level 2d texture.
+    * 
+    * This can still be improved upon.
     */
+   if (util_same_surface(src, dst) ||
+       src->texture->target != PIPE_TEXTURE_2D ||
+       src->texture->last_level != 0)
+   {
+      struct pipe_texture texTemp;
+      struct pipe_surface *texSurf;
+      const int srcLeft = MIN2(srcX0, srcX1);
+      const int srcTop = MIN2(srcY0, srcY1);
+
+      if (srcLeft != srcX0) {
+         /* left-right flip */
+         int tmp = dstX0;
+         dstX0 = dstX1;
+         dstX1 = tmp;
+      }
+
+      if (srcTop != srcY0) {
+         /* up-down flip */
+         int tmp = dstY0;
+         dstY0 = dstY1;
+         dstY1 = tmp;
+      }
+
+      /* create temp texture */
+      memset(&texTemp, 0, sizeof(texTemp));
+      texTemp.target = PIPE_TEXTURE_2D;
+      texTemp.format = src->format;
+      texTemp.last_level = 0;
+      texTemp.width[0] = srcW;
+      texTemp.height[0] = srcH;
+      texTemp.depth[0] = 1;
+      pf_get_block(src->format, &texTemp.block);
+
+      tex = screen->texture_create(screen, &texTemp);
+      if (!tex)
+         return;
+
+      texSurf = screen->get_tex_surface(screen, tex, 0, 0, 0, 
+                                        PIPE_BUFFER_USAGE_GPU_WRITE);
+
+      /* load temp texture */
+      if (pipe->surface_copy) {
+         pipe->surface_copy(pipe,
+                            texSurf, 0, 0,   /* dest */
+                            src, srcLeft, srcTop, /* src */
+                            srcW, srcH);     /* size */
+      } else {
+         util_surface_copy(pipe, FALSE,
+                           texSurf, 0, 0,   /* dest */
+                           src, srcLeft, srcTop, /* src */
+                           srcW, srcH);     /* size */
+      }
+
+      /* free the surface, update the texture if necessary.
+       */
+      pipe_surface_reference(&texSurf, NULL);
+      s0 = 0.0f; 
+      s1 = 1.0f;
+      t0 = 0.0f;
+      t1 = 1.0f;
+   }
+   else {
+      pipe_texture_reference(&tex, src->texture);
+      s0 = srcX0 / (float)tex->width[0];
+      s1 = srcX1 / (float)tex->width[0];
+      t0 = srcY0 / (float)tex->height[0];
+      t1 = srcY1 / (float)tex->height[0];
+   }
 
-   /* create temp texture */
-   memset(&texTemp, 0, sizeof(texTemp));
-   texTemp.target = PIPE_TEXTURE_2D;
-   texTemp.format = src->format;
-   texTemp.last_level = 0;
-   texTemp.width[0] = srcW;
-   texTemp.height[0] = srcH;
-   texTemp.depth[0] = 1;
-   pf_get_block(src->format, &texTemp.block);
-
-   tex = screen->texture_create(screen, &texTemp);
-   if (!tex)
-      return;
-
-   texSurf = screen->get_tex_surface(screen, tex, 0, 0, 0, 
-                                     PIPE_BUFFER_USAGE_GPU_WRITE);
-
-   /* load temp texture */
-   pipe->surface_copy(pipe,
-                      texSurf, 0, 0,   /* dest */
-                      src, srcLeft, srcTop, /* src */
-                      srcW, srcH);     /* size */
-
-   /* free the surface, update the texture if necessary.
-    */
-   pipe_surface_reference(&texSurf, NULL);
 
    /* save state (restored below) */
    cso_save_blend(ctx->cso);
@@ -447,9 +436,12 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    cso_set_framebuffer(ctx->cso, &fb);
 
    /* draw quad */
-   offset = setup_vertex_data(ctx,
-                              (float) dstX0, (float) dstY0, 
-                              (float) dstX1, (float) dstY1, z);
+   offset = setup_vertex_data_tex(ctx,
+                                  (float) dstX0, (float) dstY0, 
+                                  (float) dstX1, (float) dstY1,
+                                  s0, t0,
+                                  s1, t1,
+                                  z);
 
    util_draw_vertex_buffer(ctx->pipe, ctx->vbuf, offset,
                            PIPE_PRIM_TRIANGLE_FAN,
diff --git a/src/gallium/auxiliary/util/u_clear.h b/src/gallium/auxiliary/util/u_clear.h
index 7c16b32cf9..1e65a035ae 100644
--- a/src/gallium/auxiliary/util/u_clear.h
+++ b/src/gallium/auxiliary/util/u_clear.h
@@ -32,6 +32,7 @@
 #include "pipe/p_context.h"
 #include "pipe/p_state.h"
 #include "util/u_pack_color.h"
+#include "util/u_rect.h"
 
 
 /**
@@ -48,13 +49,22 @@ util_clear(struct pipe_context *pipe,
       unsigned color;
 
       util_pack_color(rgba, ps->format, &color);
-      pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, color);
+      if (pipe->surface_fill) {
+         pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, color);
+      } else {
+         util_surface_fill(pipe, ps, 0, 0, ps->width, ps->height, color);
+      }
    }
 
    if (buffers & PIPE_CLEAR_DEPTHSTENCIL) {
       struct pipe_surface *ps = framebuffer->zsbuf;
 
-      pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height,
-                         util_pack_z_stencil(ps->format, depth, stencil));
+      if (pipe->surface_fill) {
+         pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height,
+                            util_pack_z_stencil(ps->format, depth, stencil));
+      } else {
+         util_surface_fill(pipe, ps, 0, 0, ps->width, ps->height,
+                           util_pack_z_stencil(ps->format, depth, stencil));
+      }
    }
 }
diff --git a/src/gallium/auxiliary/util/u_cpu_detect.c b/src/gallium/auxiliary/util/u_cpu_detect.c
index f78706f447..c93e0db23c 100644
--- a/src/gallium/auxiliary/util/u_cpu_detect.c
+++ b/src/gallium/auxiliary/util/u_cpu_detect.c
@@ -24,23 +24,21 @@
  * 
  **************************************************************************/
 
-/*
- * Based on the work of Eric Anholt <anholt@FreeBSD.org>
+/**
+ * @file
+ * CPU feature detection.
+ *
+ * @author Dennis Smit
+ * @author Based on the work of Eric Anholt <anholt@FreeBSD.org>
  */
 
-/* FIXME: clean this entire file up */
+#include "pipe/p_config.h"
 
+#include "u_debug.h"
 #include "u_cpu_detect.h"
 
-#ifdef __linux__
-#define OS_LINUX
-#endif
-#ifdef WIN32
-#define OS_WIN32
-#endif
-
-#if defined(ARCH_POWERPC)
-#if defined(OS_DARWIN)
+#if defined(PIPE_ARCH_PPC)
+#if defined(PIPE_OS_DARWIN)
 #include <sys/sysctl.h>
 #else
 #include <signal.h>
@@ -48,140 +46,145 @@
 #endif
 #endif
 
-#if defined(OS_NETBSD) || defined(OS_OPENBSD)
+#if defined(PIPE_OS_NETBSD) || defined(PIPE_OS_OPENBSD)
 #include <sys/param.h>
 #include <sys/sysctl.h>
 #include <machine/cpu.h>
 #endif
 
-#if defined(OS_FREEBSD)
+#if defined(PIPE_OS_FREEBSD)
 #include <sys/types.h>
 #include <sys/sysctl.h>
 #endif
 
-#if defined(OS_LINUX)
+#if defined(PIPE_OS_LINUX)
 #include <signal.h>
 #endif
 
-#if defined(OS_WIN32)
-#include <windows.h>
+#ifdef PIPE_OS_UNIX
+#include <unistd.h>
 #endif
 
-#include <stdio.h>
-#include <stdlib.h>
-#include <unistd.h>
-#include <string.h>
+#if defined(PIPE_OS_WINDOWS)
+#include <windows.h>
+#endif
 
 
-static struct cpu_detect_caps __cpu_detect_caps;
-static int __cpu_detect_initialized = 0;
+struct util_cpu_caps util_cpu_caps;
 
 static int has_cpuid(void);
-static int cpuid(unsigned int ax, unsigned int *p);
+static int cpuid(uint32_t ax, uint32_t *p);
+
+#if defined(PIPE_ARCH_X86)
 
 /* The sigill handlers */
-#if defined(ARCH_X86) /* x86 (linux katmai handler check thing) */
-#if defined(OS_LINUX) && defined(_POSIX_SOURCE) && defined(X86_FXSR_MAGIC)
-static void sigill_handler_sse(int signal, struct sigcontext sc)
+#if defined(PIPE_OS_LINUX) //&& defined(_POSIX_SOURCE) && defined(X86_FXSR_MAGIC)
+static void
+sigill_handler_sse(int signal, struct sigcontext sc)
 {
-	/* Both the "xorps %%xmm0,%%xmm0" and "divps %xmm0,%%xmm1"
-	 * instructions are 3 bytes long.  We must increment the instruction
-	 * pointer manually to avoid repeated execution of the offending
-	 * instruction.
-	 *
-	 * If the SIGILL is caused by a divide-by-zero when unmasked
-	 * exceptions aren't supported, the SIMD FPU status and control
-	 * word will be restored at the end of the test, so we don't need
-	 * to worry about doing it here.  Besides, we may not be able to...
-	 */
-	sc.eip += 3;
-
-	__cpu_detect_caps.hasSSE=0;
+   /* Both the "xorps %%xmm0,%%xmm0" and "divps %xmm0,%%xmm1"
+    * instructions are 3 bytes long.  We must increment the instruction
+    * pointer manually to avoid repeated execution of the offending
+    * instruction.
+    *
+    * If the SIGILL is caused by a divide-by-zero when unmasked
+    * exceptions aren't supported, the SIMD FPU status and control
+    * word will be restored at the end of the test, so we don't need
+    * to worry about doing it here.  Besides, we may not be able to...
+    */
+   sc.eip += 3;
+
+   util_cpu_caps.has_sse=0;
 }
 
-static void sigfpe_handler_sse(int signal, struct sigcontext sc)
+static void
+sigfpe_handler_sse(int signal, struct sigcontext sc)
 {
-	if (sc.fpstate->magic != 0xffff) {
-		/* Our signal context has the extended FPU state, so reset the
-		 * divide-by-zero exception mask and clear the divide-by-zero
-		 * exception bit.
-		 */
-		sc.fpstate->mxcsr |= 0x00000200;
-		sc.fpstate->mxcsr &= 0xfffffffb;
-	} else {
-		/* If we ever get here, we're completely hosed.
-		*/
-	}
+   if (sc.fpstate->magic != 0xffff) {
+      /* Our signal context has the extended FPU state, so reset the
+       * divide-by-zero exception mask and clear the divide-by-zero
+       * exception bit.
+       */
+      sc.fpstate->mxcsr |= 0x00000200;
+      sc.fpstate->mxcsr &= 0xfffffffb;
+   } else {
+      /* If we ever get here, we're completely hosed.
+      */
+   }
 }
-#endif
-#endif /* OS_LINUX && _POSIX_SOURCE && X86_FXSR_MAGIC */
+#endif /* PIPE_OS_LINUX && _POSIX_SOURCE && X86_FXSR_MAGIC */
 
-#if defined(OS_WIN32)
-LONG CALLBACK win32_sig_handler_sse(EXCEPTION_POINTERS* ep)
+#if defined(PIPE_OS_WINDOWS)
+static LONG CALLBACK
+win32_sig_handler_sse(EXCEPTION_POINTERS* ep)
 {
-	if(ep->ExceptionRecord->ExceptionCode==EXCEPTION_ILLEGAL_INSTRUCTION){
-		ep->ContextRecord->Eip +=3;
-		__cpu_detect_caps.hasSSE=0;
-		return EXCEPTION_CONTINUE_EXECUTION;
-	}
-	return EXCEPTION_CONTINUE_SEARCH;
+   if(ep->ExceptionRecord->ExceptionCode==EXCEPTION_ILLEGAL_INSTRUCTION){
+      ep->ContextRecord->Eip +=3;
+      util_cpu_caps.has_sse=0;
+      return EXCEPTION_CONTINUE_EXECUTION;
+   }
+   return EXCEPTION_CONTINUE_SEARCH;
 }
-#endif /* OS_WIN32 */
+#endif /* PIPE_OS_WINDOWS */
 
+#endif /* PIPE_ARCH_X86 */
 
-#if defined(ARCH_POWERPC) && !defined(OS_DARWIN)
-static sigjmp_buf __lv_powerpc_jmpbuf;
-static volatile sig_atomic_t __lv_powerpc_canjump = 0;
 
-static void sigill_handler (int sig);
+#if defined(PIPE_ARCH_PPC) && !defined(PIPE_OS_DARWIN)
+static jmp_buf  __lv_powerpc_jmpbuf;
+static volatile sig_atomic_t __lv_powerpc_canjump = 0;
 
-static void sigill_handler (int sig)
+static void
+sigill_handler(int sig)
 {
-	if (!__lv_powerpc_canjump) {
-		signal (sig, SIG_DFL);
-		raise (sig);
-	}
+   if (!__lv_powerpc_canjump) {
+      signal (sig, SIG_DFL);
+      raise (sig);
+   }
 
-	__lv_powerpc_canjump = 0;
-	siglongjmp(__lv_powerpc_jmpbuf, 1);
+   __lv_powerpc_canjump = 0;
+   longjmp(__lv_powerpc_jmpbuf, 1);
 }
+#endif
 
-static void check_os_altivec_support(void)
+#if defined(PIPE_ARCH_PPC)
+static void
+check_os_altivec_support(void)
 {
-#if defined(OS_DARWIN)
-	int sels[2] = {CTL_HW, HW_VECTORUNIT};
-	int has_vu = 0;
-	int len = sizeof (has_vu);
-	int err;
-
-	err = sysctl(sels, 2, &has_vu, &len, NULL, 0);
-
-	if (err == 0) {
-		if (has_vu != 0) {
-			__cpu_detect_caps.hasAltiVec = 1;
-		}
-	}
-#else /* !OS_DARWIN */
-	/* no Darwin, do it the brute-force way */
-	/* this is borrowed from the libmpeg2 library */
-	signal(SIGILL, sigill_handler);
-	if (sigsetjmp(__lv_powerpc_jmpbuf, 1)) {
-		signal(SIGILL, SIG_DFL);
-	} else {
-		__lv_powerpc_canjump = 1;
-
-		__asm __volatile
-			("mtspr 256, %0\n\t"
-			 "vand %%v0, %%v0, %%v0"
-			 :
-			 : "r" (-1));
-
-		signal(SIGILL, SIG_DFL);
-		__cpu_detect_caps.hasAltiVec = 1;
-	}
-#endif
+#if defined(PIPE_OS_DARWIN)
+   int sels[2] = {CTL_HW, HW_VECTORUNIT};
+   int has_vu = 0;
+   int len = sizeof (has_vu);
+   int err;
+
+   err = sysctl(sels, 2, &has_vu, &len, NULL, 0);
+
+   if (err == 0) {
+      if (has_vu != 0) {
+         util_cpu_caps.has_altivec = 1;
+      }
+   }
+#else /* !PIPE_OS_DARWIN */
+   /* no Darwin, do it the brute-force way */
+   /* this is borrowed from the libmpeg2 library */
+   signal(SIGILL, sigill_handler);
+   if (setjmp(__lv_powerpc_jmpbuf)) {
+      signal(SIGILL, SIG_DFL);
+   } else {
+      __lv_powerpc_canjump = 1;
+
+      __asm __volatile
+         ("mtspr 256, %0\n\t"
+          "vand %%v0, %%v0, %%v0"
+          :
+          : "r" (-1));
+
+      signal(SIGILL, SIG_DFL);
+      util_cpu_caps.has_altivec = 1;
+   }
+#endif /* PIPE_OS_DARWIN */
 }
-#endif
+#endif /* PIPE_ARCH_PPC */
 
 /* If we're running on a processor that can do SSE, let's see if we
  * are allowed to or not.  This will catch 2.4.0 or later kernels that
@@ -189,318 +192,329 @@ static void check_os_altivec_support(void)
  * and RedHat patched 2.2 kernels that have broken exception handling
  * support for user space apps that do SSE.
  */
-static void check_os_katmai_support(void)
+#if defined(PIPE_ARCH_X86) || defined (PIPE_ARCH_X86_64)
+static void
+check_os_katmai_support(void)
 {
-#if defined(ARCH_X86)
-#if defined(OS_FREEBSD)
-	int has_sse=0, ret;
-	int len = sizeof (has_sse);
-
-	ret = sysctlbyname("hw.instruction_sse", &has_sse, &len, NULL, 0);
-	if (ret || !has_sse)
-		__cpu_detect_caps.hasSSE=0;
-
-#elif defined(OS_NETBSD) || defined(OS_OPENBSD)
-	int has_sse, has_sse2, ret, mib[2];
-	int varlen;
-
-	mib[0] = CTL_MACHDEP;
-	mib[1] = CPU_SSE;
-	varlen = sizeof (has_sse);
-
-	ret = sysctl(mib, 2, &has_sse, &varlen, NULL, 0);
-	if (ret < 0 || !has_sse) {
-		__cpu_detect_caps.hasSSE = 0;
-	} else {
-		__cpu_detect_caps.hasSSE = 1;
-	}
-
-	mib[1] = CPU_SSE2;
-	varlen = sizeof (has_sse2);
-	ret = sysctl(mib, 2, &has_sse2, &varlen, NULL, 0);
-	if (ret < 0 || !has_sse2) {
-		__cpu_detect_caps.hasSSE2 = 0;
-	} else {
-		__cpu_detect_caps.hasSSE2 = 1;
-	}
-	__cpu_detect_caps.hasSSE = 0; /* FIXME ?!?!? */
-
-#elif defined(OS_WIN32)
-	LPTOP_LEVEL_EXCEPTION_FILTER exc_fil;
-	if (__cpu_detect_caps.hasSSE) {
-		exc_fil = SetUnhandledExceptionFilter(win32_sig_handler_sse);
-		__asm __volatile ("xorps %xmm0, %xmm0");
-		SetUnhandledExceptionFilter(exc_fil);
-	}
-#elif defined(OS_LINUX)
-	struct sigaction saved_sigill;
-	struct sigaction saved_sigfpe;
-
-	/* Save the original signal handlers.
-	*/
-	sigaction(SIGILL, NULL, &saved_sigill);
-	sigaction(SIGFPE, NULL, &saved_sigfpe);
-
-	signal(SIGILL, (void (*)(int))sigill_handler_sse);
-	signal(SIGFPE, (void (*)(int))sigfpe_handler_sse);
-
-	/* Emulate test for OSFXSR in CR4.  The OS will set this bit if it
-	 * supports the extended FPU save and restore required for SSE.  If
-	 * we execute an SSE instruction on a PIII and get a SIGILL, the OS
-	 * doesn't support Streaming SIMD Exceptions, even if the processor
-	 * does.
-	 */
-	if (__cpu_detect_caps.hasSSE) {
-		__asm __volatile ("xorps %xmm1, %xmm0");
-	}
-
-	/* Emulate test for OSXMMEXCPT in CR4.  The OS will set this bit if
-	 * it supports unmasked SIMD FPU exceptions.  If we unmask the
-	 * exceptions, do a SIMD divide-by-zero and get a SIGILL, the OS
-	 * doesn't support unmasked SIMD FPU exceptions.  If we get a SIGFPE
-	 * as expected, we're okay but we need to clean up after it.
-	 *
-	 * Are we being too stringent in our requirement that the OS support
-	 * unmasked exceptions?  Certain RedHat 2.2 kernels enable SSE by
-	 * setting CR4.OSFXSR but don't support unmasked exceptions.  Win98
-	 * doesn't even support them.  We at least know the user-space SSE
-	 * support is good in kernels that do support unmasked exceptions,
-	 * and therefore to be safe I'm going to leave this test in here.
-	 */
-	if (__cpu_detect_caps.hasSSE) {
-           /* test_os_katmai_exception_support(); */
-	}
-
-	/* Restore the original signal handlers.
-	*/
-	sigaction(SIGILL, &saved_sigill, NULL);
-	sigaction(SIGFPE, &saved_sigfpe, NULL);
+#if defined(PIPE_ARCH_X86)
+#if defined(PIPE_OS_FREEBSD)
+   int has_sse=0, ret;
+   int len = sizeof (has_sse);
+
+   ret = sysctlbyname("hw.instruction_sse", &has_sse, &len, NULL, 0);
+   if (ret || !has_sse)
+      util_cpu_caps.has_sse=0;
+
+#elif defined(PIPE_OS_NETBSD) || defined(PIPE_OS_OPENBSD)
+   int has_sse, has_sse2, ret, mib[2];
+   int varlen;
+
+   mib[0] = CTL_MACHDEP;
+   mib[1] = CPU_SSE;
+   varlen = sizeof (has_sse);
+
+   ret = sysctl(mib, 2, &has_sse, &varlen, NULL, 0);
+   if (ret < 0 || !has_sse) {
+      util_cpu_caps.has_sse = 0;
+   } else {
+      util_cpu_caps.has_sse = 1;
+   }
+
+   mib[1] = CPU_SSE2;
+   varlen = sizeof (has_sse2);
+   ret = sysctl(mib, 2, &has_sse2, &varlen, NULL, 0);
+   if (ret < 0 || !has_sse2) {
+      util_cpu_caps.has_sse2 = 0;
+   } else {
+      util_cpu_caps.has_sse2 = 1;
+   }
+   util_cpu_caps.has_sse = 0; /* FIXME ?!?!? */
+
+#elif defined(PIPE_OS_WINDOWS)
+   LPTOP_LEVEL_EXCEPTION_FILTER exc_fil;
+   if (util_cpu_caps.has_sse) {
+      exc_fil = SetUnhandledExceptionFilter(win32_sig_handler_sse);
+#if defined(PIPE_CC_GCC)
+      __asm __volatile ("xorps %xmm0, %xmm0");
+#elif defined(PIPE_CC_MSVC)
+      __asm {
+          xorps xmm0, xmm0        // executing SSE instruction
+      }
+#else
+#error Unsupported compiler
+#endif
+      SetUnhandledExceptionFilter(exc_fil);
+   }
+#elif defined(PIPE_OS_LINUX)
+   struct sigaction saved_sigill;
+   struct sigaction saved_sigfpe;
+
+   /* Save the original signal handlers.
+   */
+   sigaction(SIGILL, NULL, &saved_sigill);
+   sigaction(SIGFPE, NULL, &saved_sigfpe);
+
+   signal(SIGILL, (void (*)(int))sigill_handler_sse);
+   signal(SIGFPE, (void (*)(int))sigfpe_handler_sse);
+
+   /* Emulate test for OSFXSR in CR4.  The OS will set this bit if it
+    * supports the extended FPU save and restore required for SSE.  If
+    * we execute an SSE instruction on a PIII and get a SIGILL, the OS
+    * doesn't support Streaming SIMD Exceptions, even if the processor
+    * does.
+    */
+   if (util_cpu_caps.has_sse) {
+      __asm __volatile ("xorps %xmm1, %xmm0");
+   }
+
+   /* Emulate test for OSXMMEXCPT in CR4.  The OS will set this bit if
+    * it supports unmasked SIMD FPU exceptions.  If we unmask the
+    * exceptions, do a SIMD divide-by-zero and get a SIGILL, the OS
+    * doesn't support unmasked SIMD FPU exceptions.  If we get a SIGFPE
+    * as expected, we're okay but we need to clean up after it.
+    *
+    * Are we being too stringent in our requirement that the OS support
+    * unmasked exceptions?  Certain RedHat 2.2 kernels enable SSE by
+    * setting CR4.OSFXSR but don't support unmasked exceptions.  Win98
+    * doesn't even support them.  We at least know the user-space SSE
+    * support is good in kernels that do support unmasked exceptions,
+    * and therefore to be safe I'm going to leave this test in here.
+    */
+   if (util_cpu_caps.has_sse) {
+      //      test_os_katmai_exception_support();
+   }
+
+   /* Restore the original signal handlers.
+   */
+   sigaction(SIGILL, &saved_sigill, NULL);
+   sigaction(SIGFPE, &saved_sigfpe, NULL);
 
 #else
-	/* We can't use POSIX signal handling to test the availability of
-	 * SSE, so we disable it by default.
-	 */
-	__cpu_detect_caps.hasSSE = 0;
+   /* We can't use POSIX signal handling to test the availability of
+    * SSE, so we disable it by default.
+    */
+   util_cpu_caps.has_sse = 0;
 #endif /* __linux__ */
 #endif
+
+#if defined(PIPE_ARCH_X86_64)
+   util_cpu_caps.has_sse = 1;
+#endif
 }
 
 
 static int has_cpuid(void)
 {
-#if defined(ARCH_X86)
-	int a, c;
-
-	__asm __volatile
-		("pushf\n"
-		 "popl %0\n"
-		 "movl %0, %1\n"
-		 "xorl $0x200000, %0\n"
-		 "push %0\n"
-		 "popf\n"
-		 "pushf\n"
-		 "popl %0\n"
-		 : "=a" (a), "=c" (c)
-		 :
-		 : "cc");
-
-	return a != c;
+#if defined(PIPE_ARCH_X86)
+#if defined(PIPE_OS_GCC)
+   int a, c;
+
+   __asm __volatile
+      ("pushf\n"
+       "popl %0\n"
+       "movl %0, %1\n"
+       "xorl $0x200000, %0\n"
+       "push %0\n"
+       "popf\n"
+       "pushf\n"
+       "popl %0\n"
+       : "=a" (a), "=c" (c)
+       :
+       : "cc");
+
+   return a != c;
 #else
-	return 0;
+   /* FIXME */
+   return 1;
 #endif
-}
-
-static int cpuid(unsigned int ax, unsigned int *p)
-{
-#if defined(ARCH_X86)
-	unsigned int flags;
-
-	__asm __volatile
-		("movl %%ebx, %%esi\n\t"
-		 "cpuid\n\t"
-		 "xchgl %%ebx, %%esi"
-		 : "=a" (p[0]), "=S" (p[1]),
-		 "=c" (p[2]), "=d" (p[3])
-		 : "0" (ax));
-
-	return 0;
+#elif defined(PIPE_ARCH_X86_64)
+   return 1;
 #else
-	return -1;
+   return 0;
 #endif
 }
 
-void cpu_detect_initialize()
+
+/**
+ * @sa cpuid.h included in gcc-4.3 onwards.
+ */
+static INLINE int
+cpuid(uint32_t ax, uint32_t *p)
 {
-	unsigned int regs[4];
-	unsigned int regs2[4];
-
-	int mib[2], ncpu;
-	int len;
-
-	memset(&__cpu_detect_caps, 0, sizeof (struct cpu_detect_caps));
-
-	/* Check for arch type */
-#if defined(ARCH_MIPS)
-	__cpu_detect_caps.type = CPU_DETECT_TYPE_MIPS;
-#elif defined(ARCH_ALPHA)
-	__cpu_detect_caps.type = CPU_DETECT_TYPE_ALPHA;
-#elif defined(ARCH_SPARC)
-	__cpu_detect_caps.type = CPU_DETECT_TYPE_SPARC;
-#elif defined(ARCH_X86)
-	__cpu_detect_caps.type = CPU_DETECT_TYPE_X86;
-#elif defined(ARCH_POWERPC)
-	__cpu_detect_caps.type = CPU_DETECT_TYPE_POWERPC;
-#else
-	__cpu_detect_caps.type = CPU_DETECT_TYPE_OTHER;
+   int ret = -1;
+
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
+   __asm __volatile (
+     "xchgl %%ebx, %1\n\t"
+     "cpuid\n\t"
+     "xchgl %%ebx, %1"
+     : "=a" (p[0]),
+       "=S" (p[1]),
+       "=c" (p[2]),
+       "=d" (p[3])
+     : "0" (ax)
+   );
+   ret = 0;
+#elif defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86_64)
+   __asm __volatile (
+     "cpuid\n\t"
+     : "=a" (p[0]),
+       "=b" (p[1]),
+       "=c" (p[2]),
+       "=d" (p[3])
+     : "0" (ax)
+   );
+   ret = 0;
+#elif defined(PIPE_CC_MSVC)
+   __cpuid(ax, p);
+
+   ret = 0;
 #endif
 
-	/* Count the number of CPUs in system */
-#if !defined(OS_WIN32) && !defined(OS_UNKNOWN) && defined(_SC_NPROCESSORS_ONLN)
-	__cpu_detect_caps.nrcpu = sysconf(_SC_NPROCESSORS_ONLN);
-	if (__cpu_detect_caps.nrcpu == -1)
-		__cpu_detect_caps.nrcpu = 1;
-
-#elif defined(OS_NETBSD) || defined(OS_FREEBSD) || defined(OS_OPENBSD)
-
-	mib[0] = CTL_HW;
-	mib[1] = HW_NCPU;
-
-	len = sizeof (ncpu);
-	sysctl(mib, 2, &ncpu, &len, NULL, 0);
-	__cpu_detect_caps.nrcpu = ncpu;
+   return ret;
+}
+#endif /* X86 or X86_64 */
 
+void
+util_cpu_detect(void)
+{
+   static boolean util_cpu_detect_initialized = FALSE;
+
+   if(util_cpu_detect_initialized)
+      return;
+
+   memset(&util_cpu_caps, 0, sizeof util_cpu_caps);
+
+   /* Check for arch type */
+#if defined(PIPE_ARCH_MIPS)
+   util_cpu_caps.arch = UTIL_CPU_ARCH_MIPS;
+#elif defined(PIPE_ARCH_ALPHA)
+   util_cpu_caps.arch = UTIL_CPU_ARCH_ALPHA;
+#elif defined(PIPE_ARCH_SPARC)
+   util_cpu_caps.arch = UTIL_CPU_ARCH_SPARC;
+#elif defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+   util_cpu_caps.arch = UTIL_CPU_ARCH_X86;
+   util_cpu_caps.little_endian = 1;
+#elif defined(PIPE_ARCH_PPC)
+   util_cpu_caps.arch = UTIL_CPU_ARCH_POWERPC;
+   util_cpu_caps.little_endian = 0;
 #else
-	__cpu_detect_caps.nrcpu = 1;
+   util_cpu_caps.arch = UTIL_CPU_ARCH_UNKNOWN;
 #endif
 
-#if defined(ARCH_X86)
-	/* No cpuid, old 486 or lower */
-	if (has_cpuid() == 0)
-		return;
-
-	__cpu_detect_caps.cacheline = 32;
-
-	/* Get max cpuid level */
-	cpuid(0x00000000, regs);
-
-	if (regs[0] >= 0x00000001) {
-		unsigned int cacheline;
-
-		cpuid (0x00000001, regs2);
-
-		__cpu_detect_caps.x86cpuType = (regs2[0] >> 8) & 0xf;
-		if (__cpu_detect_caps.x86cpuType == 0xf)
-		    __cpu_detect_caps.x86cpuType = 8 + ((regs2[0] >> 20) & 255); /* use extended family (P4, IA64) */
-
-		/* general feature flags */
-		__cpu_detect_caps.hasTSC  = (regs2[3] & (1 << 8  )) >>  8; /* 0x0000010 */
-		__cpu_detect_caps.hasMMX  = (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */
-		__cpu_detect_caps.hasSSE  = (regs2[3] & (1 << 25 )) >> 25; /* 0x2000000 */
-		__cpu_detect_caps.hasSSE2 = (regs2[3] & (1 << 26 )) >> 26; /* 0x4000000 */
-		__cpu_detect_caps.hasSSE3 = (regs2[2] & (1));	       /* 0x0000001 */
-		__cpu_detect_caps.hasSSSE3 = (regs2[2] & (1 << 9 )) >> 9;   /* 0x0000020 */
-		__cpu_detect_caps.hasMMX2 = __cpu_detect_caps.hasSSE; /* SSE cpus supports mmxext too */
-
-		cacheline = ((regs2[1] >> 8) & 0xFF) * 8;
-		if (cacheline > 0)
-			__cpu_detect_caps.cacheline = cacheline;
-	}
-
-	cpuid(0x80000000, regs);
-
-	if (regs[0] >= 0x80000001) {
-
-		cpuid(0x80000001, regs2);
-
-		__cpu_detect_caps.hasMMX  |= (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */
-		__cpu_detect_caps.hasMMX2 |= (regs2[3] & (1 << 22 )) >> 22; /* 0x400000 */
-		__cpu_detect_caps.has3DNow    = (regs2[3] & (1 << 31 )) >> 31; /* 0x80000000 */
-		__cpu_detect_caps.has3DNowExt = (regs2[3] & (1 << 30 )) >> 30;
-	}
-
-	if (regs[0] >= 0x80000006) {
-		cpuid(0x80000006, regs2);
-		__cpu_detect_caps.cacheline = regs2[2] & 0xFF;
-	}
-
-
-#if defined(OS_LINUX) || defined(OS_FREEBSD) || defined(OS_NETBSD) || defined(OS_CYGWIN) || defined(OS_OPENBSD)
-	if (__cpu_detect_caps.hasSSE)
-		check_os_katmai_support();
-
-	if (!__cpu_detect_caps.hasSSE) {
-		__cpu_detect_caps.hasSSE2 = 0;
-		__cpu_detect_caps.hasSSE3 = 0;
-		__cpu_detect_caps.hasSSSE3 = 0;
-	}
+   /* Count the number of CPUs in system */
+#if defined(PIPE_OS_WINDOWS)
+   {
+      SYSTEM_INFO system_info;
+      GetSystemInfo(&system_info);
+      util_cpu_caps.nr_cpus = system_info.dwNumberOfProcessors;
+   }
+#elif defined(PIPE_OS_UNIX) && defined(_SC_NPROCESSORS_ONLN)
+   util_cpu_caps.nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
+   if (util_cpu_caps.nr_cpus == -1)
+      util_cpu_caps.nr_cpus = 1;
+#elif defined(PIPE_OS_BSD)
+   {
+      int mib[2], ncpu;
+      int len;
+
+      mib[0] = CTL_HW;
+      mib[1] = HW_NCPU;
+
+      len = sizeof (ncpu);
+      sysctl(mib, 2, &ncpu, &len, NULL, 0);
+      util_cpu_caps.nr_cpus = ncpu;
+   }
 #else
-	__cpu_detect_caps.hasSSE = 0;
-	__cpu_detect_caps.hasSSE2 = 0;
-	__cpu_detect_caps.hasSSE3 = 0;
-	__cpu_detect_caps.hasSSSE3 = 0;
+   util_cpu_caps.nr_cpus = 1;
 #endif
-#endif /* ARCH_X86 */
-
-#if defined(ARCH_POWERPC)
-	check_os_altivec_support();
-#endif /* ARCH_POWERPC */
-
-	__cpu_detect_initialized = 1;
-}
-
-struct cpu_detect_caps *cpu_detect_get_caps()
-{
-	return &__cpu_detect_caps;
-}
-
-/* The getters and setters for feature flags */
-int cpu_detect_get_tsc()
-{
-	return __cpu_detect_caps.hasTSC;
-}
-
-int cpu_detect_get_mmx()
-{
-	return __cpu_detect_caps.hasMMX;
-}
-
-int cpu_detect_get_mmx2()
-{
-	return __cpu_detect_caps.hasMMX2;
-}
 
-int cpu_detect_get_sse()
-{
-	return __cpu_detect_caps.hasSSE;
-}
-
-int cpu_detect_get_sse2()
-{
-	return __cpu_detect_caps.hasSSE2;
-}
-
-int cpu_detect_get_sse3()
-{
-	return __cpu_detect_caps.hasSSE3;
-}
-
-int cpu_detect_get_ssse3()
-{
-	return __cpu_detect_caps.hasSSSE3;
-}
-
-int cpu_detect_get_3dnow()
-{
-	return __cpu_detect_caps.has3DNow;
-}
-
-int cpu_detect_get_3dnow2()
-{
-	return __cpu_detect_caps.has3DNowExt;
-}
+#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+   if (has_cpuid()) {
+      uint32_t regs[4];
+      uint32_t regs2[4];
+
+      util_cpu_caps.cacheline = 32;
+
+      /* Get max cpuid level */
+      cpuid(0x00000000, regs);
+
+      if (regs[0] >= 0x00000001) {
+         unsigned int cacheline;
+
+         cpuid (0x00000001, regs2);
+
+         util_cpu_caps.x86_cpu_type = (regs2[0] >> 8) & 0xf;
+         if (util_cpu_caps.x86_cpu_type == 0xf)
+             util_cpu_caps.x86_cpu_type = 8 + ((regs2[0] >> 20) & 255); /* use extended family (P4, IA64) */
+
+         /* general feature flags */
+         util_cpu_caps.has_tsc    = (regs2[3] & (1 << 8  )) >>  8; /* 0x0000010 */
+         util_cpu_caps.has_mmx    = (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */
+         util_cpu_caps.has_sse    = (regs2[3] & (1 << 25 )) >> 25; /* 0x2000000 */
+         util_cpu_caps.has_sse2   = (regs2[3] & (1 << 26 )) >> 26; /* 0x4000000 */
+         util_cpu_caps.has_sse3   = (regs2[2] & (1));          /* 0x0000001 */
+         util_cpu_caps.has_ssse3  = (regs2[2] & (1 << 9 )) >> 9;   /* 0x0000020 */
+         util_cpu_caps.has_sse4_1 = (regs2[2] & (1 << 19)) >> 19;
+         util_cpu_caps.has_mmx2   = util_cpu_caps.has_sse; /* SSE cpus supports mmxext too */
+
+         cacheline = ((regs2[1] >> 8) & 0xFF) * 8;
+         if (cacheline > 0)
+            util_cpu_caps.cacheline = cacheline;
+      }
+
+      cpuid(0x80000000, regs);
+
+      if (regs[0] >= 0x80000001) {
+
+         cpuid(0x80000001, regs2);
+
+         util_cpu_caps.has_mmx  |= (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */
+         util_cpu_caps.has_mmx2 |= (regs2[3] & (1 << 22 )) >> 22; /* 0x400000 */
+         util_cpu_caps.has_3dnow    = (regs2[3] & (1 << 31 )) >> 31; /* 0x80000000 */
+         util_cpu_caps.has_3dnow_ext = (regs2[3] & (1 << 30 )) >> 30;
+      }
+
+      if (regs[0] >= 0x80000006) {
+         cpuid(0x80000006, regs2);
+         util_cpu_caps.cacheline = regs2[2] & 0xFF;
+      }
+
+      if (util_cpu_caps.has_sse)
+         check_os_katmai_support();
+
+      if (!util_cpu_caps.has_sse) {
+         util_cpu_caps.has_sse2 = 0;
+         util_cpu_caps.has_sse3 = 0;
+         util_cpu_caps.has_ssse3 = 0;
+         util_cpu_caps.has_sse4_1 = 0;
+      }
+   }
+#endif /* PIPE_ARCH_X86 || PIPE_ARCH_X86_64 */
+
+#if defined(PIPE_ARCH_PPC)
+   check_os_altivec_support();
+#endif /* PIPE_ARCH_PPC */
+
+#ifdef DEBUG
+   debug_printf("util_cpu_caps.arch = %i\n", util_cpu_caps.arch);
+   debug_printf("util_cpu_caps.nr_cpus = %u\n", util_cpu_caps.nr_cpus);
+
+   debug_printf("util_cpu_caps.x86_cpu_type = %u\n", util_cpu_caps.x86_cpu_type);
+   debug_printf("util_cpu_caps.cacheline = %u\n", util_cpu_caps.cacheline);
+
+   debug_printf("util_cpu_caps.has_tsc = %u\n", util_cpu_caps.has_tsc);
+   debug_printf("util_cpu_caps.has_mmx = %u\n", util_cpu_caps.has_mmx);
+   debug_printf("util_cpu_caps.has_mmx2 = %u\n", util_cpu_caps.has_mmx2);
+   debug_printf("util_cpu_caps.has_sse = %u\n", util_cpu_caps.has_sse);
+   debug_printf("util_cpu_caps.has_sse2 = %u\n", util_cpu_caps.has_sse2);
+   debug_printf("util_cpu_caps.has_sse3 = %u\n", util_cpu_caps.has_sse3);
+   debug_printf("util_cpu_caps.has_ssse3 = %u\n", util_cpu_caps.has_ssse3);
+   debug_printf("util_cpu_caps.has_sse4_1 = %u\n", util_cpu_caps.has_sse4_1);
+   debug_printf("util_cpu_caps.has_3dnow = %u\n", util_cpu_caps.has_3dnow);
+   debug_printf("util_cpu_caps.has_3dnow_ext = %u\n", util_cpu_caps.has_3dnow_ext);
+   debug_printf("util_cpu_caps.has_altivec = %u\n", util_cpu_caps.has_altivec);
+#endif
 
-int cpu_detect_get_altivec()
-{
-	return __cpu_detect_caps.hasAltiVec;
+   util_cpu_detect_initialized = TRUE;
 }
-
diff --git a/src/gallium/auxiliary/util/u_cpu_detect.h b/src/gallium/auxiliary/util/u_cpu_detect.h
index 1612d49286..4b3dc39c34 100644
--- a/src/gallium/auxiliary/util/u_cpu_detect.h
+++ b/src/gallium/auxiliary/util/u_cpu_detect.h
@@ -24,55 +24,55 @@
  *
  ***************************************************************************/
 
-/*
- * Based on the work of Eric Anholt <anholt@FreeBSD.org>
+/**
+ * @file
+ * CPU feature detection.
+ *
+ * @author Dennis Smit
+ * @author Based on the work of Eric Anholt <anholt@FreeBSD.org>
  */
 
-#ifndef _CPU_DETECT_H
-#define _CPU_DETECT_H
+#ifndef _UTIL_CPU_DETECT_H
+#define _UTIL_CPU_DETECT_H
+
+#include "pipe/p_compiler.h"
 
-typedef enum {
-	CPU_DETECT_TYPE_MIPS,
-	CPU_DETECT_TYPE_ALPHA,
-	CPU_DETECT_TYPE_SPARC,
-	CPU_DETECT_TYPE_X86,
-	CPU_DETECT_TYPE_POWERPC,
-	CPU_DETECT_TYPE_OTHER
-} cpu_detect_type;
+enum util_cpu_arch {
+   UTIL_CPU_ARCH_UNKNOWN = 0,
+   UTIL_CPU_ARCH_MIPS,
+   UTIL_CPU_ARCH_ALPHA,
+   UTIL_CPU_ARCH_SPARC,
+   UTIL_CPU_ARCH_X86,
+   UTIL_CPU_ARCH_POWERPC
+};
 
-struct cpu_detect_caps {
-	cpu_detect_type	type;
-	int		nrcpu;
+struct util_cpu_caps {
+   enum util_cpu_arch arch;
+   unsigned nr_cpus;
 
-	/* Feature flags */
-	int		x86cpuType;
-	int		cacheline;
+   /* Feature flags */
+   int x86_cpu_type;
+   unsigned cacheline;
 
-	int		hasTSC;
-	int		hasMMX;
-	int		hasMMX2;
-	int		hasSSE;
-	int		hasSSE2;
-	int		hasSSE3;
-	int		hasSSSE3;
-	int		has3DNow;
-	int		has3DNowExt;
-	int		hasAltiVec;
+   unsigned little_endian:1;
+
+   unsigned has_tsc:1;
+   unsigned has_mmx:1;
+   unsigned has_mmx2:1;
+   unsigned has_sse:1;
+   unsigned has_sse2:1;
+   unsigned has_sse3:1;
+   unsigned has_ssse3:1;
+   unsigned has_sse4_1:1;
+   unsigned has_3dnow:1;
+   unsigned has_3dnow_ext:1;
+   unsigned has_altivec:1;
 };
 
-/* prototypes */
-void cpu_detect_initialize(void);
-struct cpu_detect_caps *cpu_detect_get_caps(void);
+extern struct util_cpu_caps
+util_cpu_caps;
+
+void util_cpu_detect(void);
 
-int cpu_detect_get_tsc(void);
-int cpu_detect_get_mmx(void);
-int cpu_detect_get_mmx2(void);
-int cpu_detect_get_sse(void);
-int cpu_detect_get_sse2(void);
-int cpu_detect_get_sse3(void);
-int cpu_detect_get_ssse3(void);
-int cpu_detect_get_3dnow(void);
-int cpu_detect_get_3dnow2(void);
-int cpu_detect_get_altivec(void);
 
-#endif /* _CPU_DETECT_H */
+#endif /* _UTIL_CPU_DETECT_H */
diff --git a/src/gallium/auxiliary/util/u_debug.h b/src/gallium/auxiliary/util/u_debug.h
index 1380d98d7e..b8c56fd600 100644
--- a/src/gallium/auxiliary/util/u_debug.h
+++ b/src/gallium/auxiliary/util/u_debug.h
@@ -65,6 +65,11 @@ extern "C" {
 #define __FUNCTION__ "???"
 #endif
 
+#if defined(__GNUC__)
+#define _util_printf_format(fmt, list) __attribute__ ((format (printf, fmt, list)))
+#else
+#define _util_printf_format(fmt, list)
+#endif
 
 void _debug_vprintf(const char *format, va_list ap);
    
@@ -82,14 +87,17 @@ _debug_printf(const char *format, ...)
 /**
  * Print debug messages.
  *
- * The actual channel used to output debug message is platform specific. To 
- * avoid misformating or truncation, follow these rules of thumb:   
+ * The actual channel used to output debug message is platform specific. To
+ * avoid misformating or truncation, follow these rules of thumb:
  * - output whole lines
- * - avoid outputing large strings (512 bytes is the current maximum length 
+ * - avoid outputing large strings (512 bytes is the current maximum length
  * that is guaranteed to be printed in all platforms)
  */
 #if !defined(PIPE_OS_HAIKU)
 static INLINE void
+debug_printf(const char *format, ...) _util_printf_format(1,2);
+
+static INLINE void
 debug_printf(const char *format, ...)
 {
 #ifdef DEBUG
@@ -173,11 +181,14 @@ void _debug_assert_fail(const char *expr,
  * 
  * Do not expect that the assert call terminates -- errors must be handled 
  * regardless of assert behavior.
+ *
+ * For non debug builds the assert macro will expand to a no-op, so do not
+ * call functions with side effects in the assert expression.
  */
 #ifdef DEBUG
 #define debug_assert(expr) ((expr) ? (void)0 : _debug_assert_fail(#expr, __FILE__, __LINE__, __FUNCTION__))
 #else
-#define debug_assert(expr) ((void)(expr))
+#define debug_assert(expr) ((void)0)
 #endif
 
 
diff --git a/src/gallium/auxiliary/util/u_fifo.h b/src/gallium/auxiliary/util/u_fifo.h
new file mode 100644
index 0000000000..9e007de1ad
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_fifo.h
@@ -0,0 +1,94 @@
+/**************************************************************************
+ *
+ * Copyright © 2009 Jakob Bornecrantz
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#ifndef U_FIFO_H
+#define U_FIFO_H
+
+#include "util/u_memory.h"
+
+struct util_fifo
+{
+   size_t head;
+   size_t tail;
+   size_t num;
+   size_t size;
+};
+
+static INLINE struct util_fifo *
+u_fifo_create(size_t size)
+{
+   struct util_fifo *fifo;
+   fifo = MALLOC(sizeof(*fifo) + size * sizeof(void*));
+
+   fifo->head = 0;
+   fifo->tail = 0;
+   fifo->num = 0;
+   fifo->size = size;
+
+   return fifo;
+}
+
+static INLINE boolean
+u_fifo_add(struct util_fifo *fifo, void *ptr)
+{
+   void **array = (void**)&fifo[1];
+   if (fifo->num >= fifo->size)
+      return FALSE;
+
+   if (++fifo->head >= fifo->size)
+      fifo->head = 0;
+
+   array[fifo->head] = ptr;
+
+   ++fifo->num;
+
+   return TRUE;
+}
+
+static INLINE boolean
+u_fifo_pop(struct util_fifo *fifo, void **ptr)
+{
+   void **array = (void**)&fifo[1];
+
+   if (!fifo->num)
+      return FALSE;
+
+   if (++fifo->tail >= fifo->size)
+      fifo->tail = 0;
+
+   *ptr = array[fifo->tail];
+
+   ++fifo->num;
+
+   return TRUE;
+}
+
+static INLINE void
+u_fifo_destroy(struct util_fifo *fifo)
+{
+   FREE(fifo);
+}
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index 00a46d0cc4..f1bf94f17d 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -2,7 +2,7 @@ PIPE_FORMAT_A8R8G8B8_UNORM        , arith , 1, 1, un8 , un8 , un8 , un8 , zyxw,
 PIPE_FORMAT_X8R8G8B8_UNORM        , arith , 1, 1, un8 , un8 , un8 , un8 , zyx1, rgb
 PIPE_FORMAT_B8G8R8A8_UNORM        , arith , 1, 1, un8 , un8 , un8 , un8 , yzwx, rgb
 PIPE_FORMAT_B8G8R8X8_UNORM        , arith , 1, 1, un8 , un8 , un8 , un8 , yzw1, rgb
-PIPE_FORMAT_A1R5G5B5_UNORM        , arith , 1, 1, un1 , un5 , un5 , un5 , zyxw, rgb
+PIPE_FORMAT_A1R5G5B5_UNORM        , arith , 1, 1, un5 , un5 , un5 , un1 , zyxw, rgb
 PIPE_FORMAT_A4R4G4B4_UNORM        , arith , 1, 1, un4 , un4 , un4 , un4 , zyxw, rgb
 PIPE_FORMAT_R5G6B5_UNORM          , arith , 1, 1, un5 , un6 , un5 ,     , zyx1, rgb
 PIPE_FORMAT_A2B10G10R10_UNORM     , arith , 1, 1, un10, un10, un10, un2 , xyzw, rgb
@@ -14,10 +14,10 @@ PIPE_FORMAT_L16_UNORM             , arith , 1, 1, un16,     ,     ,     , xxx1,
 PIPE_FORMAT_Z16_UNORM             , array , 1, 1, un16,     ,     ,     , x___, zs 
 PIPE_FORMAT_Z32_UNORM             , array , 1, 1, un32,     ,     ,     , x___, zs 
 PIPE_FORMAT_Z32_FLOAT             , array , 1, 1, f32 ,     ,     ,     , x___, zs 
-PIPE_FORMAT_S8Z24_UNORM           , arith , 1, 1, un8 , un24,     ,     , yx__, zs 
-PIPE_FORMAT_Z24S8_UNORM           , arith , 1, 1, un24, un8 ,     ,     , xy__, zs 
-PIPE_FORMAT_X8Z24_UNORM           , arith , 1, 1, un8 , un24,     ,     , y___, zs 
-PIPE_FORMAT_Z24X8_UNORM           , arith , 1, 1, un24, un8 ,     ,     , x___, zs 
+PIPE_FORMAT_S8Z24_UNORM           , arith , 1, 1, un24, un8 ,     ,     , xy__, zs 
+PIPE_FORMAT_Z24S8_UNORM           , arith , 1, 1, un8 , un24,     ,     , yx__, zs 
+PIPE_FORMAT_X8Z24_UNORM           , arith , 1, 1, un24, un8 ,     ,     , x___, zs 
+PIPE_FORMAT_Z24X8_UNORM           , arith , 1, 1, un8 , un24,     ,     , y___, zs 
 PIPE_FORMAT_S8_UNORM              , array , 1, 1, un8 ,     ,     ,     , _x__, zs 
 PIPE_FORMAT_R64_FLOAT             , array , 1, 1, f64 ,     ,     ,     , x001, rgb
 PIPE_FORMAT_R64G64_FLOAT          , array , 1, 1, f64 , f64 ,     ,     , xy01, rgb
diff --git a/src/gallium/auxiliary/util/u_hash_table.c b/src/gallium/auxiliary/util/u_hash_table.c
index 8c2a8f454c..5604e3ac37 100644
--- a/src/gallium/auxiliary/util/u_hash_table.c
+++ b/src/gallium/auxiliary/util/u_hash_table.c
@@ -47,7 +47,7 @@
 #include "util/u_hash_table.h"
 
 
-struct hash_table
+struct util_hash_table
 {
    struct cso_hash *cso;   
    
@@ -61,27 +61,27 @@ struct hash_table
 };
 
 
-struct hash_table_item
+struct util_hash_table_item
 {
    void *key;
    void *value;
 };
 
 
-static INLINE struct hash_table_item *
-hash_table_item(struct cso_hash_iter iter)
+static INLINE struct util_hash_table_item *
+util_hash_table_item(struct cso_hash_iter iter)
 {
-   return (struct hash_table_item *)cso_hash_iter_data(iter);
+   return (struct util_hash_table_item *)cso_hash_iter_data(iter);
 }
 
 
-struct hash_table *
-hash_table_create(unsigned (*hash)(void *key),
-                  int (*compare)(void *key1, void *key2))
+struct util_hash_table *
+util_hash_table_create(unsigned (*hash)(void *key),
+                       int (*compare)(void *key1, void *key2))
 {
-   struct hash_table *ht;
+   struct util_hash_table *ht;
    
-   ht = MALLOC_STRUCT(hash_table);
+   ht = MALLOC_STRUCT(util_hash_table);
    if(!ht)
       return NULL;
    
@@ -99,16 +99,16 @@ hash_table_create(unsigned (*hash)(void *key),
 
 
 static INLINE struct cso_hash_iter
-hash_table_find_iter(struct hash_table *ht,
-                     void *key, 
-                     unsigned key_hash)
+util_hash_table_find_iter(struct util_hash_table *ht,
+                          void *key,
+                          unsigned key_hash)
 {
    struct cso_hash_iter iter;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
    
    iter = cso_hash_find(ht->cso, key_hash);
    while (!cso_hash_iter_is_null(iter)) {
-      item = (struct hash_table_item *)cso_hash_iter_data(iter);
+      item = (struct util_hash_table_item *)cso_hash_iter_data(iter);
       if (!ht->compare(item->key, key))
          break;
       iter = cso_hash_iter_next(iter);
@@ -118,17 +118,17 @@ hash_table_find_iter(struct hash_table *ht,
 }
 
 
-static INLINE struct hash_table_item *
-hash_table_find_item(struct hash_table *ht,
-                     void *key, 
-                     unsigned key_hash)
+static INLINE struct util_hash_table_item *
+util_hash_table_find_item(struct util_hash_table *ht,
+                          void *key,
+                          unsigned key_hash)
 {
    struct cso_hash_iter iter;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
    
    iter = cso_hash_find(ht->cso, key_hash);
    while (!cso_hash_iter_is_null(iter)) {
-      item = (struct hash_table_item *)cso_hash_iter_data(iter);
+      item = (struct util_hash_table_item *)cso_hash_iter_data(iter);
       if (!ht->compare(item->key, key))
          return item;
       iter = cso_hash_iter_next(iter);
@@ -139,12 +139,12 @@ hash_table_find_item(struct hash_table *ht,
 
 
 enum pipe_error
-hash_table_set(struct hash_table *ht,
-               void *key,
-               void *value)
+util_hash_table_set(struct util_hash_table *ht,
+                    void *key,
+                    void *value)
 {
    unsigned key_hash;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
    struct cso_hash_iter iter;
 
    assert(ht);
@@ -153,14 +153,14 @@ hash_table_set(struct hash_table *ht,
 
    key_hash = ht->hash(key);
 
-   item = hash_table_find_item(ht, key, key_hash);
+   item = util_hash_table_find_item(ht, key, key_hash);
    if(item) {
       /* TODO: key/value destruction? */
       item->value = value;
       return PIPE_OK;
    }
    
-   item = MALLOC_STRUCT(hash_table_item);
+   item = MALLOC_STRUCT(util_hash_table_item);
    if(!item)
       return PIPE_ERROR_OUT_OF_MEMORY;
    
@@ -178,11 +178,11 @@ hash_table_set(struct hash_table *ht,
 
 
 void *
-hash_table_get(struct hash_table *ht, 
-               void *key)
+util_hash_table_get(struct util_hash_table *ht,
+                    void *key)
 {
    unsigned key_hash;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
 
    assert(ht);
    if (!ht)
@@ -190,7 +190,7 @@ hash_table_get(struct hash_table *ht,
 
    key_hash = ht->hash(key);
 
-   item = hash_table_find_item(ht, key, key_hash);
+   item = util_hash_table_find_item(ht, key, key_hash);
    if(!item)
       return NULL;
    
@@ -199,12 +199,12 @@ hash_table_get(struct hash_table *ht,
 
 
 void
-hash_table_remove(struct hash_table *ht, 
-                  void *key)
+util_hash_table_remove(struct util_hash_table *ht,
+                       void *key)
 {
    unsigned key_hash;
    struct cso_hash_iter iter;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
 
    assert(ht);
    if (!ht)
@@ -212,11 +212,11 @@ hash_table_remove(struct hash_table *ht,
 
    key_hash = ht->hash(key);
 
-   iter = hash_table_find_iter(ht, key, key_hash);
+   iter = util_hash_table_find_iter(ht, key, key_hash);
    if(cso_hash_iter_is_null(iter))
       return;
    
-   item = hash_table_item(iter);
+   item = util_hash_table_item(iter);
    assert(item);
    FREE(item);
    
@@ -225,10 +225,10 @@ hash_table_remove(struct hash_table *ht,
 
 
 void 
-hash_table_clear(struct hash_table *ht)
+util_hash_table_clear(struct util_hash_table *ht)
 {
    struct cso_hash_iter iter;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
 
    assert(ht);
    if (!ht)
@@ -236,7 +236,7 @@ hash_table_clear(struct hash_table *ht)
 
    iter = cso_hash_first_node(ht->cso);
    while (!cso_hash_iter_is_null(iter)) {
-      item = (struct hash_table_item *)cso_hash_take(ht->cso, cso_hash_iter_key(iter));
+      item = (struct util_hash_table_item *)cso_hash_take(ht->cso, cso_hash_iter_key(iter));
       FREE(item);
       iter = cso_hash_first_node(ht->cso);
    }
@@ -244,12 +244,13 @@ hash_table_clear(struct hash_table *ht)
 
 
 enum pipe_error
-hash_table_foreach(struct hash_table *ht,
-                   enum pipe_error (*callback)(void *key, void *value, void *data),
-                   void *data)
+util_hash_table_foreach(struct util_hash_table *ht,
+                     enum pipe_error (*callback)
+                        (void *key, void *value, void *data),
+                     void *data)
 {
    struct cso_hash_iter iter;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
    enum pipe_error result;
 
    assert(ht);
@@ -258,7 +259,7 @@ hash_table_foreach(struct hash_table *ht,
 
    iter = cso_hash_first_node(ht->cso);
    while (!cso_hash_iter_is_null(iter)) {
-      item = (struct hash_table_item *)cso_hash_iter_data(iter);
+      item = (struct util_hash_table_item *)cso_hash_iter_data(iter);
       result = callback(item->key, item->value, data);
       if(result != PIPE_OK)
 	 return result;
@@ -270,10 +271,10 @@ hash_table_foreach(struct hash_table *ht,
 
 
 void
-hash_table_destroy(struct hash_table *ht)
+util_hash_table_destroy(struct util_hash_table *ht)
 {
    struct cso_hash_iter iter;
-   struct hash_table_item *item;
+   struct util_hash_table_item *item;
 
    assert(ht);
    if (!ht)
@@ -281,7 +282,7 @@ hash_table_destroy(struct hash_table *ht)
 
    iter = cso_hash_first_node(ht->cso);
    while (!cso_hash_iter_is_null(iter)) {
-      item = (struct hash_table_item *)cso_hash_iter_data(iter);
+      item = (struct util_hash_table_item *)cso_hash_iter_data(iter);
       FREE(item);
       iter = cso_hash_iter_next(iter);
    }
diff --git a/src/gallium/auxiliary/util/u_hash_table.h b/src/gallium/auxiliary/util/u_hash_table.h
index feee881582..258a31aec8 100644
--- a/src/gallium/auxiliary/util/u_hash_table.h
+++ b/src/gallium/auxiliary/util/u_hash_table.h
@@ -46,7 +46,7 @@ extern "C" {
 /**
  * Generic purpose hash table.
  */
-struct hash_table;
+struct util_hash_table;
 
 
 /**
@@ -55,37 +55,38 @@ struct hash_table;
  * @param hash hash function
  * @param compare should return 0 for two equal keys.
  */
-struct hash_table *
-hash_table_create(unsigned (*hash)(void *key),
-                  int (*compare)(void *key1, void *key2));
+struct util_hash_table *
+util_hash_table_create(unsigned (*hash)(void *key),
+                       int (*compare)(void *key1, void *key2));
 
 
 enum pipe_error
-hash_table_set(struct hash_table *ht,
-               void *key,
-               void *value);
+util_hash_table_set(struct util_hash_table *ht,
+                    void *key,
+                    void *value);
 
 void *
-hash_table_get(struct hash_table *ht, 
-               void *key);
+util_hash_table_get(struct util_hash_table *ht,
+                    void *key);
 
 
 void
-hash_table_remove(struct hash_table *ht, 
-                  void *key);
+util_hash_table_remove(struct util_hash_table *ht,
+                       void *key);
 
 
 void
-hash_table_clear(struct hash_table *ht);
+util_hash_table_clear(struct util_hash_table *ht);
 
 
 enum pipe_error
-hash_table_foreach(struct hash_table *ht,
-                   enum pipe_error (*callback)(void *key, void *value, void *data),
-                   void *data);
+util_hash_table_foreach(struct util_hash_table *ht,
+                        enum pipe_error (*callback)
+                        (void *key, void *value, void *data),
+                        void *data);
 
 void
-hash_table_destroy(struct hash_table *ht);
+util_hash_table_destroy(struct util_hash_table *ht);
 
 
 #ifdef __cplusplus
diff --git a/src/gallium/auxiliary/util/u_keymap.c b/src/gallium/auxiliary/util/u_keymap.c
index 508a2ee063..f856395ca9 100644
--- a/src/gallium/auxiliary/util/u_keymap.c
+++ b/src/gallium/auxiliary/util/u_keymap.c
@@ -28,7 +28,7 @@
 /**
  * Key lookup/associative container.
  *
- * Like Jose's u_hash_table, based on CSO cache code for now.
+ * Like Jose's util_hash_table, based on CSO cache code for now.
  *
  * Author: Brian Paul
  */
diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h
index 4c6c2bc00e..75b075f160 100644
--- a/src/gallium/auxiliary/util/u_math.h
+++ b/src/gallium/auxiliary/util/u_math.h
@@ -283,6 +283,14 @@ util_fast_pow(float x, float y)
    return util_fast_exp2(util_fast_log2(x) * y);
 }
 
+/* Note that this counts zero as a power of two.
+ */
+static INLINE boolean
+util_is_power_of_two( unsigned v )
+{
+   return (v & (v-1)) == 0;
+}
+
 
 /**
  * Floor(x), returned as int.
@@ -341,10 +349,22 @@ util_is_inf_or_nan(float x)
 
 
 /**
+ * Test whether x is a power of two.
+ */
+static INLINE boolean
+util_is_pot(unsigned x)
+{
+   return (x & (x - 1)) == 0;
+}
+
+
+/**
  * Find first bit set in word.  Least significant bit is 1.
  * Return 0 if no bits set.
  */
-#if defined(_MSC_VER) && _MSC_VER >= 1300
+#if defined(_MSC_VER) && _MSC_VER >= 1300 && (_M_IX86 || _M_AMD64 || _M_IA64)
+unsigned char _BitScanForward(unsigned long* Index, unsigned long Mask);
+#pragma intrinsic(_BitScanForward)
 static INLINE
 unsigned long ffs( unsigned long u )
 {
@@ -451,6 +471,26 @@ util_logbase2(unsigned n)
 
 
 /**
+ * Returns the smallest power of two >= x
+ */
+static INLINE unsigned
+util_next_power_of_two(unsigned x)
+{
+   unsigned i;
+
+   if (x == 0)
+      return 1;
+
+   --x;
+
+   for (i = 1; i < sizeof(unsigned) * 8; i <<= 1)
+      x |= x >> i;
+
+   return x + 1;
+}
+
+
+/**
  * Clamp X to [MIN, MAX].
  * This is a macro to allow float, int, uint, etc. types.
  */
diff --git a/src/gallium/auxiliary/util/u_network.c b/src/gallium/auxiliary/util/u_network.c
index bc4b758406..6269c72e12 100644
--- a/src/gallium/auxiliary/util/u_network.c
+++ b/src/gallium/auxiliary/util/u_network.c
@@ -6,7 +6,7 @@
 #if defined(PIPE_SUBSYSTEM_WINDOWS_USER)
 #  include <winsock2.h>
 #  include <windows.h>
-#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU)
+#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD)
 #  include <sys/socket.h>
 #  include <netinet/in.h>
 #  include <unistd.h>
@@ -54,7 +54,7 @@ u_socket_close(int s)
    if (s < 0)
       return;
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD)
    shutdown(s, SHUT_RDWR);
    close(s);
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
@@ -169,7 +169,7 @@ u_socket_listen_on_port(uint16_t portnum)
 void
 u_socket_block(int s, boolean block)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD)
    int old = fcntl(s, F_GETFL, 0);
    if (old == -1)
       return;
diff --git a/src/gallium/auxiliary/util/u_network.h b/src/gallium/auxiliary/util/u_network.h
index 8c778f492c..0aa898b967 100644
--- a/src/gallium/auxiliary/util/u_network.h
+++ b/src/gallium/auxiliary/util/u_network.h
@@ -6,7 +6,7 @@
 
 #if defined(PIPE_SUBSYSTEM_WINDOWS_USER)
 #  define PIPE_HAVE_SOCKETS
-#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU)
+#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD)
 #  define PIPE_HAVE_SOCKETS
 #endif
 
diff --git a/src/gallium/auxiliary/util/u_simple_screen.c b/src/gallium/auxiliary/util/u_simple_screen.c
index f01296b40f..5238299015 100644
--- a/src/gallium/auxiliary/util/u_simple_screen.c
+++ b/src/gallium/auxiliary/util/u_simple_screen.c
@@ -52,8 +52,7 @@ pass_user_buffer_create(struct pipe_screen *screen,
                         unsigned bytes)
 {
    struct pipe_buffer *buffer =
-      screen->winsys->user_buffer_create(screen->winsys,
-                                             ptr, bytes);
+      screen->winsys->user_buffer_create(screen->winsys, ptr, bytes);
 
    buffer->screen = screen;
 
@@ -69,9 +68,8 @@ pass_surface_buffer_create(struct pipe_screen *screen,
                            unsigned *stride)
 {
    struct pipe_buffer *buffer =
-      screen->winsys->surface_buffer_create(screen->winsys,
-                                                width, height,
-                                                format, usage, tex_usage, stride);
+      screen->winsys->surface_buffer_create(screen->winsys, width, height,
+                                            format, usage, tex_usage, stride);
 
    buffer->screen = screen;
 
@@ -83,8 +81,7 @@ pass_buffer_map(struct pipe_screen *screen,
                 struct pipe_buffer *buf,
                 unsigned usage)
 {
-   return screen->winsys->buffer_map(screen->winsys,
-                                     buf, usage);
+   return screen->winsys->buffer_map(screen->winsys, buf, usage);
 }
 
 static void
@@ -106,8 +103,7 @@ pass_flush_frontbuffer(struct pipe_screen *screen,
                        struct pipe_surface *surf,
                        void *context_private)
 {
-   screen->winsys->flush_frontbuffer(screen->winsys,
-                                     surf, context_private);
+   screen->winsys->flush_frontbuffer(screen->winsys, surf, context_private);
 }
 
 static void
@@ -115,8 +111,7 @@ pass_fence_reference(struct pipe_screen *screen,
                      struct pipe_fence_handle **ptr,
                      struct pipe_fence_handle *fence)
 {
-   screen->winsys->fence_reference(screen->winsys,
-                                   ptr, fence);
+   screen->winsys->fence_reference(screen->winsys, ptr, fence);
 }
 
 static int
@@ -124,8 +119,7 @@ pass_fence_signalled(struct pipe_screen *screen,
                      struct pipe_fence_handle *fence,
                      unsigned flag)
 {
-   return screen->winsys->fence_signalled(screen->winsys,
-                                          fence, flag);
+   return screen->winsys->fence_signalled(screen->winsys, fence, flag);
 }
 
 static int
@@ -133,11 +127,11 @@ pass_fence_finish(struct pipe_screen *screen,
                   struct pipe_fence_handle *fence,
                   unsigned flag)
 {
-   return screen->winsys->fence_finish(screen->winsys,
-                                       fence, flag);
+   return screen->winsys->fence_finish(screen->winsys, fence, flag);
 }
 
-void u_simple_screen_init(struct pipe_screen *screen)
+void
+u_simple_screen_init(struct pipe_screen *screen)
 {
    screen->buffer_create = pass_buffer_create;
    screen->user_buffer_create = pass_user_buffer_create;
@@ -152,7 +146,8 @@ void u_simple_screen_init(struct pipe_screen *screen)
    screen->fence_finish = pass_fence_finish;
 }
 
-const char* u_simple_screen_winsys_name(struct pipe_screen *screen)
+const char *
+u_simple_screen_winsys_name(struct pipe_screen *screen)
 {
    return screen->winsys->get_name(screen->winsys);
 }
diff --git a/src/gallium/auxiliary/util/u_simple_shaders.c b/src/gallium/auxiliary/util/u_simple_shaders.c
index ab754296fa..1c8b157d91 100644
--- a/src/gallium/auxiliary/util/u_simple_shaders.c
+++ b/src/gallium/auxiliary/util/u_simple_shaders.c
@@ -34,14 +34,8 @@
 
 
 #include "pipe/p_context.h"
-#include "util/u_debug.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_screen.h"
 #include "pipe/p_shader_tokens.h"
-
-#include "util/u_memory.h"
 #include "util/u_simple_shaders.h"
-
 #include "tgsi/tgsi_ureg.h"
 
 
@@ -67,9 +61,7 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
       struct ureg_src src;
       struct ureg_dst dst;
 
-      src = ureg_DECL_vs_input( ureg,
-                                semantic_names[i],
-                                semantic_indexes[i]);
+      src = ureg_DECL_vs_input( ureg, i );
       
       dst = ureg_DECL_output( ureg,
                               semantic_names[i],
@@ -116,7 +108,15 @@ util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
                            TGSI_SEMANTIC_COLOR,
                            0 );
 
-   ureg_TEX( ureg, out, TGSI_TEXTURE_2D, tex, sampler );
+   if (writemask != TGSI_WRITEMASK_XYZW) {
+      struct ureg_src imm = ureg_imm4f( ureg, 0, 0, 0, 1 );
+
+      ureg_MOV( ureg, out, imm );
+   }
+
+   ureg_TEX( ureg, 
+             ureg_writemask(out, writemask),
+             TGSI_TEXTURE_2D, tex, sampler );
    ureg_END( ureg );
 
    return ureg_create_shader_and_destroy( ureg, pipe );
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 1235a67d26..8a22f584be 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -170,7 +170,7 @@ x8r8g8b8_get_tile_rgba(const unsigned *src,
          pRow[0] = ubyte_to_float((pixel >> 16) & 0xff);
          pRow[1] = ubyte_to_float((pixel >>  8) & 0xff);
          pRow[2] = ubyte_to_float((pixel >>  0) & 0xff);
-         pRow[3] = ubyte_to_float(0xff);
+         pRow[3] = 1.0F;
       }
       p += dst_stride;
    }
@@ -394,6 +394,52 @@ r5g6b5_put_tile_rgba(ushort *dst,
 
 
 
+/*** PIPE_FORMAT_R8G8B8_UNORM ***/
+
+static void
+r8g8b8_get_tile_rgba(const ubyte *src,
+                     unsigned w, unsigned h,
+                     float *p,
+                     unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         pRow[0] = ubyte_to_float(src[0]);
+         pRow[1] = ubyte_to_float(src[1]);
+         pRow[2] = ubyte_to_float(src[2]);
+         pRow[3] = 1.0f;
+         src += 3;
+      }
+      p += dst_stride;
+   }
+}
+
+
+static void
+r8g8b8_put_tile_rgba(ubyte *dst,
+                     unsigned w, unsigned h,
+                     const float *p,
+                     unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         dst[0] = float_to_ubyte(pRow[0]);
+         dst[1] = float_to_ubyte(pRow[1]);
+         dst[2] = float_to_ubyte(pRow[2]);
+         dst += 3;
+      }
+      p += src_stride;
+   }
+}
+
+
+
 /*** PIPE_FORMAT_Z16_UNORM ***/
 
 /**
@@ -1106,6 +1152,9 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
    case PIPE_FORMAT_R5G6B5_UNORM:
       r5g6b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
       break;
+   case PIPE_FORMAT_R8G8B8_UNORM:
+      r8g8b8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
+      break;
    case PIPE_FORMAT_L8_UNORM:
       l8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
       break;
@@ -1222,6 +1271,9 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
    case PIPE_FORMAT_R5G6B5_UNORM:
       r5g6b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
       break;
+   case PIPE_FORMAT_R8G8B8_UNORM:
+      r8g8b8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
+      break;
    case PIPE_FORMAT_R8G8B8A8_UNORM:
       assert(0);
       break;
@@ -1400,7 +1452,7 @@ pipe_put_tile_z(struct pipe_transfer *pt,
    case PIPE_FORMAT_S8Z24_UNORM:
       {
          uint *pDest = (uint *) (map + y * pt->stride + x*4);
-         assert(pt->usage == PIPE_TRANSFER_READ_WRITE);
+         assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);
          for (i = 0; i < h; i++) {
             for (j = 0; j < w; j++) {
                /* convert 32-bit Z to 24-bit Z, preserve stencil */
@@ -1427,7 +1479,7 @@ pipe_put_tile_z(struct pipe_transfer *pt,
    case PIPE_FORMAT_Z24S8_UNORM:
       {
          uint *pDest = (uint *) (map + y * pt->stride + x*4);
-         assert(pt->usage == PIPE_TRANSFER_READ_WRITE);
+         assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);
          for (i = 0; i < h; i++) {
             for (j = 0; j < w; j++) {
                /* convert 32-bit Z to 24-bit Z, preserve stencil */
diff --git a/src/gallium/auxiliary/vl/Makefile b/src/gallium/auxiliary/vl/Makefile
new file mode 100644
index 0000000000..4314c1e8d6
--- /dev/null
+++ b/src/gallium/auxiliary/vl/Makefile
@@ -0,0 +1,13 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = vl
+
+C_SOURCES = \
+	vl_bitstream_parser.c \
+	vl_mpeg12_mc_renderer.c \
+	vl_compositor.c \
+        vl_csc.c \
+	vl_shader_build.c
+
+include ../../Makefile.template
diff --git a/src/gallium/auxiliary/vl/SConscript b/src/gallium/auxiliary/vl/SConscript
new file mode 100644
index 0000000000..aed69f5efe
--- /dev/null
+++ b/src/gallium/auxiliary/vl/SConscript
@@ -0,0 +1,13 @@
+Import('*')
+
+vl = env.ConvenienceLibrary(
+	target = 'vl',
+	source = [
+		'vl_bitstream_parser.c',
+		'vl_mpeg12_mc_renderer.c',
+		'vl_compositor.c',
+                'vl_csc.c',
+		'vl_shader_build.c',
+	])
+
+auxiliaries.insert(0, vl)
diff --git a/src/gallium/auxiliary/vl/vl_bitstream_parser.c b/src/gallium/auxiliary/vl/vl_bitstream_parser.c
new file mode 100644
index 0000000000..3193ea5f41
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_bitstream_parser.c
@@ -0,0 +1,167 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "vl_bitstream_parser.h"
+#include <assert.h>
+#include <limits.h>
+#include <util/u_memory.h>
+
+static unsigned
+grab_bits(unsigned cursor, unsigned how_many_bits, unsigned bitstream_elt)
+{
+   unsigned excess_bits = sizeof(unsigned) * CHAR_BIT - how_many_bits - cursor;
+	
+   assert(cursor < sizeof(unsigned) * CHAR_BIT);
+   assert(how_many_bits > 0 && how_many_bits <= sizeof(unsigned) * CHAR_BIT);
+   assert(cursor + how_many_bits <= sizeof(unsigned) * CHAR_BIT);
+
+   return (bitstream_elt << excess_bits) >> (excess_bits + cursor);
+}
+
+static unsigned
+show_bits(unsigned cursor, unsigned how_many_bits, const unsigned *bitstream)
+{	
+   unsigned cur_int = cursor / (sizeof(unsigned) * CHAR_BIT);
+   unsigned cur_bit = cursor % (sizeof(unsigned) * CHAR_BIT);
+	
+   assert(bitstream);
+	
+   if (cur_bit + how_many_bits > sizeof(unsigned) * CHAR_BIT) {
+      unsigned lower = grab_bits(cur_bit, sizeof(unsigned) * CHAR_BIT - cur_bit,
+                                 bitstream[cur_int]);
+      unsigned upper = grab_bits(0, cur_bit + how_many_bits - sizeof(unsigned) * CHAR_BIT,
+                                 bitstream[cur_int + 1]);
+      return lower | upper << (sizeof(unsigned) * CHAR_BIT - cur_bit);
+   }
+   else
+      return grab_bits(cur_bit, how_many_bits, bitstream[cur_int]);
+}
+
+bool vl_bitstream_parser_init(struct vl_bitstream_parser *parser,
+                              unsigned num_bitstreams,
+                              const void **bitstreams,
+                              const unsigned *sizes)
+{
+   assert(parser);
+   assert(num_bitstreams);
+   assert(bitstreams);
+   assert(sizes);
+
+   parser->num_bitstreams = num_bitstreams;
+   parser->bitstreams = (const unsigned**)bitstreams;
+   parser->sizes = sizes;
+   parser->cur_bitstream = 0;
+   parser->cursor = 0;
+
+   return true;
+}
+
+void vl_bitstream_parser_cleanup(struct vl_bitstream_parser *parser)
+{
+   assert(parser);
+}
+
+unsigned
+vl_bitstream_parser_get_bits(struct vl_bitstream_parser *parser,
+                             unsigned how_many_bits)
+{
+   unsigned bits;
+
+   assert(parser);
+
+   bits = vl_bitstream_parser_show_bits(parser, how_many_bits);
+
+   vl_bitstream_parser_forward(parser, how_many_bits);
+
+   return bits;
+}
+
+unsigned
+vl_bitstream_parser_show_bits(struct vl_bitstream_parser *parser,
+                              unsigned how_many_bits)
+{	
+   unsigned bits = 0;
+   unsigned shift = 0;
+   unsigned cursor;
+   unsigned cur_bitstream;
+
+   assert(parser);
+
+   cursor = parser->cursor;
+   cur_bitstream = parser->cur_bitstream;
+
+   while (1) {
+      unsigned bits_left = parser->sizes[cur_bitstream] * CHAR_BIT - cursor;
+      unsigned bits_to_show = how_many_bits > bits_left ? bits_left : how_many_bits;
+
+      bits |= show_bits(cursor, bits_to_show,
+                        parser->bitstreams[cur_bitstream]) << shift;
+		
+      if (how_many_bits > bits_to_show) {
+         how_many_bits -= bits_to_show;
+         cursor = 0;
+         ++cur_bitstream;
+         shift += bits_to_show;
+      }
+      else
+         break;
+   }
+
+   return bits;
+}
+
+void vl_bitstream_parser_forward(struct vl_bitstream_parser *parser,
+                                 unsigned how_many_bits)
+{
+   assert(parser);
+   assert(how_many_bits);
+
+   parser->cursor += how_many_bits;
+
+   while (parser->cursor > parser->sizes[parser->cur_bitstream] * CHAR_BIT) {
+      parser->cursor -= parser->sizes[parser->cur_bitstream++] * CHAR_BIT;
+      assert(parser->cur_bitstream < parser->num_bitstreams);
+   }
+}
+
+void vl_bitstream_parser_rewind(struct vl_bitstream_parser *parser,
+                                unsigned how_many_bits)
+{
+   signed c;
+	
+   assert(parser);
+   assert(how_many_bits);
+	
+   c = parser->cursor - how_many_bits;
+
+   while (c < 0) {
+      c += parser->sizes[parser->cur_bitstream--] * CHAR_BIT;
+      assert(parser->cur_bitstream < parser->num_bitstreams);
+   }
+
+   parser->cursor = (unsigned)c;
+}
diff --git a/src/gallium/drivers/softpipe/sp_quad_occlusion.c b/src/gallium/auxiliary/vl/vl_bitstream_parser.h
index dfa7ff3b1d..30ec743fa7 100644
--- a/src/gallium/drivers/softpipe/sp_quad_occlusion.c
+++ b/src/gallium/auxiliary/vl/vl_bitstream_parser.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2009 Younes Manton.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,61 +25,39 @@
  * 
  **************************************************************************/
 
+#ifndef vl_bitstream_parser_h
+#define vl_bitstream_parser_h
 
-/**
- * \brief  Quad occlusion counter stage
- * \author  Brian Paul
- */
+#include "pipe/p_compiler.h"
 
-
-#include "pipe/p_defines.h"
-#include "util/u_memory.h"
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_surface.h"
-#include "sp_quad_pipe.h"
-
-static unsigned count_bits( unsigned val )
-{
-   unsigned i;
-
-   for (i = 0; val ; val >>= 1)
-      i += (val & 1);
-
-   return i;
-}
-
-static void
-occlusion_count_quad(struct quad_stage *qs, struct quad_header *quad)
+struct vl_bitstream_parser
 {
-   struct softpipe_context *softpipe = qs->softpipe;
+   unsigned num_bitstreams;
+   const unsigned **bitstreams;
+   const unsigned *sizes;
+   unsigned cur_bitstream;
+   unsigned cursor;
+};
 
-   softpipe->occlusion_count += count_bits(quad->inout.mask);
+bool vl_bitstream_parser_init(struct vl_bitstream_parser *parser,
+                              unsigned num_bitstreams,
+                              const void **bitstreams,
+                              const unsigned *sizes);
 
-   qs->next->run(qs->next, quad);
-}
+void vl_bitstream_parser_cleanup(struct vl_bitstream_parser *parser);
 
+unsigned
+vl_bitstream_parser_get_bits(struct vl_bitstream_parser *parser,
+                             unsigned how_many_bits);
 
-static void occlusion_begin(struct quad_stage *qs)
-{
-   qs->next->begin(qs->next);
-}
-
-
-static void occlusion_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
+unsigned
+vl_bitstream_parser_show_bits(struct vl_bitstream_parser *parser,
+                              unsigned how_many_bits);
 
-struct quad_stage *sp_quad_occlusion_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
+void vl_bitstream_parser_forward(struct vl_bitstream_parser *parser,
+                                 unsigned how_many_bits);
 
-   stage->softpipe = softpipe;
-   stage->begin = occlusion_begin;
-   stage->run = occlusion_count_quad;
-   stage->destroy = occlusion_destroy;
+void vl_bitstream_parser_rewind(struct vl_bitstream_parser *parser,
+                                unsigned how_many_bits);
 
-   return stage;
-}
+#endif /* vl_bitstream_parser_h */
diff --git a/src/gallium/auxiliary/vl/vl_compositor.c b/src/gallium/auxiliary/vl/vl_compositor.c
new file mode 100644
index 0000000000..b36dbeb208
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_compositor.c
@@ -0,0 +1,532 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "vl_compositor.h"
+#include <assert.h>
+#include <pipe/p_context.h>
+#include <pipe/p_inlines.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+#include <util/u_memory.h>
+#include "vl_csc.h"
+#include "vl_shader_build.h"
+
+struct vertex2f
+{
+   float x, y;
+};
+
+struct vertex4f
+{
+   float x, y, z, w;
+};
+
+struct vertex_shader_consts
+{
+   struct vertex4f dst_scale;
+   struct vertex4f dst_trans;
+   struct vertex4f src_scale;
+   struct vertex4f src_trans;
+};
+
+struct fragment_shader_consts
+{
+   float matrix[16];
+};
+
+/*
+ * Represents 2 triangles in a strip in normalized coords.
+ * Used to render the surface onto the frame buffer.
+ */
+static const struct vertex2f surface_verts[4] =
+{
+   {0.0f, 0.0f},
+   {0.0f, 1.0f},
+   {1.0f, 0.0f},
+   {1.0f, 1.0f}
+};
+
+/*
+ * Represents texcoords for the above. We can use the position values directly.
+ * TODO: Duplicate these in the shader, no need to create a buffer.
+ */
+static const struct vertex2f *surface_texcoords = surface_verts;
+
+static void
+create_vert_shader(struct vl_compositor *c)
+{
+   const unsigned max_tokens = 50;
+
+   struct pipe_shader_state vs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(c);
+
+   tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header*)&tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+   ti = 3;
+
+   /*
+    * decl i0             ; Vertex pos
+    * decl i1             ; Vertex texcoords
+    */
+   for (i = 0; i < 2; i++) {
+      decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * decl c0             ; Scaling vector to scale vertex pos rect to destination size
+    * decl c1             ; Translation vector to move vertex pos rect into position
+    * decl c2             ; Scaling vector to scale texcoord rect to source size
+    * decl c3             ; Translation vector to move texcoord rect into position
+    */
+   decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * decl o0             ; Vertex pos
+    * decl o1             ; Vertex texcoords
+    */
+   for (i = 0; i < 2; i++) {
+      decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* decl t0, t1 */
+   decl = vl_decl_temps(0, 1);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * mad o0, i0, c0, c1  ; Scale and translate unit output rect to destination size and pos
+    * mad o1, i1, c2, c3  ; Scale and translate unit texcoord rect to source size and pos
+    */
+   for (i = 0; i < 2; ++i) {
+      inst = vl_inst4(TGSI_OPCODE_MAD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   vs.tokens = tokens;
+   c->vertex_shader = c->pipe->create_vs_state(c->pipe, &vs);
+   FREE(tokens);
+}
+
+static void
+create_frag_shader(struct vl_compositor *c)
+{
+   const unsigned max_tokens = 50;
+
+   struct pipe_shader_state fs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(c);
+
+   tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header*)&tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+   ti = 3;
+
+   /* decl i0             ; Texcoords for s0 */
+   decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * decl c0-c3          ; CSC matrix c0-c3
+    */
+   decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl o0             ; Fragment color */
+   decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl t0 */
+   decl = vl_decl_temps(0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl s0             ; Sampler for tex containing picture to display */
+   decl = vl_decl_samplers(0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* tex2d t0, i0, s0    ; Read src pixel */
+   inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * dp4 o0.x, t0, c0    ; Multiply pixel by the color conversion matrix
+    * dp4 o0.y, t0, c1
+    * dp4 o0.z, t0, c2
+    * dp4 o0.w, t0, c3
+    */
+   for (i = 0; i < 4; ++i) {
+      inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i);
+      inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	
+   assert(ti <= max_tokens);
+
+   fs.tokens = tokens;
+   c->fragment_shader = c->pipe->create_fs_state(c->pipe, &fs);
+   FREE(tokens);
+}
+
+static bool
+init_pipe_state(struct vl_compositor *c)
+{
+   struct pipe_sampler_state sampler;
+
+   assert(c);
+
+   c->fb_state.nr_cbufs = 1;
+   c->fb_state.zsbuf = NULL;
+
+   sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+   sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+   sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+   sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR;
+   sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+   sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR;
+   sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
+   sampler.compare_func = PIPE_FUNC_ALWAYS;
+   sampler.normalized_coords = 1;
+   /*sampler.prefilter = ;*/
+   /*sampler.lod_bias = ;*/
+   /*sampler.min_lod = ;*/
+   /*sampler.max_lod = ;*/
+   /*sampler.border_color[i] = ;*/
+   /*sampler.max_anisotropy = ;*/
+   c->sampler = c->pipe->create_sampler_state(c->pipe, &sampler);
+	
+   return true;
+}
+
+static void cleanup_pipe_state(struct vl_compositor *c)
+{
+   assert(c);
+	
+   c->pipe->delete_sampler_state(c->pipe, c->sampler);
+}
+
+static bool
+init_shaders(struct vl_compositor *c)
+{
+   assert(c);
+
+   create_vert_shader(c);
+   create_frag_shader(c);
+
+   return true;
+}
+
+static void cleanup_shaders(struct vl_compositor *c)
+{
+   assert(c);
+	
+   c->pipe->delete_vs_state(c->pipe, c->vertex_shader);
+   c->pipe->delete_fs_state(c->pipe, c->fragment_shader);
+}
+
+static bool
+init_buffers(struct vl_compositor *c)
+{
+   struct fragment_shader_consts fsc;
+
+   assert(c);
+	
+   /*
+    * Create our vertex buffer and vertex buffer element
+    * VB contains 4 vertices that render a quad covering the entire window
+    * to display a rendered surface
+    * Quad is rendered as a tri strip
+    */
+   c->vertex_bufs[0].stride = sizeof(struct vertex2f);
+   c->vertex_bufs[0].max_index = 3;
+   c->vertex_bufs[0].buffer_offset = 0;
+   c->vertex_bufs[0].buffer = pipe_buffer_create
+   (
+      c->pipe->screen,
+      1,
+      PIPE_BUFFER_USAGE_VERTEX,
+      sizeof(struct vertex2f) * 4
+   );
+
+   memcpy
+   (
+      pipe_buffer_map(c->pipe->screen, c->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+      surface_verts,
+      sizeof(struct vertex2f) * 4
+   );
+
+   pipe_buffer_unmap(c->pipe->screen, c->vertex_bufs[0].buffer);
+
+   c->vertex_elems[0].src_offset = 0;
+   c->vertex_elems[0].vertex_buffer_index = 0;
+   c->vertex_elems[0].nr_components = 2;
+   c->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /*
+    * Create our texcoord buffer and texcoord buffer element
+    * Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices
+    */
+   c->vertex_bufs[1].stride = sizeof(struct vertex2f);
+   c->vertex_bufs[1].max_index = 3;
+   c->vertex_bufs[1].buffer_offset = 0;
+   c->vertex_bufs[1].buffer = pipe_buffer_create
+   (
+      c->pipe->screen,
+      1,
+      PIPE_BUFFER_USAGE_VERTEX,
+      sizeof(struct vertex2f) * 4
+   );
+
+   memcpy
+   (
+      pipe_buffer_map(c->pipe->screen, c->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+      surface_texcoords,
+      sizeof(struct vertex2f) * 4
+   );
+
+   pipe_buffer_unmap(c->pipe->screen, c->vertex_bufs[1].buffer);
+
+   c->vertex_elems[1].src_offset = 0;
+   c->vertex_elems[1].vertex_buffer_index = 1;
+   c->vertex_elems[1].nr_components = 2;
+   c->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /*
+    * Create our vertex shader's constant buffer
+    * Const buffer contains scaling and translation vectors
+    */
+   c->vs_const_buf.buffer = pipe_buffer_create
+   (
+      c->pipe->screen,
+      1,
+      PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD,
+      sizeof(struct vertex_shader_consts)
+   );
+
+   /*
+    * Create our fragment shader's constant buffer
+    * Const buffer contains the color conversion matrix and bias vectors
+    */
+   c->fs_const_buf.buffer = pipe_buffer_create
+   (
+      c->pipe->screen,
+      1,
+      PIPE_BUFFER_USAGE_CONSTANT,
+      sizeof(struct fragment_shader_consts)
+   );
+
+   vl_csc_get_matrix(VL_CSC_COLOR_STANDARD_IDENTITY, NULL, true, fsc.matrix);
+
+   vl_compositor_set_csc_matrix(c, fsc.matrix);
+
+   return true;
+}
+
+static void
+cleanup_buffers(struct vl_compositor *c)
+{
+   unsigned i;
+
+   assert(c);
+	
+   for (i = 0; i < 2; ++i)
+      pipe_buffer_reference(&c->vertex_bufs[i].buffer, NULL);
+
+   pipe_buffer_reference(&c->vs_const_buf.buffer, NULL);
+   pipe_buffer_reference(&c->fs_const_buf.buffer, NULL);
+}
+
+bool vl_compositor_init(struct vl_compositor *compositor, struct pipe_context *pipe)
+{
+   assert(compositor);
+
+   memset(compositor, 0, sizeof(struct vl_compositor));
+
+   compositor->pipe = pipe;
+
+   if (!init_pipe_state(compositor))
+      return false;
+   if (!init_shaders(compositor)) {
+      cleanup_pipe_state(compositor);
+      return false;
+   }
+   if (!init_buffers(compositor)) {
+      cleanup_shaders(compositor);
+      cleanup_pipe_state(compositor);
+      return false;
+   }
+
+   return true;
+}
+
+void vl_compositor_cleanup(struct vl_compositor *compositor)
+{
+   assert(compositor);
+	
+   cleanup_buffers(compositor);
+   cleanup_shaders(compositor);
+   cleanup_pipe_state(compositor);
+}
+
+void vl_compositor_render(struct vl_compositor          *compositor,
+                          /*struct pipe_texture         *backround,
+                          struct pipe_video_rect        *backround_area,*/
+                          struct pipe_texture           *src_surface,
+                          enum pipe_mpeg12_picture_type picture_type,
+                          /*unsigned                    num_past_surfaces,
+                          struct pipe_texture           *past_surfaces,
+                          unsigned                      num_future_surfaces,
+                          struct pipe_texture           *future_surfaces,*/
+                          struct pipe_video_rect        *src_area,
+                          struct pipe_texture           *dst_surface,
+                          struct pipe_video_rect        *dst_area,
+                          /*unsigned                      num_layers,
+                          struct pipe_texture           *layers,
+                          struct pipe_video_rect        *layer_src_areas,
+                          struct pipe_video_rect        *layer_dst_areas*/
+                          struct pipe_fence_handle      **fence)
+{
+   struct vertex_shader_consts *vs_consts;
+
+   assert(compositor);
+   assert(src_surface);
+   assert(src_area);
+   assert(dst_surface);
+   assert(dst_area);
+   assert(picture_type == PIPE_MPEG12_PICTURE_TYPE_FRAME);
+
+   compositor->fb_state.width = dst_surface->width[0];
+   compositor->fb_state.height = dst_surface->height[0];
+   compositor->fb_state.cbufs[0] = compositor->pipe->screen->get_tex_surface
+   (
+      compositor->pipe->screen,
+      dst_surface,
+      0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
+   );
+
+   compositor->viewport.scale[0] = compositor->fb_state.width;
+   compositor->viewport.scale[1] = compositor->fb_state.height;
+   compositor->viewport.scale[2] = 1;
+   compositor->viewport.scale[3] = 1;
+   compositor->viewport.translate[0] = 0;
+   compositor->viewport.translate[1] = 0;
+   compositor->viewport.translate[2] = 0;
+   compositor->viewport.translate[3] = 0;
+
+   compositor->pipe->set_framebuffer_state(compositor->pipe, &compositor->fb_state);
+   compositor->pipe->set_viewport_state(compositor->pipe, &compositor->viewport);
+   compositor->pipe->bind_sampler_states(compositor->pipe, 1, &compositor->sampler);
+   compositor->pipe->set_sampler_textures(compositor->pipe, 1, &src_surface);
+   compositor->pipe->bind_vs_state(compositor->pipe, compositor->vertex_shader);
+   compositor->pipe->bind_fs_state(compositor->pipe, compositor->fragment_shader);
+   compositor->pipe->set_vertex_buffers(compositor->pipe, 2, compositor->vertex_bufs);
+   compositor->pipe->set_vertex_elements(compositor->pipe, 2, compositor->vertex_elems);
+   compositor->pipe->set_constant_buffer(compositor->pipe, PIPE_SHADER_VERTEX, 0, &compositor->vs_const_buf);
+   compositor->pipe->set_constant_buffer(compositor->pipe, PIPE_SHADER_FRAGMENT, 0, &compositor->fs_const_buf);
+
+   vs_consts = pipe_buffer_map
+   (
+      compositor->pipe->screen,
+      compositor->vs_const_buf.buffer,
+      PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
+   );
+
+   vs_consts->dst_scale.x = dst_area->w / (float)compositor->fb_state.cbufs[0]->width;
+   vs_consts->dst_scale.y = dst_area->h / (float)compositor->fb_state.cbufs[0]->height;
+   vs_consts->dst_scale.z = 1;
+   vs_consts->dst_scale.w = 1;
+   vs_consts->dst_trans.x = dst_area->x / (float)compositor->fb_state.cbufs[0]->width;
+   vs_consts->dst_trans.y = dst_area->y / (float)compositor->fb_state.cbufs[0]->height;
+   vs_consts->dst_trans.z = 0;
+   vs_consts->dst_trans.w = 0;
+
+   vs_consts->src_scale.x = src_area->w / (float)src_surface->width[0];
+   vs_consts->src_scale.y = src_area->h / (float)src_surface->height[0];
+   vs_consts->src_scale.z = 1;
+   vs_consts->src_scale.w = 1;
+   vs_consts->src_trans.x = src_area->x / (float)src_surface->width[0];
+   vs_consts->src_trans.y = src_area->y / (float)src_surface->height[0];
+   vs_consts->src_trans.z = 0;
+   vs_consts->src_trans.w = 0;
+
+   pipe_buffer_unmap(compositor->pipe->screen, compositor->vs_const_buf.buffer);
+
+   compositor->pipe->draw_arrays(compositor->pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4);
+   compositor->pipe->flush(compositor->pipe, PIPE_FLUSH_RENDER_CACHE, fence);
+
+   pipe_surface_reference(&compositor->fb_state.cbufs[0], NULL);
+}
+
+void vl_compositor_set_csc_matrix(struct vl_compositor *compositor, const float *mat)
+{
+   assert(compositor);
+
+   memcpy
+   (
+      pipe_buffer_map(compositor->pipe->screen, compositor->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+      mat,
+      sizeof(struct fragment_shader_consts)
+   );
+
+   pipe_buffer_unmap(compositor->pipe->screen, compositor->fs_const_buf.buffer);
+}
diff --git a/src/gallium/auxiliary/vl/vl_compositor.h b/src/gallium/auxiliary/vl/vl_compositor.h
new file mode 100644
index 0000000000..17e2afd353
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_compositor.h
@@ -0,0 +1,76 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef vl_compositor_h
+#define vl_compositor_h
+
+#include <pipe/p_compiler.h>
+#include <pipe/p_state.h>
+#include <pipe/p_video_state.h>
+
+struct pipe_context;
+struct pipe_texture;
+
+struct vl_compositor
+{
+   struct pipe_context *pipe;
+
+   struct pipe_framebuffer_state fb_state;
+   void *sampler;
+   void *vertex_shader;
+   void *fragment_shader;
+   struct pipe_viewport_state viewport;
+   struct pipe_vertex_buffer vertex_bufs[2];
+   struct pipe_vertex_element vertex_elems[2];
+   struct pipe_constant_buffer vs_const_buf, fs_const_buf;
+};
+
+bool vl_compositor_init(struct vl_compositor *compositor, struct pipe_context *pipe);
+
+void vl_compositor_cleanup(struct vl_compositor *compositor);
+
+void vl_compositor_render(struct vl_compositor          *compositor,
+                          /*struct pipe_texture         *backround,
+                          struct pipe_video_rect        *backround_area,*/
+                          struct pipe_texture           *src_surface,
+                          enum pipe_mpeg12_picture_type picture_type,
+                          /*unsigned                    num_past_surfaces,
+                          struct pipe_texture           *past_surfaces,
+                          unsigned                      num_future_surfaces,
+                          struct pipe_texture           *future_surfaces,*/
+                          struct pipe_video_rect        *src_area,
+                          struct pipe_texture           *dst_surface,
+                          struct pipe_video_rect        *dst_area,
+                          /*unsigned                      num_layers,
+                          struct pipe_texture           *layers,
+                          struct pipe_video_rect        *layer_src_areas,
+                          struct pipe_video_rect        *layer_dst_areas,*/
+                          struct pipe_fence_handle      **fence);
+
+void vl_compositor_set_csc_matrix(struct vl_compositor *compositor, const float *mat);
+
+#endif /* vl_compositor_h */
diff --git a/src/gallium/auxiliary/vl/vl_csc.c b/src/gallium/auxiliary/vl/vl_csc.c
new file mode 100644
index 0000000000..5ecc43a5fa
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_csc.c
@@ -0,0 +1,206 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "vl_csc.h"
+#include <util/u_math.h>
+#include <util/u_debug.h>
+
+/*
+ * Color space conversion formulas
+ *
+ * To convert YCbCr to RGB,
+ *    vec4  ycbcr, rgb
+ *    mat44 csc
+ *    rgb = csc * ycbcr
+ *
+ * To calculate the color space conversion matrix csc with ProcAmp adjustments,
+ *    mat44 csc, cstd, procamp, bias
+ *    csc = cstd * (procamp * bias)
+ *
+ * Where cstd is a matrix corresponding to one of the color standards (BT.601, BT.709, etc)
+ * adjusted for the kind of YCbCr -> RGB mapping wanted (1:1, full),
+ * bias is a matrix corresponding to the kind of YCbCr -> RGB mapping wanted (1:1, full)
+ *
+ * To calculate procamp,
+ *    mat44 procamp, hue, saturation, brightness, contrast
+ *    procamp = brightness * (saturation * (contrast * hue))
+ * Alternatively,
+ *    procamp = saturation * (brightness * (contrast * hue))
+ *
+ * contrast
+ * [ c, 0, 0, 0]
+ * [ 0, c, 0, 0]
+ * [ 0, 0, c, 0]
+ * [ 0, 0, 0, 1]
+ *
+ * brightness
+ * [ 1, 0, 0, b]
+ * [ 0, 1, 0, 0]
+ * [ 0, 0, 1, 0]
+ * [ 0, 0, 0, 1]
+ *
+ * saturation
+ * [ 1, 0, 0, 0]
+ * [ 0, s, 0, 0]
+ * [ 0, 0, s, 0]
+ * [ 0, 0, 0, 1]
+ *
+ * hue
+ * [ 1,       0,      0, 0]
+ * [ 0,  cos(h), sin(h), 0]
+ * [ 0, -sin(h), cos(h), 0]
+ * [ 0,       0,      0, 1]
+ *
+ * procamp
+ * [ c,           0,          0, b]
+ * [ 0,  c*s*cos(h), c*s*sin(h), 0]
+ * [ 0, -c*s*sin(h), c*s*cos(h), 0]
+ * [ 0,           0,          0, 1]
+ *
+ * bias
+ * [ 1, 0, 0,  ybias]
+ * [ 0, 1, 0, cbbias]
+ * [ 0, 0, 1, crbias]
+ * [ 0, 0, 0,      1]
+ *
+ * csc
+ * [ c*cstd[ 0], c*cstd[ 1]*s*cos(h) - c*cstd[ 2]*s*sin(h), c*cstd[ 2]*s*cos(h) + c*cstd[ 1]*s*sin(h), cstd[ 3] + cstd[ 0]*(b + c*ybias) + cstd[ 1]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[ 2]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))]
+ * [ c*cstd[ 4], c*cstd[ 5]*s*cos(h) - c*cstd[ 6]*s*sin(h), c*cstd[ 6]*s*cos(h) + c*cstd[ 5]*s*sin(h), cstd[ 7] + cstd[ 4]*(b + c*ybias) + cstd[ 5]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[ 6]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))]
+ * [ c*cstd[ 8], c*cstd[ 9]*s*cos(h) - c*cstd[10]*s*sin(h), c*cstd[10]*s*cos(h) + c*cstd[ 9]*s*sin(h), cstd[11] + cstd[ 8]*(b + c*ybias) + cstd[ 9]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[10]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))]
+ * [ c*cstd[12], c*cstd[13]*s*cos(h) - c*cstd[14]*s*sin(h), c*cstd[14]*s*cos(h) + c*cstd[13]*s*sin(h), cstd[15] + cstd[12]*(b + c*ybias) + cstd[13]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[14]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))]
+ */
+
+/*
+ * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [16,235]
+ */
+static const float bt_601[16] =
+{
+   1.0f,  0.0f,    1.371f, 0.0f,
+   1.0f, -0.336f, -0.698f, 0.0f,
+   1.0f,  1.732f,  0.0f,   0.0f,
+   0.0f,  0.0f,    0.0f,   1.0f
+};
+
+/*
+ * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [0,255]
+ */
+static const float bt_601_full[16] =
+{
+   1.164f,  0.0f,    1.596f, 0.0f,
+   1.164f, -0.391f, -0.813f, 0.0f,
+   1.164f,  2.018f,  0.0f,   0.0f,
+   0.0f,    0.0f,    0.0f,   1.0f
+};
+
+/*
+ * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [16,235]
+ */
+static const float bt_709[16] =
+{
+   1.0f,  0.0f,    1.540f, 0.0f,
+   1.0f, -0.183f, -0.459f, 0.0f,
+   1.0f,  1.816f,  0.0f,   0.0f,
+   0.0f,  0.0f,    0.0f,   1.0f
+};
+
+/*
+ * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [0,255]
+ */
+static const float bt_709_full[16] =
+{
+   1.164f,  0.0f,    1.793f, 0.0f,
+   1.164f, -0.213f, -0.534f, 0.0f,
+   1.164f,  2.115f,  0.0f,   0.0f,
+   0.0f,    0.0f,    0.0f,   1.0f
+};
+
+static const float identity[16] =
+{
+   1.0f, 0.0f, 0.0f, 0.0f,
+   0.0f, 1.0f, 0.0f, 0.0f,
+   0.0f, 0.0f, 1.0f, 0.0f,
+   0.0f, 0.0f, 0.0f, 1.0f
+};
+
+void vl_csc_get_matrix(enum VL_CSC_COLOR_STANDARD cs,
+                       struct vl_procamp *procamp,
+                       bool full_range,
+                       float *matrix)
+{
+   float ybias = full_range ? -16.0f/255.0f : 0.0f;
+   float cbbias = -128.0f/255.0f;
+   float crbias = -128.0f/255.0f;
+   float c = procamp ? procamp->contrast : 1.0f;
+   float s = procamp ? procamp->saturation : 1.0f;
+   float b = procamp ? procamp->brightness : 0.0f;
+   float h = procamp ? procamp->hue : 0.0f;
+   const float *cstd;
+
+   assert(matrix);
+
+   switch (cs) {
+      case VL_CSC_COLOR_STANDARD_BT_601:
+         cstd = full_range ? &bt_601_full[0] : &bt_601[0];
+         break;
+      case VL_CSC_COLOR_STANDARD_BT_709:
+         cstd = full_range ? &bt_709_full[0] : &bt_709[0];
+         break;
+      case VL_CSC_COLOR_STANDARD_IDENTITY:
+      default:
+         assert(cs == VL_CSC_COLOR_STANDARD_IDENTITY);
+         memcpy(matrix, &identity[0], sizeof(float) * 16);
+         return;
+   }
+
+   matrix[ 0] = c*cstd[ 0];
+   matrix[ 1] = c*cstd[ 1]*s*cosf(h) - c*cstd[ 2]*s*sinf(h);
+   matrix[ 2] = c*cstd[ 2]*s*cosf(h) + c*cstd[ 1]*s*sinf(h);
+   matrix[ 3] = cstd[ 3] + cstd[ 0]*(b + c*ybias) + cstd[ 1]*(c*cbbias*s*cosf(h) + c*crbias*s*sinf(h)) + cstd[ 2]*(c*crbias*s*cosf(h) - c*cbbias*s*sinf(h));
+
+   matrix[ 4] = c*cstd[ 4];
+   matrix[ 5] = c*cstd[ 5]*s*cosf(h) - c*cstd[ 6]*s*sinf(h);
+   matrix[ 6] = c*cstd[ 6]*s*cosf(h) + c*cstd[ 5]*s*sinf(h);
+   matrix[ 7] = cstd[ 7] + cstd[ 4]*(b + c*ybias) + cstd[ 5]*(c*cbbias*s*cosf(h) + c*crbias*s*sinf(h)) + cstd[ 6]*(c*crbias*s*cosf(h) - c*cbbias*s*sinf(h));
+
+   matrix[ 8] = c*cstd[ 8];
+   matrix[ 9] = c*cstd[ 9]*s*cosf(h) - c*cstd[10]*s*sinf(h);
+   matrix[10] = c*cstd[10]*s*cosf(h) + c*cstd[ 9]*s*sinf(h);
+   matrix[11] = cstd[11] + cstd[ 8]*(b + c*ybias) + cstd[ 9]*(c*cbbias*s*cosf(h) + c*crbias*s*sinf(h)) + cstd[10]*(c*crbias*s*cosf(h) - c*cbbias*s*sinf(h));
+
+   matrix[12] = c*cstd[12];
+   matrix[13] = c*cstd[13]*s*cos(h) - c*cstd[14]*s*sin(h);
+   matrix[14] = c*cstd[14]*s*cos(h) + c*cstd[13]*s*sin(h);
+   matrix[15] = cstd[15] + cstd[12]*(b + c*ybias) + cstd[13]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[14]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h));
+}
diff --git a/src/gallium/drivers/i965simple/brw_screen.h b/src/gallium/auxiliary/vl/vl_csc.h
index d3c70387e6..722ca35f33 100644
--- a/src/gallium/drivers/i965simple/brw_screen.h
+++ b/src/gallium/auxiliary/vl/vl_csc.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2009 Younes Manton.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,44 +25,29 @@
  * 
  **************************************************************************/
 
+#ifndef vl_csc_h
+#define vl_csc_h
 
-#ifndef BRW_SCREEN_H
-#define BRW_SCREEN_H
+#include <pipe/p_compiler.h>
 
-
-#include "pipe/p_screen.h"
-
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-
-/**
- * Subclass of pipe_screen
- */
-struct brw_screen
+struct vl_procamp
 {
-   struct pipe_screen screen;
-
-   uint pci_id;
+   float brightness;
+   float contrast;
+   float saturation;
+   float hue;
 };
 
-
-/** cast wrapper */
-static INLINE struct brw_screen *
-brw_screen(struct pipe_screen *pscreen)
+enum VL_CSC_COLOR_STANDARD
 {
-   return (struct brw_screen *) pscreen;
-}
-
-
-extern struct pipe_screen *
-brw_create_screen(struct pipe_winsys *winsys, uint pci_id);
-
+   VL_CSC_COLOR_STANDARD_IDENTITY,
+   VL_CSC_COLOR_STANDARD_BT_601,
+   VL_CSC_COLOR_STANDARD_BT_709
+};
 
-#ifdef __cplusplus
-}
-#endif
+void vl_csc_get_matrix(enum VL_CSC_COLOR_STANDARD cs,
+                       struct vl_procamp *procamp,
+                       bool full_range,
+                       float *matrix);
 
-#endif /* BRW_SCREEN_H */
+#endif /* vl_csc_h */
diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
new file mode 100644
index 0000000000..6b3614821c
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
@@ -0,0 +1,1654 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "vl_mpeg12_mc_renderer.h"
+#include <assert.h>
+#include <pipe/p_context.h>
+#include <pipe/p_inlines.h>
+#include <util/u_math.h>
+#include <util/u_memory.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+#include "vl_shader_build.h"
+
+#define DEFAULT_BUF_ALIGNMENT 1
+#define MACROBLOCK_WIDTH 16
+#define MACROBLOCK_HEIGHT 16
+#define BLOCK_WIDTH 8
+#define BLOCK_HEIGHT 8
+#define ZERO_BLOCK_NIL -1.0f
+#define ZERO_BLOCK_IS_NIL(zb) ((zb).x < 0.0f)
+
+struct vertex2f
+{
+   float x, y;
+};
+
+struct vertex4f
+{
+   float x, y, z, w;
+};
+
+struct vertex_shader_consts
+{
+   struct vertex4f denorm;
+};
+
+struct fragment_shader_consts
+{
+   struct vertex4f multiplier;
+   struct vertex4f div;
+};
+
+/*
+ * Muliplier renormalizes block samples from 16 bits to 12 bits.
+ * Divider is used when calculating Y % 2 for choosing top or bottom
+ * field for P or B macroblocks.
+ * TODO: Use immediates.
+ */
+static const struct fragment_shader_consts fs_consts = {
+   {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f},
+   {0.5f, 2.0f, 0.0f, 0.0f}
+};
+
+struct vert_stream_0
+{
+   struct vertex2f pos;
+   struct vertex2f luma_tc;
+   struct vertex2f cb_tc;
+   struct vertex2f cr_tc;
+};
+
+enum MACROBLOCK_TYPE
+{
+   MACROBLOCK_TYPE_INTRA,
+   MACROBLOCK_TYPE_FWD_FRAME_PRED,
+   MACROBLOCK_TYPE_FWD_FIELD_PRED,
+   MACROBLOCK_TYPE_BKWD_FRAME_PRED,
+   MACROBLOCK_TYPE_BKWD_FIELD_PRED,
+   MACROBLOCK_TYPE_BI_FRAME_PRED,
+   MACROBLOCK_TYPE_BI_FIELD_PRED,
+
+   NUM_MACROBLOCK_TYPES
+};
+
+static void
+create_intra_vert_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   const unsigned max_tokens = 50;
+
+   struct pipe_shader_state vs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(r);
+
+   tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header *) &tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+   ti = 3;
+
+   /*
+    * decl i0              ; Vertex pos
+    * decl i1              ; Luma texcoords
+    * decl i2              ; Chroma Cb texcoords
+    * decl i3              ; Chroma Cr texcoords
+    */
+   for (i = 0; i < 4; i++) {
+      decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * decl o0              ; Vertex pos
+    * decl o1              ; Luma texcoords
+    * decl o2              ; Chroma Cb texcoords
+    * decl o3              ; Chroma Cr texcoords
+    */
+   for (i = 0; i < 4; i++) {
+      decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * mov o0, i0           ; Move input vertex pos to output
+    * mov o1, i1           ; Move input luma texcoords to output
+    * mov o2, i2           ; Move input chroma Cb texcoords to output
+    * mov o3, i3           ; Move input chroma Cr texcoords to output
+    */
+   for (i = 0; i < 4; ++i) {
+      inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   vs.tokens = tokens;
+   r->i_vs = r->pipe->create_vs_state(r->pipe, &vs);
+   free(tokens);
+}
+
+static void
+create_intra_frag_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   const unsigned max_tokens = 100;
+
+   struct pipe_shader_state fs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(r);
+
+   tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header *) &tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+   ti = 3;
+
+   /*
+    * decl i0                      ; Luma texcoords
+    * decl i1                      ; Chroma Cb texcoords
+    * decl i2                      ; Chroma Cr texcoords
+    */
+   for (i = 0; i < 3; ++i) {
+      decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* decl c0                      ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
+   decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl o0                      ; Fragment color */
+   decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl t0, t1 */
+   decl = vl_decl_temps(0, 1);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * decl s0                      ; Sampler for luma texture
+    * decl s1                      ; Sampler for chroma Cb texture
+    * decl s2                      ; Sampler for chroma Cr texture
+    */
+   for (i = 0; i < 3; ++i) {
+      decl = vl_decl_samplers(i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * tex2d t1, i0, s0             ; Read texel from luma texture
+    * mov t0.x, t1.x               ; Move luma sample into .x component
+    * tex2d t1, i1, s1             ; Read texel from chroma Cb texture
+    * mov t0.y, t1.x               ; Move Cb sample into .y component
+    * tex2d t1, i2, s2             ; Read texel from chroma Cr texture
+    * mov t0.z, t1.x               ; Move Cr sample into .z component
+    */
+   for (i = 0; i < 3; ++i) {
+      inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+      inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+      inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* mul o0, t0, c0               ; Rescale texel to correct range */
+   inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   fs.tokens = tokens;
+   r->i_fs = r->pipe->create_fs_state(r->pipe, &fs);
+   free(tokens);
+}
+
+static void
+create_frame_pred_vert_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   const unsigned max_tokens = 100;
+
+   struct pipe_shader_state vs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(r);
+
+   tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header *) &tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+   ti = 3;
+
+   /*
+    * decl i0              ; Vertex pos
+    * decl i1              ; Luma texcoords
+    * decl i2              ; Chroma Cb texcoords
+    * decl i3              ; Chroma Cr texcoords
+    * decl i4              ; Ref surface top field texcoords
+    * decl i5              ; Ref surface bottom field texcoords (unused, packed in the same stream)
+    */
+   for (i = 0; i < 6; i++) {
+      decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * decl o0              ; Vertex pos
+    * decl o1              ; Luma texcoords
+    * decl o2              ; Chroma Cb texcoords
+    * decl o3              ; Chroma Cr texcoords
+    * decl o4              ; Ref macroblock texcoords
+    */
+   for (i = 0; i < 5; i++) {
+      decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * mov o0, i0           ; Move input vertex pos to output
+    * mov o1, i1           ; Move input luma texcoords to output
+    * mov o2, i2           ; Move input chroma Cb texcoords to output
+    * mov o3, i3           ; Move input chroma Cr texcoords to output
+    */
+   for (i = 0; i < 4; ++i) {
+        inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+        ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* add o4, i0, i4       ; Translate vertex pos by motion vec to form ref macroblock texcoords */
+   inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 4);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   vs.tokens = tokens;
+   r->p_vs[0] = r->pipe->create_vs_state(r->pipe, &vs);
+   free(tokens);
+}
+
+static void
+create_field_pred_vert_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   assert(false);
+}
+
+static void
+create_frame_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   const unsigned max_tokens = 100;
+
+   struct pipe_shader_state fs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(r);
+
+   tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header *) &tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+   ti = 3;
+
+   /*
+    * decl i0                      ; Luma texcoords
+    * decl i1                      ; Chroma Cb texcoords
+    * decl i2                      ; Chroma Cr texcoords
+    * decl i3                      ; Ref macroblock texcoords
+    */
+   for (i = 0; i < 4; ++i) {
+      decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* decl c0                      ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
+   decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl o0                      ; Fragment color */
+   decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl t0, t1 */
+   decl = vl_decl_temps(0, 1);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * decl s0                      ; Sampler for luma texture
+    * decl s1                      ; Sampler for chroma Cb texture
+    * decl s2                      ; Sampler for chroma Cr texture
+    * decl s3                      ; Sampler for ref surface texture
+    */
+   for (i = 0; i < 4; ++i) {
+      decl = vl_decl_samplers(i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * tex2d t1, i0, s0             ; Read texel from luma texture
+    * mov t0.x, t1.x               ; Move luma sample into .x component
+    * tex2d t1, i1, s1             ; Read texel from chroma Cb texture
+    * mov t0.y, t1.x               ; Move Cb sample into .y component
+    * tex2d t1, i2, s2             ; Read texel from chroma Cr texture
+    * mov t0.z, t1.x               ; Move Cr sample into .z component
+    */
+   for (i = 0; i < 3; ++i) {
+      inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+      inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+      inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* mul t0, t0, c0               ; Rescale texel to correct range */
+   inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* tex2d t1, i3, s3             ; Read texel from ref macroblock */
+   inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 3, TGSI_FILE_SAMPLER, 3);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* add o0, t0, t1               ; Add ref and differential to form final output */
+   inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   fs.tokens = tokens;
+   r->p_fs[0] = r->pipe->create_fs_state(r->pipe, &fs);
+   free(tokens);
+}
+
+static void
+create_field_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   assert(false);
+}
+
+static void
+create_frame_bi_pred_vert_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   const unsigned max_tokens = 100;
+
+   struct pipe_shader_state vs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(r);
+
+   tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header *) &tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+   ti = 3;
+
+   /*
+    * decl i0              ; Vertex pos
+    * decl i1              ; Luma texcoords
+    * decl i2              ; Chroma Cb texcoords
+    * decl i3              ; Chroma Cr texcoords
+    * decl i4              ; First ref macroblock top field texcoords
+    * decl i5              ; First ref macroblock bottom field texcoords (unused, packed in the same stream)
+    * decl i6              ; Second ref macroblock top field texcoords
+    * decl i7              ; Second ref macroblock bottom field texcoords (unused, packed in the same stream)
+    */
+   for (i = 0; i < 8; i++) {
+      decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * decl o0              ; Vertex pos
+    * decl o1              ; Luma texcoords
+    * decl o2              ; Chroma Cb texcoords
+    * decl o3              ; Chroma Cr texcoords
+    * decl o4              ; First ref macroblock texcoords
+    * decl o5              ; Second ref macroblock texcoords
+    */
+   for (i = 0; i < 6; i++) {
+      decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * mov o0, i0           ; Move input vertex pos to output
+    * mov o1, i1           ; Move input luma texcoords to output
+    * mov o2, i2           ; Move input chroma Cb texcoords to output
+    * mov o3, i3           ; Move input chroma Cr texcoords to output
+    */
+   for (i = 0; i < 4; ++i) {
+      inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * add o4, i0, i4       ; Translate vertex pos by motion vec to form first ref macroblock texcoords
+    * add o5, i0, i6       ; Translate vertex pos by motion vec to form second ref macroblock texcoords
+    */
+   for (i = 0; i < 2; ++i) {
+      inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 2) * 2);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   vs.tokens = tokens;
+   r->b_vs[0] = r->pipe->create_vs_state(r->pipe, &vs);
+   free(tokens);
+}
+
+static void
+create_field_bi_pred_vert_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   assert(false);
+}
+
+static void
+create_frame_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   const unsigned max_tokens = 100;
+
+   struct pipe_shader_state fs;
+   struct tgsi_token *tokens;
+   struct tgsi_header *header;
+
+   struct tgsi_full_declaration decl;
+   struct tgsi_full_instruction inst;
+
+   unsigned ti;
+
+   unsigned i;
+
+   assert(r);
+
+   tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
+   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   header = (struct tgsi_header *) &tokens[1];
+   *header = tgsi_build_header();
+   *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+   ti = 3;
+
+   /*
+    * decl i0                      ; Luma texcoords
+    * decl i1                      ; Chroma Cb texcoords
+    * decl i2                      ; Chroma Cr texcoords
+    * decl i3                      ; First ref macroblock texcoords
+    * decl i4                      ; Second ref macroblock texcoords
+    */
+   for (i = 0; i < 5; ++i) {
+      decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * decl c0                      ; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+    * decl c1                      ; Constant 1/2 in .x channel to use as weight to blend past and future texels
+    */
+   decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl o0                      ; Fragment color */
+   decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /* decl t0-t2 */
+   decl = vl_decl_temps(0, 2);
+   ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * decl s0                      ; Sampler for luma texture
+    * decl s1                      ; Sampler for chroma Cb texture
+    * decl s2                      ; Sampler for chroma Cr texture
+    * decl s3                      ; Sampler for first ref surface texture
+    * decl s4                      ; Sampler for second ref surface texture
+    */
+   for (i = 0; i < 5; ++i) {
+      decl = vl_decl_samplers(i, i);
+      ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /*
+    * tex2d t1, i0, s0             ; Read texel from luma texture
+    * mov t0.x, t1.x               ; Move luma sample into .x component
+    * tex2d t1, i1, s1             ; Read texel from chroma Cb texture
+    * mov t0.y, t1.x               ; Move Cb sample into .y component
+    * tex2d t1, i2, s2             ; Read texel from chroma Cr texture
+    * mov t0.z, t1.x               ; Move Cr sample into .z component
+    */
+   for (i = 0; i < 3; ++i) {
+      inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+      inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+      inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+      inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* mul t0, t0, c0               ; Rescale texel to correct range */
+   inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /*
+    * tex2d t1, i3, s3             ; Read texel from first ref macroblock
+    * tex2d t2, i4, s4             ; Read texel from second ref macroblock
+    */
+   for (i = 0; i < 2; ++i) {
+      inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, i + 3);
+      ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+   }
+
+   /* lerp t1, c1.x, t1, t2        ; Blend past and future texels */
+   inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+   inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+   inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+   inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+   inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* add o0, t0, t1               ; Add past/future ref and differential to form final output */
+   inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   /* end */
+   inst = vl_end();
+   ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+   assert(ti <= max_tokens);
+
+   fs.tokens = tokens;
+   r->b_fs[0] = r->pipe->create_fs_state(r->pipe, &fs);
+   free(tokens);
+}
+
+static void
+create_field_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
+{
+   assert(false);
+}
+
+static void
+xfer_buffers_map(struct vl_mpeg12_mc_renderer *r)
+{
+   unsigned i;
+
+   assert(r);
+
+   for (i = 0; i < 3; ++i) {
+      r->tex_transfer[i] = r->pipe->screen->get_tex_transfer
+      (
+         r->pipe->screen, r->textures.all[i],
+         0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0,
+         r->textures.all[i]->width[0], r->textures.all[i]->height[0]
+      );
+
+      r->texels[i] = r->pipe->screen->transfer_map(r->pipe->screen, r->tex_transfer[i]);
+   }
+}
+
+static void
+xfer_buffers_unmap(struct vl_mpeg12_mc_renderer *r)
+{
+   unsigned i;
+
+   assert(r);
+
+   for (i = 0; i < 3; ++i) {
+      r->pipe->screen->transfer_unmap(r->pipe->screen, r->tex_transfer[i]);
+      r->pipe->screen->tex_transfer_destroy(r->tex_transfer[i]);
+   }
+}
+
+static bool
+init_pipe_state(struct vl_mpeg12_mc_renderer *r)
+{
+   struct pipe_sampler_state sampler;
+   unsigned filters[5];
+   unsigned i;
+
+   assert(r);
+
+   r->viewport.scale[0] = r->pot_buffers ?
+      util_next_power_of_two(r->picture_width) : r->picture_width;
+   r->viewport.scale[1] = r->pot_buffers ?
+      util_next_power_of_two(r->picture_height) : r->picture_height;
+   r->viewport.scale[2] = 1;
+   r->viewport.scale[3] = 1;
+   r->viewport.translate[0] = 0;
+   r->viewport.translate[1] = 0;
+   r->viewport.translate[2] = 0;
+   r->viewport.translate[3] = 0;
+
+   r->fb_state.width = r->pot_buffers ?
+      util_next_power_of_two(r->picture_width) : r->picture_width;
+   r->fb_state.height = r->pot_buffers ?
+      util_next_power_of_two(r->picture_height) : r->picture_height;
+   r->fb_state.nr_cbufs = 1;
+   r->fb_state.zsbuf = NULL;
+
+   /* Luma filter */
+   filters[0] = PIPE_TEX_FILTER_NEAREST;
+   /* Chroma filters */
+   if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_444 ||
+       r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) {
+      filters[1] = PIPE_TEX_FILTER_NEAREST;
+      filters[2] = PIPE_TEX_FILTER_NEAREST;
+   }
+   else {
+      filters[1] = PIPE_TEX_FILTER_LINEAR;
+      filters[2] = PIPE_TEX_FILTER_LINEAR;
+   }
+   /* Fwd, bkwd ref filters */
+   filters[3] = PIPE_TEX_FILTER_LINEAR;
+   filters[4] = PIPE_TEX_FILTER_LINEAR;
+
+   for (i = 0; i < 5; ++i) {
+      sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+      sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+      sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+      sampler.min_img_filter = filters[i];
+      sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+      sampler.mag_img_filter = filters[i];
+      sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
+      sampler.compare_func = PIPE_FUNC_ALWAYS;
+      sampler.normalized_coords = 1;
+      /*sampler.prefilter = ; */
+      /*sampler.shadow_ambient = ; */
+      /*sampler.lod_bias = ; */
+      sampler.min_lod = 0;
+      /*sampler.max_lod = ; */
+      /*sampler.border_color[i] = ; */
+      /*sampler.max_anisotropy = ; */
+      r->samplers.all[i] = r->pipe->create_sampler_state(r->pipe, &sampler);
+   }
+
+   return true;
+}
+
+static void
+cleanup_pipe_state(struct vl_mpeg12_mc_renderer *r)
+{
+   unsigned i;
+
+   assert(r);
+
+   for (i = 0; i < 5; ++i)
+      r->pipe->delete_sampler_state(r->pipe, r->samplers.all[i]);
+}
+
+static bool
+init_shaders(struct vl_mpeg12_mc_renderer *r)
+{
+   assert(r);
+
+   create_intra_vert_shader(r);
+   create_intra_frag_shader(r);
+   create_frame_pred_vert_shader(r);
+   create_frame_pred_frag_shader(r);
+   create_frame_bi_pred_vert_shader(r);
+   create_frame_bi_pred_frag_shader(r);
+
+   return true;
+}
+
+static void
+cleanup_shaders(struct vl_mpeg12_mc_renderer *r)
+{
+   assert(r);
+
+   r->pipe->delete_vs_state(r->pipe, r->i_vs);
+   r->pipe->delete_fs_state(r->pipe, r->i_fs);
+   r->pipe->delete_vs_state(r->pipe, r->p_vs[0]);
+   r->pipe->delete_fs_state(r->pipe, r->p_fs[0]);
+   r->pipe->delete_vs_state(r->pipe, r->b_vs[0]);
+   r->pipe->delete_fs_state(r->pipe, r->b_fs[0]);
+}
+
+static bool
+init_buffers(struct vl_mpeg12_mc_renderer *r)
+{
+   struct pipe_texture template;
+
+   const unsigned mbw =
+      align(r->picture_width, MACROBLOCK_WIDTH) / MACROBLOCK_WIDTH;
+   const unsigned mbh =
+      align(r->picture_height, MACROBLOCK_HEIGHT) / MACROBLOCK_HEIGHT;
+
+   unsigned i;
+
+   assert(r);
+
+   r->macroblocks_per_batch =
+      mbw * (r->bufmode == VL_MPEG12_MC_RENDERER_BUFFER_PICTURE ? mbh : 1);
+   r->num_macroblocks = 0;
+   r->macroblock_buf = MALLOC(r->macroblocks_per_batch * sizeof(struct pipe_mpeg12_macroblock));
+
+   memset(&template, 0, sizeof(struct pipe_texture));
+   template.target = PIPE_TEXTURE_2D;
+   /* TODO: Accomodate HW that can't do this and also for cases when this isn't precise enough */
+   template.format = PIPE_FORMAT_R16_SNORM;
+   template.last_level = 0;
+   template.width[0] = r->pot_buffers ?
+      util_next_power_of_two(r->picture_width) : r->picture_width;
+   template.height[0] = r->pot_buffers ?
+      util_next_power_of_two(r->picture_height) : r->picture_height;
+   template.depth[0] = 1;
+   pf_get_block(template.format, &template.block);
+   template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_DYNAMIC;
+
+   r->textures.individual.y = r->pipe->screen->texture_create(r->pipe->screen, &template);
+
+   if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420) {
+      template.width[0] = r->pot_buffers ?
+         util_next_power_of_two(r->picture_width / 2) :
+         r->picture_width / 2;
+      template.height[0] = r->pot_buffers ?
+         util_next_power_of_two(r->picture_height / 2) :
+         r->picture_height / 2;
+   }
+   else if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_422)
+      template.height[0] = r->pot_buffers ?
+         util_next_power_of_two(r->picture_height / 2) :
+         r->picture_height / 2;
+
+   r->textures.individual.cb =
+      r->pipe->screen->texture_create(r->pipe->screen, &template);
+   r->textures.individual.cr =
+      r->pipe->screen->texture_create(r->pipe->screen, &template);
+
+   r->vertex_bufs.individual.ycbcr.stride = sizeof(struct vertex2f) * 4;
+   r->vertex_bufs.individual.ycbcr.max_index = 24 * r->macroblocks_per_batch - 1;
+   r->vertex_bufs.individual.ycbcr.buffer_offset = 0;
+   r->vertex_bufs.individual.ycbcr.buffer = pipe_buffer_create
+   (
+      r->pipe->screen,
+      DEFAULT_BUF_ALIGNMENT,
+      PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD,
+      sizeof(struct vertex2f) * 4 * 24 * r->macroblocks_per_batch
+   );
+
+   for (i = 1; i < 3; ++i) {
+      r->vertex_bufs.all[i].stride = sizeof(struct vertex2f) * 2;
+      r->vertex_bufs.all[i].max_index = 24 * r->macroblocks_per_batch - 1;
+      r->vertex_bufs.all[i].buffer_offset = 0;
+      r->vertex_bufs.all[i].buffer = pipe_buffer_create
+      (
+         r->pipe->screen,
+         DEFAULT_BUF_ALIGNMENT,
+         PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD,
+         sizeof(struct vertex2f) * 2 * 24 * r->macroblocks_per_batch
+      );
+   }
+
+   /* Position element */
+   r->vertex_elems[0].src_offset = 0;
+   r->vertex_elems[0].vertex_buffer_index = 0;
+   r->vertex_elems[0].nr_components = 2;
+   r->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* Luma, texcoord element */
+   r->vertex_elems[1].src_offset = sizeof(struct vertex2f);
+   r->vertex_elems[1].vertex_buffer_index = 0;
+   r->vertex_elems[1].nr_components = 2;
+   r->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* Chroma Cr texcoord element */
+   r->vertex_elems[2].src_offset = sizeof(struct vertex2f) * 2;
+   r->vertex_elems[2].vertex_buffer_index = 0;
+   r->vertex_elems[2].nr_components = 2;
+   r->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* Chroma Cb texcoord element */
+   r->vertex_elems[3].src_offset = sizeof(struct vertex2f) * 3;
+   r->vertex_elems[3].vertex_buffer_index = 0;
+   r->vertex_elems[3].nr_components = 2;
+   r->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* First ref surface top field texcoord element */
+   r->vertex_elems[4].src_offset = 0;
+   r->vertex_elems[4].vertex_buffer_index = 1;
+   r->vertex_elems[4].nr_components = 2;
+   r->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* First ref surface bottom field texcoord element */
+   r->vertex_elems[5].src_offset = sizeof(struct vertex2f);
+   r->vertex_elems[5].vertex_buffer_index = 1;
+   r->vertex_elems[5].nr_components = 2;
+   r->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* Second ref surface top field texcoord element */
+   r->vertex_elems[6].src_offset = 0;
+   r->vertex_elems[6].vertex_buffer_index = 2;
+   r->vertex_elems[6].nr_components = 2;
+   r->vertex_elems[6].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   /* Second ref surface bottom field texcoord element */
+   r->vertex_elems[7].src_offset = sizeof(struct vertex2f);
+   r->vertex_elems[7].vertex_buffer_index = 2;
+   r->vertex_elems[7].nr_components = 2;
+   r->vertex_elems[7].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+   r->vs_const_buf.buffer = pipe_buffer_create
+   (
+      r->pipe->screen,
+      DEFAULT_BUF_ALIGNMENT,
+      PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD,
+      sizeof(struct vertex_shader_consts)
+   );
+
+   r->fs_const_buf.buffer = pipe_buffer_create
+   (
+      r->pipe->screen,
+      DEFAULT_BUF_ALIGNMENT,
+      PIPE_BUFFER_USAGE_CONSTANT, sizeof(struct fragment_shader_consts)
+   );
+
+   memcpy
+   (
+      pipe_buffer_map(r->pipe->screen, r->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+      &fs_consts, sizeof(struct fragment_shader_consts)
+   );
+
+   pipe_buffer_unmap(r->pipe->screen, r->fs_const_buf.buffer);
+
+   return true;
+}
+
+static void
+cleanup_buffers(struct vl_mpeg12_mc_renderer *r)
+{
+   unsigned i;
+
+   assert(r);
+
+   pipe_buffer_reference(&r->vs_const_buf.buffer, NULL);
+   pipe_buffer_reference(&r->fs_const_buf.buffer, NULL);
+
+   for (i = 0; i < 3; ++i)
+      pipe_buffer_reference(&r->vertex_bufs.all[i].buffer, NULL);
+
+   for (i = 0; i < 3; ++i)
+      pipe_texture_reference(&r->textures.all[i], NULL);
+
+   FREE(r->macroblock_buf);
+}
+
+static enum MACROBLOCK_TYPE
+get_macroblock_type(struct pipe_mpeg12_macroblock *mb)
+{
+   assert(mb);
+
+   switch (mb->mb_type) {
+      case PIPE_MPEG12_MACROBLOCK_TYPE_INTRA:
+         return MACROBLOCK_TYPE_INTRA;
+      case PIPE_MPEG12_MACROBLOCK_TYPE_FWD:
+         return mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ?
+            MACROBLOCK_TYPE_FWD_FRAME_PRED : MACROBLOCK_TYPE_FWD_FIELD_PRED;
+      case PIPE_MPEG12_MACROBLOCK_TYPE_BKWD:
+         return mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ?
+            MACROBLOCK_TYPE_BKWD_FRAME_PRED : MACROBLOCK_TYPE_BKWD_FIELD_PRED;
+      case PIPE_MPEG12_MACROBLOCK_TYPE_BI:
+         return mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ?
+            MACROBLOCK_TYPE_BI_FRAME_PRED : MACROBLOCK_TYPE_BI_FIELD_PRED;
+      default:
+         assert(0);
+   }
+
+   /* Unreachable */
+   return -1;
+}
+
+/* XXX: One of these days this will have to be killed with fire */
+#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, use_zb, zb)				\
+	do {															\
+	(vb)[0].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[0].pos.y = (mby) * (unity) + (ofsy);			\
+	(vb)[1].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
+	(vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].pos.y = (mby) * (unity) + (ofsy);			\
+	(vb)[3].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].pos.y = (mby) * (unity) + (ofsy);			\
+	(vb)[4].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
+	(vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
+																\
+	if (!use_zb || (cbp) & (lm))												\
+	{															\
+		(vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].luma_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].luma_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].luma_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+	}															\
+	else															\
+	{															\
+		(vb)[0].luma_tc.x = (zb)[0].x;		(vb)[0].luma_tc.y = (zb)[0].y;						\
+		(vb)[1].luma_tc.x = (zb)[0].x;		(vb)[1].luma_tc.y = (zb)[0].y + (hy);					\
+		(vb)[2].luma_tc.x = (zb)[0].x + (hx);	(vb)[2].luma_tc.y = (zb)[0].y;						\
+		(vb)[3].luma_tc.x = (zb)[0].x + (hx);	(vb)[3].luma_tc.y = (zb)[0].y;						\
+		(vb)[4].luma_tc.x = (zb)[0].x;		(vb)[4].luma_tc.y = (zb)[0].y + (hy);					\
+		(vb)[5].luma_tc.x = (zb)[0].x + (hx);	(vb)[5].luma_tc.y = (zb)[0].y + (hy);					\
+	}															\
+																\
+	if (!use_zb || (cbp) & (cbm))												\
+	{															\
+		(vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].cb_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].cb_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].cb_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+	}															\
+	else															\
+	{															\
+		(vb)[0].cb_tc.x = (zb)[1].x;		(vb)[0].cb_tc.y = (zb)[1].y;						\
+		(vb)[1].cb_tc.x = (zb)[1].x;		(vb)[1].cb_tc.y = (zb)[1].y + (hy);					\
+		(vb)[2].cb_tc.x = (zb)[1].x + (hx);	(vb)[2].cb_tc.y = (zb)[1].y;						\
+		(vb)[3].cb_tc.x = (zb)[1].x + (hx);	(vb)[3].cb_tc.y = (zb)[1].y;						\
+		(vb)[4].cb_tc.x = (zb)[1].x;		(vb)[4].cb_tc.y = (zb)[1].y + (hy);					\
+		(vb)[5].cb_tc.x = (zb)[1].x + (hx);	(vb)[5].cb_tc.y = (zb)[1].y + (hy);					\
+	}															\
+																\
+	if (!use_zb || (cbp) & (crm))												\
+	{															\
+		(vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].cr_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].cr_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].cr_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+	}															\
+	else															\
+	{															\
+		(vb)[0].cr_tc.x = (zb)[2].x;		(vb)[0].cr_tc.y = (zb)[2].y;						\
+		(vb)[1].cr_tc.x = (zb)[2].x;		(vb)[1].cr_tc.y = (zb)[2].y + (hy);					\
+		(vb)[2].cr_tc.x = (zb)[2].x + (hx);	(vb)[2].cr_tc.y = (zb)[2].y;						\
+		(vb)[3].cr_tc.x = (zb)[2].x + (hx);	(vb)[3].cr_tc.y = (zb)[2].y;						\
+		(vb)[4].cr_tc.x = (zb)[2].x;		(vb)[4].cr_tc.y = (zb)[2].y + (hy);					\
+		(vb)[5].cr_tc.x = (zb)[2].x + (hx);	(vb)[5].cr_tc.y = (zb)[2].y + (hy);					\
+	}															\
+	} while (0)
+
+static void
+gen_macroblock_verts(struct vl_mpeg12_mc_renderer *r,
+                     struct pipe_mpeg12_macroblock *mb, unsigned pos,
+                     struct vert_stream_0 *ycbcr_vb, struct vertex2f **ref_vb)
+{
+   struct vertex2f mo_vec[2];
+
+   unsigned i;
+
+   assert(r);
+   assert(mb);
+   assert(ycbcr_vb);
+   assert(pos < r->macroblocks_per_batch);
+
+   switch (mb->mb_type) {
+      case PIPE_MPEG12_MACROBLOCK_TYPE_BI:
+      {
+         struct vertex2f *vb;
+
+         assert(ref_vb && ref_vb[1]);
+
+         vb = ref_vb[1] + pos * 2 * 24;
+
+         mo_vec[0].x = mb->pmv[0][1][0] * 0.5f * r->surface_tex_inv_size.x;
+         mo_vec[0].y = mb->pmv[0][1][1] * 0.5f * r->surface_tex_inv_size.y;
+
+         if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME) {
+            for (i = 0; i < 24 * 2; i += 2) {
+               vb[i].x = mo_vec[0].x;
+               vb[i].y = mo_vec[0].y;
+            }
+         }
+         else {
+            mo_vec[1].x = mb->pmv[1][1][0] * 0.5f * r->surface_tex_inv_size.x;
+            mo_vec[1].y = mb->pmv[1][1][1] * 0.5f * r->surface_tex_inv_size.y;
+
+            for (i = 0; i < 24 * 2; i += 2) {
+               vb[i].x = mo_vec[0].x;
+               vb[i].y = mo_vec[0].y;
+               vb[i + 1].x = mo_vec[1].x;
+               vb[i + 1].y = mo_vec[1].y;
+            }
+         }
+
+         /* fall-through */
+      }
+      case PIPE_MPEG12_MACROBLOCK_TYPE_FWD:
+      case PIPE_MPEG12_MACROBLOCK_TYPE_BKWD:
+      {
+         struct vertex2f *vb;
+
+         assert(ref_vb && ref_vb[0]);
+
+         vb = ref_vb[0] + pos * 2 * 24;
+
+         if (mb->mb_type == PIPE_MPEG12_MACROBLOCK_TYPE_BKWD) {
+             mo_vec[0].x = mb->pmv[0][1][0] * 0.5f * r->surface_tex_inv_size.x;
+             mo_vec[0].y = mb->pmv[0][1][1] * 0.5f * r->surface_tex_inv_size.y;
+
+             if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FIELD) {
+                mo_vec[1].x = mb->pmv[1][1][0] * 0.5f * r->surface_tex_inv_size.x;
+                mo_vec[1].y = mb->pmv[1][1][1] * 0.5f * r->surface_tex_inv_size.y;
+             }
+         }
+         else {
+            mo_vec[0].x = mb->pmv[0][0][0] * 0.5f * r->surface_tex_inv_size.x;
+            mo_vec[0].y = mb->pmv[0][0][1] * 0.5f * r->surface_tex_inv_size.y;
+
+            if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FIELD) {
+               mo_vec[1].x = mb->pmv[1][0][0] * 0.5f * r->surface_tex_inv_size.x;
+               mo_vec[1].y = mb->pmv[1][0][1] * 0.5f * r->surface_tex_inv_size.y;
+            }
+         }
+
+         if (mb->mb_type == PIPE_MPEG12_MOTION_TYPE_FRAME) {
+            for (i = 0; i < 24 * 2; i += 2) {
+               vb[i].x = mo_vec[0].x;
+               vb[i].y = mo_vec[0].y;
+            }
+         }
+         else {
+            for (i = 0; i < 24 * 2; i += 2) {
+               vb[i].x = mo_vec[0].x;
+               vb[i].y = mo_vec[0].y;
+               vb[i + 1].x = mo_vec[1].x;
+               vb[i + 1].y = mo_vec[1].y;
+            }
+         }
+
+         /* fall-through */
+      }
+      case PIPE_MPEG12_MACROBLOCK_TYPE_INTRA:
+      {
+         const struct vertex2f unit =
+         {
+            r->surface_tex_inv_size.x * MACROBLOCK_WIDTH,
+            r->surface_tex_inv_size.y * MACROBLOCK_HEIGHT
+         };
+         const struct vertex2f half =
+         {
+            r->surface_tex_inv_size.x * (MACROBLOCK_WIDTH / 2),
+            r->surface_tex_inv_size.y * (MACROBLOCK_HEIGHT / 2)
+         };
+         const bool use_zb = r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE;
+
+         struct vert_stream_0 *vb = ycbcr_vb + pos * 24;
+
+         SET_BLOCK(vb, mb->cbp, mb->mbx, mb->mby,
+                   unit.x, unit.y, 0, 0, half.x, half.y,
+                   32, 2, 1, use_zb, r->zero_block);
+
+         SET_BLOCK(vb + 6, mb->cbp, mb->mbx, mb->mby,
+                   unit.x, unit.y, half.x, 0, half.x, half.y,
+                   16, 2, 1, use_zb, r->zero_block);
+
+         SET_BLOCK(vb + 12, mb->cbp, mb->mbx, mb->mby,
+                   unit.x, unit.y, 0, half.y, half.x, half.y,
+                   8, 2, 1, use_zb, r->zero_block);
+
+         SET_BLOCK(vb + 18, mb->cbp, mb->mbx, mb->mby,
+                   unit.x, unit.y, half.x, half.y, half.x, half.y,
+                   4, 2, 1, use_zb, r->zero_block);
+
+         break;
+      }
+      default:
+         assert(0);
+   }
+}
+
+static void
+gen_macroblock_stream(struct vl_mpeg12_mc_renderer *r,
+                      unsigned *num_macroblocks)
+{
+   unsigned offset[NUM_MACROBLOCK_TYPES];
+   struct vert_stream_0 *ycbcr_vb;
+   struct vertex2f *ref_vb[2];
+   unsigned i;
+
+   assert(r);
+   assert(num_macroblocks);
+
+   for (i = 0; i < r->num_macroblocks; ++i) {
+      enum MACROBLOCK_TYPE mb_type = get_macroblock_type(&r->macroblock_buf[i]);
+      ++num_macroblocks[mb_type];
+   }
+
+   offset[0] = 0;
+
+   for (i = 1; i < NUM_MACROBLOCK_TYPES; ++i)
+      offset[i] = offset[i - 1] + num_macroblocks[i - 1];
+
+   ycbcr_vb = (struct vert_stream_0 *)pipe_buffer_map
+   (
+      r->pipe->screen,
+      r->vertex_bufs.individual.ycbcr.buffer,
+      PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
+   );
+
+   for (i = 0; i < 2; ++i)
+      ref_vb[i] = (struct vertex2f *)pipe_buffer_map
+      (
+         r->pipe->screen,
+         r->vertex_bufs.individual.ref[i].buffer,
+         PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
+      );
+
+   for (i = 0; i < r->num_macroblocks; ++i) {
+      enum MACROBLOCK_TYPE mb_type = get_macroblock_type(&r->macroblock_buf[i]);
+
+      gen_macroblock_verts(r, &r->macroblock_buf[i], offset[mb_type],
+                           ycbcr_vb, ref_vb);
+
+      ++offset[mb_type];
+   }
+
+   pipe_buffer_unmap(r->pipe->screen, r->vertex_bufs.individual.ycbcr.buffer);
+   for (i = 0; i < 2; ++i)
+      pipe_buffer_unmap(r->pipe->screen, r->vertex_bufs.individual.ref[i].buffer);
+}
+
+static void
+flush(struct vl_mpeg12_mc_renderer *r)
+{
+   unsigned num_macroblocks[NUM_MACROBLOCK_TYPES] = { 0 };
+   unsigned vb_start = 0;
+   struct vertex_shader_consts *vs_consts;
+   unsigned i;
+
+   assert(r);
+   assert(r->num_macroblocks == r->macroblocks_per_batch);
+
+   gen_macroblock_stream(r, num_macroblocks);
+
+   r->fb_state.cbufs[0] = r->pipe->screen->get_tex_surface
+   (
+      r->pipe->screen, r->surface,
+      0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE
+   );
+
+   r->pipe->set_framebuffer_state(r->pipe, &r->fb_state);
+   r->pipe->set_viewport_state(r->pipe, &r->viewport);
+
+   vs_consts = pipe_buffer_map
+   (
+      r->pipe->screen, r->vs_const_buf.buffer,
+      PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
+   );
+
+   vs_consts->denorm.x = r->surface->width[0];
+   vs_consts->denorm.y = r->surface->height[0];
+
+   pipe_buffer_unmap(r->pipe->screen, r->vs_const_buf.buffer);
+
+   r->pipe->set_constant_buffer(r->pipe, PIPE_SHADER_VERTEX, 0,
+                                &r->vs_const_buf);
+   r->pipe->set_constant_buffer(r->pipe, PIPE_SHADER_FRAGMENT, 0,
+                                &r->fs_const_buf);
+
+   if (num_macroblocks[MACROBLOCK_TYPE_INTRA] > 0) {
+      r->pipe->set_vertex_buffers(r->pipe, 1, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 4, r->vertex_elems);
+      r->pipe->set_sampler_textures(r->pipe, 3, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 3, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->i_vs);
+      r->pipe->bind_fs_state(r->pipe, r->i_fs);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_INTRA] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_INTRA] * 24;
+   }
+
+   if (num_macroblocks[MACROBLOCK_TYPE_FWD_FRAME_PRED] > 0) {
+      r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
+      r->textures.individual.ref[0] = r->past;
+      r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->p_vs[0]);
+      r->pipe->bind_fs_state(r->pipe, r->p_fs[0]);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_FWD_FRAME_PRED] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_FWD_FRAME_PRED] * 24;
+   }
+
+   if (false /*num_macroblocks[MACROBLOCK_TYPE_FWD_FIELD_PRED] > 0 */ ) {
+      r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
+      r->textures.individual.ref[0] = r->past;
+      r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->p_vs[1]);
+      r->pipe->bind_fs_state(r->pipe, r->p_fs[1]);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_FWD_FIELD_PRED] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_FWD_FIELD_PRED] * 24;
+   }
+
+   if (num_macroblocks[MACROBLOCK_TYPE_BKWD_FRAME_PRED] > 0) {
+      r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
+      r->textures.individual.ref[0] = r->future;
+      r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->p_vs[0]);
+      r->pipe->bind_fs_state(r->pipe, r->p_fs[0]);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_BKWD_FRAME_PRED] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_BKWD_FRAME_PRED] * 24;
+   }
+
+   if (false /*num_macroblocks[MACROBLOCK_TYPE_BKWD_FIELD_PRED] > 0 */ ) {
+      r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
+      r->textures.individual.ref[0] = r->future;
+      r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->p_vs[1]);
+      r->pipe->bind_fs_state(r->pipe, r->p_fs[1]);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_BKWD_FIELD_PRED] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_BKWD_FIELD_PRED] * 24;
+   }
+
+   if (num_macroblocks[MACROBLOCK_TYPE_BI_FRAME_PRED] > 0) {
+      r->pipe->set_vertex_buffers(r->pipe, 3, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 8, r->vertex_elems);
+      r->textures.individual.ref[0] = r->past;
+      r->textures.individual.ref[1] = r->future;
+      r->pipe->set_sampler_textures(r->pipe, 5, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 5, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->b_vs[0]);
+      r->pipe->bind_fs_state(r->pipe, r->b_fs[0]);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_BI_FRAME_PRED] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_BI_FRAME_PRED] * 24;
+   }
+
+   if (false /*num_macroblocks[MACROBLOCK_TYPE_BI_FIELD_PRED] > 0 */ ) {
+      r->pipe->set_vertex_buffers(r->pipe, 3, r->vertex_bufs.all);
+      r->pipe->set_vertex_elements(r->pipe, 8, r->vertex_elems);
+      r->textures.individual.ref[0] = r->past;
+      r->textures.individual.ref[1] = r->future;
+      r->pipe->set_sampler_textures(r->pipe, 5, r->textures.all);
+      r->pipe->bind_sampler_states(r->pipe, 5, r->samplers.all);
+      r->pipe->bind_vs_state(r->pipe, r->b_vs[1]);
+      r->pipe->bind_fs_state(r->pipe, r->b_fs[1]);
+
+      r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start,
+                           num_macroblocks[MACROBLOCK_TYPE_BI_FIELD_PRED] * 24);
+      vb_start += num_macroblocks[MACROBLOCK_TYPE_BI_FIELD_PRED] * 24;
+   }
+
+   r->pipe->flush(r->pipe, PIPE_FLUSH_RENDER_CACHE, r->fence);
+   pipe_surface_reference(&r->fb_state.cbufs[0], NULL);
+
+   if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE)
+      for (i = 0; i < 3; ++i)
+         r->zero_block[i].x = ZERO_BLOCK_NIL;
+
+   r->num_macroblocks = 0;
+}
+
+static void
+grab_frame_coded_block(short *src, short *dst, unsigned dst_pitch)
+{
+   unsigned y;
+
+   assert(src);
+   assert(dst);
+
+   for (y = 0; y < BLOCK_HEIGHT; ++y)
+      memcpy(dst + y * dst_pitch, src + y * BLOCK_WIDTH, BLOCK_WIDTH * 2);
+}
+
+static void
+grab_field_coded_block(short *src, short *dst, unsigned dst_pitch)
+{
+   unsigned y;
+
+   assert(src);
+   assert(dst);
+
+   for (y = 0; y < BLOCK_HEIGHT; ++y)
+      memcpy(dst + y * dst_pitch * 2, src + y * BLOCK_WIDTH, BLOCK_WIDTH * 2);
+}
+
+static void
+fill_zero_block(short *dst, unsigned dst_pitch)
+{
+   unsigned y;
+
+   assert(dst);
+
+   for (y = 0; y < BLOCK_HEIGHT; ++y)
+      memset(dst + y * dst_pitch, 0, BLOCK_WIDTH * 2);
+}
+
+static void
+grab_blocks(struct vl_mpeg12_mc_renderer *r, unsigned mbx, unsigned mby,
+            enum pipe_mpeg12_dct_type dct_type, unsigned cbp, short *blocks)
+{
+   unsigned tex_pitch;
+   short *texels;
+   unsigned tb = 0, sb = 0;
+   unsigned mbpx = mbx * MACROBLOCK_WIDTH, mbpy = mby * MACROBLOCK_HEIGHT;
+   unsigned x, y;
+
+   assert(r);
+   assert(blocks);
+
+   tex_pitch = r->tex_transfer[0]->stride / r->tex_transfer[0]->block.size;
+   texels = r->texels[0] + mbpy * tex_pitch + mbpx;
+
+   for (y = 0; y < 2; ++y) {
+      for (x = 0; x < 2; ++x, ++tb) {
+         if ((cbp >> (5 - tb)) & 1) {
+            if (dct_type == PIPE_MPEG12_DCT_TYPE_FRAME) {
+               grab_frame_coded_block(blocks + sb * BLOCK_WIDTH * BLOCK_HEIGHT,
+                                      texels + y * tex_pitch * BLOCK_WIDTH +
+                                      x * BLOCK_WIDTH, tex_pitch);
+            }
+            else {
+               grab_field_coded_block(blocks + sb * BLOCK_WIDTH * BLOCK_HEIGHT,
+                                      texels + y * tex_pitch + x * BLOCK_WIDTH,
+                                      tex_pitch);
+            }
+
+            ++sb;
+         }
+         else if (r->eb_handling != VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE) {
+            if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ALL ||
+                ZERO_BLOCK_IS_NIL(r->zero_block[0])) {
+               fill_zero_block(texels + y * tex_pitch * BLOCK_WIDTH + x * BLOCK_WIDTH, tex_pitch);
+               if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) {
+                  r->zero_block[0].x = (mbpx + x * 8) * r->surface_tex_inv_size.x;
+                  r->zero_block[0].y = (mbpy + y * 8) * r->surface_tex_inv_size.y;
+               }
+            }
+         }
+      }
+   }
+
+   /* TODO: Implement 422, 444 */
+   assert(r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420);
+
+   mbpx /= 2;
+   mbpy /= 2;
+
+   for (tb = 0; tb < 2; ++tb) {
+      tex_pitch = r->tex_transfer[tb + 1]->stride / r->tex_transfer[tb + 1]->block.size;
+      texels = r->texels[tb + 1] + mbpy * tex_pitch + mbpx;
+
+      if ((cbp >> (1 - tb)) & 1) {
+         grab_frame_coded_block(blocks + sb * BLOCK_WIDTH * BLOCK_HEIGHT, texels, tex_pitch);
+         ++sb;
+      }
+      else if (r->eb_handling != VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE) {
+         if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ALL ||
+             ZERO_BLOCK_IS_NIL(r->zero_block[tb + 1])) {
+            fill_zero_block(texels, tex_pitch);
+            if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) {
+               r->zero_block[tb + 1].x = (mbpx << 1) * r->surface_tex_inv_size.x;
+               r->zero_block[tb + 1].y = (mbpy << 1) * r->surface_tex_inv_size.y;
+            }
+         }
+      }
+   }
+}
+
+static void
+grab_macroblock(struct vl_mpeg12_mc_renderer *r,
+                struct pipe_mpeg12_macroblock *mb)
+{
+   assert(r);
+   assert(mb);
+   assert(r->num_macroblocks < r->macroblocks_per_batch);
+
+   memcpy(&r->macroblock_buf[r->num_macroblocks], mb,
+          sizeof(struct pipe_mpeg12_macroblock));
+
+   grab_blocks(r, mb->mbx, mb->mby, mb->dct_type, mb->cbp, mb->blocks);
+
+   ++r->num_macroblocks;
+}
+
+bool
+vl_mpeg12_mc_renderer_init(struct vl_mpeg12_mc_renderer *renderer,
+                           struct pipe_context *pipe,
+                           unsigned picture_width,
+                           unsigned picture_height,
+                           enum pipe_video_chroma_format chroma_format,
+                           enum VL_MPEG12_MC_RENDERER_BUFFER_MODE bufmode,
+                           enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK eb_handling,
+                           bool pot_buffers)
+{
+   unsigned i;
+
+   assert(renderer);
+   assert(pipe);
+   /* TODO: Implement other policies */
+   assert(bufmode == VL_MPEG12_MC_RENDERER_BUFFER_PICTURE);
+   /* TODO: Implement this */
+   /* XXX: XFER_ALL sampling issue at block edges when using bilinear filtering */
+   assert(eb_handling != VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE);
+   /* TODO: Non-pot buffers untested, probably doesn't work without changes to texcoord generation, vert shader, etc */
+   assert(pot_buffers);
+
+   memset(renderer, 0, sizeof(struct vl_mpeg12_mc_renderer));
+
+   renderer->pipe = pipe;
+   renderer->picture_width = picture_width;
+   renderer->picture_height = picture_height;
+   renderer->chroma_format = chroma_format;
+   renderer->bufmode = bufmode;
+   renderer->eb_handling = eb_handling;
+   renderer->pot_buffers = pot_buffers;
+
+   if (!init_pipe_state(renderer))
+      return false;
+   if (!init_shaders(renderer)) {
+      cleanup_pipe_state(renderer);
+      return false;
+   }
+   if (!init_buffers(renderer)) {
+      cleanup_shaders(renderer);
+      cleanup_pipe_state(renderer);
+      return false;
+   }
+
+   renderer->surface = NULL;
+   renderer->past = NULL;
+   renderer->future = NULL;
+   for (i = 0; i < 3; ++i)
+      renderer->zero_block[i].x = ZERO_BLOCK_NIL;
+   renderer->num_macroblocks = 0;
+
+   xfer_buffers_map(renderer);
+
+   return true;
+}
+
+void
+vl_mpeg12_mc_renderer_cleanup(struct vl_mpeg12_mc_renderer *renderer)
+{
+   assert(renderer);
+
+   xfer_buffers_unmap(renderer);
+
+   cleanup_pipe_state(renderer);
+   cleanup_shaders(renderer);
+   cleanup_buffers(renderer);
+}
+
+void
+vl_mpeg12_mc_renderer_render_macroblocks(struct vl_mpeg12_mc_renderer
+                                         *renderer,
+                                         struct pipe_texture *surface,
+                                         struct pipe_texture *past,
+                                         struct pipe_texture *future,
+                                         unsigned num_macroblocks,
+                                         struct pipe_mpeg12_macroblock
+                                         *mpeg12_macroblocks,
+                                         struct pipe_fence_handle **fence)
+{
+   bool new_surface = false;
+
+   assert(renderer);
+   assert(surface);
+   assert(num_macroblocks);
+   assert(mpeg12_macroblocks);
+
+   if (renderer->surface) {
+      if (surface != renderer->surface) {
+         if (renderer->num_macroblocks > 0) {
+            xfer_buffers_unmap(renderer);
+            flush(renderer);
+         }
+         
+         new_surface = true;
+      }
+
+      /* If the surface we're rendering hasn't changed the ref frames shouldn't change. */
+      assert(surface != renderer->surface || renderer->past == past);
+      assert(surface != renderer->surface || renderer->future == future);
+   }
+   else
+      new_surface = true;
+
+   if (new_surface) {
+      renderer->surface = surface;
+      renderer->past = past;
+      renderer->future = future;
+      renderer->fence = fence;
+      renderer->surface_tex_inv_size.x = 1.0f / surface->width[0];
+      renderer->surface_tex_inv_size.y = 1.0f / surface->height[0];
+   }
+
+   while (num_macroblocks) {
+      unsigned left_in_batch = renderer->macroblocks_per_batch - renderer->num_macroblocks;
+      unsigned num_to_submit = MIN2(num_macroblocks, left_in_batch);
+      unsigned i;
+
+      for (i = 0; i < num_to_submit; ++i) {
+         assert(mpeg12_macroblocks[i].base.codec == PIPE_VIDEO_CODEC_MPEG12);
+         grab_macroblock(renderer, &mpeg12_macroblocks[i]);
+      }
+
+      num_macroblocks -= num_to_submit;
+
+      if (renderer->num_macroblocks == renderer->macroblocks_per_batch) {
+         xfer_buffers_unmap(renderer);
+         flush(renderer);
+         xfer_buffers_map(renderer);
+         /* Next time we get this surface it may have new ref frames */
+         renderer->surface = NULL;
+      }
+   }
+}
diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h
new file mode 100644
index 0000000000..5d2c1273ee
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h
@@ -0,0 +1,120 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef vl_mpeg12_mc_renderer_h
+#define vl_mpeg12_mc_renderer_h
+
+#include <pipe/p_compiler.h>
+#include <pipe/p_state.h>
+#include <pipe/p_video_state.h>
+
+struct pipe_context;
+struct pipe_video_surface;
+struct pipe_macroblock;
+
+/* A slice is video-width (rounded up to a multiple of macroblock width) x macroblock height */
+enum VL_MPEG12_MC_RENDERER_BUFFER_MODE
+{
+   VL_MPEG12_MC_RENDERER_BUFFER_SLICE,  /* Saves memory at the cost of smaller batches */
+   VL_MPEG12_MC_RENDERER_BUFFER_PICTURE /* Larger batches, more memory */
+};
+
+enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK
+{
+   VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ALL, /* Waste of memory bandwidth */
+   VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE, /* Can only do point-filtering when interpolating subsampled chroma channels */
+   VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE /* Needs conditional texel fetch! */
+};
+
+struct vl_mpeg12_mc_renderer
+{
+   struct pipe_context *pipe;
+   unsigned picture_width;
+   unsigned picture_height;
+   enum pipe_video_chroma_format chroma_format;
+   enum VL_MPEG12_MC_RENDERER_BUFFER_MODE bufmode;
+   enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK eb_handling;
+   bool pot_buffers;
+   unsigned macroblocks_per_batch;
+
+   struct pipe_viewport_state viewport;
+   struct pipe_constant_buffer vs_const_buf;
+   struct pipe_constant_buffer fs_const_buf;
+   struct pipe_framebuffer_state fb_state;
+   struct pipe_vertex_element vertex_elems[8];
+	
+   union
+   {
+      void *all[5];
+      struct { void *y, *cb, *cr, *ref[2]; } individual;
+   } samplers;
+	
+   void *i_vs, *p_vs[2], *b_vs[2];
+   void *i_fs, *p_fs[2], *b_fs[2];
+	
+   union
+   {
+      struct pipe_texture *all[5];
+      struct { struct pipe_texture *y, *cb, *cr, *ref[2]; } individual;
+   } textures;
+
+   union
+   {
+      struct pipe_vertex_buffer all[3];
+      struct { struct pipe_vertex_buffer ycbcr, ref[2]; } individual;
+   } vertex_bufs;
+	
+   struct pipe_texture *surface, *past, *future;
+   struct pipe_fence_handle **fence;
+   unsigned num_macroblocks;
+   struct pipe_mpeg12_macroblock *macroblock_buf;
+   struct pipe_transfer *tex_transfer[3];
+   short *texels[3];
+   struct { float x, y; } surface_tex_inv_size;
+   struct { float x, y; } zero_block[3];
+};
+
+bool vl_mpeg12_mc_renderer_init(struct vl_mpeg12_mc_renderer *renderer,
+                                struct pipe_context *pipe,
+                                unsigned picture_width,
+                                unsigned picture_height,
+                                enum pipe_video_chroma_format chroma_format,
+                                enum VL_MPEG12_MC_RENDERER_BUFFER_MODE bufmode,
+                                enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK eb_handling,
+                                bool pot_buffers);
+
+void vl_mpeg12_mc_renderer_cleanup(struct vl_mpeg12_mc_renderer *renderer);
+
+void vl_mpeg12_mc_renderer_render_macroblocks(struct vl_mpeg12_mc_renderer *renderer,
+                                              struct pipe_texture *surface,
+                                              struct pipe_texture *past,
+                                              struct pipe_texture *future,
+                                              unsigned num_macroblocks,
+                                              struct pipe_mpeg12_macroblock *mpeg12_macroblocks,
+                                              struct pipe_fence_handle **fence);
+
+#endif /* vl_mpeg12_mc_renderer_h */
diff --git a/src/gallium/auxiliary/vl/vl_shader_build.c b/src/gallium/auxiliary/vl/vl_shader_build.c
new file mode 100644
index 0000000000..faa20a903c
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_shader_build.c
@@ -0,0 +1,242 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "vl_shader_build.h"
+#include <assert.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+
+struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
+{
+   struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+   decl.Declaration.File = TGSI_FILE_INPUT;
+   decl.Declaration.Semantic = 1;
+   decl.Semantic.SemanticName = name;
+   decl.Semantic.SemanticIndex = index;
+   decl.DeclarationRange.First = first;
+   decl.DeclarationRange.Last = last;
+
+   return decl;
+}
+
+struct tgsi_full_declaration vl_decl_interpolated_input
+(
+   unsigned int name,
+   unsigned int index,
+   unsigned int first,
+   unsigned int last,
+   int interpolation
+)
+{
+   struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+   assert
+   (
+      interpolation == TGSI_INTERPOLATE_CONSTANT ||
+      interpolation == TGSI_INTERPOLATE_LINEAR ||
+      interpolation == TGSI_INTERPOLATE_PERSPECTIVE
+   );
+
+   decl.Declaration.File = TGSI_FILE_INPUT;
+   decl.Declaration.Semantic = 1;
+   decl.Semantic.SemanticName = name;
+   decl.Semantic.SemanticIndex = index;
+   decl.Declaration.Interpolate = interpolation;;
+   decl.DeclarationRange.First = first;
+   decl.DeclarationRange.Last = last;
+
+   return decl;
+}
+
+struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
+{
+   struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+   decl.Declaration.File = TGSI_FILE_CONSTANT;
+   decl.Declaration.Semantic = 1;
+   decl.Semantic.SemanticName = name;
+   decl.Semantic.SemanticIndex = index;
+   decl.DeclarationRange.First = first;
+   decl.DeclarationRange.Last = last;
+
+   return decl;
+}
+
+struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
+{
+   struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+   decl.Declaration.File = TGSI_FILE_OUTPUT;
+   decl.Declaration.Semantic = 1;
+   decl.Semantic.SemanticName = name;
+   decl.Semantic.SemanticIndex = index;
+   decl.DeclarationRange.First = first;
+   decl.DeclarationRange.Last = last;
+
+   return decl;
+}
+
+struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last)
+{
+   struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+   decl = tgsi_default_full_declaration();
+   decl.Declaration.File = TGSI_FILE_TEMPORARY;
+   decl.DeclarationRange.First = first;
+   decl.DeclarationRange.Last = last;
+
+   return decl;
+}
+
+struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last)
+{
+   struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+   decl = tgsi_default_full_declaration();
+   decl.Declaration.File = TGSI_FILE_SAMPLER;
+   decl.DeclarationRange.First = first;
+   decl.DeclarationRange.Last = last;
+
+   return decl;
+}
+
+struct tgsi_full_instruction vl_inst2
+(
+   int opcode,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src_file,
+   unsigned int src_index
+)
+{
+   struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+   inst.Instruction.Opcode = opcode;
+   inst.Instruction.NumDstRegs = 1;
+   inst.FullDstRegisters[0].DstRegister.File = dst_file;
+   inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+   inst.Instruction.NumSrcRegs = 1;
+   inst.FullSrcRegisters[0].SrcRegister.File = src_file;
+   inst.FullSrcRegisters[0].SrcRegister.Index = src_index;
+
+   return inst;
+}
+
+struct tgsi_full_instruction vl_inst3
+(
+   int opcode,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src1_file,
+   unsigned int src1_index,
+   enum tgsi_file_type src2_file,
+   unsigned int src2_index
+)
+{
+   struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+   inst.Instruction.Opcode = opcode;
+   inst.Instruction.NumDstRegs = 1;
+   inst.FullDstRegisters[0].DstRegister.File = dst_file;
+   inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+   inst.Instruction.NumSrcRegs = 2;
+   inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
+   inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
+   inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
+   inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+
+   return inst;
+}
+
+struct tgsi_full_instruction vl_tex
+(
+   int tex,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src1_file,
+   unsigned int src1_index,
+   enum tgsi_file_type src2_file,
+   unsigned int src2_index
+)
+{
+   struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+   inst.Instruction.Opcode = TGSI_OPCODE_TEX;
+   inst.Instruction.NumDstRegs = 1;
+   inst.FullDstRegisters[0].DstRegister.File = dst_file;
+   inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+   inst.Instruction.NumSrcRegs = 2;
+   inst.InstructionExtTexture.Texture = tex;
+   inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
+   inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
+   inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
+   inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+
+   return inst;
+}
+
+struct tgsi_full_instruction vl_inst4
+(
+   int opcode,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src1_file,
+   unsigned int src1_index,
+   enum tgsi_file_type src2_file,
+   unsigned int src2_index,
+   enum tgsi_file_type src3_file,
+   unsigned int src3_index
+)
+{
+   struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+   inst.Instruction.Opcode = opcode;
+   inst.Instruction.NumDstRegs = 1;
+   inst.FullDstRegisters[0].DstRegister.File = dst_file;
+   inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+   inst.Instruction.NumSrcRegs = 3;
+   inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
+   inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
+   inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
+   inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+   inst.FullSrcRegisters[2].SrcRegister.File = src3_file;
+   inst.FullSrcRegisters[2].SrcRegister.Index = src3_index;
+
+   return inst;
+}
+
+struct tgsi_full_instruction vl_end(void)
+{
+   struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+   inst.Instruction.Opcode = TGSI_OPCODE_END;
+   inst.Instruction.NumDstRegs = 0;
+   inst.Instruction.NumSrcRegs = 0;
+
+   return inst;
+}
diff --git a/src/gallium/auxiliary/vl/vl_shader_build.h b/src/gallium/auxiliary/vl/vl_shader_build.h
new file mode 100644
index 0000000000..5da71f8e13
--- /dev/null
+++ b/src/gallium/auxiliary/vl/vl_shader_build.h
@@ -0,0 +1,88 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef vl_shader_build_h
+#define vl_shader_build_h
+
+#include <pipe/p_shader_tokens.h>
+
+struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_interpolated_input
+(
+   unsigned int name,
+   unsigned int index,
+   unsigned int first,
+   unsigned int last,
+   int interpolation
+);
+struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last);
+struct tgsi_full_instruction vl_inst2
+(
+   int opcode,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src_file,
+   unsigned int src_index
+);
+struct tgsi_full_instruction vl_inst3
+(
+   int opcode,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src1_file,
+   unsigned int src1_index,
+   enum tgsi_file_type src2_file,
+   unsigned int src2_index
+);
+struct tgsi_full_instruction vl_tex
+(
+   int tex,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src1_file,
+   unsigned int src1_index,
+   enum tgsi_file_type src2_file,
+   unsigned int src2_index
+);
+struct tgsi_full_instruction vl_inst4
+(
+   int opcode,
+   enum tgsi_file_type dst_file,
+   unsigned int dst_index,
+   enum tgsi_file_type src1_file,
+   unsigned int src1_index,
+   enum tgsi_file_type src2_file,
+   unsigned int src2_index,
+   enum tgsi_file_type src3_file,
+   unsigned int src3_index
+);
+struct tgsi_full_instruction vl_end(void);
+
+#endif
diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c
index bd48ce7005..d185c6b849 100644
--- a/src/gallium/drivers/cell/ppu/cell_screen.c
+++ b/src/gallium/drivers/cell/ppu/cell_screen.c
@@ -41,7 +41,7 @@
 static const char *
 cell_get_vendor(struct pipe_screen *screen)
 {
-   return "Tungsten Graphics, Inc.";
+   return "VMware, Inc.";
 }
 
 
@@ -64,8 +64,6 @@ cell_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_GLSL:
       return 1;
-   case PIPE_CAP_S3TC:
-      return 0;
    case PIPE_CAP_ANISOTROPIC_FILTER:
       return 0;
    case PIPE_CAP_POINT_SPRITE:
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c
index 6a63a0e6ce..ae4c61efb3 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.c
+++ b/src/gallium/drivers/cell/ppu/cell_texture.c
@@ -389,22 +389,14 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
    const uint texWidth = pt->width[level];
    const uint texHeight = pt->height[level];
    const uint stride = ct->stride[level];
-   unsigned flags = 0x0;
    unsigned size;
 
    assert(transfer->texture);
 
-   if (transfer->usage != PIPE_TRANSFER_READ) {
-      flags |= PIPE_BUFFER_USAGE_CPU_WRITE;
-   }
-
-   if (transfer->usage != PIPE_TRANSFER_WRITE) {
-      flags |= PIPE_BUFFER_USAGE_CPU_READ;
-   }
-
    if (!ct->mapped) {
       /* map now */
-      ct->mapped = pipe_buffer_map(screen, ct->buffer, flags);
+      ct->mapped = pipe_buffer_map(screen, ct->buffer,
+                                   pipe_transfer_buffer_flags(transfer));
    }
 
    /*
@@ -417,8 +409,7 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
    if (!ctrans->map)
       return NULL; /* out of memory */
 
-   if (transfer->usage == PIPE_TRANSFER_READ ||
-       transfer->usage == PIPE_TRANSFER_READ_WRITE) {
+   if (transfer->usage & PIPE_TRANSFER_READ) {
       /* need to untwiddle the texture to make a linear version */
       const uint bpp = pf_get_size(ct->base.format);
       if (bpp == 4) {
@@ -459,8 +450,7 @@ cell_transfer_unmap(struct pipe_screen *screen,
                                    PIPE_BUFFER_USAGE_CPU_READ);
    }
 
-   if (transfer->usage == PIPE_TRANSFER_WRITE ||
-       transfer->usage == PIPE_TRANSFER_READ_WRITE) {
+   if (transfer->usage & PIPE_TRANSFER_WRITE) {
       /* The user wrote new texture data into the mapped buffer.
        * We need to convert the new linear data into the twiddled/tiled format.
        */
diff --git a/src/gallium/drivers/i915simple/Makefile b/src/gallium/drivers/i915/Makefile
index fb533c1796..e33c74d02f 100644
--- a/src/gallium/drivers/i915simple/Makefile
+++ b/src/gallium/drivers/i915/Makefile
@@ -1,7 +1,7 @@
 TOP = ../../../..
 include $(TOP)/configs/current
 
-LIBNAME = i915simple
+LIBNAME = i915
 
 C_SOURCES = \
 	i915_blit.c \
diff --git a/src/gallium/drivers/i915simple/SConscript b/src/gallium/drivers/i915/SConscript
index 778c4ed0fd..5a1c47c88d 100644
--- a/src/gallium/drivers/i915simple/SConscript
+++ b/src/gallium/drivers/i915/SConscript
@@ -2,8 +2,8 @@ Import('*')
 
 env = env.Clone()
 
-i915simple = env.ConvenienceLibrary(
-	target = 'i915simple',
+i915 = env.ConvenienceLibrary(
+	target = 'i915',
 	source = [
 		'i915_blit.c',
 		'i915_buffer.c',
@@ -27,4 +27,4 @@ i915simple = env.ConvenienceLibrary(
 		'i915_texture.c',
 	])
 
-Export('i915simple')
+Export('i915')
diff --git a/src/gallium/drivers/i915simple/i915_batch.h b/src/gallium/drivers/i915/i915_batch.h
index b813784723..b813784723 100644
--- a/src/gallium/drivers/i915simple/i915_batch.h
+++ b/src/gallium/drivers/i915/i915_batch.h
diff --git a/src/gallium/drivers/i915simple/i915_blit.c b/src/gallium/drivers/i915/i915_blit.c
index 83dfc33528..83dfc33528 100644
--- a/src/gallium/drivers/i915simple/i915_blit.c
+++ b/src/gallium/drivers/i915/i915_blit.c
diff --git a/src/gallium/drivers/i915simple/i915_blit.h b/src/gallium/drivers/i915/i915_blit.h
index 8ce3220cfd..8ce3220cfd 100644
--- a/src/gallium/drivers/i915simple/i915_blit.h
+++ b/src/gallium/drivers/i915/i915_blit.h
diff --git a/src/gallium/drivers/i915simple/i915_buffer.c b/src/gallium/drivers/i915/i915_buffer.c
index effeba1297..effeba1297 100644
--- a/src/gallium/drivers/i915simple/i915_buffer.c
+++ b/src/gallium/drivers/i915/i915_buffer.c
diff --git a/src/gallium/drivers/i915simple/i915_buffer.h b/src/gallium/drivers/i915/i915_buffer.h
index 80fda7c62f..80fda7c62f 100644
--- a/src/gallium/drivers/i915simple/i915_buffer.h
+++ b/src/gallium/drivers/i915/i915_buffer.h
diff --git a/src/gallium/drivers/i915simple/i915_clear.c b/src/gallium/drivers/i915/i915_clear.c
index 90530f2826..90530f2826 100644
--- a/src/gallium/drivers/i915simple/i915_clear.c
+++ b/src/gallium/drivers/i915/i915_clear.c
diff --git a/src/gallium/drivers/i915simple/i915_context.c b/src/gallium/drivers/i915/i915_context.c
index b43f735245..e745f3342d 100644
--- a/src/gallium/drivers/i915simple/i915_context.c
+++ b/src/gallium/drivers/i915/i915_context.c
@@ -175,12 +175,19 @@ i915_is_buffer_referenced(struct pipe_context *pipe,
 static void i915_destroy(struct pipe_context *pipe)
 {
    struct i915_context *i915 = i915_context(pipe);
+   int i;
 
    draw_destroy(i915->draw);
    
    if(i915->batch)
       i915->iws->batchbuffer_destroy(i915->batch);
 
+   /* unbind framebuffer */
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      pipe_surface_reference(&i915->framebuffer.cbufs[i], NULL);
+   }
+   pipe_surface_reference(&i915->framebuffer.zsbuf, NULL);
+
    FREE(i915);
 }
 
diff --git a/src/gallium/drivers/i915simple/i915_context.h b/src/gallium/drivers/i915/i915_context.h
index 234b441ce6..234b441ce6 100644
--- a/src/gallium/drivers/i915simple/i915_context.h
+++ b/src/gallium/drivers/i915/i915_context.h
diff --git a/src/gallium/drivers/i915simple/i915_debug.c b/src/gallium/drivers/i915/i915_debug.c
index ce92d1af9a..e6640e587b 100644
--- a/src/gallium/drivers/i915simple/i915_debug.c
+++ b/src/gallium/drivers/i915/i915_debug.c
@@ -880,7 +880,7 @@ i915_dump_batchbuffer( struct intel_batchbuffer *batch )
       return;
    }
    
-   debug_printf( "\n\nBATCH: (%d)\n", bytes / 4);
+   debug_printf( "\n\nBATCH: (%d)\n", (int)bytes / 4);
 
    while (!done &&
 	  stream.offset < bytes)
diff --git a/src/gallium/drivers/i915simple/i915_debug.h b/src/gallium/drivers/i915/i915_debug.h
index dd9b86e17b..dd9b86e17b 100644
--- a/src/gallium/drivers/i915simple/i915_debug.h
+++ b/src/gallium/drivers/i915/i915_debug.h
diff --git a/src/gallium/drivers/i915simple/i915_debug_fp.c b/src/gallium/drivers/i915/i915_debug_fp.c
index 9c5b117b6d..9c5b117b6d 100644
--- a/src/gallium/drivers/i915simple/i915_debug_fp.c
+++ b/src/gallium/drivers/i915/i915_debug_fp.c
diff --git a/src/gallium/drivers/i915simple/i915_flush.c b/src/gallium/drivers/i915/i915_flush.c
index 1582168eba..1582168eba 100644
--- a/src/gallium/drivers/i915simple/i915_flush.c
+++ b/src/gallium/drivers/i915/i915_flush.c
diff --git a/src/gallium/drivers/i915simple/i915_fpc.h b/src/gallium/drivers/i915/i915_fpc.h
index 2f0f99d046..2f0f99d046 100644
--- a/src/gallium/drivers/i915simple/i915_fpc.h
+++ b/src/gallium/drivers/i915/i915_fpc.h
diff --git a/src/gallium/drivers/i915simple/i915_fpc_emit.c b/src/gallium/drivers/i915/i915_fpc_emit.c
index b054ce41d3..b054ce41d3 100644
--- a/src/gallium/drivers/i915simple/i915_fpc_emit.c
+++ b/src/gallium/drivers/i915/i915_fpc_emit.c
diff --git a/src/gallium/drivers/i915simple/i915_fpc_translate.c b/src/gallium/drivers/i915/i915_fpc_translate.c
index 89504ced27..1fe5cda956 100644
--- a/src/gallium/drivers/i915simple/i915_fpc_translate.c
+++ b/src/gallium/drivers/i915/i915_fpc_translate.c
@@ -127,7 +127,7 @@ i915_program_error(struct i915_fp_compile *p, const char *msg, ...)
    va_start( args, msg );  
    util_vsnprintf( buffer, sizeof(buffer), msg, args );
    va_end( args );
-   debug_printf(buffer);
+   debug_printf("%s", buffer);
    debug_printf("\n");
 
    p->error = 1;
diff --git a/src/gallium/drivers/i915simple/i915_prim_emit.c b/src/gallium/drivers/i915/i915_prim_emit.c
index d9a5c40ab9..d9a5c40ab9 100644
--- a/src/gallium/drivers/i915simple/i915_prim_emit.c
+++ b/src/gallium/drivers/i915/i915_prim_emit.c
diff --git a/src/gallium/drivers/i915simple/i915_prim_vbuf.c b/src/gallium/drivers/i915/i915_prim_vbuf.c
index 508f4560e4..6b832140a8 100644
--- a/src/gallium/drivers/i915simple/i915_prim_vbuf.c
+++ b/src/gallium/drivers/i915/i915_prim_vbuf.c
@@ -44,6 +44,7 @@
 #include "pipe/p_inlines.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
+#include "util/u_fifo.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
@@ -51,6 +52,9 @@
 #include "i915_state.h"
 
 
+#undef VBUF_USE_FIFO
+#undef VBUF_MAP_BUFFER
+
 /**
  * Primitive renderer for i915.
  */
@@ -73,11 +77,25 @@ struct i915_vbuf_render {
 
    /* Stuff for the vbo */
    struct intel_buffer *vbo;
-   size_t vbo_size;
+   size_t vbo_size; /**< current size of allocated buffer */
+   size_t vbo_alloc_size; /**< minimum buffer size to allocate */
    size_t vbo_offset;
    void *vbo_ptr;
-   size_t vbo_alloc_size;
    size_t vbo_max_used;
+
+#ifndef VBUF_MAP_BUFFER
+   size_t map_used_start;
+   size_t map_used_end;
+   size_t map_size;
+#endif
+
+#ifdef VBUF_USE_FIFO
+   /* Stuff for the pool */
+   struct util_fifo *pool_fifo;
+   unsigned pool_used;
+   unsigned pool_buffer_size;
+   boolean pool_not_used;
+#endif
 };
 
 
@@ -106,33 +124,90 @@ i915_vbuf_render_get_vertex_info(struct vbuf_render *render)
 }
 
 static boolean
+i915_vbuf_render_reserve(struct i915_vbuf_render *i915_render, size_t size)
+{
+   struct i915_context *i915 = i915_render->i915;
+
+   if (i915_render->vbo_size < size + i915_render->vbo_offset)
+      return FALSE;
+
+   if (i915->vbo_flushed)
+      return FALSE;
+
+   return TRUE;
+}
+
+static void
+i915_vbuf_render_new_buf(struct i915_vbuf_render *i915_render, size_t size)
+{
+   struct i915_context *i915 = i915_render->i915;
+   struct intel_winsys *iws = i915->iws;
+
+   if (i915_render->vbo) {
+#ifdef VBUF_USE_FIFO
+      if (i915_render->pool_not_used)
+         iws->buffer_destroy(iws, i915_render->vbo);
+      else
+         u_fifo_add(i915_render->pool_fifo, i915_render->vbo);
+      i915_render->vbo = NULL;
+#else
+      iws->buffer_destroy(iws, i915_render->vbo);
+#endif
+   }
+
+   i915->vbo_flushed = 0;
+
+   i915_render->vbo_size = MAX2(size, i915_render->vbo_alloc_size);
+   i915_render->vbo_offset = 0;
+
+#ifndef VBUF_MAP_BUFFER
+   if (i915_render->vbo_size > i915_render->map_size) {
+      i915_render->map_size = i915_render->vbo_size;
+      FREE(i915_render->vbo_ptr);
+      i915_render->vbo_ptr = MALLOC(i915_render->map_size);
+   }
+#endif
+
+#ifdef VBUF_USE_FIFO
+   if (i915_render->vbo_size != i915_render->pool_buffer_size) {
+      i915_render->pool_not_used = TRUE;
+      i915_render->vbo = iws->buffer_create(iws, i915_render->vbo_size, 64,
+            INTEL_NEW_VERTEX);
+   } else {
+      i915_render->pool_not_used = FALSE;
+
+      if (i915_render->pool_used >= 2) {
+         FLUSH_BATCH(NULL);
+         i915->vbo_flushed = 0;
+         i915_render->pool_used = 0;
+      }
+      u_fifo_pop(i915_render->pool_fifo, (void**)&i915_render->vbo);
+   }
+#else
+   i915_render->vbo = iws->buffer_create(iws, i915_render->vbo_size,
+                                         64, INTEL_NEW_VERTEX);
+#endif
+}
+
+static boolean
 i915_vbuf_render_allocate_vertices(struct vbuf_render *render,
                                    ushort vertex_size,
                                    ushort nr_vertices)
 {
    struct i915_vbuf_render *i915_render = i915_vbuf_render(render);
    struct i915_context *i915 = i915_render->i915;
-   struct intel_winsys *iws = i915->iws;
    size_t size = (size_t)vertex_size * (size_t)nr_vertices;
 
    /* FIXME: handle failure */
    assert(!i915->vbo);
 
-   if (i915_render->vbo_size > size + i915_render->vbo_offset && !i915->vbo_flushed) {
-   } else {
-      i915->vbo_flushed = 0;
-      if (i915_render->vbo) {
-         iws->buffer_destroy(iws, i915_render->vbo);
-         i915_render->vbo = NULL;
-      }
-   }
-
-   if (!i915_render->vbo) {
-      i915_render->vbo_size = MAX2(size, i915_render->vbo_alloc_size);
-      i915_render->vbo_offset = 0;
-      i915_render->vbo = iws->buffer_create(iws, i915_render->vbo_size, 64,
-                                            INTEL_NEW_VERTEX);
-
+   if (!i915_vbuf_render_reserve(i915_render, size)) {
+#ifdef VBUF_USE_FIFO
+      /* incase we flushed reset the number of pool buffers used */
+      if (i915->vbo_flushed)
+         i915_render->pool_used = 0;
+#endif
+      i915_vbuf_render_new_buf(i915_render, size);
    }
 
    i915_render->vertex_size = vertex_size;
@@ -153,11 +228,15 @@ i915_vbuf_render_map_vertices(struct vbuf_render *render)
    struct intel_winsys *iws = i915->iws;
 
    if (i915->vbo_flushed)
-      debug_printf("%s bad vbo flush occured stalling on hw\n");
+      debug_printf("%s bad vbo flush occured stalling on hw\n", __FUNCTION__);
 
+#ifdef VBUF_MAP_BUFFER
    i915_render->vbo_ptr = iws->buffer_map(iws, i915_render->vbo, TRUE);
-
-   return (unsigned char *)i915_render->vbo_ptr + i915->vbo_offset;
+   return (unsigned char *)i915_render->vbo_ptr + i915_render->vbo_offset;
+#else
+   (void)iws;
+   return (unsigned char *)i915_render->vbo_ptr;
+#endif
 }
 
 static void
@@ -170,7 +249,17 @@ i915_vbuf_render_unmap_vertices(struct vbuf_render *render,
    struct intel_winsys *iws = i915->iws;
 
    i915_render->vbo_max_used = MAX2(i915_render->vbo_max_used, i915_render->vertex_size * (max_index + 1));
+#ifdef VBUF_MAP_BUFFER
    iws->buffer_unmap(iws, i915_render->vbo);
+#else
+   i915_render->map_used_start = i915_render->vertex_size * min_index;
+   i915_render->map_used_end = i915_render->vertex_size * (max_index + 1);
+   iws->buffer_write(iws, i915_render->vbo,
+                     i915_render->map_used_start + i915_render->vbo_offset,
+                     i915_render->map_used_end - i915_render->map_used_start,
+                     (unsigned char *)i915_render->vbo_ptr + i915_render->map_used_start);
+
+#endif
 }
 
 static boolean
@@ -344,14 +433,43 @@ i915_vbuf_render_draw_arrays(struct vbuf_render *render,
                              uint nr)
 {
    struct i915_vbuf_render *i915_render = i915_vbuf_render(render);
+   struct i915_context *i915 = i915_render->i915;
 
    if (i915_render->fallback) {
       draw_arrays_fallback(render, start, nr);
       return;
    }
 
-   /* JB: TODO submit direct cmds */
-   draw_arrays_fallback(render, start, nr);
+   if (i915->dirty)
+      i915_update_derived(i915);
+
+   if (i915->hardware_dirty)
+      i915_emit_hardware_state(i915);
+
+   if (!BEGIN_BATCH(2, 0)) {
+      FLUSH_BATCH(NULL);
+
+      /* Make sure state is re-emitted after a flush:
+       */
+      i915_update_derived(i915);
+      i915_emit_hardware_state(i915);
+      i915->vbo_flushed = 1;
+
+      if (!BEGIN_BATCH(2, 0)) {
+         assert(0);
+         goto out;
+      }
+   }
+
+   OUT_BATCH(_3DPRIMITIVE |
+             PRIM_INDIRECT |
+             PRIM_INDIRECT_SEQUENTIAL |
+             i915_render->hwprim |
+             nr);
+   OUT_BATCH(start); /* Beginning vertex index */
+
+out:
+   return;
 }
 
 /**
@@ -504,11 +622,12 @@ i915_vbuf_render_create(struct i915_context *i915)
 {
    struct i915_vbuf_render *i915_render = CALLOC_STRUCT(i915_vbuf_render);
    struct intel_winsys *iws = i915->iws;
+   int i;
 
    i915_render->i915 = i915;
-   
-   i915_render->base.max_vertex_buffer_bytes = 128*1024;
-   
+
+   i915_render->base.max_vertex_buffer_bytes = 16*4096;
+
    /* NOTE: it must be such that state and vertices indices fit in a single 
     * batch buffer.
     */
@@ -524,14 +643,30 @@ i915_vbuf_render_create(struct i915_context *i915)
    i915_render->base.release_vertices = i915_vbuf_render_release_vertices;
    i915_render->base.destroy = i915_vbuf_render_destroy;
 
-   i915_render->vbo_alloc_size = 128 * 4096;
-   i915_render->vbo_size = i915_render->vbo_alloc_size;
+#ifndef VBUF_MAP_BUFFER
+   i915_render->map_size = 0;
+   i915_render->map_used_start = 0;
+   i915_render->map_used_end = 0;
+#endif
+
+   i915_render->vbo = NULL;
+   i915_render->vbo_ptr = NULL;
+   i915_render->vbo_size = 0;
    i915_render->vbo_offset = 0;
-   i915_render->vbo = iws->buffer_create(iws, i915_render->vbo_size, 64,
-                                         INTEL_NEW_VERTEX);
-   /* TODO JB: is this realy needed? */
-   i915_render->vbo_ptr = iws->buffer_map(iws, i915_render->vbo, TRUE);
-   iws->buffer_unmap(iws, i915_render->vbo);
+   i915_render->vbo_alloc_size = i915_render->base.max_vertex_buffer_bytes * 4;
+
+#ifdef VBUF_USE_POOL
+   i915_render->pool_used = FALSE;
+   i915_render->pool_buffer_size = i915_render->vbo_alloc_size;
+   i915_render->pool_fifo = u_fifo_create(6);
+   for (i = 0; i < 6; i++)
+      u_fifo_add(i915_render->pool_fifo,
+                 iws->buffer_create(iws, i915_render->pool_buffer_size, 64,
+                                    INTEL_NEW_VERTEX));
+#else
+   (void)i;
+   (void)iws;
+#endif
 
    return &i915_render->base;
 }
diff --git a/src/gallium/drivers/i915simple/i915_reg.h b/src/gallium/drivers/i915/i915_reg.h
index 04620fec68..04620fec68 100644
--- a/src/gallium/drivers/i915simple/i915_reg.h
+++ b/src/gallium/drivers/i915/i915_reg.h
diff --git a/src/gallium/drivers/i915simple/i915_screen.c b/src/gallium/drivers/i915/i915_screen.c
index 9f017a14cc..c66558c320 100644
--- a/src/gallium/drivers/i915simple/i915_screen.c
+++ b/src/gallium/drivers/i915/i915_screen.c
@@ -46,7 +46,7 @@
 static const char *
 i915_get_vendor(struct pipe_screen *screen)
 {
-   return "Tungsten Graphics, Inc.";
+   return "VMware, Inc.";
 }
 
 static const char *
@@ -101,8 +101,6 @@ i915_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_GLSL:
       return 0;
-   case PIPE_CAP_S3TC:
-      return 0;
    case PIPE_CAP_ANISOTROPIC_FILTER:
       return 0;
    case PIPE_CAP_POINT_SPRITE:
diff --git a/src/gallium/drivers/i915simple/i915_screen.h b/src/gallium/drivers/i915/i915_screen.h
index 5126485caa..5126485caa 100644
--- a/src/gallium/drivers/i915simple/i915_screen.h
+++ b/src/gallium/drivers/i915/i915_screen.h
diff --git a/src/gallium/drivers/i915simple/i915_state.c b/src/gallium/drivers/i915/i915_state.c
index 0087dfa410..7d48e6e84d 100644
--- a/src/gallium/drivers/i915simple/i915_state.c
+++ b/src/gallium/drivers/i915/i915_state.c
@@ -588,9 +588,17 @@ static void i915_set_framebuffer_state(struct pipe_context *pipe,
 				       const struct pipe_framebuffer_state *fb)
 {
    struct i915_context *i915 = i915_context(pipe);
+   int i;
+
    draw_flush(i915->draw);
 
-   i915->framebuffer = *fb; /* struct copy */
+   i915->framebuffer.width = fb->width;
+   i915->framebuffer.height = fb->height;
+   i915->framebuffer.nr_cbufs = fb->nr_cbufs;
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      pipe_surface_reference(&i915->framebuffer.cbufs[i], fb->cbufs[i]);
+   }
+   pipe_surface_reference(&i915->framebuffer.zsbuf, fb->zsbuf);
 
    i915->dirty |= I915_NEW_FRAMEBUFFER;
 }
diff --git a/src/gallium/drivers/i915simple/i915_state.h b/src/gallium/drivers/i915/i915_state.h
index 86c6b0027d..86c6b0027d 100644
--- a/src/gallium/drivers/i915simple/i915_state.h
+++ b/src/gallium/drivers/i915/i915_state.h
diff --git a/src/gallium/drivers/i915simple/i915_state_derived.c b/src/gallium/drivers/i915/i915_state_derived.c
index 178d4e8781..178d4e8781 100644
--- a/src/gallium/drivers/i915simple/i915_state_derived.c
+++ b/src/gallium/drivers/i915/i915_state_derived.c
diff --git a/src/gallium/drivers/i915simple/i915_state_dynamic.c b/src/gallium/drivers/i915/i915_state_dynamic.c
index 86126a5a15..86126a5a15 100644
--- a/src/gallium/drivers/i915simple/i915_state_dynamic.c
+++ b/src/gallium/drivers/i915/i915_state_dynamic.c
diff --git a/src/gallium/drivers/i915simple/i915_state_emit.c b/src/gallium/drivers/i915/i915_state_emit.c
index a3d4e3b04e..a3d4e3b04e 100644
--- a/src/gallium/drivers/i915simple/i915_state_emit.c
+++ b/src/gallium/drivers/i915/i915_state_emit.c
diff --git a/src/gallium/drivers/i915simple/i915_state_immediate.c b/src/gallium/drivers/i915/i915_state_immediate.c
index 8c16bb4e27..8c16bb4e27 100644
--- a/src/gallium/drivers/i915simple/i915_state_immediate.c
+++ b/src/gallium/drivers/i915/i915_state_immediate.c
diff --git a/src/gallium/drivers/i915simple/i915_state_inlines.h b/src/gallium/drivers/i915/i915_state_inlines.h
index 378de8f9c4..378de8f9c4 100644
--- a/src/gallium/drivers/i915simple/i915_state_inlines.h
+++ b/src/gallium/drivers/i915/i915_state_inlines.h
diff --git a/src/gallium/drivers/i915simple/i915_state_sampler.c b/src/gallium/drivers/i915/i915_state_sampler.c
index c5e9084d12..c5e9084d12 100644
--- a/src/gallium/drivers/i915simple/i915_state_sampler.c
+++ b/src/gallium/drivers/i915/i915_state_sampler.c
diff --git a/src/gallium/drivers/i915simple/i915_surface.c b/src/gallium/drivers/i915/i915_surface.c
index ab8331f3e6..ab8331f3e6 100644
--- a/src/gallium/drivers/i915simple/i915_surface.c
+++ b/src/gallium/drivers/i915/i915_surface.c
diff --git a/src/gallium/drivers/i915simple/i915_texture.c b/src/gallium/drivers/i915/i915_texture.c
index 6a6c654271..286c9ace8e 100644
--- a/src/gallium/drivers/i915simple/i915_texture.c
+++ b/src/gallium/drivers/i915/i915_texture.c
@@ -165,7 +165,7 @@ i915_scanout_layout(struct i915_texture *tex)
    struct pipe_texture *pt = &tex->base;
 
    if (pt->last_level > 0 || pt->block.size != 4)
-      return 0;
+      return FALSE;
 
    i915_miptree_set_level_info(tex, 0, 1,
                                tex->base.width[0],
@@ -191,6 +191,38 @@ i915_scanout_layout(struct i915_texture *tex)
    return TRUE;
 }
 
+/**
+ * Special case to deal with shared textures.
+ */
+static boolean
+i915_display_target_layout(struct i915_texture *tex)
+{
+   struct pipe_texture *pt = &tex->base;
+
+   if (pt->last_level > 0 || pt->block.size != 4)
+      return FALSE;
+
+   /* fallback to normal textures for small textures */
+   if (tex->base.width[0] < 240)
+      return FALSE;
+
+   i915_miptree_set_level_info(tex, 0, 1,
+                               tex->base.width[0],
+                               tex->base.height[0],
+                               1);
+   i915_miptree_set_image_offset(tex, 0, 0, 0, 0);
+
+   tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size);
+   tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8);
+   tex->hw_tiled = INTEL_TILE_X;
+
+   debug_printf("%s size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__,
+      tex->base.width[0], tex->base.height[0], pt->block.size,
+      tex->stride, tex->total_nblocksy, tex->stride * tex->total_nblocksy);
+
+   return TRUE;
+}
+
 static void
 i915_miptree_layout_2d(struct i915_texture *tex)
 {
@@ -201,6 +233,16 @@ i915_miptree_layout_2d(struct i915_texture *tex)
    unsigned nblocksx = pt->nblocksx[0];
    unsigned nblocksy = pt->nblocksy[0];
 
+   /* used for scanouts that need special layouts */
+   if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_PRIMARY)
+      if (i915_scanout_layout(tex))
+         return;
+
+   /* for shared buffers we use some very like scanout */
+   if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET)
+      if (i915_display_target_layout(tex))
+         return;
+
    tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4);
    tex->total_nblocksy = 0;
 
@@ -351,6 +393,11 @@ i945_miptree_layout_2d(struct i915_texture *tex)
       if (i915_scanout_layout(tex))
          return;
 
+   /* for shared buffers we use some very like scanout */
+   if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET)
+      if (i915_display_target_layout(tex))
+         return;
+
    tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4);
 
    /* May need to adjust pitch to accomodate the placement of
@@ -812,7 +859,7 @@ i915_transfer_map(struct pipe_screen *screen,
    char *map;
    boolean write = FALSE;
 
-   if (transfer->usage != PIPE_TRANSFER_READ)
+   if (transfer->usage & PIPE_TRANSFER_WRITE)
       write = TRUE;
 
    map = iws->buffer_map(iws, tex->buffer, write);
diff --git a/src/gallium/drivers/i915simple/i915_texture.h b/src/gallium/drivers/i915/i915_texture.h
index 51a1dd984c..51a1dd984c 100644
--- a/src/gallium/drivers/i915simple/i915_texture.h
+++ b/src/gallium/drivers/i915/i915_texture.h
diff --git a/src/gallium/drivers/i915simple/intel_batchbuffer.h b/src/gallium/drivers/i915/intel_batchbuffer.h
index db12dfd2ac..db12dfd2ac 100644
--- a/src/gallium/drivers/i915simple/intel_batchbuffer.h
+++ b/src/gallium/drivers/i915/intel_batchbuffer.h
diff --git a/src/gallium/drivers/i915simple/intel_winsys.h b/src/gallium/drivers/i915/intel_winsys.h
index f949f52a9c..2c8dc63f3f 100644
--- a/src/gallium/drivers/i915simple/intel_winsys.h
+++ b/src/gallium/drivers/i915/intel_winsys.h
@@ -150,6 +150,17 @@ struct intel_winsys {
    void (*buffer_unmap)(struct intel_winsys *iws,
                         struct intel_buffer *buffer);
 
+   /**
+    * Write to a buffer.
+    *
+    * Arguments follows pipe_buffer_write.
+    */
+   int (*buffer_write)(struct intel_winsys *iws,
+                       struct intel_buffer *dst,
+                       size_t offset,
+                       size_t size,
+                       const void *data);
+
    void (*buffer_destroy)(struct intel_winsys *iws,
                           struct intel_buffer *buffer);
    /*@}*/
diff --git a/src/gallium/drivers/i965simple/Makefile b/src/gallium/drivers/i965simple/Makefile
deleted file mode 100644
index 19182afa75..0000000000
--- a/src/gallium/drivers/i965simple/Makefile
+++ /dev/null
@@ -1,52 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = i965simple
-
-C_SOURCES = \
-	brw_blit.c \
-	brw_flush.c \
-	brw_screen.c \
-	brw_surface.c \
-	brw_cc.c \
-	brw_clip.c \
-	brw_clip_line.c \
-	brw_clip_point.c \
-	brw_clip_state.c \
-	brw_clip_tri.c \
-	brw_clip_util.c \
-	brw_context.c \
-	brw_curbe.c \
-	brw_draw.c \
-	brw_draw_upload.c \
-	brw_eu.c \
-	brw_eu_debug.c \
-	brw_eu_emit.c \
-	brw_eu_util.c \
-	brw_gs.c \
-	brw_gs_emit.c \
-	brw_gs_state.c \
-	brw_misc_state.c \
-	brw_sf.c \
-	brw_sf_emit.c \
-	brw_sf_state.c \
-	brw_state.c \
-	brw_state_batch.c \
-	brw_state_cache.c \
-	brw_state_pool.c \
-	brw_state_upload.c \
-	brw_tex_layout.c \
-	brw_urb.c \
-	brw_util.c \
-	brw_vs.c \
-	brw_vs_emit.c \
-	brw_vs_state.c \
-	brw_wm.c \
-	brw_wm_iz.c \
-	brw_wm_decl.c \
-	brw_wm_glsl.c \
-	brw_wm_sampler_state.c \
-	brw_wm_state.c \
-	brw_wm_surface_state.c
-
-include ../../Makefile.template
diff --git a/src/gallium/drivers/i965simple/SConscript b/src/gallium/drivers/i965simple/SConscript
deleted file mode 100644
index 43fc2a4005..0000000000
--- a/src/gallium/drivers/i965simple/SConscript
+++ /dev/null
@@ -1,54 +0,0 @@
-Import('*')
-
-env = env.Clone()
-
-i965simple = env.ConvenienceLibrary(
-	target = 'i965simple',
-	source = [
-		'brw_blit.c',
-		'brw_cc.c',
-		'brw_clip.c',
-		'brw_clip_line.c',
-		'brw_clip_point.c',
-		'brw_clip_state.c',
-		'brw_clip_tri.c',
-		'brw_clip_util.c',
-		'brw_context.c',
-		'brw_curbe.c',
-		'brw_draw.c',
-		'brw_draw_upload.c',
-		'brw_eu.c',
-		'brw_eu_debug.c',
-		'brw_eu_emit.c',
-		'brw_eu_util.c',
-		'brw_flush.c',
-		'brw_gs.c',
-		'brw_gs_emit.c',
-		'brw_gs_state.c',
-		'brw_misc_state.c',
-		'brw_screen.c',
-		'brw_sf.c',
-		'brw_sf_emit.c',
-		'brw_sf_state.c',
-		'brw_state.c',
-		'brw_state_batch.c',
-		'brw_state_cache.c',
-		'brw_state_pool.c',
-		'brw_state_upload.c',
-		'brw_surface.c',
-		'brw_tex_layout.c',
-		'brw_urb.c',
-		'brw_util.c',
-		'brw_vs.c',
-		'brw_vs_emit.c',
-		'brw_vs_state.c',
-		'brw_wm.c',
-		'brw_wm_decl.c',
-		'brw_wm_glsl.c',
-		'brw_wm_iz.c',
-		'brw_wm_sampler_state.c',
-		'brw_wm_state.c',
-		'brw_wm_surface_state.c',
-	])
-
-Export('i965simple')
diff --git a/src/gallium/drivers/i965simple/brw_blit.c b/src/gallium/drivers/i965simple/brw_blit.c
deleted file mode 100644
index 4d11f8d2ab..0000000000
--- a/src/gallium/drivers/i965simple/brw_blit.c
+++ /dev/null
@@ -1,218 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-
-#include <stdio.h>
-#include <errno.h>
-
-#include "brw_batch.h"
-#include "brw_blit.h"
-#include "brw_context.h"
-#include "brw_reg.h"
-
-#include "pipe/p_context.h"
-#include "pipe/internal/p_winsys_screen.h"
-
-#define FILE_DEBUG_FLAG DEBUG_BLIT
-
-void brw_fill_blit(struct brw_context *brw,
-                   unsigned cpp,
-                   short dst_pitch,
-                   struct pipe_buffer *dst_buffer,
-                   unsigned dst_offset,
-                   boolean dst_tiled,
-                   short x, short y,
-                   short w, short h,
-                   unsigned color)
-{
-   unsigned BR13, CMD;
-   BATCH_LOCALS;
-
-   dst_pitch *= cpp;
-
-   switch(cpp) {
-   case 1:
-   case 2:
-   case 3:
-      BR13 = (0xF0 << 16) | (1<<24);
-      CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (0xF0 << 16) | (1<<24) | (1<<25);
-      CMD = XY_COLOR_BLT_CMD | XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
-      break;
-   default:
-      return;
-   }
-
-   if (dst_tiled) {
-      CMD |= XY_DST_TILED;
-      dst_pitch /= 4;
-   }
-
-   BEGIN_BATCH(6, INTEL_BATCH_NO_CLIPRECTS);
-   OUT_BATCH( CMD );
-   OUT_BATCH( dst_pitch | BR13 );
-   OUT_BATCH( (y << 16) | x );
-   OUT_BATCH( ((y+h) << 16) | (x+w) );
-   OUT_RELOC( dst_buffer, BRW_BUFFER_ACCESS_WRITE, dst_offset );
-   OUT_BATCH( color );
-   ADVANCE_BATCH();
-}
-
-static unsigned translate_raster_op(unsigned logicop)
-{
-   switch(logicop) {
-   case PIPE_LOGICOP_CLEAR: return 0x00;
-   case PIPE_LOGICOP_AND: return 0x88;
-   case PIPE_LOGICOP_AND_REVERSE: return 0x44;
-   case PIPE_LOGICOP_COPY: return 0xCC;
-   case PIPE_LOGICOP_AND_INVERTED: return 0x22;
-   case PIPE_LOGICOP_NOOP: return 0xAA;
-   case PIPE_LOGICOP_XOR: return 0x66;
-   case PIPE_LOGICOP_OR: return 0xEE;
-   case PIPE_LOGICOP_NOR: return 0x11;
-   case PIPE_LOGICOP_EQUIV: return 0x99;
-   case PIPE_LOGICOP_INVERT: return 0x55;
-   case PIPE_LOGICOP_OR_REVERSE: return 0xDD;
-   case PIPE_LOGICOP_COPY_INVERTED: return 0x33;
-   case PIPE_LOGICOP_OR_INVERTED: return 0xBB;
-   case PIPE_LOGICOP_NAND: return 0x77;
-   case PIPE_LOGICOP_SET: return 0xFF;
-   default: return 0;
-   }
-}
-
-
-/* Copy BitBlt
- */
-void brw_copy_blit(struct brw_context *brw,
-                   unsigned do_flip,
-                   unsigned cpp,
-                   short src_pitch,
-                   struct pipe_buffer *src_buffer,
-                   unsigned  src_offset,
-                   boolean src_tiled,
-                   short dst_pitch,
-                   struct pipe_buffer *dst_buffer,
-                   unsigned  dst_offset,
-                   boolean dst_tiled,
-                   short src_x, short src_y,
-                   short dst_x, short dst_y,
-                   short w, short h,
-                   unsigned logic_op)
-{
-   unsigned CMD, BR13;
-   int dst_y2 = dst_y + h;
-   int dst_x2 = dst_x + w;
-   BATCH_LOCALS;
-
-
-   DBG("%s src:buf(%d)/%d %d,%d dst:buf(%d)/%d %d,%d sz:%dx%d op:%d\n",
-       __FUNCTION__,
-       src_buffer, src_pitch, src_x, src_y,
-       dst_buffer, dst_pitch, dst_x, dst_y,
-       w,h,logic_op);
-
-   assert( logic_op - PIPE_LOGICOP_CLEAR >= 0 );
-   assert( logic_op - PIPE_LOGICOP_CLEAR < 0x10 );
-
-   src_pitch *= cpp;
-   dst_pitch *= cpp;
-
-   switch(cpp) {
-   case 1:
-   case 2:
-   case 3:
-      BR13 = (translate_raster_op(logic_op) << 16) | (1<<24);
-      CMD = XY_SRC_COPY_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (translate_raster_op(logic_op) << 16) | (1<<24) |
-	  (1<<25);
-      CMD = XY_SRC_COPY_BLT_CMD | XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
-      break;
-   default:
-      return;
-   }
-
-   if (src_tiled) {
-      CMD |= XY_SRC_TILED;
-      src_pitch /= 4;
-   }
-
-   if (dst_tiled) {
-      CMD |= XY_DST_TILED;
-      dst_pitch /= 4;
-   }
-
-   if (dst_y2 < dst_y ||
-       dst_x2 < dst_x) {
-      return;
-   }
-
-   dst_pitch &= 0xffff;
-   src_pitch &= 0xffff;
-
-   /* Initial y values don't seem to work with negative pitches.  If
-    * we adjust the offsets manually (below), it seems to work fine.
-    *
-    * On the other hand, if we always adjust, the hardware doesn't
-    * know which blit directions to use, so overlapping copypixels get
-    * the wrong result.
-    */
-   if (dst_pitch > 0 && src_pitch > 0) {
-      BEGIN_BATCH(8, INTEL_BATCH_NO_CLIPRECTS);
-      OUT_BATCH( CMD );
-      OUT_BATCH( dst_pitch | BR13 );
-      OUT_BATCH( (dst_y << 16) | dst_x );
-      OUT_BATCH( (dst_y2 << 16) | dst_x2 );
-      OUT_RELOC( dst_buffer, BRW_BUFFER_ACCESS_WRITE,
-		 dst_offset );
-      OUT_BATCH( (src_y << 16) | src_x );
-      OUT_BATCH( src_pitch );
-      OUT_RELOC( src_buffer, BRW_BUFFER_ACCESS_READ,
-		 src_offset );
-      ADVANCE_BATCH();
-   }
-   else {
-      BEGIN_BATCH(8, INTEL_BATCH_NO_CLIPRECTS);
-      OUT_BATCH( CMD );
-      OUT_BATCH( (dst_pitch & 0xffff) | BR13 );
-      OUT_BATCH( (0 << 16) | dst_x );
-      OUT_BATCH( (h << 16) | dst_x2 );
-      OUT_RELOC( dst_buffer, BRW_BUFFER_ACCESS_WRITE,
-		 dst_offset + dst_y * dst_pitch );
-      OUT_BATCH( (src_pitch & 0xffff) );
-      OUT_RELOC( src_buffer, BRW_BUFFER_ACCESS_READ,
-		 src_offset + src_y * src_pitch );
-      ADVANCE_BATCH();
-   }
-}
-
-
-
diff --git a/src/gallium/drivers/i965simple/brw_blit.h b/src/gallium/drivers/i965simple/brw_blit.h
deleted file mode 100644
index 111c5d91d3..0000000000
--- a/src/gallium/drivers/i965simple/brw_blit.h
+++ /dev/null
@@ -1,33 +0,0 @@
-#ifndef BRW_BLIT_H
-#define BRW_BLIT_H
-
-#include "pipe/p_compiler.h"
-
-struct pipe_buffer;
-struct brw_context;
-
-void brw_fill_blit(struct brw_context *intel,
-                   unsigned cpp,
-                   short dst_pitch,
-                   struct pipe_buffer *dst_buffer,
-                   unsigned dst_offset,
-                   boolean dst_tiled,
-                   short x, short y,
-                   short w, short h,
-                   unsigned color);
-void brw_copy_blit(struct brw_context *intel,
-                   unsigned do_flip,
-                   unsigned cpp,
-                   short src_pitch,
-                   struct pipe_buffer *src_buffer,
-                   unsigned  src_offset,
-                   boolean src_tiled,
-                   short dst_pitch,
-                   struct pipe_buffer *dst_buffer,
-                   unsigned  dst_offset,
-                   boolean dst_tiled,
-                   short src_x, short src_y,
-                   short dst_x, short dst_y,
-                   short w, short h,
-                   unsigned logic_op);
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_cc.c b/src/gallium/drivers/i965simple/brw_cc.c
deleted file mode 100644
index 3668123e2e..0000000000
--- a/src/gallium/drivers/i965simple/brw_cc.c
+++ /dev/null
@@ -1,269 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-#include "brw_util.h"
-
-
-static int brw_translate_compare_func(int func)
-{
-   switch(func) {
-   case PIPE_FUNC_NEVER:
-      return BRW_COMPAREFUNCTION_NEVER;
-   case PIPE_FUNC_LESS:
-      return BRW_COMPAREFUNCTION_LESS;
-   case PIPE_FUNC_LEQUAL:
-      return BRW_COMPAREFUNCTION_LEQUAL;
-   case PIPE_FUNC_GREATER:
-      return BRW_COMPAREFUNCTION_GREATER;
-   case PIPE_FUNC_GEQUAL:
-      return BRW_COMPAREFUNCTION_GEQUAL;
-   case PIPE_FUNC_NOTEQUAL:
-      return BRW_COMPAREFUNCTION_NOTEQUAL;
-   case PIPE_FUNC_EQUAL:
-      return BRW_COMPAREFUNCTION_EQUAL;
-   case PIPE_FUNC_ALWAYS:
-      return BRW_COMPAREFUNCTION_ALWAYS;
-   }
-
-   debug_printf("Unknown value in %s: %x\n", __FUNCTION__, func);
-   return BRW_COMPAREFUNCTION_ALWAYS;
-}
-
-static int brw_translate_stencil_op(int op)
-{
-   switch(op) {
-   case PIPE_STENCIL_OP_KEEP:
-      return BRW_STENCILOP_KEEP;
-   case PIPE_STENCIL_OP_ZERO:
-      return BRW_STENCILOP_ZERO;
-   case PIPE_STENCIL_OP_REPLACE:
-      return BRW_STENCILOP_REPLACE;
-   case PIPE_STENCIL_OP_INCR:
-      return BRW_STENCILOP_INCRSAT;
-   case PIPE_STENCIL_OP_DECR:
-      return BRW_STENCILOP_DECRSAT;
-   case PIPE_STENCIL_OP_INCR_WRAP:
-      return BRW_STENCILOP_INCR;
-   case PIPE_STENCIL_OP_DECR_WRAP:
-      return BRW_STENCILOP_DECR;
-   case PIPE_STENCIL_OP_INVERT:
-      return BRW_STENCILOP_INVERT;
-   default:
-      return BRW_STENCILOP_ZERO;
-   }
-}
-
-
-static int brw_translate_logic_op(int opcode)
-{
-   switch(opcode) {
-   case PIPE_LOGICOP_CLEAR:
-      return BRW_LOGICOPFUNCTION_CLEAR;
-   case PIPE_LOGICOP_AND:
-      return BRW_LOGICOPFUNCTION_AND;
-   case PIPE_LOGICOP_AND_REVERSE:
-      return BRW_LOGICOPFUNCTION_AND_REVERSE;
-   case PIPE_LOGICOP_COPY:
-      return BRW_LOGICOPFUNCTION_COPY;
-   case PIPE_LOGICOP_COPY_INVERTED:
-      return BRW_LOGICOPFUNCTION_COPY_INVERTED;
-   case PIPE_LOGICOP_AND_INVERTED:
-      return BRW_LOGICOPFUNCTION_AND_INVERTED;
-   case PIPE_LOGICOP_NOOP:
-      return BRW_LOGICOPFUNCTION_NOOP;
-   case PIPE_LOGICOP_XOR:
-      return BRW_LOGICOPFUNCTION_XOR;
-   case PIPE_LOGICOP_OR:
-      return BRW_LOGICOPFUNCTION_OR;
-   case PIPE_LOGICOP_OR_INVERTED:
-      return BRW_LOGICOPFUNCTION_OR_INVERTED;
-   case PIPE_LOGICOP_NOR:
-      return BRW_LOGICOPFUNCTION_NOR;
-   case PIPE_LOGICOP_EQUIV:
-      return BRW_LOGICOPFUNCTION_EQUIV;
-   case PIPE_LOGICOP_INVERT:
-      return BRW_LOGICOPFUNCTION_INVERT;
-   case PIPE_LOGICOP_OR_REVERSE:
-      return BRW_LOGICOPFUNCTION_OR_REVERSE;
-   case PIPE_LOGICOP_NAND:
-      return BRW_LOGICOPFUNCTION_NAND;
-   case PIPE_LOGICOP_SET:
-      return BRW_LOGICOPFUNCTION_SET;
-   default:
-      return BRW_LOGICOPFUNCTION_SET;
-   }
-}
-
-
-static void upload_cc_vp( struct brw_context *brw )
-{
-   struct brw_cc_viewport ccv;
-
-   memset(&ccv, 0, sizeof(ccv));
-
-   ccv.min_depth = 0.0;
-   ccv.max_depth = 1.0;
-
-   brw->cc.vp_gs_offset = brw_cache_data( &brw->cache[BRW_CC_VP], &ccv );
-}
-
-const struct brw_tracked_state brw_cc_vp = {
-   .dirty = {
-      .brw = BRW_NEW_SCENE,
-      .cache = 0
-   },
-   .update = upload_cc_vp
-};
-
-
-static void upload_cc_unit( struct brw_context *brw )
-{
-   struct brw_cc_unit_state cc;
-
-   memset(&cc, 0, sizeof(cc));
-
-   /* BRW_NEW_DEPTH_STENCIL */
-   if (brw->attribs.DepthStencil->stencil[0].enabled) {
-      cc.cc0.stencil_enable = brw->attribs.DepthStencil->stencil[0].enabled;
-      cc.cc0.stencil_func = brw_translate_compare_func(brw->attribs.DepthStencil->stencil[0].func);
-      cc.cc0.stencil_fail_op = brw_translate_stencil_op(brw->attribs.DepthStencil->stencil[0].fail_op);
-      cc.cc0.stencil_pass_depth_fail_op = brw_translate_stencil_op(
-         brw->attribs.DepthStencil->stencil[0].zfail_op);
-      cc.cc0.stencil_pass_depth_pass_op = brw_translate_stencil_op(
-         brw->attribs.DepthStencil->stencil[0].zpass_op);
-      cc.cc1.stencil_ref = brw->attribs.DepthStencil->stencil[0].ref_value;
-      cc.cc1.stencil_write_mask = brw->attribs.DepthStencil->stencil[0].writemask;
-      cc.cc1.stencil_test_mask = brw->attribs.DepthStencil->stencil[0].valuemask;
-
-      if (brw->attribs.DepthStencil->stencil[1].enabled) {
-	 cc.cc0.bf_stencil_enable = brw->attribs.DepthStencil->stencil[1].enabled;
-	 cc.cc0.bf_stencil_func = brw_translate_compare_func(
-            brw->attribs.DepthStencil->stencil[1].func);
-	 cc.cc0.bf_stencil_fail_op = brw_translate_stencil_op(
-            brw->attribs.DepthStencil->stencil[1].fail_op);
-	 cc.cc0.bf_stencil_pass_depth_fail_op = brw_translate_stencil_op(
-            brw->attribs.DepthStencil->stencil[1].zfail_op);
-	 cc.cc0.bf_stencil_pass_depth_pass_op = brw_translate_stencil_op(
-            brw->attribs.DepthStencil->stencil[1].zpass_op);
-	 cc.cc1.bf_stencil_ref = brw->attribs.DepthStencil->stencil[1].ref_value;
-	 cc.cc2.bf_stencil_write_mask = brw->attribs.DepthStencil->stencil[1].writemask;
-	 cc.cc2.bf_stencil_test_mask = brw->attribs.DepthStencil->stencil[1].valuemask;
-      }
-
-      /* Not really sure about this:
-       */
-      if (brw->attribs.DepthStencil->stencil[0].writemask ||
-	  brw->attribs.DepthStencil->stencil[1].writemask)
-	 cc.cc0.stencil_write_enable = 1;
-   }
-
-   /* BRW_NEW_BLEND */
-   if (brw->attribs.Blend->logicop_enable) {
-      cc.cc2.logicop_enable = 1;
-      cc.cc5.logicop_func = brw_translate_logic_op( brw->attribs.Blend->logicop_func );
-   }
-   else if (brw->attribs.Blend->blend_enable) {
-      int eqRGB = brw->attribs.Blend->rgb_func;
-      int eqA = brw->attribs.Blend->alpha_func;
-      int srcRGB = brw->attribs.Blend->rgb_src_factor;
-      int dstRGB = brw->attribs.Blend->rgb_dst_factor;
-      int srcA = brw->attribs.Blend->alpha_src_factor;
-      int dstA = brw->attribs.Blend->alpha_dst_factor;
-
-      if (eqRGB == PIPE_BLEND_MIN || eqRGB == PIPE_BLEND_MAX) {
-	 srcRGB = dstRGB = PIPE_BLENDFACTOR_ONE;
-      }
-
-      if (eqA == PIPE_BLEND_MIN || eqA == PIPE_BLEND_MAX) {
-	 srcA = dstA = PIPE_BLENDFACTOR_ONE;
-      }
-
-      cc.cc6.dest_blend_factor = brw_translate_blend_factor(dstRGB);
-      cc.cc6.src_blend_factor = brw_translate_blend_factor(srcRGB);
-      cc.cc6.blend_function = brw_translate_blend_equation( eqRGB );
-
-      cc.cc5.ia_dest_blend_factor = brw_translate_blend_factor(dstA);
-      cc.cc5.ia_src_blend_factor = brw_translate_blend_factor(srcA);
-      cc.cc5.ia_blend_function = brw_translate_blend_equation( eqA );
-
-      cc.cc3.blend_enable = 1;
-      cc.cc3.ia_blend_enable = (srcA != srcRGB ||
-				dstA != dstRGB ||
-				eqA != eqRGB);
-   }
-   
-   /* BRW_NEW_ALPHATEST
-    */
-   if (brw->attribs.DepthStencil->alpha.enabled) {
-      cc.cc3.alpha_test = 1;
-      cc.cc3.alpha_test_func = 
-	 brw_translate_compare_func(brw->attribs.DepthStencil->alpha.func);
-
-      cc.cc7.alpha_ref.ub[0] = float_to_ubyte(brw->attribs.DepthStencil->alpha.ref_value);
-
-      cc.cc3.alpha_test_format = BRW_ALPHATEST_FORMAT_UNORM8;
-   }
-
-   if (brw->attribs.Blend->dither) {
-      cc.cc5.dither_enable = 1;
-      cc.cc6.y_dither_offset = 0;
-      cc.cc6.x_dither_offset = 0;
-   }
-
-   if (brw->attribs.DepthStencil->depth.enabled) {
-      cc.cc2.depth_test = brw->attribs.DepthStencil->depth.enabled;
-      cc.cc2.depth_test_function = brw_translate_compare_func(brw->attribs.DepthStencil->depth.func);
-      cc.cc2.depth_write_enable = brw->attribs.DepthStencil->depth.writemask;
-   }
-
-   /* CACHE_NEW_CC_VP */
-   cc.cc4.cc_viewport_state_offset =  brw->cc.vp_gs_offset >> 5;
-
-   if (BRW_DEBUG & DEBUG_STATS)
-      cc.cc5.statistics_enable = 1;
-
-   brw->cc.state_gs_offset = brw_cache_data( &brw->cache[BRW_CC_UNIT], &cc );
-}
-
-const struct brw_tracked_state brw_cc_unit = {
-   .dirty = {
-      .brw = BRW_NEW_DEPTH_STENCIL | BRW_NEW_BLEND | BRW_NEW_ALPHA_TEST,
-      .cache = CACHE_NEW_CC_VP
-   },
-   .update = upload_cc_unit
-};
-
diff --git a/src/gallium/drivers/i965simple/brw_clip.c b/src/gallium/drivers/i965simple/brw_clip.c
deleted file mode 100644
index 268124cc53..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip.c
+++ /dev/null
@@ -1,206 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_state.h"
-#include "brw_clip.h"
-
-#define FRONT_UNFILLED_BIT  0x1
-#define BACK_UNFILLED_BIT   0x2
-
-
-static void compile_clip_prog( struct brw_context *brw,
-			     struct brw_clip_prog_key *key )
-{
-   struct brw_clip_compile c;
-   const unsigned *program;
-   unsigned program_size;
-   unsigned delta;
-   unsigned i;
-
-   memset(&c, 0, sizeof(c));
-
-   /* Begin the compilation:
-    */
-   brw_init_compile(&c.func);
-
-   c.func.single_program_flow = 1;
-
-   c.key = *key;
-
-
-   /* Need to locate the two positions present in vertex + header.
-    * These are currently hardcoded:
-    */
-   c.header_position_offset = ATTR_SIZE;
-
-   for (i = 0, delta = REG_SIZE; i < PIPE_MAX_SHADER_OUTPUTS; i++)
-      if (c.key.attrs & (1<<i)) {
-	 c.offset[i] = delta;
-	 delta += ATTR_SIZE;
-      }
-
-   c.nr_attrs = brw_count_bits(c.key.attrs);
-   c.nr_regs = (c.nr_attrs + 1) / 2 + 1;  /* are vertices packed, or reg-aligned? */
-   c.nr_bytes = c.nr_regs * REG_SIZE;
-
-   c.prog_data.clip_mode = c.key.clip_mode; /* XXX */
-
-   /* For some reason the thread is spawned with only 4 channels
-    * unmasked.
-    */
-   brw_set_mask_control(&c.func, BRW_MASK_DISABLE);
-
-
-   /* Would ideally have the option of producing a program which could
-    * do all three:
-    */
-   switch (key->primitive) {
-   case PIPE_PRIM_TRIANGLES:
-#if 0
-      if (key->do_unfilled)
-	 brw_emit_unfilled_clip( &c );
-      else
-#endif
-	 brw_emit_tri_clip( &c );
-      break;
-   case PIPE_PRIM_LINES:
-      brw_emit_line_clip( &c );
-      break;
-   case PIPE_PRIM_POINTS:
-      brw_emit_point_clip( &c );
-      break;
-   default:
-      assert(0);
-      return;
-   }
-
-
-
-   /* get the program
-    */
-   program = brw_get_program(&c.func, &program_size);
-
-   /* Upload
-    */
-   brw->clip.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_CLIP_PROG],
-						&c.key,
-						sizeof(c.key),
-						program,
-						program_size,
-						&c.prog_data,
-						&brw->clip.prog_data );
-}
-
-
-static boolean search_cache( struct brw_context *brw,
-			       struct brw_clip_prog_key *key )
-{
-   return brw_search_cache(&brw->cache[BRW_CLIP_PROG],
-			   key, sizeof(*key),
-			   &brw->clip.prog_data,
-			   &brw->clip.prog_gs_offset);
-}
-
-
-
-
-/* Calculate interpolants for triangle and line rasterization.
- */
-static void upload_clip_prog(struct brw_context *brw)
-{
-   struct brw_clip_prog_key key;
-
-   memset(&key, 0, sizeof(key));
-
-   /* Populate the key:
-    */
-   /* BRW_NEW_REDUCED_PRIMITIVE */
-   key.primitive = brw->reduced_primitive;
-   /* CACHE_NEW_VS_PROG */
-   key.attrs = brw->vs.prog_data->outputs_written;
-   /* BRW_NEW_RASTER */
-   key.do_flat_shading = (brw->attribs.Raster->flatshade);
-   /* BRW_NEW_CLIP */
-   key.nr_userclip = brw->attribs.Clip.nr; /* XXX */
-
-#if 0
-   key.clip_mode = BRW_CLIPMODE_NORMAL;
-
-   if (key.primitive == PIPE_PRIM_TRIANGLES) {
-      if (brw->attribs.Raster->cull_mode == PIPE_WINDING_BOTH)
-	 key.clip_mode = BRW_CLIPMODE_REJECT_ALL;
-      else {
-         if (brw->attribs.Raster->fill_cw != PIPE_POLYGON_MODE_FILL ||
-             brw->attribs.Raster->fill_ccw != PIPE_POLYGON_MODE_FILL)
-            key.do_unfilled = 1;
-
-	 /* Most cases the fixed function units will handle.  Cases where
-	  * one or more polygon faces are unfilled will require help:
-	  */
-	 if (key.do_unfilled) {
-	    key.clip_mode = BRW_CLIPMODE_CLIP_NON_REJECTED;
-
-	    if (brw->attribs.Raster->offset_cw ||
-                brw->attribs.Raster->offset_ccw) {
-	       key.offset_units = brw->attribs.Raster->offset_units;
-	       key.offset_factor = brw->attribs.Raster->offset_scale;
-	    }
-            key.fill_ccw = brw->attribs.Raster->fill_ccw;
-            key.fill_cw = brw->attribs.Raster->fill_cw;
-            key.offset_ccw = brw->attribs.Raster->offset_ccw;
-            key.offset_cw = brw->attribs.Raster->offset_cw;
-            if (brw->attribs.Raster->light_twoside &&
-                key.fill_cw != CLIP_CULL)
-               key.copy_bfc_cw = 1;
-	 }
-      }
-   }
-#else
-   key.clip_mode = BRW_CLIPMODE_ACCEPT_ALL;
-#endif
-
-   if (!search_cache(brw, &key))
-      compile_clip_prog( brw, &key );
-}
-
-const struct brw_tracked_state brw_clip_prog = {
-   .dirty = {
-      .brw   = (BRW_NEW_RASTERIZER |
-		BRW_NEW_CLIP |
-		BRW_NEW_REDUCED_PRIMITIVE),
-      .cache = CACHE_NEW_VS_PROG
-   },
-   .update = upload_clip_prog
-};
diff --git a/src/gallium/drivers/i965simple/brw_clip.h b/src/gallium/drivers/i965simple/brw_clip.h
deleted file mode 100644
index d70fc094ff..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip.h
+++ /dev/null
@@ -1,170 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#ifndef BRW_CLIP_H
-#define BRW_CLIP_H
-
-
-#include "brw_context.h"
-#include "brw_eu.h"
-
-#define MAX_VERTS (3+6+6)	
-
-/* Note that if unfilled primitives are being emitted, we have to fix
- * up polygon offset and flatshading at this point:
- */
-struct brw_clip_prog_key {
-   unsigned attrs:32;		
-   unsigned primitive:4;
-   unsigned nr_userclip:3;
-   unsigned do_flat_shading:1;
-   unsigned do_unfilled:1;
-   unsigned fill_cw:2;		/* includes cull information */
-   unsigned fill_ccw:2;		/* includes cull information */
-   unsigned offset_cw:1;
-   unsigned offset_ccw:1;
-   unsigned pad0:17;
-
-   unsigned copy_bfc_cw:1;
-   unsigned copy_bfc_ccw:1;
-   unsigned clip_mode:3;
-   unsigned pad1:27;
-   
-   float offset_factor;
-   float offset_units;
-};
-
-
-#define CLIP_LINE   0
-#define CLIP_POINT  1
-#define CLIP_FILL   2
-#define CLIP_CULL   3
-
-
-#define PRIM_MASK  (0x1f)
-
-struct brw_clip_compile {
-   struct brw_compile func;
-   struct brw_clip_prog_key key;
-   struct brw_clip_prog_data prog_data;
-   
-   struct {
-      struct brw_reg R0;
-      struct brw_reg vertex[MAX_VERTS];
-
-      struct brw_reg t;
-      struct brw_reg t0, t1;
-      struct brw_reg dp0, dp1;
-
-      struct brw_reg dpPrev;
-      struct brw_reg dp;
-      struct brw_reg loopcount;
-      struct brw_reg nr_verts;
-      struct brw_reg planemask;
-
-      struct brw_reg inlist;
-      struct brw_reg outlist;
-      struct brw_reg freelist;
-
-      struct brw_reg dir;
-      struct brw_reg tmp0, tmp1;
-      struct brw_reg offset;
-      
-      struct brw_reg fixed_planes;
-      struct brw_reg plane_equation;
-   } reg;
-
-   /* 3 different ways of expressing vertex size:
-    */
-   unsigned nr_attrs;
-   unsigned nr_regs;
-   unsigned nr_bytes;
-
-   unsigned first_tmp;
-   unsigned last_tmp;
-
-   boolean need_direction;
-
-   unsigned last_mrf;
-
-   unsigned header_position_offset;
-   unsigned offset[PIPE_MAX_ATTRIBS];
-};
-
-#define ATTR_SIZE  (4*4)
-
-/* Points are only culled, so no need for a clip routine, however it
- * works out easier to have a dummy one.
- */
-void brw_emit_unfilled_clip( struct brw_clip_compile *c );
-void brw_emit_tri_clip( struct brw_clip_compile *c );
-void brw_emit_line_clip( struct brw_clip_compile *c );
-void brw_emit_point_clip( struct brw_clip_compile *c );
-
-/* brw_clip_tri.c, for use by the unfilled clip routine:
- */
-void brw_clip_tri_init_vertices( struct brw_clip_compile *c );
-void brw_clip_tri_flat_shade( struct brw_clip_compile *c );
-void brw_clip_tri( struct brw_clip_compile *c );
-void brw_clip_tri_emit_polygon( struct brw_clip_compile *c );
-void brw_clip_tri_alloc_regs( struct brw_clip_compile *c, 
-			      unsigned nr_verts );
-
-
-/* Utils:
- */
-
-void brw_clip_interp_vertex( struct brw_clip_compile *c,
-			     struct brw_indirect dest_ptr,
-			     struct brw_indirect v0_ptr, /* from */
-			     struct brw_indirect v1_ptr, /* to */
-			     struct brw_reg t0,
-			     boolean force_edgeflag );
-
-void brw_clip_init_planes( struct brw_clip_compile *c );
-
-void brw_clip_emit_vue(struct brw_clip_compile *c, 
-		       struct brw_indirect vert,
-		       boolean allocate,
-		       boolean eot,
-		       unsigned header);
-
-void brw_clip_kill_thread(struct brw_clip_compile *c);
-
-struct brw_reg brw_clip_plane_stride( struct brw_clip_compile *c );
-struct brw_reg brw_clip_plane0_address( struct brw_clip_compile *c );
-
-void brw_clip_copy_colors( struct brw_clip_compile *c,
-			   unsigned to, unsigned from );
-
-void brw_clip_init_clipmask( struct brw_clip_compile *c );
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_clip_line.c b/src/gallium/drivers/i965simple/brw_clip_line.c
deleted file mode 100644
index 75d9e5fcda..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip_line.c
+++ /dev/null
@@ -1,245 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_clip.h"
-
-
-
-static void brw_clip_line_alloc_regs( struct brw_clip_compile *c )
-{
-   unsigned i = 0,j;
-
-   /* Register usage is static, precompute here:
-    */
-   c->reg.R0 = retype(brw_vec8_grf(i, 0), BRW_REGISTER_TYPE_UD); i++;
-
-   if (c->key.nr_userclip) {
-      c->reg.fixed_planes = brw_vec4_grf(i, 0);
-      i += (6 + c->key.nr_userclip + 1) / 2;
-
-      c->prog_data.curb_read_length = (6 + c->key.nr_userclip + 1) / 2;
-   }
-   else
-      c->prog_data.curb_read_length = 0;
-
-
-   /* Payload vertices plus space for more generated vertices:
-    */
-   for (j = 0; j < 4; j++) {
-      c->reg.vertex[j] = brw_vec4_grf(i, 0);
-      i += c->nr_regs;
-   }
-
-   c->reg.t           = brw_vec1_grf(i, 0);
-   c->reg.t0          = brw_vec1_grf(i, 1);
-   c->reg.t1          = brw_vec1_grf(i, 2);
-   c->reg.planemask   = retype(brw_vec1_grf(i, 3), BRW_REGISTER_TYPE_UD);
-   c->reg.plane_equation = brw_vec4_grf(i, 4);
-   i++;
-
-   c->reg.dp0         = brw_vec1_grf(i, 0); /* fixme - dp4 will clobber r.1,2,3 */
-   c->reg.dp1         = brw_vec1_grf(i, 4);
-   i++;
-
-   if (!c->key.nr_userclip) {
-      c->reg.fixed_planes = brw_vec8_grf(i, 0);
-      i++;
-   }
-
-
-   c->first_tmp = i;
-   c->last_tmp = i;
-
-   c->prog_data.urb_read_length = c->nr_regs; /* ? */
-   c->prog_data.total_grf = i;
-}
-
-
-
-/* Line clipping, more or less following the following algorithm:
- *
- *  for (p=0;p<MAX_PLANES;p++) {
- *     if (clipmask & (1 << p)) {
- *        float dp0 = DOTPROD( vtx0, plane[p] );
- *        float dp1 = DOTPROD( vtx1, plane[p] );
- *
- *        if (IS_NEGATIVE(dp1)) {
- *           float t = dp1 / (dp1 - dp0);
- *           if (t > t1) t1 = t;
- *        } else {
- *           float t = dp0 / (dp0 - dp1);
- *           if (t > t0) t0 = t;
- *        }
- *
- *        if (t0 + t1 >= 1.0)
- *           return;
- *     }
- *  }
- *
- *  interp( ctx, newvtx0, vtx0, vtx1, t0 );
- *  interp( ctx, newvtx1, vtx1, vtx0, t1 );
- *
- */
-static void clip_and_emit_line( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_indirect vtx0     = brw_indirect(0, 0);
-   struct brw_indirect vtx1      = brw_indirect(1, 0);
-   struct brw_indirect newvtx0   = brw_indirect(2, 0);
-   struct brw_indirect newvtx1   = brw_indirect(3, 0);
-   struct brw_indirect plane_ptr = brw_indirect(4, 0);
-   struct brw_instruction *plane_loop;
-   struct brw_instruction *plane_active;
-   struct brw_instruction *is_negative;
-   struct brw_instruction *is_neg2;
-   struct brw_instruction *not_culled;
-   struct brw_reg v1_null_ud = retype(vec1(brw_null_reg()), BRW_REGISTER_TYPE_UD);
-
-   brw_MOV(p, get_addr_reg(vtx0),      brw_address(c->reg.vertex[0]));
-   brw_MOV(p, get_addr_reg(vtx1),      brw_address(c->reg.vertex[1]));
-   brw_MOV(p, get_addr_reg(newvtx0),   brw_address(c->reg.vertex[2]));
-   brw_MOV(p, get_addr_reg(newvtx1),   brw_address(c->reg.vertex[3]));
-   brw_MOV(p, get_addr_reg(plane_ptr), brw_clip_plane0_address(c));
-
-   /* Note: init t0, t1 together:
-    */
-   brw_MOV(p, vec2(c->reg.t0), brw_imm_f(0));
-
-   brw_clip_init_planes(c);
-   brw_clip_init_clipmask(c);
-
-   /* -ve rhw workaround */
-   brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-   brw_AND(p, brw_null_reg(), get_element_ud(c->reg.R0, 2),
-	   brw_imm_ud(1<<20));
-   brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(0x3f));
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-   plane_loop = brw_DO(p, BRW_EXECUTE_1);
-   {
-      /* if (planemask & 1)
-       */
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-      brw_AND(p, v1_null_ud, c->reg.planemask, brw_imm_ud(1));
-
-      plane_active = brw_IF(p, BRW_EXECUTE_1);
-      {
-	 if (c->key.nr_userclip)
-	    brw_MOV(p, c->reg.plane_equation, deref_4f(plane_ptr, 0));
-	 else
-	    brw_MOV(p, c->reg.plane_equation, deref_4b(plane_ptr, 0));
-
-#if 0
-	 /* dp = DP4(vtx->position, plane)
-	  */
-	 brw_DP4(p, vec4(c->reg.dp0), deref_4f(vtx0, c->offset[VERT_RESULT_HPOS]), c->reg.plane_equation);
-
-	 /* if (IS_NEGATIVE(dp1))
-	  */
-	 brw_set_conditionalmod(p, BRW_CONDITIONAL_L);
-	 brw_DP4(p, vec4(c->reg.dp1), deref_4f(vtx1, c->offset[VERT_RESULT_HPOS]), c->reg.plane_equation);
-#else
-         #warning "disabled"
-#endif
-	 is_negative = brw_IF(p, BRW_EXECUTE_1);
-	 {
-	    brw_ADD(p, c->reg.t, c->reg.dp1, negate(c->reg.dp0));
-	    brw_math_invert(p, c->reg.t, c->reg.t);
-	    brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp1);
-
-	    brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t1 );
-	    brw_MOV(p, c->reg.t1, c->reg.t);
-	    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-	 }
-	 is_negative = brw_ELSE(p, is_negative);
-	 {
-	    /* Coming back in.  We know that both cannot be negative
-	     * because the line would have been culled in that case.
-	     */
-
-	    /* If both are positive, do nothing */
-             brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.dp0, brw_imm_f(0.0));
-             is_neg2 = brw_IF(p, BRW_EXECUTE_1);
-             {
-		brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
-		brw_math_invert(p, c->reg.t, c->reg.t);
-		brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
-
-		brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
-		brw_MOV(p, c->reg.t0, c->reg.t);
-		brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-	     }
-	     brw_ENDIF(p, is_neg2);
-	 }
-	 brw_ENDIF(p, is_negative);
-      }
-      brw_ENDIF(p, plane_active);
-
-      /* plane_ptr++;
-       */
-      brw_ADD(p, get_addr_reg(plane_ptr), get_addr_reg(plane_ptr), brw_clip_plane_stride(c));
-
-      /* while (planemask>>=1) != 0
-       */
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-      brw_SHR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(1));
-   }
-   brw_WHILE(p, plane_loop);
-
-   brw_ADD(p, c->reg.t, c->reg.t0, c->reg.t1);
-   brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.t, brw_imm_f(1.0));
-   not_culled = brw_IF(p, BRW_EXECUTE_1);
-   {
-      brw_clip_interp_vertex(c, newvtx0, vtx0, vtx1, c->reg.t0, FALSE);
-      brw_clip_interp_vertex(c, newvtx1, vtx1, vtx0, c->reg.t1, FALSE);
-
-      brw_clip_emit_vue(c, newvtx0, 1, 0, (_3DPRIM_LINESTRIP << 2) | R02_PRIM_START);
-      brw_clip_emit_vue(c, newvtx1, 0, 1, (_3DPRIM_LINESTRIP << 2) | R02_PRIM_END);
-   }
-   brw_ENDIF(p, not_culled);
-   brw_clip_kill_thread(c);
-}
-
-
-
-void brw_emit_line_clip( struct brw_clip_compile *c )
-{
-   brw_clip_line_alloc_regs(c);
-
-   if (c->key.do_flat_shading)
-      brw_clip_copy_colors(c, 0, 1);
-
-   clip_and_emit_line(c);
-}
diff --git a/src/gallium/drivers/i965simple/brw_clip_point.c b/src/gallium/drivers/i965simple/brw_clip_point.c
deleted file mode 100644
index 6fce7210d1..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip_point.c
+++ /dev/null
@@ -1,47 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_clip.h"
-
-
-/* Point clipping, nothing to do?
- */
-void brw_emit_point_clip( struct brw_clip_compile *c )
-{
-   /* Send an empty message to kill the thread:
-    */
-   brw_clip_tri_alloc_regs(c, 0);
-   brw_clip_kill_thread(c);
-}
diff --git a/src/gallium/drivers/i965simple/brw_clip_state.c b/src/gallium/drivers/i965simple/brw_clip_state.c
deleted file mode 100644
index 8e78dd51be..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip_state.c
+++ /dev/null
@@ -1,93 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-
-static void upload_clip_unit( struct brw_context *brw )
-{
-   struct brw_clip_unit_state clip;
-
-   memset(&clip, 0, sizeof(clip));
-
-   /* CACHE_NEW_CLIP_PROG */
-   clip.thread0.grf_reg_count =
-      align(brw->clip.prog_data->total_grf, 16) / 16 - 1;
-   clip.thread0.kernel_start_pointer = brw->clip.prog_gs_offset >> 6;
-   clip.thread3.urb_entry_read_length = brw->clip.prog_data->urb_read_length;
-   clip.thread3.const_urb_entry_read_length = brw->clip.prog_data->curb_read_length;
-   clip.clip5.clip_mode = brw->clip.prog_data->clip_mode;
-
-   /* BRW_NEW_CURBE_OFFSETS */
-   clip.thread3.const_urb_entry_read_offset = brw->curbe.clip_start * 2;
-
-   /* BRW_NEW_URB_FENCE */
-   clip.thread4.nr_urb_entries = brw->urb.nr_clip_entries; 
-   clip.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
-   clip.thread4.max_threads = 1; /* 2 threads */
-
-   if (BRW_DEBUG & DEBUG_STATS)
-      clip.thread4.stats_enable = 1; 
-
-   /* CONSTANT */
-   clip.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
-   clip.thread1.single_program_flow = 1;
-   clip.thread3.dispatch_grf_start_reg = 1;
-   clip.thread3.urb_entry_read_offset = 0;
-   clip.clip5.userclip_enable_flags = 0x7f;
-   clip.clip5.userclip_must_clip = 1;
-   clip.clip5.guard_band_enable = 0;
-   clip.clip5.viewport_z_clip_enable = 1;
-   clip.clip5.viewport_xy_clip_enable = 1;
-   clip.clip5.vertex_position_space = BRW_CLIP_NDCSPACE;
-   clip.clip5.api_mode = BRW_CLIP_API_OGL;   
-   clip.clip6.clipper_viewport_state_ptr = 0;
-   clip.viewport_xmin = -1;
-   clip.viewport_xmax = 1;
-   clip.viewport_ymin = -1;
-   clip.viewport_ymax = 1;
-
-   brw->clip.state_gs_offset = brw_cache_data( &brw->cache[BRW_CLIP_UNIT], &clip );
-}
-
-
-const struct brw_tracked_state brw_clip_unit = {
-   .dirty = {
-      .brw   = (BRW_NEW_CURBE_OFFSETS |
-		BRW_NEW_URB_FENCE),
-      .cache = CACHE_NEW_CLIP_PROG
-   },
-   .update = upload_clip_unit
-};
diff --git a/src/gallium/drivers/i965simple/brw_clip_tri.c b/src/gallium/drivers/i965simple/brw_clip_tri.c
deleted file mode 100644
index c5da7b825e..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip_tri.c
+++ /dev/null
@@ -1,566 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_clip.h"
-
-static struct brw_reg get_tmp( struct brw_clip_compile *c )
-{
-   struct brw_reg tmp = brw_vec4_grf(c->last_tmp, 0);
-
-   if (++c->last_tmp > c->prog_data.total_grf)
-      c->prog_data.total_grf = c->last_tmp;
-
-   return tmp;
-}
-
-static void release_tmps( struct brw_clip_compile *c )
-{
-   c->last_tmp = c->first_tmp;
-}
-
-
-void brw_clip_tri_alloc_regs( struct brw_clip_compile *c,
-			      unsigned nr_verts )
-{
-   unsigned i = 0,j;
-
-   /* Register usage is static, precompute here:
-    */
-   c->reg.R0 = retype(brw_vec8_grf(i, 0), BRW_REGISTER_TYPE_UD); i++;
-
-   if (c->key.nr_userclip) {
-      c->reg.fixed_planes = brw_vec4_grf(i, 0);
-      i += (6 + c->key.nr_userclip + 1) / 2;
-
-      c->prog_data.curb_read_length = (6 + c->key.nr_userclip + 1) / 2;
-   }
-   else
-      c->prog_data.curb_read_length = 0;
-
-
-   /* Payload vertices plus space for more generated vertices:
-    */
-   for (j = 0; j < nr_verts; j++) {
-      c->reg.vertex[j] = brw_vec4_grf(i, 0);
-      i += c->nr_regs;
-   }
-
-   if (c->nr_attrs & 1) {
-      for (j = 0; j < 3; j++) {
-	 unsigned delta = c->nr_attrs*16 + 32;
-	 brw_MOV(&c->func, byte_offset(c->reg.vertex[j], delta), brw_imm_f(0));
-      }
-   }
-
-   c->reg.t          = brw_vec1_grf(i, 0);
-   c->reg.loopcount  = retype(brw_vec1_grf(i, 1), BRW_REGISTER_TYPE_UD);
-   c->reg.nr_verts   = retype(brw_vec1_grf(i, 2), BRW_REGISTER_TYPE_UD);
-   c->reg.planemask  = retype(brw_vec1_grf(i, 3), BRW_REGISTER_TYPE_UD);
-   c->reg.plane_equation = brw_vec4_grf(i, 4);
-   i++;
-
-   c->reg.dpPrev     = brw_vec1_grf(i, 0); /* fixme - dp4 will clobber r.1,2,3 */
-   c->reg.dp         = brw_vec1_grf(i, 4);
-   i++;
-
-   c->reg.inlist     = brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, i, 0);
-   i++;
-
-   c->reg.outlist    = brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, i, 0);
-   i++;
-
-   c->reg.freelist   = brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, i, 0);
-   i++;
-
-   if (!c->key.nr_userclip) {
-      c->reg.fixed_planes = brw_vec8_grf(i, 0);
-      i++;
-   }
-
-   if (c->key.do_unfilled) {
-      c->reg.dir     = brw_vec4_grf(i, 0);
-      c->reg.offset  = brw_vec4_grf(i, 4);
-      i++;
-      c->reg.tmp0    = brw_vec4_grf(i, 0);
-      c->reg.tmp1    = brw_vec4_grf(i, 4);
-      i++;
-   }
-
-   c->first_tmp = i;
-   c->last_tmp = i;
-
-   c->prog_data.urb_read_length = c->nr_regs; /* ? */
-   c->prog_data.total_grf = i;
-}
-
-
-
-void brw_clip_tri_init_vertices( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp0 = c->reg.loopcount; /* handy temporary */
-   struct brw_instruction *is_rev;
-
-   /* Initial list of indices for incoming vertexes:
-    */
-   brw_AND(p, tmp0, get_element_ud(c->reg.R0, 2), brw_imm_ud(PRIM_MASK));
-   brw_CMP(p,
-	   vec1(brw_null_reg()),
-	   BRW_CONDITIONAL_EQ,
-	   tmp0,
-	   brw_imm_ud(_3DPRIM_TRISTRIP_REVERSE));
-
-   /* XXX: Is there an easier way to do this?  Need to reverse every
-    * second tristrip element:  Can ignore sometimes?
-    */
-   is_rev = brw_IF(p, BRW_EXECUTE_1);
-   {
-      brw_MOV(p, get_element(c->reg.inlist, 0),  brw_address(c->reg.vertex[1]) );
-      brw_MOV(p, get_element(c->reg.inlist, 1),  brw_address(c->reg.vertex[0]) );
-      if (c->need_direction)
-	 brw_MOV(p, c->reg.dir, brw_imm_f(-1));
-   }
-   is_rev = brw_ELSE(p, is_rev);
-   {
-      brw_MOV(p, get_element(c->reg.inlist, 0),  brw_address(c->reg.vertex[0]) );
-      brw_MOV(p, get_element(c->reg.inlist, 1),  brw_address(c->reg.vertex[1]) );
-      if (c->need_direction)
-	 brw_MOV(p, c->reg.dir, brw_imm_f(1));
-   }
-   brw_ENDIF(p, is_rev);
-
-   brw_MOV(p, get_element(c->reg.inlist, 2),  brw_address(c->reg.vertex[2]) );
-   brw_MOV(p, brw_vec8_grf(c->reg.outlist.nr, 0), brw_imm_f(0));
-   brw_MOV(p, c->reg.nr_verts, brw_imm_ud(3));
-}
-
-
-
-void brw_clip_tri_flat_shade( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *is_poly;
-   struct brw_reg tmp0 = c->reg.loopcount; /* handy temporary */
-
-   brw_AND(p, tmp0, get_element_ud(c->reg.R0, 2), brw_imm_ud(PRIM_MASK));
-   brw_CMP(p,
-	   vec1(brw_null_reg()),
-	   BRW_CONDITIONAL_EQ,
-	   tmp0,
-	   brw_imm_ud(_3DPRIM_POLYGON));
-
-   is_poly = brw_IF(p, BRW_EXECUTE_1);
-   {
-      brw_clip_copy_colors(c, 1, 0);
-      brw_clip_copy_colors(c, 2, 0);
-   }
-   is_poly = brw_ELSE(p, is_poly);
-   {
-      brw_clip_copy_colors(c, 0, 2);
-      brw_clip_copy_colors(c, 1, 2);
-   }
-   brw_ENDIF(p, is_poly);
-}
-
-
-
-/* Use mesa's clipping algorithms, translated to GEN4 assembly.
- */
-void brw_clip_tri( struct brw_clip_compile *c )
-{
-#if 0
-   struct brw_compile *p = &c->func;
-   struct brw_indirect vtx = brw_indirect(0, 0);
-   struct brw_indirect vtxPrev = brw_indirect(1, 0);
-   struct brw_indirect vtxOut = brw_indirect(2, 0);
-   struct brw_indirect plane_ptr = brw_indirect(3, 0);
-   struct brw_indirect inlist_ptr = brw_indirect(4, 0);
-   struct brw_indirect outlist_ptr = brw_indirect(5, 0);
-   struct brw_indirect freelist_ptr = brw_indirect(6, 0);
-   struct brw_instruction *plane_loop;
-   struct brw_instruction *plane_active;
-   struct brw_instruction *vertex_loop;
-   struct brw_instruction *next_test;
-   struct brw_instruction *prev_test;
-
-   brw_MOV(p, get_addr_reg(vtxPrev),     brw_address(c->reg.vertex[2]) );
-   brw_MOV(p, get_addr_reg(plane_ptr),   brw_clip_plane0_address(c));
-   brw_MOV(p, get_addr_reg(inlist_ptr),  brw_address(c->reg.inlist));
-   brw_MOV(p, get_addr_reg(outlist_ptr), brw_address(c->reg.outlist));
-
-   brw_MOV(p, get_addr_reg(freelist_ptr), brw_address(c->reg.vertex[3]) );
-
-   plane_loop = brw_DO(p, BRW_EXECUTE_1);
-   {
-      /* if (planemask & 1)
-       */
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-      brw_AND(p, vec1(brw_null_reg()), c->reg.planemask, brw_imm_ud(1));
-
-      plane_active = brw_IF(p, BRW_EXECUTE_1);
-      {
-	 /* vtxOut = freelist_ptr++
-	  */
-	 brw_MOV(p, get_addr_reg(vtxOut),       get_addr_reg(freelist_ptr) );
-	 brw_ADD(p, get_addr_reg(freelist_ptr), get_addr_reg(freelist_ptr), brw_imm_uw(c->nr_regs * REG_SIZE));
-
-	 if (c->key.nr_userclip)
-	    brw_MOV(p, c->reg.plane_equation, deref_4f(plane_ptr, 0));
-	 else
-	    brw_MOV(p, c->reg.plane_equation, deref_4b(plane_ptr, 0));
-
-	 brw_MOV(p, c->reg.loopcount, c->reg.nr_verts);
-	 brw_MOV(p, c->reg.nr_verts, brw_imm_ud(0));
-
-	 vertex_loop = brw_DO(p, BRW_EXECUTE_1);
-	 {
-	    /* vtx = *input_ptr;
-	     */
-	    brw_MOV(p, get_addr_reg(vtx), deref_1uw(inlist_ptr, 0));
-
-	    /* IS_NEGATIVE(prev) */
-	    brw_set_conditionalmod(p, BRW_CONDITIONAL_L);
-	    brw_DP4(p, vec4(c->reg.dpPrev), deref_4f(vtxPrev, c->offset[VERT_RESULT_HPOS]), c->reg.plane_equation);
-	    prev_test = brw_IF(p, BRW_EXECUTE_1);
-	    {
-	       /* IS_POSITIVE(next)
-		*/
-	       brw_set_conditionalmod(p, BRW_CONDITIONAL_GE);
-	       brw_DP4(p, vec4(c->reg.dp), deref_4f(vtx, c->offset[VERT_RESULT_HPOS]), c->reg.plane_equation);
-	       next_test = brw_IF(p, BRW_EXECUTE_1);
-	       {
-
-		  /* Coming back in.
-		   */
-		  brw_ADD(p, c->reg.t, c->reg.dpPrev, negate(c->reg.dp));
-		  brw_math_invert(p, c->reg.t, c->reg.t);
-		  brw_MUL(p, c->reg.t, c->reg.t, c->reg.dpPrev);
-
-		  /* If (vtxOut == 0) vtxOut = vtxPrev
-		   */
-		  brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_EQ, get_addr_reg(vtxOut), brw_imm_uw(0) );
-		  brw_MOV(p, get_addr_reg(vtxOut), get_addr_reg(vtxPrev) );
-		  brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-		  brw_clip_interp_vertex(c, vtxOut, vtxPrev, vtx, c->reg.t, FALSE);
-
-		  /* *outlist_ptr++ = vtxOut;
-		   * nr_verts++;
-		   * vtxOut = 0;
-		   */
-		  brw_MOV(p, deref_1uw(outlist_ptr, 0), get_addr_reg(vtxOut));
-		  brw_ADD(p, get_addr_reg(outlist_ptr), get_addr_reg(outlist_ptr), brw_imm_uw(sizeof(short)));
-		  brw_ADD(p, c->reg.nr_verts, c->reg.nr_verts, brw_imm_ud(1));
-		  brw_MOV(p, get_addr_reg(vtxOut), brw_imm_uw(0) );
-	       }
-	       brw_ENDIF(p, next_test);
-
-	    }
-	    prev_test = brw_ELSE(p, prev_test);
-	    {
-	       /* *outlist_ptr++ = vtxPrev;
-		* nr_verts++;
-		*/
-	       brw_MOV(p, deref_1uw(outlist_ptr, 0), get_addr_reg(vtxPrev));
-	       brw_ADD(p, get_addr_reg(outlist_ptr), get_addr_reg(outlist_ptr), brw_imm_uw(sizeof(short)));
-	       brw_ADD(p, c->reg.nr_verts, c->reg.nr_verts, brw_imm_ud(1));
-
-	       /* IS_NEGATIVE(next)
-		*/
-	       brw_set_conditionalmod(p, BRW_CONDITIONAL_L);
-	       brw_DP4(p, vec4(c->reg.dp), deref_4f(vtx, c->offset[VERT_RESULT_HPOS]), c->reg.plane_equation);
-	       next_test = brw_IF(p, BRW_EXECUTE_1);
-	       {
-		  /* Going out of bounds.  Avoid division by zero as we
-		   * know dp != dpPrev from DIFFERENT_SIGNS, above.
-		   */
-		  brw_ADD(p, c->reg.t, c->reg.dp, negate(c->reg.dpPrev));
-		  brw_math_invert(p, c->reg.t, c->reg.t);
-		  brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp);
-
-		  /* If (vtxOut == 0) vtxOut = vtx
-		   */
-		  brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_EQ, get_addr_reg(vtxOut), brw_imm_uw(0) );
-		  brw_MOV(p, get_addr_reg(vtxOut), get_addr_reg(vtx) );
-		  brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-		  brw_clip_interp_vertex(c, vtxOut, vtx, vtxPrev, c->reg.t, TRUE);
-
-		  /* *outlist_ptr++ = vtxOut;
-		   * nr_verts++;
-		   * vtxOut = 0;
-		   */
-		  brw_MOV(p, deref_1uw(outlist_ptr, 0), get_addr_reg(vtxOut));
-		  brw_ADD(p, get_addr_reg(outlist_ptr), get_addr_reg(outlist_ptr), brw_imm_uw(sizeof(short)));
-		  brw_ADD(p, c->reg.nr_verts, c->reg.nr_verts, brw_imm_ud(1));
-		  brw_MOV(p, get_addr_reg(vtxOut), brw_imm_uw(0) );
-	       }
-	       brw_ENDIF(p, next_test);
-	    }
-	    brw_ENDIF(p, prev_test);
-
-	    /* vtxPrev = vtx;
-	     * inlist_ptr++;
-	     */
-	    brw_MOV(p, get_addr_reg(vtxPrev), get_addr_reg(vtx));
-	    brw_ADD(p, get_addr_reg(inlist_ptr), get_addr_reg(inlist_ptr), brw_imm_uw(sizeof(short)));
-
-	    /* while (--loopcount != 0)
-	     */
-	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-	    brw_ADD(p, c->reg.loopcount, c->reg.loopcount, brw_imm_d(-1));
-	 }
-	 brw_WHILE(p, vertex_loop);
-
-	 /* vtxPrev = *(outlist_ptr-1)  OR: outlist[nr_verts-1]
-	  * inlist = outlist
-	  * inlist_ptr = &inlist[0]
-	  * outlist_ptr = &outlist[0]
-	  */
-	 brw_ADD(p, get_addr_reg(outlist_ptr), get_addr_reg(outlist_ptr), brw_imm_w(-2));
-	 brw_MOV(p, get_addr_reg(vtxPrev), deref_1uw(outlist_ptr, 0));
-	 brw_MOV(p, brw_vec8_grf(c->reg.inlist.nr, 0), brw_vec8_grf(c->reg.outlist.nr, 0));
-	 brw_MOV(p, get_addr_reg(inlist_ptr), brw_address(c->reg.inlist));
-	 brw_MOV(p, get_addr_reg(outlist_ptr), brw_address(c->reg.outlist));
-      }
-      brw_ENDIF(p, plane_active);
-
-      /* plane_ptr++;
-       */
-      brw_ADD(p, get_addr_reg(plane_ptr), get_addr_reg(plane_ptr), brw_clip_plane_stride(c));
-
-      /* nr_verts >= 3
-       */
-      brw_CMP(p,
-	      vec1(brw_null_reg()),
-	      BRW_CONDITIONAL_GE,
-	      c->reg.nr_verts,
-	      brw_imm_ud(3));
-
-      /* && (planemask>>=1) != 0
-       */
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-      brw_SHR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(1));
-   }
-   brw_WHILE(p, plane_loop);
-#else
-         #warning "disabled"
-#endif
-}
-
-
-
-void brw_clip_tri_emit_polygon(struct brw_clip_compile *c)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *loop, *if_insn;
-
-   /* for (loopcount = nr_verts-2; loopcount > 0; loopcount--)
-    */
-   brw_set_conditionalmod(p, BRW_CONDITIONAL_G);
-   brw_ADD(p,
-	   c->reg.loopcount,
-	   c->reg.nr_verts,
-	   brw_imm_d(-2));
-
-   if_insn = brw_IF(p, BRW_EXECUTE_1);
-   {
-      struct brw_indirect v0 = brw_indirect(0, 0);
-      struct brw_indirect vptr = brw_indirect(1, 0);
-
-      brw_MOV(p, get_addr_reg(vptr), brw_address(c->reg.inlist));
-      brw_MOV(p, get_addr_reg(v0), deref_1uw(vptr, 0));
-
-      brw_clip_emit_vue(c, v0, 1, 0, ((_3DPRIM_TRIFAN << 2) | R02_PRIM_START));
-
-      brw_ADD(p, get_addr_reg(vptr), get_addr_reg(vptr), brw_imm_uw(2));
-      brw_MOV(p, get_addr_reg(v0), deref_1uw(vptr, 0));
-
-      loop = brw_DO(p, BRW_EXECUTE_1);
-      {
-	 brw_clip_emit_vue(c, v0, 1, 0, (_3DPRIM_TRIFAN << 2));
-
-	 brw_ADD(p, get_addr_reg(vptr), get_addr_reg(vptr), brw_imm_uw(2));
-	 brw_MOV(p, get_addr_reg(v0), deref_1uw(vptr, 0));
-
-	 brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-	 brw_ADD(p, c->reg.loopcount, c->reg.loopcount, brw_imm_d(-1));
-      }
-      brw_WHILE(p, loop);
-
-      brw_clip_emit_vue(c, v0, 0, 1, ((_3DPRIM_TRIFAN << 2) | R02_PRIM_END));
-   }
-   brw_ENDIF(p, if_insn);
-}
-
-static void do_clip_tri( struct brw_clip_compile *c )
-{
-   brw_clip_init_planes(c);
-
-   brw_clip_tri(c);
-}
-
-
-static void maybe_do_clip_tri( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *do_clip;
-
-   brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_NZ, c->reg.planemask, brw_imm_ud(0));
-   do_clip = brw_IF(p, BRW_EXECUTE_1);
-   {
-      do_clip_tri(c);
-   }
-   brw_ENDIF(p, do_clip);
-}
-
-static void brw_clip_test( struct brw_clip_compile *c )
-{
-#if 0
-    struct brw_reg t = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
-    struct brw_reg t1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
-    struct brw_reg t2 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
-    struct brw_reg t3 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
-
-    struct brw_reg v0 = get_tmp(c);
-    struct brw_reg v1 = get_tmp(c);
-    struct brw_reg v2 = get_tmp(c);
-
-    struct brw_indirect vt0 = brw_indirect(0, 0);
-    struct brw_indirect vt1 = brw_indirect(1, 0);
-    struct brw_indirect vt2 = brw_indirect(2, 0);
-
-    struct brw_compile *p = &c->func;
-
-    brw_MOV(p, get_addr_reg(vt0), brw_address(c->reg.vertex[0]));
-    brw_MOV(p, get_addr_reg(vt1), brw_address(c->reg.vertex[1]));
-    brw_MOV(p, get_addr_reg(vt2), brw_address(c->reg.vertex[2]));
-    brw_MOV(p, v0, deref_4f(vt0, c->offset[VERT_RESULT_HPOS]));
-    brw_MOV(p, v1, deref_4f(vt1, c->offset[VERT_RESULT_HPOS]));
-    brw_MOV(p, v2, deref_4f(vt2, c->offset[VERT_RESULT_HPOS]));
-
-    /* test nearz, xmin, ymin plane */
-    brw_CMP(p, t1, BRW_CONDITIONAL_LE, negate(v0), get_element(v0, 3));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t2, BRW_CONDITIONAL_LE, negate(v1), get_element(v1, 3));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t3, BRW_CONDITIONAL_LE, negate(v2), get_element(v2, 3));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_XOR(p, t, t1, t2);
-    brw_XOR(p, t1, t2, t3);
-    brw_OR(p, t, t, t1);
-
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
-	    get_element(t, 0), brw_imm_ud(0));
-    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<5)));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
-	    get_element(t, 1), brw_imm_ud(0));
-    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<3)));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
-	    get_element(t, 2), brw_imm_ud(0));
-    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<1)));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-    /* test farz, xmax, ymax plane */
-    brw_CMP(p, t1, BRW_CONDITIONAL_L, v0, get_element(v0, 3));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t2, BRW_CONDITIONAL_L, v1, get_element(v1, 3));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t3, BRW_CONDITIONAL_L, v2, get_element(v2, 3));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-    brw_XOR(p, t, t1, t2);
-    brw_XOR(p, t1, t2, t3);
-    brw_OR(p, t, t, t1);
-
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
-	    get_element(t, 0), brw_imm_ud(0));
-    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<4)));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
-	    get_element(t, 1), brw_imm_ud(0));
-    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<2)));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
-	    get_element(t, 2), brw_imm_ud(0));
-    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<0)));
-    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-    release_tmps(c);
-#else
-         #warning "disabled"
-#endif
-}
-
-
-void brw_emit_tri_clip( struct brw_clip_compile *c )
-{
-   struct brw_instruction *neg_rhw;
-   struct brw_compile *p = &c->func;
-   brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6);
-   brw_clip_tri_init_vertices(c);
-   brw_clip_init_clipmask(c);
-
-   /* if -ve rhw workaround bit is set,
-      do cliptest */
-   brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-   brw_AND(p, brw_null_reg(), get_element_ud(c->reg.R0, 2),
-	   brw_imm_ud(1<<20));
-   neg_rhw = brw_IF(p, BRW_EXECUTE_1);
-   {
-       brw_clip_test(c);
-   }
-   brw_ENDIF(p, neg_rhw);
-
-   /* Can't push into do_clip_tri because with polygon (or quad)
-    * flatshading, need to apply the flatshade here because we don't
-    * respect the PV when converting to trifan for emit:
-    */
-   if (c->key.do_flat_shading)
-      brw_clip_tri_flat_shade(c);
-
-   if (c->key.clip_mode == BRW_CLIPMODE_NORMAL)
-      do_clip_tri(c);
-   else
-      maybe_do_clip_tri(c);
-
-   brw_clip_tri_emit_polygon(c);
-
-   /* Send an empty message to kill the thread:
-    */
-   brw_clip_kill_thread(c);
-}
diff --git a/src/gallium/drivers/i965simple/brw_clip_unfilled.c b/src/gallium/drivers/i965simple/brw_clip_unfilled.c
deleted file mode 100644
index b774a76dd6..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip_unfilled.c
+++ /dev/null
@@ -1,477 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_clip.h"
-
-
-
-/* This is performed against the original triangles, so no indirection
- * required:
-BZZZT!
- */
-static void compute_tri_direction( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg e = c->reg.tmp0;
-   struct brw_reg f = c->reg.tmp1;
-   struct brw_reg v0 = byte_offset(c->reg.vertex[0], c->offset[VERT_RESULT_HPOS]); 
-   struct brw_reg v1 = byte_offset(c->reg.vertex[1], c->offset[VERT_RESULT_HPOS]); 
-   struct brw_reg v2 = byte_offset(c->reg.vertex[2], c->offset[VERT_RESULT_HPOS]); 
-
-
-   /* Calculate the vectors of two edges of the triangle:
-    */
-   brw_ADD(p, e, v0, negate(v2)); 
-   brw_ADD(p, f, v1, negate(v2)); 
-
-   /* Take their crossproduct:
-    */
-   brw_set_access_mode(p, BRW_ALIGN_16);
-   brw_MUL(p, vec4(brw_null_reg()), brw_swizzle(e, 1,2,0,3),  brw_swizzle(f,2,0,1,3));
-   brw_MAC(p, vec4(e),  negate(brw_swizzle(e, 2,0,1,3)), brw_swizzle(f,1,2,0,3));
-   brw_set_access_mode(p, BRW_ALIGN_1);
-
-   brw_MUL(p, c->reg.dir, c->reg.dir, vec4(e));
-}
-
-
-static void cull_direction( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *ccw;
-   unsigned conditional;
-
-   assert (!(c->key.fill_ccw == CLIP_CULL &&
-	     c->key.fill_cw == CLIP_CULL));
-
-   if (c->key.fill_ccw == CLIP_CULL)
-      conditional = BRW_CONDITIONAL_GE;
-   else
-      conditional = BRW_CONDITIONAL_L;
-
-   brw_CMP(p,
-	   vec1(brw_null_reg()),
-	   conditional,
-	   get_element(c->reg.dir, 2),
-	   brw_imm_f(0));
-   
-   ccw = brw_IF(p, BRW_EXECUTE_1);
-   {
-      brw_clip_kill_thread(c);
-   }
-   brw_ENDIF(p, ccw);
-}
-
-
-
-static void copy_bfc( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *ccw;
-   unsigned conditional;
-
-   /* Do we have any colors to copy? 
-    */
-   if (!(c->offset[VERT_RESULT_COL0] && c->offset[VERT_RESULT_BFC0]) &&
-       !(c->offset[VERT_RESULT_COL1] && c->offset[VERT_RESULT_BFC1]))
-      return;
-
-   /* In some wierd degnerate cases we can end up testing the
-    * direction twice, once for culling and once for bfc copying.  Oh
-    * well, that's what you get for setting wierd GL state.
-    */
-   if (c->key.copy_bfc_ccw)
-      conditional = BRW_CONDITIONAL_GE;
-   else
-      conditional = BRW_CONDITIONAL_L;
-
-   brw_CMP(p,
-	   vec1(brw_null_reg()),
-	   conditional,
-	   get_element(c->reg.dir, 2),
-	   brw_imm_f(0));
-   
-   ccw = brw_IF(p, BRW_EXECUTE_1);
-   {
-      unsigned i;
-
-      for (i = 0; i < 3; i++) {
-	 if (c->offset[VERT_RESULT_COL0] && c->offset[VERT_RESULT_BFC0])
-	    brw_MOV(p, 
-		    byte_offset(c->reg.vertex[i], c->offset[VERT_RESULT_COL0]),
-		    byte_offset(c->reg.vertex[i], c->offset[VERT_RESULT_BFC0]));
-
-	 if (c->offset[VERT_RESULT_COL1] && c->offset[VERT_RESULT_BFC1])
-	    brw_MOV(p, 
-		    byte_offset(c->reg.vertex[i], c->offset[VERT_RESULT_COL1]),
-		    byte_offset(c->reg.vertex[i], c->offset[VERT_RESULT_BFC1]));
-      }
-   }
-   brw_ENDIF(p, ccw);
-}
-
-
-
-
-/*
-  float iz	= 1.0 / dir.z;
-  float ac	= dir.x * iz;
-  float bc	= dir.y * iz;
-  offset = ctx->Polygon.OffsetUnits * DEPTH_SCALE;
-  offset += MAX2( abs(ac), abs(bc) ) * ctx->Polygon.OffsetFactor;
-  offset *= MRD;
-*/
-static void compute_offset( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg off = c->reg.offset;
-   struct brw_reg dir = c->reg.dir;
-   
-   brw_math_invert(p, get_element(off, 2), get_element(dir, 2));
-   brw_MUL(p, vec2(off), dir, get_element(off, 2));
-
-   brw_CMP(p, 
-	   vec1(brw_null_reg()), 
-	   BRW_CONDITIONAL_GE,
-	   brw_abs(get_element(off, 0)), 
-	   brw_abs(get_element(off, 1)));
-
-   brw_SEL(p, vec1(off), brw_abs(get_element(off, 0)), brw_abs(get_element(off, 1)));
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-   brw_MUL(p, vec1(off), off, brw_imm_f(c->key.offset_factor));
-   brw_ADD(p, vec1(off), off, brw_imm_f(c->key.offset_units));
-}
-
-
-static void merge_edgeflags( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *is_poly;
-   struct brw_reg tmp0 = get_element_ud(c->reg.tmp0, 0);
-
-   brw_AND(p, tmp0, get_element_ud(c->reg.R0, 2), brw_imm_ud(PRIM_MASK)); 
-   brw_CMP(p, 
-	   vec1(brw_null_reg()), 
-	   BRW_CONDITIONAL_EQ, 
-	   tmp0,
-	   brw_imm_ud(_3DPRIM_POLYGON));
-
-   /* Get away with using reg.vertex because we know that this is not
-    * a _3DPRIM_TRISTRIP_REVERSE:
-    */
-   is_poly = brw_IF(p, BRW_EXECUTE_1);
-   {   
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_EQ);
-      brw_AND(p, vec1(brw_null_reg()), get_element_ud(c->reg.R0, 2), brw_imm_ud(1<<8));
-      brw_MOV(p, byte_offset(c->reg.vertex[0], c->offset[VERT_RESULT_EDGE]), brw_imm_f(0));
-      brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_EQ);
-      brw_AND(p, vec1(brw_null_reg()), get_element_ud(c->reg.R0, 2), brw_imm_ud(1<<9));
-      brw_MOV(p, byte_offset(c->reg.vertex[2], c->offset[VERT_RESULT_EDGE]), brw_imm_f(0));
-      brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-   }
-   brw_ENDIF(p, is_poly);
-}
-
-
-
-static void apply_one_offset( struct brw_clip_compile *c,
-			  struct brw_indirect vert )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg pos = deref_4f(vert, c->offset[VERT_RESULT_HPOS]);
-   struct brw_reg z = get_element(pos, 2);
-
-   brw_ADD(p, z, z, vec1(c->reg.offset));
-}
-
-
-
-/***********************************************************************
- * Output clipped polygon as an unfilled primitive:
- */
-static void emit_lines(struct brw_clip_compile *c,
-		       boolean do_offset)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *loop;
-   struct brw_instruction *draw_edge;
-   struct brw_indirect v0 = brw_indirect(0, 0);
-   struct brw_indirect v1 = brw_indirect(1, 0);
-   struct brw_indirect v0ptr = brw_indirect(2, 0);
-   struct brw_indirect v1ptr = brw_indirect(3, 0);
-
-   /* Need a seperate loop for offset:
-    */
-   if (do_offset) {
-      brw_MOV(p, c->reg.loopcount, c->reg.nr_verts);
-      brw_MOV(p, get_addr_reg(v0ptr), brw_address(c->reg.inlist));
-
-      loop = brw_DO(p, BRW_EXECUTE_1);
-      {
-	 brw_MOV(p, get_addr_reg(v0), deref_1uw(v0ptr, 0));
-	 brw_ADD(p, get_addr_reg(v0ptr), get_addr_reg(v0ptr), brw_imm_uw(2));
-	    
-	 apply_one_offset(c, v0);
-	    
-	 brw_set_conditionalmod(p, BRW_CONDITIONAL_G);
-	 brw_ADD(p, c->reg.loopcount, c->reg.loopcount, brw_imm_d(-1));
-      }
-      brw_WHILE(p, loop);
-   }
-
-   /* v1ptr = &inlist[nr_verts]
-    * *v1ptr = v0
-    */
-   brw_MOV(p, c->reg.loopcount, c->reg.nr_verts);
-   brw_MOV(p, get_addr_reg(v0ptr), brw_address(c->reg.inlist));
-   brw_ADD(p, get_addr_reg(v1ptr), get_addr_reg(v0ptr), retype(c->reg.nr_verts, BRW_REGISTER_TYPE_UW));
-   brw_ADD(p, get_addr_reg(v1ptr), get_addr_reg(v1ptr), retype(c->reg.nr_verts, BRW_REGISTER_TYPE_UW));
-   brw_MOV(p, deref_1uw(v1ptr, 0), deref_1uw(v0ptr, 0));
-
-   loop = brw_DO(p, BRW_EXECUTE_1);
-   {
-      brw_MOV(p, get_addr_reg(v0), deref_1uw(v0ptr, 0));
-      brw_MOV(p, get_addr_reg(v1), deref_1uw(v0ptr, 2));
-      brw_ADD(p, get_addr_reg(v0ptr), get_addr_reg(v0ptr), brw_imm_uw(2));
-
-      /* draw edge if edgeflag != 0 */
-      brw_CMP(p, 
-	      vec1(brw_null_reg()), BRW_CONDITIONAL_NZ, 
-	      deref_1f(v0, c->offset[VERT_RESULT_EDGE]),
-	      brw_imm_f(0));
-      draw_edge = brw_IF(p, BRW_EXECUTE_1);
-      {
-	 brw_clip_emit_vue(c, v0, 1, 0, (_3DPRIM_LINESTRIP << 2) | R02_PRIM_START);
-	 brw_clip_emit_vue(c, v1, 1, 0, (_3DPRIM_LINESTRIP << 2) | R02_PRIM_END);
-      }
-      brw_ENDIF(p, draw_edge);
-
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-      brw_ADD(p, c->reg.loopcount, c->reg.loopcount, brw_imm_d(-1));
-   }
-   brw_WHILE(p, loop);
-}
-
-
-
-static void emit_points(struct brw_clip_compile *c,
-			boolean do_offset )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *loop;
-   struct brw_instruction *draw_point;
-
-   struct brw_indirect v0 = brw_indirect(0, 0);
-   struct brw_indirect v0ptr = brw_indirect(2, 0);
-
-   brw_MOV(p, c->reg.loopcount, c->reg.nr_verts);
-   brw_MOV(p, get_addr_reg(v0ptr), brw_address(c->reg.inlist));
-
-   loop = brw_DO(p, BRW_EXECUTE_1);
-   {
-      brw_MOV(p, get_addr_reg(v0), deref_1uw(v0ptr, 0));
-      brw_ADD(p, get_addr_reg(v0ptr), get_addr_reg(v0ptr), brw_imm_uw(2));
-
-      /* draw if edgeflag != 0 
-       */
-      brw_CMP(p, 
-	      vec1(brw_null_reg()), BRW_CONDITIONAL_NZ, 
-	      deref_1f(v0, c->offset[VERT_RESULT_EDGE]),
-	      brw_imm_f(0));
-      draw_point = brw_IF(p, BRW_EXECUTE_1);
-      {
-	 if (do_offset)
-	    apply_one_offset(c, v0);
-
-	 brw_clip_emit_vue(c, v0, 1, 0, (_3DPRIM_POINTLIST << 2) | R02_PRIM_START | R02_PRIM_END);
-      }
-      brw_ENDIF(p, draw_point);
-
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-      brw_ADD(p, c->reg.loopcount, c->reg.loopcount, brw_imm_d(-1));
-   }
-   brw_WHILE(p, loop);
-}
-
-
-
-
-
-
-
-static void emit_primitives( struct brw_clip_compile *c,
-			     unsigned mode, 
-			     boolean do_offset )
-{
-   switch (mode) {
-   case CLIP_FILL:
-      brw_clip_tri_emit_polygon(c);
-      break;
-
-   case CLIP_LINE:
-      emit_lines(c, do_offset);
-      break;
-
-   case CLIP_POINT:
-      emit_points(c, do_offset);
-      break;
-
-   case CLIP_CULL:
-      assert(0);
-      break;
-   }
-} 
-
-
-
-static void emit_unfilled_primitives( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *ccw;
-
-   /* Direction culling has already been done.
-    */
-   if (c->key.fill_ccw != c->key.fill_cw &&
-       c->key.fill_ccw != CLIP_CULL &&
-       c->key.fill_cw != CLIP_CULL)
-   {
-      brw_CMP(p,
-	      vec1(brw_null_reg()),
-	      BRW_CONDITIONAL_GE,
-	      get_element(c->reg.dir, 2),
-	      brw_imm_f(0));
-   
-      ccw = brw_IF(p, BRW_EXECUTE_1);
-      {
-	 emit_primitives(c, c->key.fill_ccw, c->key.offset_ccw);
-      }
-      ccw = brw_ELSE(p, ccw);
-      {
-	 emit_primitives(c, c->key.fill_cw, c->key.offset_cw);
-      }
-      brw_ENDIF(p, ccw);
-   }
-   else if (c->key.fill_cw != CLIP_CULL) {
-      emit_primitives(c, c->key.fill_cw, c->key.offset_cw);
-   }
-   else if (c->key.fill_ccw != CLIP_CULL) { 
-      emit_primitives(c, c->key.fill_ccw, c->key.offset_ccw);
-   }
-}
-
-
-
-
-static void check_nr_verts( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *if_insn;
-
-   brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.nr_verts, brw_imm_d(3));      
-   if_insn = brw_IF(p, BRW_EXECUTE_1);
-   {
-      brw_clip_kill_thread(c);
-   }
-   brw_ENDIF(p, if_insn);
-}
-
-
-void brw_emit_unfilled_clip( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *do_clip;
-   
-
-   c->need_direction = ((c->key.offset_ccw || c->key.offset_cw) ||
-			(c->key.fill_ccw != c->key.fill_cw) ||
-			c->key.fill_ccw == CLIP_CULL ||
-			c->key.fill_cw == CLIP_CULL ||
-			c->key.copy_bfc_cw ||
-			c->key.copy_bfc_ccw);
-
-   brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6);
-   brw_clip_tri_init_vertices(c);
-
-   assert(c->offset[VERT_RESULT_EDGE]);
-
-   if (c->key.fill_ccw == CLIP_CULL &&
-       c->key.fill_cw == CLIP_CULL) {
-      brw_clip_kill_thread(c);
-      return;
-   }
-
-   merge_edgeflags(c);
-
-   /* Need to use the inlist indirection here: 
-    */
-   if (c->need_direction) 
-      compute_tri_direction(c);
-   
-   if (c->key.fill_ccw == CLIP_CULL ||
-       c->key.fill_cw == CLIP_CULL)
-      cull_direction(c);
-
-   if (c->key.offset_ccw ||
-       c->key.offset_cw)
-      compute_offset(c);
-
-   if (c->key.copy_bfc_ccw ||
-       c->key.copy_bfc_cw)
-      copy_bfc(c);
-
-   /* Need to do this whether we clip or not:
-    */
-   if (c->key.do_flat_shading)
-      brw_clip_tri_flat_shade(c);
-   
-   brw_clip_init_clipmask(c);
-   brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_NZ, c->reg.planemask, brw_imm_ud(0));
-   do_clip = brw_IF(p, BRW_EXECUTE_1);
-   {
-      brw_clip_init_planes(c);
-      brw_clip_tri(c);
-      check_nr_verts(c);
-   }
-   brw_ENDIF(p, do_clip);
-   
-   emit_unfilled_primitives(c);
-   brw_clip_kill_thread(c);
-}
-
-
-
diff --git a/src/gallium/drivers/i965simple/brw_clip_util.c b/src/gallium/drivers/i965simple/brw_clip_util.c
deleted file mode 100644
index 6d58ceafff..0000000000
--- a/src/gallium/drivers/i965simple/brw_clip_util.c
+++ /dev/null
@@ -1,351 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_clip.h"
-
-
-
-
-
-static struct brw_reg get_tmp( struct brw_clip_compile *c )
-{
-   struct brw_reg tmp = brw_vec4_grf(c->last_tmp, 0);
-
-   if (++c->last_tmp > c->prog_data.total_grf)
-      c->prog_data.total_grf = c->last_tmp;
-
-   return tmp;
-}
-
-static void release_tmp( struct brw_clip_compile *c, struct brw_reg tmp )
-{
-   if (tmp.nr == c->last_tmp-1)
-      c->last_tmp--;
-}
-
-
-static struct brw_reg make_plane_ud(unsigned x, unsigned y, unsigned z, unsigned w)
-{
-   return brw_imm_ud((w<<24) | (z<<16) | (y<<8) | x);
-}
-
-
-void brw_clip_init_planes( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-
-   if (!c->key.nr_userclip) {
-      brw_MOV(p, get_element_ud(c->reg.fixed_planes, 0), make_plane_ud( 0,    0, 0xff, 1));
-      brw_MOV(p, get_element_ud(c->reg.fixed_planes, 1), make_plane_ud( 0,    0,    1, 1));
-      brw_MOV(p, get_element_ud(c->reg.fixed_planes, 2), make_plane_ud( 0, 0xff,    0, 1));
-      brw_MOV(p, get_element_ud(c->reg.fixed_planes, 3), make_plane_ud( 0,    1,    0, 1));
-      brw_MOV(p, get_element_ud(c->reg.fixed_planes, 4), make_plane_ud(0xff,  0,    0, 1));
-      brw_MOV(p, get_element_ud(c->reg.fixed_planes, 5), make_plane_ud( 1,    0,    0, 1));
-   }
-}
-
-
-
-#define W 3
-
-/* Project 'pos' to screen space (or back again), overwrite with results:
- */
-static void brw_clip_project_position(struct brw_clip_compile *c, struct brw_reg pos )
-{
-   struct brw_compile *p = &c->func;
-
-   /* calc rhw
-    */
-   brw_math_invert(p, get_element(pos, W), get_element(pos, W));
-
-   /* value.xyz *= value.rhw
-    */
-   brw_set_access_mode(p, BRW_ALIGN_16);
-   brw_MUL(p, brw_writemask(pos, TGSI_WRITEMASK_XYZ), pos, brw_swizzle1(pos, W));
-   brw_set_access_mode(p, BRW_ALIGN_1);
-}
-
-
-static void brw_clip_project_vertex( struct brw_clip_compile *c,
-				     struct brw_indirect vert_addr )
-{
-#if 0
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = get_tmp(c);
-
-   /* Fixup position.  Extract from the original vertex and re-project
-    * to screen space:
-    */
-   brw_MOV(p, tmp, deref_4f(vert_addr, c->offset[VERT_RESULT_HPOS]));
-   brw_clip_project_position(c, tmp);
-   brw_MOV(p, deref_4f(vert_addr, c->header_position_offset), tmp);
-
-   release_tmp(c, tmp);
-#else
-         #warning "disabled"
-#endif
-}
-
-
-
-
-/* Interpolate between two vertices and put the result into a0.0.
- * Increment a0.0 accordingly.
- */
-void brw_clip_interp_vertex( struct brw_clip_compile *c,
-			     struct brw_indirect dest_ptr,
-			     struct brw_indirect v0_ptr, /* from */
-			     struct brw_indirect v1_ptr, /* to */
-			     struct brw_reg t0,
-			     boolean force_edgeflag)
-{
-#if 0
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = get_tmp(c);
-   unsigned i;
-
-   /* Just copy the vertex header:
-    */
-   brw_copy_indirect_to_indirect(p, dest_ptr, v0_ptr, 1);
-
-   /* Iterate over each attribute (could be done in pairs?)
-    */
-   for (i = 0; i < c->nr_attrs; i++) {
-      unsigned delta = i*16 + 32;
-
-      if (delta == c->offset[VERT_RESULT_EDGE]) {
-	 if (force_edgeflag)
-	    brw_MOV(p, deref_4f(dest_ptr, delta), brw_imm_f(1));
-	 else
-	    brw_MOV(p, deref_4f(dest_ptr, delta), deref_4f(v0_ptr, delta));
-      }
-      else {
-	 /* Interpolate:
-	  *
-	  *        New = attr0 + t*attr1 - t*attr0
-	  */
-	 brw_MUL(p,
-		 vec4(brw_null_reg()),
-		 deref_4f(v1_ptr, delta),
-		 t0);
-
-	 brw_MAC(p,
-		 tmp,
-		 negate(deref_4f(v0_ptr, delta)),
-		 t0);
-
-	 brw_ADD(p,
-		 deref_4f(dest_ptr, delta),
-		 deref_4f(v0_ptr, delta),
-		 tmp);
-      }
-   }
-
-   if (i & 1) {
-      unsigned delta = i*16 + 32;
-      brw_MOV(p, deref_4f(dest_ptr, delta), brw_imm_f(0));
-   }
-
-   release_tmp(c, tmp);
-
-   /* Recreate the projected (NDC) coordinate in the new vertex
-    * header:
-    */
-   brw_clip_project_vertex(c, dest_ptr );
-#else
-         #warning "disabled"
-#endif
-}
-
-
-
-
-#define MAX_MRF 16
-
-void brw_clip_emit_vue(struct brw_clip_compile *c,
-		       struct brw_indirect vert,
-		       boolean allocate,
-		       boolean eot,
-		       unsigned header)
-{
-   struct brw_compile *p = &c->func;
-   unsigned start = c->last_mrf;
-
-   assert(!(allocate && eot));
-
-   /* Cycle through mrf regs - probably futile as we have to wait for
-    * the allocation response anyway.  Also, the order this function
-    * is invoked doesn't correspond to the order the instructions will
-    * be executed, so it won't have any effect in many cases.
-    */
-#if 0
-   if (start + c->nr_regs + 1 >= MAX_MRF)
-      start = 0;
-
-   c->last_mrf = start + c->nr_regs + 1;
-#endif
-
-   /* Copy the vertex from vertn into m1..mN+1:
-    */
-   brw_copy_from_indirect(p, brw_message_reg(start+1), vert, c->nr_regs);
-
-   /* Overwrite PrimType and PrimStart in the message header, for
-    * each vertex in turn:
-    */
-   brw_MOV(p, get_element_ud(c->reg.R0, 2), brw_imm_ud(header));
-
-
-   /* Send each vertex as a seperate write to the urb.  This
-    * is different to the concept in brw_sf_emit.c, where
-    * subsequent writes are used to build up a single urb
-    * entry.  Each of these writes instantiates a seperate
-    * urb entry - (I think... what about 'allocate'?)
-    */
-   brw_urb_WRITE(p,
-		 allocate ? c->reg.R0 : retype(brw_null_reg(), BRW_REGISTER_TYPE_UD),
-		 start,
-		 c->reg.R0,
-		 allocate,
-		 1,		/* used */
-		 c->nr_regs + 1, /* msg length */
-		 allocate ? 1 : 0, /* response_length */
-		 eot,		/* eot */
-		 1,		/* writes_complete */
-		 0,		/* urb offset */
-		 BRW_URB_SWIZZLE_NONE);
-}
-
-
-
-void brw_clip_kill_thread(struct brw_clip_compile *c)
-{
-   struct brw_compile *p = &c->func;
-
-   /* Send an empty message to kill the thread and release any
-    * allocated urb entry:
-    */
-   brw_urb_WRITE(p,
-		 retype(brw_null_reg(), BRW_REGISTER_TYPE_UD),
-		 0,
-		 c->reg.R0,
-		 0,		/* allocate */
-		 0,		/* used */
-		 0, 		/* msg len */
-		 0, 		/* response len */
-		 1, 		/* eot */
-		 1,		/* writes complete */
-		 0,
-		 BRW_URB_SWIZZLE_NONE);
-}
-
-
-
-
-struct brw_reg brw_clip_plane0_address( struct brw_clip_compile *c )
-{
-   return brw_address(c->reg.fixed_planes);
-}
-
-
-struct brw_reg brw_clip_plane_stride( struct brw_clip_compile *c )
-{
-   if (c->key.nr_userclip) {
-      return brw_imm_uw(16);
-   }
-   else {
-      return brw_imm_uw(4);
-   }
-}
-
-
-/* If flatshading, distribute color from provoking vertex prior to
- * clipping.
- */
-void brw_clip_copy_colors( struct brw_clip_compile *c,
-			   unsigned to, unsigned from )
-{
-#if 0
-   struct brw_compile *p = &c->func;
-
-   if (c->offset[VERT_RESULT_COL0])
-      brw_MOV(p,
-	      byte_offset(c->reg.vertex[to], c->offset[VERT_RESULT_COL0]),
-	      byte_offset(c->reg.vertex[from], c->offset[VERT_RESULT_COL0]));
-
-   if (c->offset[VERT_RESULT_COL1])
-      brw_MOV(p,
-	      byte_offset(c->reg.vertex[to], c->offset[VERT_RESULT_COL1]),
-	      byte_offset(c->reg.vertex[from], c->offset[VERT_RESULT_COL1]));
-
-   if (c->offset[VERT_RESULT_BFC0])
-      brw_MOV(p,
-	      byte_offset(c->reg.vertex[to], c->offset[VERT_RESULT_BFC0]),
-	      byte_offset(c->reg.vertex[from], c->offset[VERT_RESULT_BFC0]));
-
-   if (c->offset[VERT_RESULT_BFC1])
-      brw_MOV(p,
-	      byte_offset(c->reg.vertex[to], c->offset[VERT_RESULT_BFC1]),
-	      byte_offset(c->reg.vertex[from], c->offset[VERT_RESULT_BFC1]));
-#else
-         #warning "disabled"
-#endif
-}
-
-
-
-void brw_clip_init_clipmask( struct brw_clip_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg incoming = get_element_ud(c->reg.R0, 2);
-
-   /* Shift so that lowest outcode bit is rightmost:
-    */
-   brw_SHR(p, c->reg.planemask, incoming, brw_imm_ud(26));
-
-   if (c->key.nr_userclip) {
-      struct brw_reg tmp = retype(vec1(get_tmp(c)), BRW_REGISTER_TYPE_UD);
-
-      /* Rearrange userclip outcodes so that they come directly after
-       * the fixed plane bits.
-       */
-      brw_AND(p, tmp, incoming, brw_imm_ud(0x3f<<14));
-      brw_SHR(p, tmp, tmp, brw_imm_ud(8));
-      brw_OR(p, c->reg.planemask, c->reg.planemask, tmp);
-
-      release_tmp(c, tmp);
-   }
-}
-
diff --git a/src/gallium/drivers/i965simple/brw_context.c b/src/gallium/drivers/i965simple/brw_context.c
deleted file mode 100644
index 9b33285bc7..0000000000
--- a/src/gallium/drivers/i965simple/brw_context.c
+++ /dev/null
@@ -1,139 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_draw.h"
-#include "brw_vs.h"
-#include "brw_tex_layout.h"
-#include "brw_winsys.h"
-
-#include "pipe/internal/p_winsys_screen.h"
-#include "pipe/p_context.h"
-#include "util/u_memory.h"
-#include "pipe/p_screen.h"
-
-
-#ifndef BRW_DEBUG
-int BRW_DEBUG = (0);
-#endif
-
-
-static void brw_destroy(struct pipe_context *pipe)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   if(brw->winsys->destroy)
-      brw->winsys->destroy(brw->winsys);
-   
-   FREE(brw);
-}
-
-
-static void brw_clear(struct pipe_context *pipe, struct pipe_surface *ps,
-                      unsigned clearValue)
-{
-   int x, y, w, h;
-   /* FIXME: corny... */
-
-   x = 0;
-   y = 0;
-   w = ps->width;
-   h = ps->height;
-
-   pipe->surface_fill(pipe, ps, x, y, w, h, clearValue);
-}
-
-static unsigned int
-brw_is_texture_referenced( struct pipe_context *pipe,
-			   struct pipe_texture *texture,
-			   unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-brw_is_buffer_referenced( struct pipe_context *pipe,
-			  struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-struct pipe_context *brw_create(struct pipe_screen *screen,
-                                struct brw_winsys *brw_winsys,
-                                unsigned pci_id)
-{
-   struct brw_context *brw;
-
-   debug_printf("%s: creating brw_context with pci id 0x%x\n",
-                __FUNCTION__, pci_id);
-
-   brw = CALLOC_STRUCT(brw_context);
-   if (brw == NULL)
-      return NULL;
-
-   brw->winsys = brw_winsys;
-   brw->pipe.winsys = screen->winsys;
-   brw->pipe.screen = screen;
-
-   brw->pipe.destroy = brw_destroy;
-   brw->pipe.clear = brw_clear;
-
-   brw->pipe.is_texture_referenced = brw_is_texture_referenced;
-   brw->pipe.is_buffer_referenced = brw_is_buffer_referenced;
-
-   brw_init_surface_functions(brw);
-   brw_init_texture_functions(brw);
-   brw_init_state_functions(brw);
-   brw_init_flush_functions(brw);
-   brw_init_draw_functions( brw );
-
-
-   brw_init_state( brw );
-
-   brw->pci_id = pci_id;
-   brw->dirty = ~0;
-   brw->hardware_dirty = ~0;
-
-   memset(&brw->wm.bind, ~0, sizeof(brw->wm.bind));
-
-   return &brw->pipe;
-}
-
diff --git a/src/gallium/drivers/i965simple/brw_context.h b/src/gallium/drivers/i965simple/brw_context.h
deleted file mode 100644
index 3079485180..0000000000
--- a/src/gallium/drivers/i965simple/brw_context.h
+++ /dev/null
@@ -1,684 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#ifndef BRWCONTEXT_INC
-#define BRWCONTEXT_INC
-
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_state.h"
-
-#include "tgsi/tgsi_scan.h"
-
-#include "brw_structs.h"
-#include "brw_winsys.h"
-
-
-/* Glossary:
- *
- * URB - uniform resource buffer.  A mid-sized buffer which is
- * partitioned between the fixed function units and used for passing
- * values (vertices, primitives, constants) between them.
- *
- * CURBE - constant URB entry.  An urb region (entry) used to hold
- * constant values which the fixed function units can be instructed to
- * preload into the GRF when spawining a thread.
- *
- * VUE - vertex URB entry.  An urb entry holding a vertex and usually
- * a vertex header.  The header contains control information and
- * things like primitive type, Begin/end flags and clip codes.
- *
- * PUE - primitive URB entry.  An urb entry produced by the setup (SF)
- * unit holding rasterization and interpolation parameters.
- *
- * GRF - general register file.  One of several register files
- * addressable by programmed threads.  The inputs (r0, payload, curbe,
- * urb) of the thread are preloaded to this area before the thread is
- * spawned.  The registers are individually 8 dwords wide and suitable
- * for general usage.  Registers holding thread input values are not
- * special and may be overwritten.
- *
- * MRF - message register file.  Threads communicate (and terminate)
- * by sending messages.  Message parameters are placed in contigous
- * MRF registers.  All program output is via these messages.  URB
- * entries are populated by sending a message to the shared URB
- * function containing the new data, together with a control word,
- * often an unmodified copy of R0.
- *
- * R0 - GRF register 0.  Typically holds control information used when
- * sending messages to other threads.
- *
- * EU or GEN4 EU: The name of the programmable subsystem of the
- * i965 hardware.  Threads are executed by the EU, the registers
- * described above are part of the EU architecture.
- *
- * Fixed function units:
- *
- * CS - Command streamer.  Notional first unit, little software
- * interaction.  Holds the URB entries used for constant data, ie the
- * CURBEs.
- *
- * VF/VS - Vertex Fetch / Vertex Shader.  The fixed function part of
- * this unit is responsible for pulling vertices out of vertex buffers
- * in vram and injecting them into the processing pipe as VUEs.  If
- * enabled, it first passes them to a VS thread which is a good place
- * for the driver to implement any active vertex shader.
- *
- * GS - Geometry Shader.  This corresponds to a new DX10 concept.  If
- * enabled, incoming strips etc are passed to GS threads in individual
- * line/triangle/point units.  The GS thread may perform arbitary
- * computation and emit whatever primtives with whatever vertices it
- * chooses.  This makes GS an excellent place to implement GL's
- * unfilled polygon modes, though of course it is capable of much
- * more.  Additionally, GS is used to translate away primitives not
- * handled by latter units, including Quads and Lineloops.
- *
- * CS - Clipper.  Mesa's clipping algorithms are imported to run on
- * this unit.  The fixed function part performs cliptesting against
- * the 6 fixed clipplanes and makes descisions on whether or not the
- * incoming primitive needs to be passed to a thread for clipping.
- * User clip planes are handled via cooperation with the VS thread.
- *
- * SF - Strips Fans or Setup: Triangles are prepared for
- * rasterization.  Interpolation coefficients are calculated.
- * Flatshading and two-side lighting usually performed here.
- *
- * WM - Windower.  Interpolation of vertex attributes performed here.
- * Fragment shader implemented here.  SIMD aspects of EU taken full
- * advantage of, as pixels are processed in blocks of 16.
- *
- * CC - Color Calculator.  No EU threads associated with this unit.
- * Handles blending and (presumably) depth and stencil testing.
- */
-
-#define BRW_MAX_CURBE                    (32*16)
-
-struct brw_context;
-struct brw_winsys;
-
-
-/* Raised when we receive new state across the pipe interface:
- */
-#define BRW_NEW_VIEWPORT                0x1
-#define BRW_NEW_RASTERIZER              0x2
-#define BRW_NEW_FS                      0x4
-#define BRW_NEW_BLEND                   0x8
-#define BRW_NEW_CLIP                    0x10
-#define BRW_NEW_SCISSOR                 0x20
-#define BRW_NEW_STIPPLE                 0x40
-#define BRW_NEW_FRAMEBUFFER             0x80
-#define BRW_NEW_ALPHA_TEST              0x100
-#define BRW_NEW_DEPTH_STENCIL           0x200
-#define BRW_NEW_SAMPLER                 0x400
-#define BRW_NEW_TEXTURE                 0x800
-#define BRW_NEW_CONSTANTS               0x1000
-#define BRW_NEW_VBO                     0x2000
-#define BRW_NEW_VS                      0x4000
-
-/* Raised for other internal events:
- */
-#define BRW_NEW_URB_FENCE               0x10000
-#define BRW_NEW_PSP                     0x20000
-#define BRW_NEW_CURBE_OFFSETS           0x40000
-#define BRW_NEW_REDUCED_PRIMITIVE       0x80000
-#define BRW_NEW_PRIMITIVE               0x100000
-#define BRW_NEW_SCENE                 0x200000
-#define BRW_NEW_SF_LINKAGE              0x400000
-
-extern int BRW_DEBUG;
-
-#define DEBUG_TEXTURE	0x1
-#define DEBUG_STATE	0x2
-#define DEBUG_IOCTL	0x4
-#define DEBUG_PRIMS	0x8
-#define DEBUG_VERTS	0x10
-#define DEBUG_FALLBACKS	0x20
-#define DEBUG_VERBOSE	0x40
-#define DEBUG_DRI       0x80
-#define DEBUG_DMA       0x100
-#define DEBUG_SANITY    0x200
-#define DEBUG_SYNC      0x400
-#define DEBUG_SLEEP     0x800
-#define DEBUG_PIXEL     0x1000
-#define DEBUG_STATS     0x2000
-#define DEBUG_TILE      0x4000
-#define DEBUG_SINGLE_THREAD   0x8000
-#define DEBUG_WM        0x10000
-#define DEBUG_URB       0x20000
-#define DEBUG_VS        0x40000
-#define DEBUG_BATCH	0x80000
-#define DEBUG_BUFMGR	0x100000
-#define DEBUG_BLIT	0x200000
-#define DEBUG_REGION	0x400000
-#define DEBUG_MIPTREE	0x800000
-
-#define DBG(...) do {						\
-   if (BRW_DEBUG & FILE_DEBUG_FLAG)				\
-      debug_printf(__VA_ARGS__);				\
-} while(0)
-
-#define PRINT(...) do {						\
-   debug_printf(__VA_ARGS__);			                \
-} while(0)
-
-struct brw_state_flags {
-   unsigned cache;
-   unsigned brw;
-};
-
-
-struct brw_vertex_program {
-   struct pipe_shader_state program;
-   struct tgsi_shader_info info;
-   int id;
-};
-
-
-struct brw_fragment_program {
-   struct pipe_shader_state program;
-   struct tgsi_shader_info info;
-   
-   boolean UsesDepth; /* XXX add this to tgsi_shader_info? */
-   int id;
-};
-
-
-struct pipe_setup_linkage {
-   struct {
-      unsigned vp_output:5;
-      unsigned interp_mode:4;
-      unsigned bf_vp_output:5;
-   } fp_input[PIPE_MAX_SHADER_INPUTS];
-
-   unsigned fp_input_count:5;
-   unsigned max_vp_output:5;
-};
-   
-
-
-struct brw_texture {
-   struct pipe_texture base;
-
-   /* Derived from the above:
-    */
-   unsigned stride;
-   unsigned depth_pitch;          /* per-image on i945? */
-   unsigned total_nblocksy;
-
-   unsigned nr_images[PIPE_MAX_TEXTURE_LEVELS];
-
-   /* Explicitly store the offset of each image for each cube face or
-    * depth value.  Pretty much have to accept that hardware formats
-    * are going to be so diverse that there is no unified way to
-    * compute the offsets of depth/cube images within a mipmap level,
-    * so have to store them as a lookup table:
-    */
-   unsigned *image_offset[PIPE_MAX_TEXTURE_LEVELS];   /**< array [depth] of offsets */
-
-   /* Includes image offset tables:
-    */
-   unsigned level_offset[PIPE_MAX_TEXTURE_LEVELS];
-
-   /* The data is held here:
-    */
-   struct pipe_buffer *buffer;
-};
-
-/* Data about a particular attempt to compile a program.  Note that
- * there can be many of these, each in a different GL state
- * corresponding to a different brw_wm_prog_key struct, with different
- * compiled programs:
- */
-/* Data about a particular attempt to compile a program.  Note that
- * there can be many of these, each in a different GL state
- * corresponding to a different brw_wm_prog_key struct, with different
- * compiled programs:
- */
-
-struct brw_wm_prog_data {
-   unsigned curb_read_length;
-   unsigned urb_read_length;
-
-   unsigned first_curbe_grf;
-   unsigned total_grf;
-   unsigned total_scratch;
-
-   /* Internally generated constants for the CURBE.  These are loaded
-    * ahead of the data from the constant buffer.
-    */
-   const float internal_const[8];
-   unsigned nr_internal_consts;
-   unsigned max_const;
-
-   boolean error;
-};
-
-struct brw_sf_prog_data {
-   unsigned urb_read_length;
-   unsigned total_grf;
-
-   /* Each vertex may have upto 12 attributes, 4 components each,
-    * except WPOS which requires only 2.  (11*4 + 2) == 44 ==> 11
-    * rows.
-    *
-    * Actually we use 4 for each, so call it 12 rows.
-    */
-   unsigned urb_entry_size;
-};
-
-struct brw_clip_prog_data {
-   unsigned curb_read_length;	/* user planes? */
-   unsigned clip_mode;
-   unsigned urb_read_length;
-   unsigned total_grf;
-};
-
-struct brw_gs_prog_data {
-   unsigned urb_read_length;
-   unsigned total_grf;
-};
-
-struct brw_vs_prog_data {
-   unsigned curb_read_length;
-   unsigned urb_read_length;
-   unsigned total_grf;
-   unsigned outputs_written;
-
-   unsigned inputs_read;
-
-   unsigned max_const;
-
-   float    imm_buf[PIPE_MAX_CONSTANT][4];
-   unsigned num_imm;
-   unsigned num_consts;
-
-   /* Used for calculating urb partitions:
-    */
-   unsigned urb_entry_size;
-};
-
-
-#define BRW_MAX_TEX_UNIT 8
-#define BRW_WM_MAX_SURF BRW_MAX_TEX_UNIT + 1
-
-/* Create a fixed sized struct for caching binding tables:
- */
-struct brw_surface_binding_table {
-   unsigned surf_ss_offset[BRW_WM_MAX_SURF];
-};
-
-
-struct brw_cache;
-
-struct brw_mem_pool {
-   struct pipe_buffer *buffer;
-
-   unsigned size;
-   unsigned offset;		/* offset of first free byte */
-
-   struct brw_context *brw;
-};
-
-struct brw_cache_item {
-   unsigned hash;
-   unsigned key_size;		/* for variable-sized keys */
-   const void *key;
-
-   unsigned offset;		/* offset within pool's buffer */
-   unsigned data_size;
-
-   struct brw_cache_item *next;
-};
-
-
-
-struct brw_cache {
-   unsigned id;
-
-   const char *name;
-
-   struct brw_context *brw;
-   struct brw_mem_pool *pool;
-
-   struct brw_cache_item **items;
-   unsigned size, n_items;
-
-   unsigned key_size;		/* for fixed-size keys */
-   unsigned aux_size;
-
-   unsigned last_addr;			/* offset of active item */
-};
-
-
-
-
-/* Considered adding a member to this struct to document which flags
- * an update might raise so that ordering of the state atoms can be
- * checked or derived at runtime.  Dropped the idea in favor of having
- * a debug mode where the state is monitored for flags which are
- * raised that have already been tested against.
- */
-struct brw_tracked_state {
-   struct brw_state_flags dirty;
-   void (*update)( struct brw_context *brw );
-};
-
-
-/* Flags for brw->state.cache.
- */
-#define CACHE_NEW_CC_VP                  (1<<BRW_CC_VP)
-#define CACHE_NEW_CC_UNIT                (1<<BRW_CC_UNIT)
-#define CACHE_NEW_WM_PROG                (1<<BRW_WM_PROG)
-#define CACHE_NEW_SAMPLER_DEFAULT_COLOR  (1<<BRW_SAMPLER_DEFAULT_COLOR)
-#define CACHE_NEW_SAMPLER                (1<<BRW_SAMPLER)
-#define CACHE_NEW_WM_UNIT                (1<<BRW_WM_UNIT)
-#define CACHE_NEW_SF_PROG                (1<<BRW_SF_PROG)
-#define CACHE_NEW_SF_VP                  (1<<BRW_SF_VP)
-#define CACHE_NEW_SF_UNIT                (1<<BRW_SF_UNIT)
-#define CACHE_NEW_VS_UNIT                (1<<BRW_VS_UNIT)
-#define CACHE_NEW_VS_PROG                (1<<BRW_VS_PROG)
-#define CACHE_NEW_GS_UNIT                (1<<BRW_GS_UNIT)
-#define CACHE_NEW_GS_PROG                (1<<BRW_GS_PROG)
-#define CACHE_NEW_CLIP_VP                (1<<BRW_CLIP_VP)
-#define CACHE_NEW_CLIP_UNIT              (1<<BRW_CLIP_UNIT)
-#define CACHE_NEW_CLIP_PROG              (1<<BRW_CLIP_PROG)
-#define CACHE_NEW_SURFACE                (1<<BRW_SS_SURFACE)
-#define CACHE_NEW_SURF_BIND              (1<<BRW_SS_SURF_BIND)
-
-
-
-
-enum brw_mempool_id {
-   BRW_GS_POOL,
-   BRW_SS_POOL,
-   BRW_MAX_POOL
-};
-
-
-struct brw_cached_batch_item {
-   struct header *header;
-   unsigned sz;
-   struct brw_cached_batch_item *next;
-};
-
-
-
-/* Protect against a future where PIPE_MAX_ATTRIBS > 32.  Wouldn't life
- * be easier if C allowed arrays of packed elements?
- */
-#define ATTRIB_BIT_DWORDS  ((PIPE_MAX_ATTRIBS+31)/32)
-
-
-
-
-struct brw_vertex_info {
-   unsigned varying;  /* varying:1[PIPE_MAX_ATTRIBS] */
-   unsigned sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[PIPE_MAX_ATTRIBS] */
-};
-
-
-
-
-
-struct brw_context
-{
-   struct pipe_context pipe;
-   struct brw_winsys *winsys;
-
-   unsigned primitive;
-   unsigned reduced_primitive;
-
-   boolean emit_state_always;
-
-   struct {
-      struct brw_state_flags dirty;
-   } state;
-
-
-   struct {
-      const struct pipe_blend_state         *Blend;
-      const struct pipe_depth_stencil_alpha_state *DepthStencil;
-      const struct pipe_poly_stipple        *PolygonStipple;
-      const struct pipe_rasterizer_state    *Raster;
-      const struct pipe_sampler_state       *Samplers[PIPE_MAX_SAMPLERS];
-      const struct brw_vertex_program       *VertexProgram;
-      const struct brw_fragment_program     *FragmentProgram;
-
-      struct pipe_clip_state          Clip;
-      struct pipe_blend_color         BlendColor;
-      struct pipe_scissor_state       Scissor;
-      struct pipe_viewport_state      Viewport;
-      struct pipe_framebuffer_state   FrameBuffer;
-
-      const struct pipe_constant_buffer *Constants[2];
-      const struct brw_texture          *Texture[PIPE_MAX_SAMPLERS];
-   } attribs;
-
-   unsigned num_samplers;
-   unsigned num_textures;
-
-   struct brw_mem_pool pool[BRW_MAX_POOL];
-   struct brw_cache cache[BRW_MAX_CACHE];
-   struct brw_cached_batch_item *cached_batch_items;
-
-   struct {
-
-      /* Arrays with buffer objects to copy non-bufferobj arrays into
-       * for upload:
-       */
-      const struct pipe_vertex_buffer *vbo_array[PIPE_MAX_ATTRIBS];
-
-      struct brw_vertex_element_state inputs[PIPE_MAX_ATTRIBS];
-
-#define BRW_NR_UPLOAD_BUFS 17
-#define BRW_UPLOAD_INIT_SIZE (128*1024)
-
-      /* Summary of size and varying of active arrays, so we can check
-       * for changes to this state:
-       */
-      struct brw_vertex_info info;
-   } vb;
-
-
-   unsigned hardware_dirty;
-   unsigned dirty;
-   unsigned pci_id;
-   /* BRW_NEW_URB_ALLOCATIONS:
-    */
-   struct {
-      unsigned vsize;		/* vertex size plus header in urb registers */
-      unsigned csize;		/* constant buffer size in urb registers */
-      unsigned sfsize;		/* setup data size in urb registers */
-
-      boolean constrained;
-
-      unsigned nr_vs_entries;
-      unsigned nr_gs_entries;
-      unsigned nr_clip_entries;
-      unsigned nr_sf_entries;
-      unsigned nr_cs_entries;
-
-/*       unsigned vs_size; */
-/*       unsigned gs_size; */
-/*       unsigned clip_size; */
-/*       unsigned sf_size; */
-/*       unsigned cs_size; */
-
-      unsigned vs_start;
-      unsigned gs_start;
-      unsigned clip_start;
-      unsigned sf_start;
-      unsigned cs_start;
-   } urb;
-
-
-   /* BRW_NEW_CURBE_OFFSETS:
-    */
-   struct {
-      unsigned wm_start;
-      unsigned wm_size;
-      unsigned clip_start;
-      unsigned clip_size;
-      unsigned vs_start;
-      unsigned vs_size;
-      unsigned total_size;
-
-      unsigned gs_offset;
-
-      float *last_buf;
-      unsigned last_bufsz;
-   } curbe;
-
-   struct {
-      struct brw_vs_prog_data *prog_data;
-
-      unsigned prog_gs_offset;
-      unsigned state_gs_offset;
-   } vs;
-
-   struct {
-      struct brw_gs_prog_data *prog_data;
-
-      boolean prog_active;
-      unsigned prog_gs_offset;
-      unsigned state_gs_offset;
-   } gs;
-
-   struct {
-      struct brw_clip_prog_data *prog_data;
-
-      unsigned prog_gs_offset;
-      unsigned vp_gs_offset;
-      unsigned state_gs_offset;
-   } clip;
-
-
-   struct {
-      struct brw_sf_prog_data *prog_data;
-
-      struct pipe_setup_linkage linkage;
-
-      unsigned prog_gs_offset;
-      unsigned vp_gs_offset;
-      unsigned state_gs_offset;
-   } sf;
-
-   struct {
-      struct brw_wm_prog_data *prog_data;
-
-//      struct brw_wm_compiler *compile_data;
-
-
-      /**
-       * Array of sampler state uploaded at sampler_gs_offset of BRW_SAMPLER
-       * cache
-       */
-      struct brw_sampler_state sampler[BRW_MAX_TEX_UNIT];
-
-      unsigned render_surf;
-      unsigned nr_surfaces;
-
-      unsigned max_threads;
-      struct pipe_buffer *scratch_buffer;
-      unsigned scratch_buffer_size;
-
-      unsigned sampler_count;
-      unsigned sampler_gs_offset;
-
-      struct brw_surface_binding_table bind;
-      unsigned bind_ss_offset;
-
-      unsigned prog_gs_offset;
-      unsigned state_gs_offset;
-   } wm;
-
-
-   struct {
-      unsigned vp_gs_offset;
-      unsigned state_gs_offset;
-   } cc;
-
-
-   /* Used to give every program string a unique id
-    */
-   unsigned program_id;
-};
-
-
-#define BRW_PACKCOLOR8888(r,g,b,a)  ((r<<24) | (g<<16) | (b<<8) | a)
-
-
-/*======================================================================
- * brw_vtbl.c
- */
-void brw_do_flush( struct brw_context *brw,
-		   unsigned flags );
-
-
-/*======================================================================
- * brw_state.c
- */
-void brw_validate_state(struct brw_context *brw);
-void brw_init_state(struct brw_context *brw);
-void brw_destroy_state(struct brw_context *brw);
-
-
-/*======================================================================
- * brw_tex.c
- */
-void brwUpdateTextureState( struct brw_context *brw );
-
-
-/* brw_urb.c
- */
-void brw_upload_urb_fence(struct brw_context *brw);
-
-void brw_upload_constant_buffer_state(struct brw_context *brw);
-
-void brw_init_surface_functions(struct brw_context *brw);
-void brw_init_state_functions(struct brw_context *brw);
-void brw_init_flush_functions(struct brw_context *brw);
-void brw_init_string_functions(struct brw_context *brw);
-
-/*======================================================================
- * Inline conversion functions.  These are better-typed than the
- * macros used previously:
- */
-static inline struct brw_context *
-brw_context( struct pipe_context *ctx )
-{
-   return (struct brw_context *)ctx;
-}
-
-#endif
-
diff --git a/src/gallium/drivers/i965simple/brw_curbe.c b/src/gallium/drivers/i965simple/brw_curbe.c
deleted file mode 100644
index 904cde8e30..0000000000
--- a/src/gallium/drivers/i965simple/brw_curbe.c
+++ /dev/null
@@ -1,369 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_state.h"
-#include "brw_batch.h"
-#include "brw_util.h"
-#include "brw_wm.h"
-#include "pipe/p_state.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-#define FILE_DEBUG_FLAG DEBUG_FALLBACKS
-
-/* Partition the CURBE between the various users of constant values:
- */
-static void calculate_curbe_offsets( struct brw_context *brw )
-{
-   /* CACHE_NEW_WM_PROG */
-   unsigned nr_fp_regs = align(brw->wm.prog_data->max_const, 16);
-
-   /* BRW_NEW_VERTEX_PROGRAM */
-   unsigned nr_vp_regs = align(brw->vs.prog_data->max_const, 16);
-   unsigned nr_clip_regs = 0;
-   unsigned total_regs;
-
-#if 0
-   /* BRW_NEW_CLIP ? */
-   if (brw->attribs.Transform->ClipPlanesEnabled) {
-      unsigned nr_planes = 6 + brw_count_bits(brw->attribs.Transform->ClipPlanesEnabled);
-      nr_clip_regs = align(nr_planes * 4, 16);
-   }
-#endif
-
-
-   total_regs = nr_fp_regs + nr_vp_regs + nr_clip_regs;
-
-   /* This can happen - what to do?  Probably rather than falling
-    * back, the best thing to do is emit programs which code the
-    * constants as immediate values.  Could do this either as a static
-    * cap on WM and VS, or adaptively.
-    *
-    * Unfortunately, this is currently dependent on the results of the
-    * program generation process (in the case of wm), so this would
-    * introduce the need to re-generate programs in the event of a
-    * curbe allocation failure.
-    */
-   /* Max size is 32 - just large enough to
-    * hold the 128 parameters allowed by
-    * the fragment and vertex program
-    * api's.  It's not clear what happens
-    * when both VP and FP want to use 128
-    * parameters, though.
-    */
-   assert(total_regs <= 32);
-
-   /* Lazy resize:
-    */
-   if (nr_fp_regs > brw->curbe.wm_size ||
-       nr_vp_regs > brw->curbe.vs_size ||
-       nr_clip_regs != brw->curbe.clip_size ||
-       (total_regs < brw->curbe.total_size / 4 &&
-	brw->curbe.total_size > 16)) {
-
-      unsigned reg = 0;
-
-      /* Calculate a new layout:
-       */
-      reg = 0;
-      brw->curbe.wm_start = reg;
-      brw->curbe.wm_size = nr_fp_regs; reg += nr_fp_regs;
-      brw->curbe.clip_start = reg;
-      brw->curbe.clip_size = nr_clip_regs; reg += nr_clip_regs;
-      brw->curbe.vs_start = reg;
-      brw->curbe.vs_size = nr_vp_regs; reg += nr_vp_regs;
-      brw->curbe.total_size = reg;
-
-#if 0
-      if (0)
-	 DBG("curbe wm %d+%d clip %d+%d vs %d+%d\n",
-		      brw->curbe.wm_start,
-		      brw->curbe.wm_size,
-		      brw->curbe.clip_start,
-		      brw->curbe.clip_size,
-		      brw->curbe.vs_start,
-		      brw->curbe.vs_size );
-#endif
-
-      brw->state.dirty.brw |= BRW_NEW_CURBE_OFFSETS;
-   }
-}
-
-
-const struct brw_tracked_state brw_curbe_offsets = {
-   .dirty = {
-      .brw  = (BRW_NEW_CLIP |
-	       BRW_NEW_VS),
-      .cache = CACHE_NEW_WM_PROG
-   },
-   .update = calculate_curbe_offsets
-};
-
-
-
-/* Define the number of curbes within CS's urb allocation.  Multiple
- * urb entries -> multiple curbes.  These will be used by
- * fixed-function hardware in a double-buffering scheme to avoid a
- * pipeline stall each time the contents of the curbe is changed.
- */
-void brw_upload_constant_buffer_state(struct brw_context *brw)
-{
-   struct brw_constant_buffer_state cbs;
-   memset(&cbs, 0, sizeof(cbs));
-
-   /* It appears that this is the state packet for the CS unit, ie. the
-    * urb entries detailed here are housed in the CS range from the
-    * URB_FENCE command.
-    */
-   cbs.header.opcode = CMD_CONST_BUFFER_STATE;
-   cbs.header.length = sizeof(cbs)/4 - 2;
-
-   /* BRW_NEW_URB_FENCE */
-   cbs.bits0.nr_urb_entries = brw->urb.nr_cs_entries;
-   cbs.bits0.urb_entry_size = brw->urb.csize - 1;
-
-   assert(brw->urb.nr_cs_entries);
-   BRW_CACHED_BATCH_STRUCT(brw, &cbs);
-}
-
-
-static float fixed_plane[6][4] = {
-   { 0,    0,   -1, 1 },
-   { 0,    0,    1, 1 },
-   { 0,   -1,    0, 1 },
-   { 0,    1,    0, 1 },
-   {-1,    0,    0, 1 },
-   { 1,    0,    0, 1 }
-};
-
-/* Upload a new set of constants.  Too much variability to go into the
- * cache mechanism, but maybe would benefit from a comparison against
- * the current uploaded set of constants.
- */
-static void upload_constant_buffer(struct brw_context *brw)
-{
-   struct brw_mem_pool *pool = &brw->pool[BRW_GS_POOL];
-   unsigned sz = brw->curbe.total_size;
-   unsigned bufsz = sz * sizeof(float);
-   float *buf;
-   unsigned i;
-
-
-   if (sz == 0) {
-      struct brw_constant_buffer cb;
-      cb.header.opcode = CMD_CONST_BUFFER;
-      cb.header.length = sizeof(cb)/4 - 2;
-      cb.header.valid = 0;
-      cb.bits0.buffer_length = 0;
-      cb.bits0.buffer_address = 0;
-      BRW_BATCH_STRUCT(brw, &cb);
-
-      if (brw->curbe.last_buf) {
-	 free(brw->curbe.last_buf);
-	 brw->curbe.last_buf = NULL;
-	 brw->curbe.last_bufsz  = 0;
-      }
-
-      return;
-   }
-
-   buf = (float *)malloc(bufsz);
-
-   memset(buf, 0, bufsz);
-
-   if (brw->curbe.wm_size) {
-      unsigned offset = brw->curbe.wm_start * 16;
-
-      /* First the constant buffer constants:
-       */
-      
-      /* Then any internally generated constants: 
-       */
-      for (i = 0; i < brw->wm.prog_data->nr_internal_consts; i++)
-	 buf[offset + i] = brw->wm.prog_data->internal_const[i];
-
-      assert(brw->wm.prog_data->max_const == 
-	     brw->wm.prog_data->nr_internal_consts);
-   }
-
-
-   /* The clipplanes are actually delivered to both CLIP and VS units.
-    * VS uses them to calculate the outcode bitmasks.
-    */
-   if (brw->curbe.clip_size) {
-      unsigned offset = brw->curbe.clip_start * 16;
-      unsigned j;
-
-      /* If any planes are going this way, send them all this way:
-       */
-      for (i = 0; i < 6; i++) {
-	 buf[offset + i * 4 + 0] = fixed_plane[i][0];
-	 buf[offset + i * 4 + 1] = fixed_plane[i][1];
-	 buf[offset + i * 4 + 2] = fixed_plane[i][2];
-	 buf[offset + i * 4 + 3] = fixed_plane[i][3];
-      }
-
-      /* Clip planes: BRW_NEW_CLIP:
-       */
-      for (j = 0; j < brw->attribs.Clip.nr; j++) {
-	 buf[offset + i * 4 + 0] = brw->attribs.Clip.ucp[j][0];
-	 buf[offset + i * 4 + 1] = brw->attribs.Clip.ucp[j][1];
-	 buf[offset + i * 4 + 2] = brw->attribs.Clip.ucp[j][2];
-	 buf[offset + i * 4 + 3] = brw->attribs.Clip.ucp[j][3];
-	 i++;
-      }
-   }
-
-
-   if (brw->curbe.vs_size) {
-      unsigned offset = brw->curbe.vs_start * 16;
-      /*unsigned nr = vp->max_const;*/
-      const struct pipe_constant_buffer *cbuffer = brw->attribs.Constants[0];
-      struct pipe_winsys *ws = brw->pipe.winsys;
-      /* FIXME: buffer size is num_consts + num_immediates */
-      if (brw->vs.prog_data->num_consts) {
-         /* map the vertex constant buffer and copy to curbe: */
-         void *data = ws->buffer_map(ws, cbuffer->buffer, 0);
-         /* FIXME: this is wrong. the cbuffer->buffer->size currently
-          * represents size of consts + immediates. so if we'll
-          * have both we'll copy over the end of the buffer
-          * with the subsequent memcpy */
-         memcpy(&buf[offset], data, cbuffer->buffer->size);
-         ws->buffer_unmap(ws, cbuffer->buffer);
-         offset += cbuffer->buffer->size;
-      }
-      /*immediates*/
-      if (brw->vs.prog_data->num_imm) {
-         memcpy(&buf[offset], brw->vs.prog_data->imm_buf,
-                brw->vs.prog_data->num_imm * 4 * sizeof(float));
-      }
-   }
-
-   if (1) {
-      for (i = 0; i < sz; i+=4)
-	 debug_printf("curbe %d.%d: %f %f %f %f\n", i/8, i&4,
-		      buf[i+0], buf[i+1], buf[i+2], buf[i+3]);
-
-      debug_printf("last_buf %p buf %p sz %d/%d cmp %d\n",
-		   brw->curbe.last_buf, buf,
-		   bufsz, brw->curbe.last_bufsz,
-		   brw->curbe.last_buf ? memcmp(buf, brw->curbe.last_buf, bufsz) : -1);
-   }
-
-   if (brw->curbe.last_buf &&
-       bufsz == brw->curbe.last_bufsz &&
-       memcmp(buf, brw->curbe.last_buf, bufsz) == 0) {
-      free(buf);
-/*       return; */
-   }
-   else {
-      if (brw->curbe.last_buf)
-	 free(brw->curbe.last_buf);
-      brw->curbe.last_buf = buf;
-      brw->curbe.last_bufsz = bufsz;
-
-
-      if (!brw_pool_alloc(pool,
-			  bufsz,
-			  1 << 6,
-			  &brw->curbe.gs_offset)) {
-	 debug_printf("out of GS memory for curbe\n");
-	 assert(0);
-	 return;
-      }
-
-
-      /* Copy data to the buffer:
-       */
-      brw->winsys->buffer_subdata_typed(brw->winsys,
-					pool->buffer, 
-					brw->curbe.gs_offset, 
-					bufsz, 
-					buf,
-					BRW_CONSTANT_BUFFER );
-   }
-
-   /* TODO: only emit the constant_buffer packet when necessary, ie:
-      - contents have changed
-      - offset has changed
-      - hw requirements due to other packets emitted.
-   */
-   {
-      struct brw_constant_buffer cb;
-
-      memset(&cb, 0, sizeof(cb));
-
-      cb.header.opcode = CMD_CONST_BUFFER;
-      cb.header.length = sizeof(cb)/4 - 2;
-      cb.header.valid = 1;
-      cb.bits0.buffer_length = sz - 1;
-      cb.bits0.buffer_address = brw->curbe.gs_offset >> 6;
-
-      /* Because this provokes an action (ie copy the constants into the
-       * URB), it shouldn't be shortcircuited if identical to the
-       * previous time - because eg. the urb destination may have
-       * changed, or the urb contents different to last time.
-       *
-       * Note that the data referred to is actually copied internally,
-       * not just used in place according to passed pointer.
-       *
-       * It appears that the CS unit takes care of using each available
-       * URB entry (Const URB Entry == CURBE) in turn, and issuing
-       * flushes as necessary when doublebuffering of CURBEs isn't
-       * possible.
-       */
-      BRW_BATCH_STRUCT(brw, &cb);
-   }
-}
-
-/* This tracked state is unique in that the state it monitors varies
- * dynamically depending on the parameters tracked by the fragment and
- * vertex programs.  This is the template used as a starting point,
- * each context will maintain a copy of this internally and update as
- * required.
- */
-const struct brw_tracked_state brw_constant_buffer = {
-   .dirty = {
-      .brw  = (BRW_NEW_CLIP |
-	       BRW_NEW_CONSTANTS |
-	       BRW_NEW_URB_FENCE | /* Implicit - hardware requires this, not used above */
-	       BRW_NEW_PSP | /* Implicit - hardware requires this, not used above */
-	       BRW_NEW_CURBE_OFFSETS),
-      .cache = (CACHE_NEW_WM_PROG)
-   },
-   .update = upload_constant_buffer
-};
-
diff --git a/src/gallium/drivers/i965simple/brw_defines.h b/src/gallium/drivers/i965simple/brw_defines.h
deleted file mode 100644
index 715d2d2d01..0000000000
--- a/src/gallium/drivers/i965simple/brw_defines.h
+++ /dev/null
@@ -1,870 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#ifndef BRW_DEFINES_H
-#define BRW_DEFINES_H
-
-/*
- */
-#define MI_NOOP                              0x00
-#define MI_USER_INTERRUPT                    0x02
-#define MI_WAIT_FOR_EVENT                    0x03
-#define MI_FLUSH                             0x04
-#define MI_REPORT_HEAD                       0x07
-#define MI_ARB_ON_OFF                        0x08
-#define MI_BATCH_BUFFER_END                  0x0A
-#define MI_OVERLAY_FLIP                      0x11
-#define MI_LOAD_SCAN_LINES_INCL              0x12
-#define MI_LOAD_SCAN_LINES_EXCL              0x13
-#define MI_DISPLAY_BUFFER_INFO               0x14
-#define MI_SET_CONTEXT                       0x18
-#define MI_STORE_DATA_IMM                    0x20
-#define MI_STORE_DATA_INDEX                  0x21
-#define MI_LOAD_REGISTER_IMM                 0x22
-#define MI_STORE_REGISTER_MEM                0x24
-#define MI_BATCH_BUFFER_START                0x31
-
-#define MI_SYNCHRONOUS_FLIP                  0x0
-#define MI_ASYNCHRONOUS_FLIP                 0x1
-
-#define MI_BUFFER_SECURE                     0x0
-#define MI_BUFFER_NONSECURE                  0x1
-
-#define MI_ARBITRATE_AT_CHAIN_POINTS         0x0
-#define MI_ARBITRATE_BETWEEN_INSTS           0x1
-#define MI_NO_ARBITRATION                    0x3
-
-#define MI_CONDITION_CODE_WAIT_DISABLED      0x0
-#define MI_CONDITION_CODE_WAIT_0             0x1
-#define MI_CONDITION_CODE_WAIT_1             0x2
-#define MI_CONDITION_CODE_WAIT_2             0x3
-#define MI_CONDITION_CODE_WAIT_3             0x4
-#define MI_CONDITION_CODE_WAIT_4             0x5
-
-#define MI_DISPLAY_PIPE_A                    0x0
-#define MI_DISPLAY_PIPE_B                    0x1
-
-#define MI_DISPLAY_PLANE_A                   0x0
-#define MI_DISPLAY_PLANE_B                   0x1
-#define MI_DISPLAY_PLANE_C                   0x2
-
-#define MI_STANDARD_FLIP                                 0x0
-#define MI_ENQUEUE_FLIP_PERFORM_BASE_FRAME_NUMBER_LOAD   0x1
-#define MI_ENQUEUE_FLIP_TARGET_FRAME_NUMBER_RELATIVE     0x2
-#define MI_ENQUEUE_FLIP_ABSOLUTE_TARGET_FRAME_NUMBER     0x3
-
-#define MI_PHYSICAL_ADDRESS                  0x0
-#define MI_VIRTUAL_ADDRESS                   0x1
-
-#define MI_BUFFER_MEMORY_MAIN                0x0
-#define MI_BUFFER_MEMORY_GTT                 0x2
-#define MI_BUFFER_MEMORY_PER_PROCESS_GTT     0x3
-
-#define MI_FLIP_CONTINUE                     0x0
-#define MI_FLIP_ON                           0x1
-#define MI_FLIP_OFF                          0x2
-
-#define MI_UNTRUSTED_REGISTER_SPACE          0x0
-#define MI_TRUSTED_REGISTER_SPACE            0x1
-
-/* 3D state:
- */
-#define _3DOP_3DSTATE_PIPELINED       0x0
-#define _3DOP_3DSTATE_NONPIPELINED    0x1
-#define _3DOP_3DCONTROL               0x2
-#define _3DOP_3DPRIMITIVE             0x3
-
-#define _3DSTATE_PIPELINED_POINTERS       0x00
-#define _3DSTATE_BINDING_TABLE_POINTERS   0x01
-#define _3DSTATE_VERTEX_BUFFERS           0x08
-#define _3DSTATE_VERTEX_ELEMENTS          0x09
-#define _3DSTATE_INDEX_BUFFER             0x0A
-#define _3DSTATE_VF_STATISTICS            0x0B
-#define _3DSTATE_DRAWING_RECTANGLE            0x00
-#define _3DSTATE_CONSTANT_COLOR               0x01
-#define _3DSTATE_SAMPLER_PALETTE_LOAD         0x02
-#define _3DSTATE_CHROMA_KEY                   0x04
-#define _3DSTATE_DEPTH_BUFFER                 0x05
-#define _3DSTATE_POLY_STIPPLE_OFFSET          0x06
-#define _3DSTATE_POLY_STIPPLE_PATTERN         0x07
-#define _3DSTATE_LINE_STIPPLE                 0x08
-#define _3DSTATE_GLOBAL_DEPTH_OFFSET_CLAMP    0x09
-#define _3DCONTROL    0x00
-#define _3DPRIMITIVE  0x00
-
-#define PIPE_CONTROL_NOWRITE          0x00
-#define PIPE_CONTROL_WRITEIMMEDIATE   0x01
-#define PIPE_CONTROL_WRITEDEPTH       0x02
-#define PIPE_CONTROL_WRITETIMESTAMP   0x03
-
-#define PIPE_CONTROL_GTTWRITE_PROCESS_LOCAL 0x00
-#define PIPE_CONTROL_GTTWRITE_GLOBAL        0x01
-
-#define _3DPRIM_POINTLIST         0x01
-#define _3DPRIM_LINELIST          0x02
-#define _3DPRIM_LINESTRIP         0x03
-#define _3DPRIM_TRILIST           0x04
-#define _3DPRIM_TRISTRIP          0x05
-#define _3DPRIM_TRIFAN            0x06
-#define _3DPRIM_QUADLIST          0x07
-#define _3DPRIM_QUADSTRIP         0x08
-#define _3DPRIM_LINELIST_ADJ      0x09
-#define _3DPRIM_LINESTRIP_ADJ     0x0A
-#define _3DPRIM_TRILIST_ADJ       0x0B
-#define _3DPRIM_TRISTRIP_ADJ      0x0C
-#define _3DPRIM_TRISTRIP_REVERSE  0x0D
-#define _3DPRIM_POLYGON           0x0E
-#define _3DPRIM_RECTLIST          0x0F
-#define _3DPRIM_LINELOOP          0x10
-#define _3DPRIM_POINTLIST_BF      0x11
-#define _3DPRIM_LINESTRIP_CONT    0x12
-#define _3DPRIM_LINESTRIP_BF      0x13
-#define _3DPRIM_LINESTRIP_CONT_BF 0x14
-#define _3DPRIM_TRIFAN_NOSTIPPLE  0x15
-
-#define _3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL 0
-#define _3DPRIM_VERTEXBUFFER_ACCESS_RANDOM     1
-
-#define BRW_ANISORATIO_2     0
-#define BRW_ANISORATIO_4     1
-#define BRW_ANISORATIO_6     2
-#define BRW_ANISORATIO_8     3
-#define BRW_ANISORATIO_10    4
-#define BRW_ANISORATIO_12    5
-#define BRW_ANISORATIO_14    6
-#define BRW_ANISORATIO_16    7
-
-#define BRW_BLENDFACTOR_ONE                 0x1
-#define BRW_BLENDFACTOR_SRC_COLOR           0x2
-#define BRW_BLENDFACTOR_SRC_ALPHA           0x3
-#define BRW_BLENDFACTOR_DST_ALPHA           0x4
-#define BRW_BLENDFACTOR_DST_COLOR           0x5
-#define BRW_BLENDFACTOR_SRC_ALPHA_SATURATE  0x6
-#define BRW_BLENDFACTOR_CONST_COLOR         0x7
-#define BRW_BLENDFACTOR_CONST_ALPHA         0x8
-#define BRW_BLENDFACTOR_SRC1_COLOR          0x9
-#define BRW_BLENDFACTOR_SRC1_ALPHA          0x0A
-#define BRW_BLENDFACTOR_ZERO                0x11
-#define BRW_BLENDFACTOR_INV_SRC_COLOR       0x12
-#define BRW_BLENDFACTOR_INV_SRC_ALPHA       0x13
-#define BRW_BLENDFACTOR_INV_DST_ALPHA       0x14
-#define BRW_BLENDFACTOR_INV_DST_COLOR       0x15
-#define BRW_BLENDFACTOR_INV_CONST_COLOR     0x17
-#define BRW_BLENDFACTOR_INV_CONST_ALPHA     0x18
-#define BRW_BLENDFACTOR_INV_SRC1_COLOR      0x19
-#define BRW_BLENDFACTOR_INV_SRC1_ALPHA      0x1A
-
-#define BRW_BLENDFUNCTION_ADD               0
-#define BRW_BLENDFUNCTION_SUBTRACT          1
-#define BRW_BLENDFUNCTION_REVERSE_SUBTRACT  2
-#define BRW_BLENDFUNCTION_MIN               3
-#define BRW_BLENDFUNCTION_MAX               4
-
-#define BRW_ALPHATEST_FORMAT_UNORM8         0
-#define BRW_ALPHATEST_FORMAT_FLOAT32        1
-
-#define BRW_CHROMAKEY_KILL_ON_ANY_MATCH  0
-#define BRW_CHROMAKEY_REPLACE_BLACK      1
-
-#define BRW_CLIP_API_OGL     0
-#define BRW_CLIP_API_DX      1
-
-#define BRW_CLIPMODE_NORMAL              0
-#define BRW_CLIPMODE_CLIP_ALL            1
-#define BRW_CLIPMODE_CLIP_NON_REJECTED   2
-#define BRW_CLIPMODE_REJECT_ALL          3
-#define BRW_CLIPMODE_ACCEPT_ALL          4
-
-#define BRW_CLIP_NDCSPACE     0
-#define BRW_CLIP_SCREENSPACE  1
-
-#define BRW_COMPAREFUNCTION_ALWAYS       0
-#define BRW_COMPAREFUNCTION_NEVER        1
-#define BRW_COMPAREFUNCTION_LESS         2
-#define BRW_COMPAREFUNCTION_EQUAL        3
-#define BRW_COMPAREFUNCTION_LEQUAL       4
-#define BRW_COMPAREFUNCTION_GREATER      5
-#define BRW_COMPAREFUNCTION_NOTEQUAL     6
-#define BRW_COMPAREFUNCTION_GEQUAL       7
-
-#define BRW_COVERAGE_PIXELS_HALF     0
-#define BRW_COVERAGE_PIXELS_1        1
-#define BRW_COVERAGE_PIXELS_2        2
-#define BRW_COVERAGE_PIXELS_4        3
-
-#define BRW_CULLMODE_BOTH        0
-#define BRW_CULLMODE_NONE        1
-#define BRW_CULLMODE_FRONT       2
-#define BRW_CULLMODE_BACK        3
-
-#define BRW_DEFAULTCOLOR_R8G8B8A8_UNORM      0
-#define BRW_DEFAULTCOLOR_R32G32B32A32_FLOAT  1
-
-#define BRW_DEPTHFORMAT_D32_FLOAT_S8X24_UINT     0
-#define BRW_DEPTHFORMAT_D32_FLOAT                1
-#define BRW_DEPTHFORMAT_D24_UNORM_S8_UINT        2
-#define BRW_DEPTHFORMAT_D16_UNORM                5
-
-#define BRW_FLOATING_POINT_IEEE_754        0
-#define BRW_FLOATING_POINT_NON_IEEE_754    1
-
-#define BRW_FRONTWINDING_CW      0
-#define BRW_FRONTWINDING_CCW     1
-
-#define BRW_SPRITE_POINT_ENABLE  16
-
-#define BRW_INDEX_BYTE     0
-#define BRW_INDEX_WORD     1
-#define BRW_INDEX_DWORD    2
-
-#define BRW_LOGICOPFUNCTION_CLEAR            0
-#define BRW_LOGICOPFUNCTION_NOR              1
-#define BRW_LOGICOPFUNCTION_AND_INVERTED     2
-#define BRW_LOGICOPFUNCTION_COPY_INVERTED    3
-#define BRW_LOGICOPFUNCTION_AND_REVERSE      4
-#define BRW_LOGICOPFUNCTION_INVERT           5
-#define BRW_LOGICOPFUNCTION_XOR              6
-#define BRW_LOGICOPFUNCTION_NAND             7
-#define BRW_LOGICOPFUNCTION_AND              8
-#define BRW_LOGICOPFUNCTION_EQUIV            9
-#define BRW_LOGICOPFUNCTION_NOOP             10
-#define BRW_LOGICOPFUNCTION_OR_INVERTED      11
-#define BRW_LOGICOPFUNCTION_COPY             12
-#define BRW_LOGICOPFUNCTION_OR_REVERSE       13
-#define BRW_LOGICOPFUNCTION_OR               14
-#define BRW_LOGICOPFUNCTION_SET              15
-
-#define BRW_MAPFILTER_NEAREST        0x0
-#define BRW_MAPFILTER_LINEAR         0x1
-#define BRW_MAPFILTER_ANISOTROPIC    0x2
-
-#define BRW_MIPFILTER_NONE        0
-#define BRW_MIPFILTER_NEAREST     1
-#define BRW_MIPFILTER_LINEAR      3
-
-#define BRW_POLYGON_FRONT_FACING     0
-#define BRW_POLYGON_BACK_FACING      1
-
-#define BRW_PREFILTER_ALWAYS     0x0
-#define BRW_PREFILTER_NEVER      0x1
-#define BRW_PREFILTER_LESS       0x2
-#define BRW_PREFILTER_EQUAL      0x3
-#define BRW_PREFILTER_LEQUAL     0x4
-#define BRW_PREFILTER_GREATER    0x5
-#define BRW_PREFILTER_NOTEQUAL   0x6
-#define BRW_PREFILTER_GEQUAL     0x7
-
-#define BRW_PROVOKING_VERTEX_0    0
-#define BRW_PROVOKING_VERTEX_1    1
-#define BRW_PROVOKING_VERTEX_2    2
-
-#define BRW_RASTRULE_UPPER_LEFT  0
-#define BRW_RASTRULE_UPPER_RIGHT 1
-/* These are listed as "Reserved, but not seen as useful"
- * in Intel documentation (page 212, "Point Rasterization Rule",
- * section 7.4 "SF Pipeline State Summary", of document
- * "Intel® 965 Express Chipset Family and Intel® G35 Express
- * Chipset Graphics Controller Programmer's Reference Manual,
- * Volume 2: 3D/Media", Revision 1.0b as of January 2008,
- * available at 
- *     http://intellinuxgraphics.org/documentation.html
- * at the time of this writing).
- *
- * These appear to be supported on at least some
- * i965-family devices, and the BRW_RASTRULE_LOWER_RIGHT
- * is useful when using OpenGL to render to a FBO
- * (which has the pixel coordinate Y orientation inverted
- * with respect to the normal OpenGL pixel coordinate system).
- */
-#define BRW_RASTRULE_LOWER_LEFT  2
-#define BRW_RASTRULE_LOWER_RIGHT 3
-
-#define BRW_RENDERTARGET_CLAMPRANGE_UNORM    0
-#define BRW_RENDERTARGET_CLAMPRANGE_SNORM    1
-#define BRW_RENDERTARGET_CLAMPRANGE_FORMAT   2
-
-#define BRW_STENCILOP_KEEP               0
-#define BRW_STENCILOP_ZERO               1
-#define BRW_STENCILOP_REPLACE            2
-#define BRW_STENCILOP_INCRSAT            3
-#define BRW_STENCILOP_DECRSAT            4
-#define BRW_STENCILOP_INCR               5
-#define BRW_STENCILOP_DECR               6
-#define BRW_STENCILOP_INVERT             7
-
-#define BRW_SURFACE_MIPMAPLAYOUT_BELOW   0
-#define BRW_SURFACE_MIPMAPLAYOUT_RIGHT   1
-
-#define BRW_SURFACEFORMAT_R32G32B32A32_FLOAT             0x000
-#define BRW_SURFACEFORMAT_R32G32B32A32_SINT              0x001
-#define BRW_SURFACEFORMAT_R32G32B32A32_UINT              0x002
-#define BRW_SURFACEFORMAT_R32G32B32A32_UNORM             0x003
-#define BRW_SURFACEFORMAT_R32G32B32A32_SNORM             0x004
-#define BRW_SURFACEFORMAT_R64G64_FLOAT                   0x005
-#define BRW_SURFACEFORMAT_R32G32B32X32_FLOAT             0x006
-#define BRW_SURFACEFORMAT_R32G32B32A32_SSCALED           0x007
-#define BRW_SURFACEFORMAT_R32G32B32A32_USCALED           0x008
-#define BRW_SURFACEFORMAT_R32G32B32_FLOAT                0x040
-#define BRW_SURFACEFORMAT_R32G32B32_SINT                 0x041
-#define BRW_SURFACEFORMAT_R32G32B32_UINT                 0x042
-#define BRW_SURFACEFORMAT_R32G32B32_UNORM                0x043
-#define BRW_SURFACEFORMAT_R32G32B32_SNORM                0x044
-#define BRW_SURFACEFORMAT_R32G32B32_SSCALED              0x045
-#define BRW_SURFACEFORMAT_R32G32B32_USCALED              0x046
-#define BRW_SURFACEFORMAT_R16G16B16A16_UNORM             0x080
-#define BRW_SURFACEFORMAT_R16G16B16A16_SNORM             0x081
-#define BRW_SURFACEFORMAT_R16G16B16A16_SINT              0x082
-#define BRW_SURFACEFORMAT_R16G16B16A16_UINT              0x083
-#define BRW_SURFACEFORMAT_R16G16B16A16_FLOAT             0x084
-#define BRW_SURFACEFORMAT_R32G32_FLOAT                   0x085
-#define BRW_SURFACEFORMAT_R32G32_SINT                    0x086
-#define BRW_SURFACEFORMAT_R32G32_UINT                    0x087
-#define BRW_SURFACEFORMAT_R32_FLOAT_X8X24_TYPELESS       0x088
-#define BRW_SURFACEFORMAT_X32_TYPELESS_G8X24_UINT        0x089
-#define BRW_SURFACEFORMAT_L32A32_FLOAT                   0x08A
-#define BRW_SURFACEFORMAT_R32G32_UNORM                   0x08B
-#define BRW_SURFACEFORMAT_R32G32_SNORM                   0x08C
-#define BRW_SURFACEFORMAT_R64_FLOAT                      0x08D
-#define BRW_SURFACEFORMAT_R16G16B16X16_UNORM             0x08E
-#define BRW_SURFACEFORMAT_R16G16B16X16_FLOAT             0x08F
-#define BRW_SURFACEFORMAT_A32X32_FLOAT                   0x090
-#define BRW_SURFACEFORMAT_L32X32_FLOAT                   0x091
-#define BRW_SURFACEFORMAT_I32X32_FLOAT                   0x092
-#define BRW_SURFACEFORMAT_R16G16B16A16_SSCALED           0x093
-#define BRW_SURFACEFORMAT_R16G16B16A16_USCALED           0x094
-#define BRW_SURFACEFORMAT_R32G32_SSCALED                 0x095
-#define BRW_SURFACEFORMAT_R32G32_USCALED                 0x096
-#define BRW_SURFACEFORMAT_B8G8R8A8_UNORM                 0x0C0
-#define BRW_SURFACEFORMAT_B8G8R8A8_UNORM_SRGB            0x0C1
-#define BRW_SURFACEFORMAT_R10G10B10A2_UNORM              0x0C2
-#define BRW_SURFACEFORMAT_R10G10B10A2_UNORM_SRGB         0x0C3
-#define BRW_SURFACEFORMAT_R10G10B10A2_UINT               0x0C4
-#define BRW_SURFACEFORMAT_R10G10B10_SNORM_A2_UNORM       0x0C5
-#define BRW_SURFACEFORMAT_R8G8B8A8_UNORM                 0x0C7
-#define BRW_SURFACEFORMAT_R8G8B8A8_UNORM_SRGB            0x0C8
-#define BRW_SURFACEFORMAT_R8G8B8A8_SNORM                 0x0C9
-#define BRW_SURFACEFORMAT_R8G8B8A8_SINT                  0x0CA
-#define BRW_SURFACEFORMAT_R8G8B8A8_UINT                  0x0CB
-#define BRW_SURFACEFORMAT_R16G16_UNORM                   0x0CC
-#define BRW_SURFACEFORMAT_R16G16_SNORM                   0x0CD
-#define BRW_SURFACEFORMAT_R16G16_SINT                    0x0CE
-#define BRW_SURFACEFORMAT_R16G16_UINT                    0x0CF
-#define BRW_SURFACEFORMAT_R16G16_FLOAT                   0x0D0
-#define BRW_SURFACEFORMAT_B10G10R10A2_UNORM              0x0D1
-#define BRW_SURFACEFORMAT_B10G10R10A2_UNORM_SRGB         0x0D2
-#define BRW_SURFACEFORMAT_R11G11B10_FLOAT                0x0D3
-#define BRW_SURFACEFORMAT_R32_SINT                       0x0D6
-#define BRW_SURFACEFORMAT_R32_UINT                       0x0D7
-#define BRW_SURFACEFORMAT_R32_FLOAT                      0x0D8
-#define BRW_SURFACEFORMAT_R24_UNORM_X8_TYPELESS          0x0D9
-#define BRW_SURFACEFORMAT_X24_TYPELESS_G8_UINT           0x0DA
-#define BRW_SURFACEFORMAT_L16A16_UNORM                   0x0DF
-#define BRW_SURFACEFORMAT_I24X8_UNORM                    0x0E0
-#define BRW_SURFACEFORMAT_L24X8_UNORM                    0x0E1
-#define BRW_SURFACEFORMAT_A24X8_UNORM                    0x0E2
-#define BRW_SURFACEFORMAT_I32_FLOAT                      0x0E3
-#define BRW_SURFACEFORMAT_L32_FLOAT                      0x0E4
-#define BRW_SURFACEFORMAT_A32_FLOAT                      0x0E5
-#define BRW_SURFACEFORMAT_B8G8R8X8_UNORM                 0x0E9
-#define BRW_SURFACEFORMAT_B8G8R8X8_UNORM_SRGB            0x0EA
-#define BRW_SURFACEFORMAT_R8G8B8X8_UNORM                 0x0EB
-#define BRW_SURFACEFORMAT_R8G8B8X8_UNORM_SRGB            0x0EC
-#define BRW_SURFACEFORMAT_R9G9B9E5_SHAREDEXP             0x0ED
-#define BRW_SURFACEFORMAT_B10G10R10X2_UNORM              0x0EE
-#define BRW_SURFACEFORMAT_L16A16_FLOAT                   0x0F0
-#define BRW_SURFACEFORMAT_R32_UNORM                      0x0F1
-#define BRW_SURFACEFORMAT_R32_SNORM                      0x0F2
-#define BRW_SURFACEFORMAT_R10G10B10X2_USCALED            0x0F3
-#define BRW_SURFACEFORMAT_R8G8B8A8_SSCALED               0x0F4
-#define BRW_SURFACEFORMAT_R8G8B8A8_USCALED               0x0F5
-#define BRW_SURFACEFORMAT_R16G16_SSCALED                 0x0F6
-#define BRW_SURFACEFORMAT_R16G16_USCALED                 0x0F7
-#define BRW_SURFACEFORMAT_R32_SSCALED                    0x0F8
-#define BRW_SURFACEFORMAT_R32_USCALED                    0x0F9
-#define BRW_SURFACEFORMAT_B5G6R5_UNORM                   0x100
-#define BRW_SURFACEFORMAT_B5G6R5_UNORM_SRGB              0x101
-#define BRW_SURFACEFORMAT_B5G5R5A1_UNORM                 0x102
-#define BRW_SURFACEFORMAT_B5G5R5A1_UNORM_SRGB            0x103
-#define BRW_SURFACEFORMAT_B4G4R4A4_UNORM                 0x104
-#define BRW_SURFACEFORMAT_B4G4R4A4_UNORM_SRGB            0x105
-#define BRW_SURFACEFORMAT_R8G8_UNORM                     0x106
-#define BRW_SURFACEFORMAT_R8G8_SNORM                     0x107
-#define BRW_SURFACEFORMAT_R8G8_SINT                      0x108
-#define BRW_SURFACEFORMAT_R8G8_UINT                      0x109
-#define BRW_SURFACEFORMAT_R16_UNORM                      0x10A
-#define BRW_SURFACEFORMAT_R16_SNORM                      0x10B
-#define BRW_SURFACEFORMAT_R16_SINT                       0x10C
-#define BRW_SURFACEFORMAT_R16_UINT                       0x10D
-#define BRW_SURFACEFORMAT_R16_FLOAT                      0x10E
-#define BRW_SURFACEFORMAT_I16_UNORM                      0x111
-#define BRW_SURFACEFORMAT_L16_UNORM                      0x112
-#define BRW_SURFACEFORMAT_A16_UNORM                      0x113
-#define BRW_SURFACEFORMAT_L8A8_UNORM                     0x114
-#define BRW_SURFACEFORMAT_I16_FLOAT                      0x115
-#define BRW_SURFACEFORMAT_L16_FLOAT                      0x116
-#define BRW_SURFACEFORMAT_A16_FLOAT                      0x117
-#define BRW_SURFACEFORMAT_R5G5_SNORM_B6_UNORM            0x119
-#define BRW_SURFACEFORMAT_B5G5R5X1_UNORM                 0x11A
-#define BRW_SURFACEFORMAT_B5G5R5X1_UNORM_SRGB            0x11B
-#define BRW_SURFACEFORMAT_R8G8_SSCALED                   0x11C
-#define BRW_SURFACEFORMAT_R8G8_USCALED                   0x11D
-#define BRW_SURFACEFORMAT_R16_SSCALED                    0x11E
-#define BRW_SURFACEFORMAT_R16_USCALED                    0x11F
-#define BRW_SURFACEFORMAT_R8_UNORM                       0x140
-#define BRW_SURFACEFORMAT_R8_SNORM                       0x141
-#define BRW_SURFACEFORMAT_R8_SINT                        0x142
-#define BRW_SURFACEFORMAT_R8_UINT                        0x143
-#define BRW_SURFACEFORMAT_A8_UNORM                       0x144
-#define BRW_SURFACEFORMAT_I8_UNORM                       0x145
-#define BRW_SURFACEFORMAT_L8_UNORM                       0x146
-#define BRW_SURFACEFORMAT_P4A4_UNORM                     0x147
-#define BRW_SURFACEFORMAT_A4P4_UNORM                     0x148
-#define BRW_SURFACEFORMAT_R8_SSCALED                     0x149
-#define BRW_SURFACEFORMAT_R8_USCALED                     0x14A
-#define BRW_SURFACEFORMAT_R1_UINT                        0x181
-#define BRW_SURFACEFORMAT_YCRCB_NORMAL                   0x182
-#define BRW_SURFACEFORMAT_YCRCB_SWAPUVY                  0x183
-#define BRW_SURFACEFORMAT_BC1_UNORM                      0x186
-#define BRW_SURFACEFORMAT_BC2_UNORM                      0x187
-#define BRW_SURFACEFORMAT_BC3_UNORM                      0x188
-#define BRW_SURFACEFORMAT_BC4_UNORM                      0x189
-#define BRW_SURFACEFORMAT_BC5_UNORM                      0x18A
-#define BRW_SURFACEFORMAT_BC1_UNORM_SRGB                 0x18B
-#define BRW_SURFACEFORMAT_BC2_UNORM_SRGB                 0x18C
-#define BRW_SURFACEFORMAT_BC3_UNORM_SRGB                 0x18D
-#define BRW_SURFACEFORMAT_MONO8                          0x18E
-#define BRW_SURFACEFORMAT_YCRCB_SWAPUV                   0x18F
-#define BRW_SURFACEFORMAT_YCRCB_SWAPY                    0x190
-#define BRW_SURFACEFORMAT_DXT1_RGB                       0x191
-#define BRW_SURFACEFORMAT_FXT1                           0x192
-#define BRW_SURFACEFORMAT_R8G8B8_UNORM                   0x193
-#define BRW_SURFACEFORMAT_R8G8B8_SNORM                   0x194
-#define BRW_SURFACEFORMAT_R8G8B8_SSCALED                 0x195
-#define BRW_SURFACEFORMAT_R8G8B8_USCALED                 0x196
-#define BRW_SURFACEFORMAT_R64G64B64A64_FLOAT             0x197
-#define BRW_SURFACEFORMAT_R64G64B64_FLOAT                0x198
-#define BRW_SURFACEFORMAT_BC4_SNORM                      0x199
-#define BRW_SURFACEFORMAT_BC5_SNORM                      0x19A
-#define BRW_SURFACEFORMAT_R16G16B16_UNORM                0x19C
-#define BRW_SURFACEFORMAT_R16G16B16_SNORM                0x19D
-#define BRW_SURFACEFORMAT_R16G16B16_SSCALED              0x19E
-#define BRW_SURFACEFORMAT_R16G16B16_USCALED              0x19F
-
-#define BRW_SURFACERETURNFORMAT_FLOAT32  0
-#define BRW_SURFACERETURNFORMAT_S1       1
-
-#define BRW_SURFACE_1D      0
-#define BRW_SURFACE_2D      1
-#define BRW_SURFACE_3D      2
-#define BRW_SURFACE_CUBE    3
-#define BRW_SURFACE_BUFFER  4
-#define BRW_SURFACE_NULL    7
-
-#define BRW_TEXCOORDMODE_WRAP            0
-#define BRW_TEXCOORDMODE_MIRROR          1
-#define BRW_TEXCOORDMODE_CLAMP           2
-#define BRW_TEXCOORDMODE_CUBE            3
-#define BRW_TEXCOORDMODE_CLAMP_BORDER    4
-#define BRW_TEXCOORDMODE_MIRROR_ONCE     5
-
-#define BRW_THREAD_PRIORITY_NORMAL   0
-#define BRW_THREAD_PRIORITY_HIGH     1
-
-#define BRW_TILEWALK_XMAJOR                 0
-#define BRW_TILEWALK_YMAJOR                 1
-
-#define BRW_VERTEX_SUBPIXEL_PRECISION_8BITS  0
-#define BRW_VERTEX_SUBPIXEL_PRECISION_4BITS  1
-
-#define BRW_VERTEXBUFFER_ACCESS_VERTEXDATA     0
-#define BRW_VERTEXBUFFER_ACCESS_INSTANCEDATA   1
-
-#define BRW_VFCOMPONENT_NOSTORE      0
-#define BRW_VFCOMPONENT_STORE_SRC    1
-#define BRW_VFCOMPONENT_STORE_0      2
-#define BRW_VFCOMPONENT_STORE_1_FLT  3
-#define BRW_VFCOMPONENT_STORE_1_INT  4
-#define BRW_VFCOMPONENT_STORE_VID    5
-#define BRW_VFCOMPONENT_STORE_IID    6
-#define BRW_VFCOMPONENT_STORE_PID    7
-
-
-
-/* Execution Unit (EU) defines
- */
-
-#define BRW_ALIGN_1   0
-#define BRW_ALIGN_16  1
-
-#define BRW_ADDRESS_DIRECT                        0
-#define BRW_ADDRESS_REGISTER_INDIRECT_REGISTER    1
-
-#define BRW_CHANNEL_X     0
-#define BRW_CHANNEL_Y     1
-#define BRW_CHANNEL_Z     2
-#define BRW_CHANNEL_W     3
-
-#define BRW_COMPRESSION_NONE          0
-#define BRW_COMPRESSION_2NDHALF       1
-#define BRW_COMPRESSION_COMPRESSED    2
-
-#define BRW_CONDITIONAL_NONE  0
-#define BRW_CONDITIONAL_Z     1
-#define BRW_CONDITIONAL_NZ    2
-#define BRW_CONDITIONAL_EQ    1	/* Z */
-#define BRW_CONDITIONAL_NEQ   2	/* NZ */
-#define BRW_CONDITIONAL_G     3
-#define BRW_CONDITIONAL_GE    4
-#define BRW_CONDITIONAL_L     5
-#define BRW_CONDITIONAL_LE    6
-#define BRW_CONDITIONAL_C     7
-#define BRW_CONDITIONAL_O     8
-
-#define BRW_DEBUG_NONE        0
-#define BRW_DEBUG_BREAKPOINT  1
-
-#define BRW_DEPENDENCY_NORMAL         0
-#define BRW_DEPENDENCY_NOTCLEARED     1
-#define BRW_DEPENDENCY_NOTCHECKED     2
-#define BRW_DEPENDENCY_DISABLE        3
-
-#define BRW_EXECUTE_1     0
-#define BRW_EXECUTE_2     1
-#define BRW_EXECUTE_4     2
-#define BRW_EXECUTE_8     3
-#define BRW_EXECUTE_16    4
-#define BRW_EXECUTE_32    5
-
-#define BRW_HORIZONTAL_STRIDE_0   0
-#define BRW_HORIZONTAL_STRIDE_1   1
-#define BRW_HORIZONTAL_STRIDE_2   2
-#define BRW_HORIZONTAL_STRIDE_4   3
-
-#define BRW_INSTRUCTION_NORMAL    0
-#define BRW_INSTRUCTION_SATURATE  1
-
-#define BRW_MASK_ENABLE   0
-#define BRW_MASK_DISABLE  1
-
-#define BRW_OPCODE_MOV        1
-#define BRW_OPCODE_SEL        2
-#define BRW_OPCODE_NOT        4
-#define BRW_OPCODE_AND        5
-#define BRW_OPCODE_OR         6
-#define BRW_OPCODE_XOR        7
-#define BRW_OPCODE_SHR        8
-#define BRW_OPCODE_SHL        9
-#define BRW_OPCODE_RSR        10
-#define BRW_OPCODE_RSL        11
-#define BRW_OPCODE_ASR        12
-#define BRW_OPCODE_CMP        16
-#define BRW_OPCODE_JMPI       32
-#define BRW_OPCODE_IF         34
-#define BRW_OPCODE_IFF        35
-#define BRW_OPCODE_ELSE       36
-#define BRW_OPCODE_ENDIF      37
-#define BRW_OPCODE_DO         38
-#define BRW_OPCODE_WHILE      39
-#define BRW_OPCODE_BREAK      40
-#define BRW_OPCODE_CONTINUE   41
-#define BRW_OPCODE_HALT       42
-#define BRW_OPCODE_MSAVE      44
-#define BRW_OPCODE_MRESTORE   45
-#define BRW_OPCODE_PUSH       46
-#define BRW_OPCODE_POP        47
-#define BRW_OPCODE_WAIT       48
-#define BRW_OPCODE_SEND       49
-#define BRW_OPCODE_ADD        64
-#define BRW_OPCODE_MUL        65
-#define BRW_OPCODE_AVG        66
-#define BRW_OPCODE_FRC        67
-#define BRW_OPCODE_RNDU       68
-#define BRW_OPCODE_RNDD       69
-#define BRW_OPCODE_RNDE       70
-#define BRW_OPCODE_RNDZ       71
-#define BRW_OPCODE_MAC        72
-#define BRW_OPCODE_MACH       73
-#define BRW_OPCODE_LZD        74
-#define BRW_OPCODE_SAD2       80
-#define BRW_OPCODE_SADA2      81
-#define BRW_OPCODE_DP4        84
-#define BRW_OPCODE_DPH        85
-#define BRW_OPCODE_DP3        86
-#define BRW_OPCODE_DP2        87
-#define BRW_OPCODE_DPA2       88
-#define BRW_OPCODE_LINE       89
-#define BRW_OPCODE_NOP        126
-
-#define BRW_PREDICATE_NONE             0
-#define BRW_PREDICATE_NORMAL           1
-#define BRW_PREDICATE_ALIGN1_ANYV             2
-#define BRW_PREDICATE_ALIGN1_ALLV             3
-#define BRW_PREDICATE_ALIGN1_ANY2H            4
-#define BRW_PREDICATE_ALIGN1_ALL2H            5
-#define BRW_PREDICATE_ALIGN1_ANY4H            6
-#define BRW_PREDICATE_ALIGN1_ALL4H            7
-#define BRW_PREDICATE_ALIGN1_ANY8H            8
-#define BRW_PREDICATE_ALIGN1_ALL8H            9
-#define BRW_PREDICATE_ALIGN1_ANY16H           10
-#define BRW_PREDICATE_ALIGN1_ALL16H           11
-#define BRW_PREDICATE_ALIGN16_REPLICATE_X     2
-#define BRW_PREDICATE_ALIGN16_REPLICATE_Y     3
-#define BRW_PREDICATE_ALIGN16_REPLICATE_Z     4
-#define BRW_PREDICATE_ALIGN16_REPLICATE_W     5
-#define BRW_PREDICATE_ALIGN16_ANY4H           6
-#define BRW_PREDICATE_ALIGN16_ALL4H           7
-
-#define BRW_ARCHITECTURE_REGISTER_FILE    0
-#define BRW_GENERAL_REGISTER_FILE         1
-#define BRW_MESSAGE_REGISTER_FILE         2
-#define BRW_IMMEDIATE_VALUE               3
-
-#define BRW_REGISTER_TYPE_UD  0
-#define BRW_REGISTER_TYPE_D   1
-#define BRW_REGISTER_TYPE_UW  2
-#define BRW_REGISTER_TYPE_W   3
-#define BRW_REGISTER_TYPE_UB  4
-#define BRW_REGISTER_TYPE_B   5
-#define BRW_REGISTER_TYPE_VF  5	/* packed float vector, immediates only? */
-#define BRW_REGISTER_TYPE_HF  6
-#define BRW_REGISTER_TYPE_V   6	/* packed int vector, immediates only, uword dest only */
-#define BRW_REGISTER_TYPE_F   7
-
-#define BRW_ARF_NULL                  0x00
-#define BRW_ARF_ADDRESS               0x10
-#define BRW_ARF_ACCUMULATOR           0x20
-#define BRW_ARF_FLAG                  0x30
-#define BRW_ARF_MASK                  0x40
-#define BRW_ARF_MASK_STACK            0x50
-#define BRW_ARF_MASK_STACK_DEPTH      0x60
-#define BRW_ARF_STATE                 0x70
-#define BRW_ARF_CONTROL               0x80
-#define BRW_ARF_NOTIFICATION_COUNT    0x90
-#define BRW_ARF_IP                    0xA0
-
-#define BRW_AMASK   0
-#define BRW_IMASK   1
-#define BRW_LMASK   2
-#define BRW_CMASK   3
-
-
-
-#define BRW_THREAD_NORMAL     0
-#define BRW_THREAD_ATOMIC     1
-#define BRW_THREAD_SWITCH     2
-
-#define BRW_VERTICAL_STRIDE_0                 0
-#define BRW_VERTICAL_STRIDE_1                 1
-#define BRW_VERTICAL_STRIDE_2                 2
-#define BRW_VERTICAL_STRIDE_4                 3
-#define BRW_VERTICAL_STRIDE_8                 4
-#define BRW_VERTICAL_STRIDE_16                5
-#define BRW_VERTICAL_STRIDE_32                6
-#define BRW_VERTICAL_STRIDE_64                7
-#define BRW_VERTICAL_STRIDE_128               8
-#define BRW_VERTICAL_STRIDE_256               9
-#define BRW_VERTICAL_STRIDE_ONE_DIMENSIONAL   0xF
-
-#define BRW_WIDTH_1       0
-#define BRW_WIDTH_2       1
-#define BRW_WIDTH_4       2
-#define BRW_WIDTH_8       3
-#define BRW_WIDTH_16      4
-
-#define BRW_STATELESS_BUFFER_BOUNDARY_1K      0
-#define BRW_STATELESS_BUFFER_BOUNDARY_2K      1
-#define BRW_STATELESS_BUFFER_BOUNDARY_4K      2
-#define BRW_STATELESS_BUFFER_BOUNDARY_8K      3
-#define BRW_STATELESS_BUFFER_BOUNDARY_16K     4
-#define BRW_STATELESS_BUFFER_BOUNDARY_32K     5
-#define BRW_STATELESS_BUFFER_BOUNDARY_64K     6
-#define BRW_STATELESS_BUFFER_BOUNDARY_128K    7
-#define BRW_STATELESS_BUFFER_BOUNDARY_256K    8
-#define BRW_STATELESS_BUFFER_BOUNDARY_512K    9
-#define BRW_STATELESS_BUFFER_BOUNDARY_1M      10
-#define BRW_STATELESS_BUFFER_BOUNDARY_2M      11
-
-#define BRW_POLYGON_FACING_FRONT      0
-#define BRW_POLYGON_FACING_BACK       1
-
-#define BRW_MESSAGE_TARGET_NULL               0
-#define BRW_MESSAGE_TARGET_MATH               1
-#define BRW_MESSAGE_TARGET_SAMPLER            2
-#define BRW_MESSAGE_TARGET_GATEWAY            3
-#define BRW_MESSAGE_TARGET_DATAPORT_READ      4
-#define BRW_MESSAGE_TARGET_DATAPORT_WRITE     5
-#define BRW_MESSAGE_TARGET_URB                6
-#define BRW_MESSAGE_TARGET_THREAD_SPAWNER     7
-
-#define BRW_SAMPLER_RETURN_FORMAT_FLOAT32     0
-#define BRW_SAMPLER_RETURN_FORMAT_UINT32      2
-#define BRW_SAMPLER_RETURN_FORMAT_SINT32      3
-
-#define BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE              0
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE             0
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS        0
-#define BRW_SAMPLER_MESSAGE_SIMD8_KILLPIX             1
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_LOD        1
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_LOD         1
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_GRADIENTS  2
-#define BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_GRADIENTS    2
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_COMPARE    0
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE     2
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_RESINFO           2
-#define BRW_SAMPLER_MESSAGE_SIMD8_RESINFO             2
-#define BRW_SAMPLER_MESSAGE_SIMD16_RESINFO            2
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_LD                3
-#define BRW_SAMPLER_MESSAGE_SIMD8_LD                  3
-#define BRW_SAMPLER_MESSAGE_SIMD16_LD                 3
-
-#define BRW_DATAPORT_OWORD_BLOCK_1_OWORDLOW   0
-#define BRW_DATAPORT_OWORD_BLOCK_1_OWORDHIGH  1
-#define BRW_DATAPORT_OWORD_BLOCK_2_OWORDS     2
-#define BRW_DATAPORT_OWORD_BLOCK_4_OWORDS     3
-#define BRW_DATAPORT_OWORD_BLOCK_8_OWORDS     4
-
-#define BRW_DATAPORT_OWORD_DUAL_BLOCK_1OWORD     0
-#define BRW_DATAPORT_OWORD_DUAL_BLOCK_4OWORDS    2
-
-#define BRW_DATAPORT_DWORD_SCATTERED_BLOCK_8DWORDS   2
-#define BRW_DATAPORT_DWORD_SCATTERED_BLOCK_16DWORDS  3
-
-#define BRW_DATAPORT_READ_MESSAGE_OWORD_BLOCK_READ          0
-#define BRW_DATAPORT_READ_MESSAGE_OWORD_DUAL_BLOCK_READ     1
-#define BRW_DATAPORT_READ_MESSAGE_DWORD_BLOCK_READ          2
-#define BRW_DATAPORT_READ_MESSAGE_DWORD_SCATTERED_READ      3
-
-#define BRW_DATAPORT_READ_TARGET_DATA_CACHE      0
-#define BRW_DATAPORT_READ_TARGET_RENDER_CACHE    1
-#define BRW_DATAPORT_READ_TARGET_SAMPLER_CACHE   2
-
-#define BRW_DATAPORT_RENDER_TARGET_WRITE_SIMD16_SINGLE_SOURCE                0
-#define BRW_DATAPORT_RENDER_TARGET_WRITE_SIMD16_SINGLE_SOURCE_REPLICATED     1
-#define BRW_DATAPORT_RENDER_TARGET_WRITE_SIMD8_DUAL_SOURCE_SUBSPAN01         2
-#define BRW_DATAPORT_RENDER_TARGET_WRITE_SIMD8_DUAL_SOURCE_SUBSPAN23         3
-#define BRW_DATAPORT_RENDER_TARGET_WRITE_SIMD8_SINGLE_SOURCE_SUBSPAN01       4
-
-#define BRW_DATAPORT_WRITE_MESSAGE_OWORD_BLOCK_WRITE                0
-#define BRW_DATAPORT_WRITE_MESSAGE_OWORD_DUAL_BLOCK_WRITE           1
-#define BRW_DATAPORT_WRITE_MESSAGE_DWORD_BLOCK_WRITE                2
-#define BRW_DATAPORT_WRITE_MESSAGE_DWORD_SCATTERED_WRITE            3
-#define BRW_DATAPORT_WRITE_MESSAGE_RENDER_TARGET_WRITE              4
-#define BRW_DATAPORT_WRITE_MESSAGE_STREAMED_VERTEX_BUFFER_WRITE     5
-#define BRW_DATAPORT_WRITE_MESSAGE_FLUSH_RENDER_CACHE               7
-
-#define BRW_MATH_FUNCTION_INV                              1
-#define BRW_MATH_FUNCTION_LOG                              2
-#define BRW_MATH_FUNCTION_EXP                              3
-#define BRW_MATH_FUNCTION_SQRT                             4
-#define BRW_MATH_FUNCTION_RSQ                              5
-#define BRW_MATH_FUNCTION_SIN                              6 /* was 7 */
-#define BRW_MATH_FUNCTION_COS                              7 /* was 8 */
-#define BRW_MATH_FUNCTION_SINCOS                           8 /* was 6 */
-#define BRW_MATH_FUNCTION_TAN                              9
-#define BRW_MATH_FUNCTION_POW                              10
-#define BRW_MATH_FUNCTION_INT_DIV_QUOTIENT_AND_REMAINDER   11
-#define BRW_MATH_FUNCTION_INT_DIV_QUOTIENT                 12
-#define BRW_MATH_FUNCTION_INT_DIV_REMAINDER                13
-
-#define BRW_MATH_INTEGER_UNSIGNED     0
-#define BRW_MATH_INTEGER_SIGNED       1
-
-#define BRW_MATH_PRECISION_FULL        0
-#define BRW_MATH_PRECISION_PARTIAL     1
-
-#define BRW_MATH_SATURATE_NONE         0
-#define BRW_MATH_SATURATE_SATURATE     1
-
-#define BRW_MATH_DATA_VECTOR  0
-#define BRW_MATH_DATA_SCALAR  1
-
-#define BRW_URB_OPCODE_WRITE  0
-
-#define BRW_URB_SWIZZLE_NONE          0
-#define BRW_URB_SWIZZLE_INTERLEAVE    1
-#define BRW_URB_SWIZZLE_TRANSPOSE     2
-
-#define BRW_SCRATCH_SPACE_SIZE_1K     0
-#define BRW_SCRATCH_SPACE_SIZE_2K     1
-#define BRW_SCRATCH_SPACE_SIZE_4K     2
-#define BRW_SCRATCH_SPACE_SIZE_8K     3
-#define BRW_SCRATCH_SPACE_SIZE_16K    4
-#define BRW_SCRATCH_SPACE_SIZE_32K    5
-#define BRW_SCRATCH_SPACE_SIZE_64K    6
-#define BRW_SCRATCH_SPACE_SIZE_128K   7
-#define BRW_SCRATCH_SPACE_SIZE_256K   8
-#define BRW_SCRATCH_SPACE_SIZE_512K   9
-#define BRW_SCRATCH_SPACE_SIZE_1M     10
-#define BRW_SCRATCH_SPACE_SIZE_2M     11
-
-
-
-
-#define CMD_URB_FENCE                 0x6000
-#define CMD_CONST_BUFFER_STATE        0x6001
-#define CMD_CONST_BUFFER              0x6002
-
-#define CMD_STATE_BASE_ADDRESS        0x6101
-#define CMD_STATE_INSN_POINTER        0x6102
-#define CMD_PIPELINE_SELECT           0x6104
-
-#define CMD_PIPELINED_STATE_POINTERS  0x7800
-#define CMD_BINDING_TABLE_PTRS        0x7801
-#define CMD_VERTEX_BUFFER             0x7808
-#define CMD_VERTEX_ELEMENT            0x7809
-#define CMD_INDEX_BUFFER              0x780a
-#define CMD_VF_STATISTICS             0x780b
-
-#define CMD_DRAW_RECT                 0x7900
-#define CMD_BLEND_CONSTANT_COLOR      0x7901
-#define CMD_CHROMA_KEY                0x7904
-#define CMD_DEPTH_BUFFER              0x7905
-#define CMD_POLY_STIPPLE_OFFSET       0x7906
-#define CMD_POLY_STIPPLE_PATTERN      0x7907
-#define CMD_LINE_STIPPLE_PATTERN      0x7908
-#define CMD_GLOBAL_DEPTH_OFFSET_CLAMP 0x7909
-
-#define CMD_PIPE_CONTROL              0x7a00
-
-#define CMD_3D_PRIM                   0x7b00
-
-#define CMD_MI_FLUSH                  0x0200
-
-
-/* Various values from the R0 vertex header:
- */
-#define R02_PRIM_END    0x1
-#define R02_PRIM_START  0x2
-
-
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_draw.c b/src/gallium/drivers/i965simple/brw_draw.c
deleted file mode 100644
index 49d80cb41c..0000000000
--- a/src/gallium/drivers/i965simple/brw_draw.c
+++ /dev/null
@@ -1,226 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-#include <stdlib.h>
-
-#include "brw_batch.h"
-#include "brw_draw.h"
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_state.h"
-
-#include "pipe/p_context.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_prim.h"
-
-static unsigned hw_prim[PIPE_PRIM_POLYGON+1] = {
-   _3DPRIM_POINTLIST,
-   _3DPRIM_LINELIST,
-   _3DPRIM_LINELOOP,
-   _3DPRIM_LINESTRIP,
-   _3DPRIM_TRILIST,
-   _3DPRIM_TRISTRIP,
-   _3DPRIM_TRIFAN,
-   _3DPRIM_QUADLIST,
-   _3DPRIM_QUADSTRIP,
-   _3DPRIM_POLYGON
-};
-
-
-/* When the primitive changes, set a state bit and re-validate.  Not
- * the nicest and would rather deal with this by having all the
- * programs be immune to the active primitive (ie. cope with all
- * possibilities).  That may not be realistic however.
- */
-static void brw_set_prim(struct brw_context *brw, int prim)
-{
-   PRINT("PRIM: %d\n", prim);
-
-   /* Slight optimization to avoid the GS program when not needed:
-    */
-   if (prim == PIPE_PRIM_QUAD_STRIP &&
-       brw->attribs.Raster->flatshade &&
-       brw->attribs.Raster->fill_cw == PIPE_POLYGON_MODE_FILL &&
-       brw->attribs.Raster->fill_ccw == PIPE_POLYGON_MODE_FILL)
-      prim = PIPE_PRIM_TRIANGLE_STRIP;
-
-   if (prim != brw->primitive) {
-      brw->primitive = prim;
-      brw->state.dirty.brw |= BRW_NEW_PRIMITIVE;
-
-      if (u_reduced_prim(prim) != brw->reduced_primitive) {
-	 brw->reduced_primitive = u_reduced_prim(prim);
-	 brw->state.dirty.brw |= BRW_NEW_REDUCED_PRIMITIVE;
-      }
-
-      brw_validate_state(brw);
-   }
-
-}
-
-
-static unsigned trim(int prim, unsigned length)
-{
-   if (prim == PIPE_PRIM_QUAD_STRIP)
-      return length > 3 ? (length - length % 2) : 0;
-   else if (prim == PIPE_PRIM_QUADS)
-      return length - length % 4;
-   else
-      return length;
-}
-
-
-
-static boolean brw_emit_prim( struct brw_context *brw,
-			      boolean indexed,
-			      unsigned start,
-			      unsigned count )
-
-{
-   struct brw_3d_primitive prim_packet;
-
-   if (BRW_DEBUG & DEBUG_PRIMS)
-      PRINT("PRIM: %d %d %d\n",  brw->primitive, start, count);
-
-   prim_packet.header.opcode = CMD_3D_PRIM;
-   prim_packet.header.length = sizeof(prim_packet)/4 - 2;
-   prim_packet.header.pad = 0;
-   prim_packet.header.topology = hw_prim[brw->primitive];
-   prim_packet.header.indexed = indexed;
-
-   prim_packet.verts_per_instance = trim(brw->primitive, count);
-   prim_packet.start_vert_location = start;
-   prim_packet.instance_count = 1;
-   prim_packet.start_instance_location = 0;
-   prim_packet.base_vert_location = 0;
-
-   if (prim_packet.verts_per_instance == 0)
-      return TRUE;
-
-   return brw_batchbuffer_data( brw->winsys,
-                                &prim_packet,
-                                sizeof(prim_packet) );
-}
-
-
-/* May fail if out of video memory for texture or vbo upload, or on
- * fallback conditions.
- */
-static boolean brw_try_draw_elements( struct pipe_context *pipe,
-				      struct pipe_buffer *index_buffer,
-				      unsigned index_size,
-				      unsigned mode,
-				      unsigned start,
-				      unsigned count )
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   /* Set the first primitive ahead of validate_state:
-    */
-   brw_set_prim(brw, mode);
-
-   /* Upload index, vertex data:
-    */
-   if (index_buffer &&
-       !brw_upload_indices( brw, index_buffer, index_size, start, count ))
-      return FALSE;
-
-   if (!brw_upload_vertex_buffers(brw))
-      return FALSE;
-
-   if (!brw_upload_vertex_elements( brw ))
-      return FALSE;
-
-   /* XXX:  Need to separate validate and upload of state.
-    */
-   if (brw->state.dirty.brw)
-      brw_validate_state( brw );
-
-   if (!brw_emit_prim(brw, index_buffer != NULL,
-                      start, count))
-      return FALSE;
-
-   return TRUE;
-}
-
-
-
-static boolean brw_draw_elements( struct pipe_context *pipe,
-				  struct pipe_buffer *indexBuffer,
-				  unsigned indexSize,
-				  unsigned mode,
-				  unsigned start,
-				  unsigned count )
-{
-   if (!brw_try_draw_elements( pipe,
-			       indexBuffer,
-			       indexSize,
-			       mode, start, count ))
-   {
-      /* flush ? */
-
-      if (!brw_try_draw_elements( pipe,
-				  indexBuffer,
-				  indexSize,
-				  mode, start,
-				  count )) {
-	 assert(0);
-	 return FALSE;
-      }
-   }
-
-   return TRUE;
-}
-
-
-
-static boolean brw_draw_arrays( struct pipe_context *pipe,
-				    unsigned mode,
-				    unsigned start,
-				    unsigned count )
-{
-   if (!brw_try_draw_elements( pipe, NULL, 0, mode, start, count )) {
-      /* flush ? */
-
-      if (!brw_try_draw_elements( pipe, NULL, 0, mode, start, count )) {
-	 assert(0);
-	 return FALSE;
-      }
-   }
-   
-   return TRUE;
-}
-
-
-
-void brw_init_draw_functions( struct brw_context *brw )
-{
-   brw->pipe.draw_arrays = brw_draw_arrays;
-   brw->pipe.draw_elements = brw_draw_elements;
-}
-
-
diff --git a/src/gallium/drivers/i965simple/brw_draw_upload.c b/src/gallium/drivers/i965simple/brw_draw_upload.c
deleted file mode 100644
index 2d9ca3f2ea..0000000000
--- a/src/gallium/drivers/i965simple/brw_draw_upload.c
+++ /dev/null
@@ -1,300 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-#include <stdlib.h>
-
-#include "brw_batch.h"
-#include "brw_draw.h"
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_state.h"
-
-
-struct brw_array_state {
-   union header_union header;
-
-   struct {
-      union {
-	 struct {
-	    unsigned pitch:11;
-	    unsigned pad:15;
-	    unsigned access_type:1;
-	    unsigned vb_index:5;
-	 } bits;
-	 unsigned dword;
-      } vb0;
-
-      struct pipe_buffer *buffer;
-      unsigned offset;
-
-      unsigned max_index;
-      unsigned instance_data_step_rate;
-
-   } vb[BRW_VBP_MAX];
-};
-
-
-
-unsigned brw_translate_surface_format( unsigned id )
-{
-   switch (id) {
-   case PIPE_FORMAT_R64_FLOAT:
-      return BRW_SURFACEFORMAT_R64_FLOAT;
-   case PIPE_FORMAT_R64G64_FLOAT:
-      return BRW_SURFACEFORMAT_R64G64_FLOAT;
-   case PIPE_FORMAT_R64G64B64_FLOAT:
-      return BRW_SURFACEFORMAT_R64G64B64_FLOAT;
-   case PIPE_FORMAT_R64G64B64A64_FLOAT:
-      return BRW_SURFACEFORMAT_R64G64B64A64_FLOAT;
-
-   case PIPE_FORMAT_R32_FLOAT:
-      return BRW_SURFACEFORMAT_R32_FLOAT;
-   case PIPE_FORMAT_R32G32_FLOAT:
-      return BRW_SURFACEFORMAT_R32G32_FLOAT;
-   case PIPE_FORMAT_R32G32B32_FLOAT:
-      return BRW_SURFACEFORMAT_R32G32B32_FLOAT;
-   case PIPE_FORMAT_R32G32B32A32_FLOAT:
-      return BRW_SURFACEFORMAT_R32G32B32A32_FLOAT;
-
-   case PIPE_FORMAT_R32_UNORM:
-      return BRW_SURFACEFORMAT_R32_UNORM;
-   case PIPE_FORMAT_R32G32_UNORM:
-      return BRW_SURFACEFORMAT_R32G32_UNORM;
-   case PIPE_FORMAT_R32G32B32_UNORM:
-      return BRW_SURFACEFORMAT_R32G32B32_UNORM;
-   case PIPE_FORMAT_R32G32B32A32_UNORM:
-      return BRW_SURFACEFORMAT_R32G32B32A32_UNORM;
-
-   case PIPE_FORMAT_R32_USCALED:
-      return BRW_SURFACEFORMAT_R32_USCALED;
-   case PIPE_FORMAT_R32G32_USCALED:
-      return BRW_SURFACEFORMAT_R32G32_USCALED;
-   case PIPE_FORMAT_R32G32B32_USCALED:
-      return BRW_SURFACEFORMAT_R32G32B32_USCALED;
-   case PIPE_FORMAT_R32G32B32A32_USCALED:
-      return BRW_SURFACEFORMAT_R32G32B32A32_USCALED;
-
-   case PIPE_FORMAT_R32_SNORM:
-      return BRW_SURFACEFORMAT_R32_SNORM;
-   case PIPE_FORMAT_R32G32_SNORM:
-      return BRW_SURFACEFORMAT_R32G32_SNORM;
-   case PIPE_FORMAT_R32G32B32_SNORM:
-      return BRW_SURFACEFORMAT_R32G32B32_SNORM;
-   case PIPE_FORMAT_R32G32B32A32_SNORM:
-      return BRW_SURFACEFORMAT_R32G32B32A32_SNORM;
-
-   case PIPE_FORMAT_R32_SSCALED:
-      return BRW_SURFACEFORMAT_R32_SSCALED;
-   case PIPE_FORMAT_R32G32_SSCALED:
-      return BRW_SURFACEFORMAT_R32G32_SSCALED;
-   case PIPE_FORMAT_R32G32B32_SSCALED:
-      return BRW_SURFACEFORMAT_R32G32B32_SSCALED;
-   case PIPE_FORMAT_R32G32B32A32_SSCALED:
-      return BRW_SURFACEFORMAT_R32G32B32A32_SSCALED;
-
-   case PIPE_FORMAT_R16_UNORM:
-      return BRW_SURFACEFORMAT_R16_UNORM;
-   case PIPE_FORMAT_R16G16_UNORM:
-      return BRW_SURFACEFORMAT_R16G16_UNORM;
-   case PIPE_FORMAT_R16G16B16_UNORM:
-      return BRW_SURFACEFORMAT_R16G16B16_UNORM;
-   case PIPE_FORMAT_R16G16B16A16_UNORM:
-      return BRW_SURFACEFORMAT_R16G16B16A16_UNORM;
-
-   case PIPE_FORMAT_R16_USCALED:
-      return BRW_SURFACEFORMAT_R16_USCALED;
-   case PIPE_FORMAT_R16G16_USCALED:
-      return BRW_SURFACEFORMAT_R16G16_USCALED;
-   case PIPE_FORMAT_R16G16B16_USCALED:
-      return BRW_SURFACEFORMAT_R16G16B16_USCALED;
-   case PIPE_FORMAT_R16G16B16A16_USCALED:
-      return BRW_SURFACEFORMAT_R16G16B16A16_USCALED;
-
-   case PIPE_FORMAT_R16_SNORM:
-      return BRW_SURFACEFORMAT_R16_SNORM;
-   case PIPE_FORMAT_R16G16_SNORM:
-      return BRW_SURFACEFORMAT_R16G16_SNORM;
-   case PIPE_FORMAT_R16G16B16_SNORM:
-      return BRW_SURFACEFORMAT_R16G16B16_SNORM;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      return BRW_SURFACEFORMAT_R16G16B16A16_SNORM;
-
-   case PIPE_FORMAT_R16_SSCALED:
-      return BRW_SURFACEFORMAT_R16_SSCALED;
-   case PIPE_FORMAT_R16G16_SSCALED:
-      return BRW_SURFACEFORMAT_R16G16_SSCALED;
-   case PIPE_FORMAT_R16G16B16_SSCALED:
-      return BRW_SURFACEFORMAT_R16G16B16_SSCALED;
-   case PIPE_FORMAT_R16G16B16A16_SSCALED:
-      return BRW_SURFACEFORMAT_R16G16B16A16_SSCALED;
-
-   case PIPE_FORMAT_R8_UNORM:
-      return BRW_SURFACEFORMAT_R8_UNORM;
-   case PIPE_FORMAT_R8G8_UNORM:
-      return BRW_SURFACEFORMAT_R8G8_UNORM;
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      return BRW_SURFACEFORMAT_R8G8B8_UNORM;
-   case PIPE_FORMAT_R8G8B8A8_UNORM:
-      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM;
-
-   case PIPE_FORMAT_R8_USCALED:
-      return BRW_SURFACEFORMAT_R8_USCALED;
-   case PIPE_FORMAT_R8G8_USCALED:
-      return BRW_SURFACEFORMAT_R8G8_USCALED;
-   case PIPE_FORMAT_R8G8B8_USCALED:
-      return BRW_SURFACEFORMAT_R8G8B8_USCALED;
-   case PIPE_FORMAT_R8G8B8A8_USCALED:
-      return BRW_SURFACEFORMAT_R8G8B8A8_USCALED;
-
-   case PIPE_FORMAT_R8_SNORM:
-      return BRW_SURFACEFORMAT_R8_SNORM;
-   case PIPE_FORMAT_R8G8_SNORM:
-      return BRW_SURFACEFORMAT_R8G8_SNORM;
-   case PIPE_FORMAT_R8G8B8_SNORM:
-      return BRW_SURFACEFORMAT_R8G8B8_SNORM;
-   case PIPE_FORMAT_R8G8B8A8_SNORM:
-      return BRW_SURFACEFORMAT_R8G8B8A8_SNORM;
-
-   case PIPE_FORMAT_R8_SSCALED:
-      return BRW_SURFACEFORMAT_R8_SSCALED;
-   case PIPE_FORMAT_R8G8_SSCALED:
-      return BRW_SURFACEFORMAT_R8G8_SSCALED;
-   case PIPE_FORMAT_R8G8B8_SSCALED:
-      return BRW_SURFACEFORMAT_R8G8B8_SSCALED;
-   case PIPE_FORMAT_R8G8B8A8_SSCALED:
-      return BRW_SURFACEFORMAT_R8G8B8A8_SSCALED;
-
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-static unsigned get_index_type(int type)
-{
-   switch (type) {
-   case 1: return BRW_INDEX_BYTE;
-   case 2: return BRW_INDEX_WORD;
-   case 4: return BRW_INDEX_DWORD;
-   default: assert(0); return 0;
-   }
-}
-
-
-boolean brw_upload_vertex_buffers( struct brw_context *brw )
-{
-   struct brw_array_state vbp;
-   unsigned nr_enabled = 0;
-   unsigned i;
-
-   memset(&vbp, 0, sizeof(vbp));
-
-   /* This is a hardware limit:
-    */
-
-   for (i = 0; i < BRW_VEP_MAX; i++)
-   {
-      if (brw->vb.vbo_array[i] == NULL) {
-	 nr_enabled = i;
-	 break;
-      }
-
-      vbp.vb[i].vb0.bits.pitch = brw->vb.vbo_array[i]->stride;
-      vbp.vb[i].vb0.bits.pad = 0;
-      vbp.vb[i].vb0.bits.access_type = BRW_VERTEXBUFFER_ACCESS_VERTEXDATA;
-      vbp.vb[i].vb0.bits.vb_index = i;
-      vbp.vb[i].offset = brw->vb.vbo_array[i]->buffer_offset;
-      vbp.vb[i].buffer = brw->vb.vbo_array[i]->buffer;
-      vbp.vb[i].max_index = brw->vb.vbo_array[i]->max_index;
-   }
-
-
-   vbp.header.bits.length = (1 + nr_enabled * 4) - 2;
-   vbp.header.bits.opcode = CMD_VERTEX_BUFFER;
-
-   BEGIN_BATCH(vbp.header.bits.length+2, 0);
-   OUT_BATCH( vbp.header.dword );
-
-   for (i = 0; i < nr_enabled; i++) {
-      OUT_BATCH( vbp.vb[i].vb0.dword );
-      OUT_RELOC( vbp.vb[i].buffer,  PIPE_BUFFER_USAGE_GPU_READ,
-		 vbp.vb[i].offset);
-      OUT_BATCH( vbp.vb[i].max_index );
-      OUT_BATCH( vbp.vb[i].instance_data_step_rate );
-   }
-   ADVANCE_BATCH();
-   return TRUE;
-}
-
-
-
-boolean brw_upload_vertex_elements( struct brw_context *brw )
-{
-   struct brw_vertex_element_packet vep;
-
-   unsigned i;
-   unsigned nr_enabled = brw->attribs.VertexProgram->info.num_inputs;
-
-   memset(&vep, 0, sizeof(vep));
-
-   for (i = 0; i < nr_enabled; i++) 
-      vep.ve[i] = brw->vb.inputs[i];
-
-
-   vep.header.length = (1 + nr_enabled * sizeof(vep.ve[0])/4) - 2;
-   vep.header.opcode = CMD_VERTEX_ELEMENT;
-   brw_cached_batch_struct(brw, &vep, 4 + nr_enabled * sizeof(vep.ve[0]));
-
-   return TRUE;
-}
-
-boolean brw_upload_indices( struct brw_context *brw,
-                            const struct pipe_buffer *index_buffer,
-                            int ib_size, int start, int count)
-{
-   /* Emit the indexbuffer packet:
-    */
-   {
-      struct brw_indexbuffer ib;
-
-      memset(&ib, 0, sizeof(ib));
-
-      ib.header.bits.opcode = CMD_INDEX_BUFFER;
-      ib.header.bits.length = sizeof(ib)/4 - 2;
-      ib.header.bits.index_format = get_index_type(ib_size);
-      ib.header.bits.cut_index_enable = 0;
-
-
-      BEGIN_BATCH(4, 0);
-      OUT_BATCH( ib.header.dword );
-      OUT_RELOC( index_buffer, PIPE_BUFFER_USAGE_GPU_READ, start);
-      OUT_RELOC( index_buffer, PIPE_BUFFER_USAGE_GPU_READ, start + count);
-      OUT_BATCH( 0 );
-      ADVANCE_BATCH();
-   }
-   return TRUE;
-}
diff --git a/src/gallium/drivers/i965simple/brw_eu.c b/src/gallium/drivers/i965simple/brw_eu.c
deleted file mode 100644
index e2002d1821..0000000000
--- a/src/gallium/drivers/i965simple/brw_eu.c
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-  
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_eu.h"
-
-
-
-/* How does predicate control work when execution_size != 8?  Do I
- * need to test/set for 0xffff when execution_size is 16?
- */
-void brw_set_predicate_control_flag_value( struct brw_compile *p, unsigned value )
-{
-   p->current->header.predicate_control = BRW_PREDICATE_NONE;
-
-   if (value != 0xff) {
-      if (value != p->flag_value) {
-	 brw_push_insn_state(p);
-	 brw_MOV(p, brw_flag_reg(), brw_imm_uw(value));
-	 p->flag_value = value;
-	 brw_pop_insn_state(p);
-      }
-
-      p->current->header.predicate_control = BRW_PREDICATE_NORMAL;
-   }   
-}
-
-void brw_set_predicate_control( struct brw_compile *p, unsigned pc )
-{
-   p->current->header.predicate_control = pc;
-}
-
-void brw_set_conditionalmod( struct brw_compile *p, unsigned conditional )
-{
-   p->current->header.destreg__conditonalmod = conditional;
-}
-
-void brw_set_access_mode( struct brw_compile *p, unsigned access_mode )
-{
-   p->current->header.access_mode = access_mode;
-}
-
-void brw_set_compression_control( struct brw_compile *p, boolean compression_control )
-{
-   p->current->header.compression_control = compression_control;
-}
-
-void brw_set_mask_control( struct brw_compile *p, unsigned value )
-{
-   p->current->header.mask_control = value;
-}
-
-void brw_set_saturate( struct brw_compile *p, unsigned value )
-{
-   p->current->header.saturate = value;
-}
-
-void brw_push_insn_state( struct brw_compile *p )
-{
-   assert(p->current != &p->stack[BRW_EU_MAX_INSN_STACK-1]);
-   memcpy(p->current+1, p->current, sizeof(struct brw_instruction));
-   p->current++;   
-}
-
-void brw_pop_insn_state( struct brw_compile *p )
-{
-   assert(p->current != p->stack);
-   p->current--;
-}
-
-
-/***********************************************************************
- */
-void brw_init_compile( struct brw_compile *p )
-{
-   p->nr_insn = 0;
-   p->current = p->stack;
-   memset(p->current, 0, sizeof(p->current[0]));
-
-   /* Some defaults?
-    */
-   brw_set_mask_control(p, BRW_MASK_ENABLE); /* what does this do? */
-   brw_set_saturate(p, 0);
-   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-   brw_set_predicate_control_flag_value(p, 0xff); 
-}
-
-
-const unsigned *brw_get_program( struct brw_compile *p,
-			       unsigned *sz )
-{
-   unsigned i;
-
-   for (i = 0; i < 8; i++)
-      brw_NOP(p);
-
-   *sz = p->nr_insn * sizeof(struct brw_instruction);
-   return (const unsigned *)p->store;
-}
-
diff --git a/src/gallium/drivers/i965simple/brw_eu.h b/src/gallium/drivers/i965simple/brw_eu.h
deleted file mode 100644
index 23151ae9ed..0000000000
--- a/src/gallium/drivers/i965simple/brw_eu.h
+++ /dev/null
@@ -1,888 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#ifndef BRW_EU_H
-#define BRW_EU_H
-
-#include "brw_structs.h"
-#include "brw_defines.h"
-
-#include "pipe/p_compiler.h"
-#include "pipe/p_shader_tokens.h"
-
-#define BRW_SWIZZLE4(a,b,c,d) (((a)<<0) | ((b)<<2) | ((c)<<4) | ((d)<<6))
-#define BRW_GET_SWZ(swz, idx) (((swz) >> ((idx)*2)) & 0x3)
-
-#define BRW_SWIZZLE_NOOP      BRW_SWIZZLE4(0,1,2,3)
-#define BRW_SWIZZLE_XYZW      BRW_SWIZZLE4(0,1,2,3)
-#define BRW_SWIZZLE_XXXX      BRW_SWIZZLE4(0,0,0,0)
-#define BRW_SWIZZLE_XYXY      BRW_SWIZZLE4(0,1,0,1)
-
-
-#define REG_SIZE (8*4)
-
-
-/* These aren't hardware structs, just something useful for us to pass around:
- *
- * Align1 operation has a lot of control over input ranges.  Used in
- * WM programs to implement shaders decomposed into "channel serial"
- * or "structure of array" form:
- */
-struct brw_reg
-{
-   unsigned type:4;
-   unsigned file:2;
-   unsigned nr:8;
-   unsigned subnr:5;		/* :1 in align16 */
-   unsigned negate:1;		/* source only */
-   unsigned abs:1;		/* source only */
-   unsigned vstride:4;		/* source only */
-   unsigned width:3;		/* src only, align1 only */
-   unsigned hstride:2;   		/* src only, align1 only */
-   unsigned address_mode:1;	/* relative addressing, hopefully! */
-   unsigned pad0:1;
-
-   union {
-      struct {
-	 unsigned swizzle:8;		/* src only, align16 only */
-	 unsigned writemask:4;		/* dest only, align16 only */
-	 int  indirect_offset:10;	/* relative addressing offset */
-	 unsigned pad1:10;		/* two dwords total */
-      } bits;
-
-      float f;
-      int   d;
-      unsigned ud;
-   } dw1;
-};
-
-
-struct brw_indirect {
-   unsigned addr_subnr:4;
-   int addr_offset:10;
-   unsigned pad:18;
-};
-
-
-#define BRW_EU_MAX_INSN_STACK 5
-#define BRW_EU_MAX_INSN 1200
-
-struct brw_compile {
-   struct brw_instruction store[BRW_EU_MAX_INSN];
-   unsigned nr_insn;
-
-   /* Allow clients to push/pop instruction state:
-    */
-   struct brw_instruction stack[BRW_EU_MAX_INSN_STACK];
-   struct brw_instruction *current;
-
-   unsigned flag_value;
-   boolean single_program_flow;
-};
-
-
-
-static __inline int type_sz( unsigned type )
-{
-   switch( type ) {
-   case BRW_REGISTER_TYPE_UD:
-   case BRW_REGISTER_TYPE_D:
-   case BRW_REGISTER_TYPE_F:
-      return 4;
-   case BRW_REGISTER_TYPE_HF:
-   case BRW_REGISTER_TYPE_UW:
-   case BRW_REGISTER_TYPE_W:
-      return 2;
-   case BRW_REGISTER_TYPE_UB:
-   case BRW_REGISTER_TYPE_B:
-      return 1;
-   default:
-      return 0;
-   }
-}
-
-static __inline struct brw_reg brw_reg( unsigned file,
-					unsigned nr,
-					unsigned subnr,
-					unsigned type,
-					unsigned vstride,
-					unsigned width,
-					unsigned hstride,
-					unsigned swizzle,
-					unsigned writemask)
-{
-
-   struct brw_reg reg;
-   reg.type = type;
-   reg.file = file;
-   reg.nr = nr;
-   reg.subnr = subnr * type_sz(type);
-   reg.negate = 0;
-   reg.abs = 0;
-   reg.vstride = vstride;
-   reg.width = width;
-   reg.hstride = hstride;
-   reg.address_mode = BRW_ADDRESS_DIRECT;
-   reg.pad0 = 0;
-
-   /* Could do better: If the reg is r5.3<0;1,0>, we probably want to
-    * set swizzle and writemask to W, as the lower bits of subnr will
-    * be lost when converted to align16.  This is probably too much to
-    * keep track of as you'd want it adjusted by suboffset(), etc.
-    * Perhaps fix up when converting to align16?
-    */
-   reg.dw1.bits.swizzle = swizzle;
-   reg.dw1.bits.writemask = writemask;
-   reg.dw1.bits.indirect_offset = 0;
-   reg.dw1.bits.pad1 = 0;
-   return reg;
-}
-
-static __inline struct brw_reg brw_vec16_reg( unsigned file,
-					      unsigned nr,
-					      unsigned subnr )
-{
-   return brw_reg(file,
-		  nr,
-		  subnr,
-		  BRW_REGISTER_TYPE_F,
-		  BRW_VERTICAL_STRIDE_16,
-		  BRW_WIDTH_16,
-		  BRW_HORIZONTAL_STRIDE_1,
-		  BRW_SWIZZLE_XYZW,
-		  TGSI_WRITEMASK_XYZW);
-}
-
-static __inline struct brw_reg brw_vec8_reg( unsigned file,
-					     unsigned nr,
-					     unsigned subnr )
-{
-   return brw_reg(file,
-		  nr,
-		  subnr,
-		  BRW_REGISTER_TYPE_F,
-		  BRW_VERTICAL_STRIDE_8,
-		  BRW_WIDTH_8,
-		  BRW_HORIZONTAL_STRIDE_1,
-		  BRW_SWIZZLE_XYZW,
-		  TGSI_WRITEMASK_XYZW);
-}
-
-
-static __inline struct brw_reg brw_vec4_reg( unsigned file,
-					      unsigned nr,
-					      unsigned subnr )
-{
-   return brw_reg(file,
-		  nr,
-		  subnr,
-		  BRW_REGISTER_TYPE_F,
-		  BRW_VERTICAL_STRIDE_4,
-		  BRW_WIDTH_4,
-		  BRW_HORIZONTAL_STRIDE_1,
-		  BRW_SWIZZLE_XYZW,
-		  TGSI_WRITEMASK_XYZW);
-}
-
-
-static __inline struct brw_reg brw_vec2_reg( unsigned file,
-					      unsigned nr,
-					      unsigned subnr )
-{
-   return brw_reg(file,
-		  nr,
-		  subnr,
-		  BRW_REGISTER_TYPE_F,
-		  BRW_VERTICAL_STRIDE_2,
-		  BRW_WIDTH_2,
-		  BRW_HORIZONTAL_STRIDE_1,
-		  BRW_SWIZZLE_XYXY,
-		  TGSI_WRITEMASK_XY);
-}
-
-static __inline struct brw_reg brw_vec1_reg( unsigned file,
-					     unsigned nr,
-					     unsigned subnr )
-{
-   return brw_reg(file,
-		  nr,
-		  subnr,
-		  BRW_REGISTER_TYPE_F,
-		  BRW_VERTICAL_STRIDE_0,
-		  BRW_WIDTH_1,
-		  BRW_HORIZONTAL_STRIDE_0,
-		  BRW_SWIZZLE_XXXX,
-		  TGSI_WRITEMASK_X);
-}
-
-
-static __inline struct brw_reg retype( struct brw_reg reg,
-				       unsigned type )
-{
-   reg.type = type;
-   return reg;
-}
-
-static __inline struct brw_reg suboffset( struct brw_reg reg,
-					  unsigned delta )
-{
-   reg.subnr += delta * type_sz(reg.type);
-   return reg;
-}
-
-
-static __inline struct brw_reg offset( struct brw_reg reg,
-				       unsigned delta )
-{
-   reg.nr += delta;
-   return reg;
-}
-
-
-static __inline struct brw_reg byte_offset( struct brw_reg reg,
-					    unsigned bytes )
-{
-   unsigned newoffset = reg.nr * REG_SIZE + reg.subnr + bytes;
-   reg.nr = newoffset / REG_SIZE;
-   reg.subnr = newoffset % REG_SIZE;
-   return reg;
-}
-
-
-static __inline struct brw_reg brw_uw16_reg( unsigned file,
-					     unsigned nr,
-					     unsigned subnr )
-{
-   return suboffset(retype(brw_vec16_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
-}
-
-static __inline struct brw_reg brw_uw8_reg( unsigned file,
-					    unsigned nr,
-					    unsigned subnr )
-{
-   return suboffset(retype(brw_vec8_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
-}
-
-static __inline struct brw_reg brw_uw1_reg( unsigned file,
-					    unsigned nr,
-					    unsigned subnr )
-{
-   return suboffset(retype(brw_vec1_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
-}
-
-static __inline struct brw_reg brw_imm_reg( unsigned type )
-{
-   return brw_reg( BRW_IMMEDIATE_VALUE,
-		   0,
-		   0,
-		   type,
-		   BRW_VERTICAL_STRIDE_0,
-		   BRW_WIDTH_1,
-		   BRW_HORIZONTAL_STRIDE_0,
-		   0,
-		   0);
-}
-
-static __inline struct brw_reg brw_imm_f( float f )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_F);
-   imm.dw1.f = f;
-   return imm;
-}
-
-static __inline struct brw_reg brw_imm_d( int d )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_D);
-   imm.dw1.d = d;
-   return imm;
-}
-
-static __inline struct brw_reg brw_imm_ud( unsigned ud )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UD);
-   imm.dw1.ud = ud;
-   return imm;
-}
-
-static __inline struct brw_reg brw_imm_uw( ushort uw )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UW);
-   imm.dw1.ud = uw;
-   return imm;
-}
-
-static __inline struct brw_reg brw_imm_w( short w )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_W);
-   imm.dw1.d = w;
-   return imm;
-}
-
-/* brw_imm_b and brw_imm_ub aren't supported by hardware - the type
- * numbers alias with _V and _VF below:
- */
-
-/* Vector of eight signed half-byte values:
- */
-static __inline struct brw_reg brw_imm_v( unsigned v )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_V);
-   imm.vstride = BRW_VERTICAL_STRIDE_0;
-   imm.width = BRW_WIDTH_8;
-   imm.hstride = BRW_HORIZONTAL_STRIDE_1;
-   imm.dw1.ud = v;
-   return imm;
-}
-
-/* Vector of four 8-bit float values:
- */
-static __inline struct brw_reg brw_imm_vf( unsigned v )
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_VF);
-   imm.vstride = BRW_VERTICAL_STRIDE_0;
-   imm.width = BRW_WIDTH_4;
-   imm.hstride = BRW_HORIZONTAL_STRIDE_1;
-   imm.dw1.ud = v;
-   return imm;
-}
-
-#define VF_ZERO 0x0
-#define VF_ONE  0x30
-#define VF_NEG  (1<<7)
-
-static __inline struct brw_reg brw_imm_vf4( unsigned v0,
-					    unsigned v1,
-					    unsigned v2,
-					    unsigned v3)
-{
-   struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_VF);
-   imm.vstride = BRW_VERTICAL_STRIDE_0;
-   imm.width = BRW_WIDTH_4;
-   imm.hstride = BRW_HORIZONTAL_STRIDE_1;
-   imm.dw1.ud = ((v0 << 0) |
-		 (v1 << 8) |
-		 (v2 << 16) |
-		 (v3 << 24));
-   return imm;
-}
-
-
-static __inline struct brw_reg brw_address( struct brw_reg reg )
-{
-   return brw_imm_uw(reg.nr * REG_SIZE + reg.subnr);
-}
-
-
-static __inline struct brw_reg brw_vec1_grf( unsigned nr,
-					       unsigned subnr )
-{
-   return brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
-}
-
-static __inline struct brw_reg brw_vec8_grf( unsigned nr,
-					     unsigned subnr )
-{
-   return brw_vec8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
-}
-
-static __inline struct brw_reg brw_vec4_grf( unsigned nr,
-					     unsigned subnr )
-{
-   return brw_vec4_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
-}
-
-
-static __inline struct brw_reg brw_vec2_grf( unsigned nr,
-					     unsigned subnr )
-{
-   return brw_vec2_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
-}
-
-static __inline struct brw_reg brw_uw8_grf( unsigned nr,
-					    unsigned subnr )
-{
-   return brw_uw8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
-}
-
-static __inline struct brw_reg brw_null_reg( void )
-{
-   return brw_vec8_reg(BRW_ARCHITECTURE_REGISTER_FILE,
-		       BRW_ARF_NULL,
-		       0);
-}
-
-static __inline struct brw_reg brw_address_reg( unsigned subnr )
-{
-   return brw_uw1_reg(BRW_ARCHITECTURE_REGISTER_FILE,
-		      BRW_ARF_ADDRESS,
-		      subnr);
-}
-
-/* If/else instructions break in align16 mode if writemask & swizzle
- * aren't xyzw.  This goes against the convention for other scalar
- * regs:
- */
-static __inline struct brw_reg brw_ip_reg( void )
-{
-   return brw_reg(BRW_ARCHITECTURE_REGISTER_FILE,
-		  BRW_ARF_IP,
-		  0,
-		  BRW_REGISTER_TYPE_UD,
-		  BRW_VERTICAL_STRIDE_4, /* ? */
-		  BRW_WIDTH_1,
-		  BRW_HORIZONTAL_STRIDE_0,
-		  BRW_SWIZZLE_XYZW, /* NOTE! */
-		  TGSI_WRITEMASK_XYZW); /* NOTE! */
-}
-
-static __inline struct brw_reg brw_acc_reg( void )
-{
-   return brw_vec8_reg(BRW_ARCHITECTURE_REGISTER_FILE,
-		       BRW_ARF_ACCUMULATOR,
-		       0);
-}
-
-
-static __inline struct brw_reg brw_flag_reg( void )
-{
-   return brw_uw1_reg(BRW_ARCHITECTURE_REGISTER_FILE,
-		      BRW_ARF_FLAG,
-		      0);
-}
-
-
-static __inline struct brw_reg brw_mask_reg( unsigned subnr )
-{
-   return brw_uw1_reg(BRW_ARCHITECTURE_REGISTER_FILE,
-		      BRW_ARF_MASK,
-		      subnr);
-}
-
-static __inline struct brw_reg brw_message_reg( unsigned nr )
-{
-   return brw_vec8_reg(BRW_MESSAGE_REGISTER_FILE,
-		       nr,
-		       0);
-}
-
-
-
-
-/* This is almost always called with a numeric constant argument, so
- * make things easy to evaluate at compile time:
- */
-static __inline unsigned cvt( unsigned val )
-{
-   switch (val) {
-   case 0: return 0;
-   case 1: return 1;
-   case 2: return 2;
-   case 4: return 3;
-   case 8: return 4;
-   case 16: return 5;
-   case 32: return 6;
-   }
-   return 0;
-}
-
-static __inline struct brw_reg stride( struct brw_reg reg,
-				       unsigned vstride,
-				       unsigned width,
-				       unsigned hstride )
-{
-
-   reg.vstride = cvt(vstride);
-   reg.width = cvt(width) - 1;
-   reg.hstride = cvt(hstride);
-   return reg;
-}
-
-static __inline struct brw_reg vec16( struct brw_reg reg )
-{
-   return stride(reg, 16,16,1);
-}
-
-static __inline struct brw_reg vec8( struct brw_reg reg )
-{
-   return stride(reg, 8,8,1);
-}
-
-static __inline struct brw_reg vec4( struct brw_reg reg )
-{
-   return stride(reg, 4,4,1);
-}
-
-static __inline struct brw_reg vec2( struct brw_reg reg )
-{
-   return stride(reg, 2,2,1);
-}
-
-static __inline struct brw_reg vec1( struct brw_reg reg )
-{
-   return stride(reg, 0,1,0);
-}
-
-static __inline struct brw_reg get_element( struct brw_reg reg, unsigned elt )
-{
-   return vec1(suboffset(reg, elt));
-}
-
-static __inline struct brw_reg get_element_ud( struct brw_reg reg, unsigned elt )
-{
-   return vec1(suboffset(retype(reg, BRW_REGISTER_TYPE_UD), elt));
-}
-
-
-static __inline struct brw_reg brw_swizzle( struct brw_reg reg,
-					    unsigned x,
-					    unsigned y,
-					    unsigned z,
-					    unsigned w)
-{
-   reg.dw1.bits.swizzle = BRW_SWIZZLE4(BRW_GET_SWZ(reg.dw1.bits.swizzle, x),
-				       BRW_GET_SWZ(reg.dw1.bits.swizzle, y),
-				       BRW_GET_SWZ(reg.dw1.bits.swizzle, z),
-				       BRW_GET_SWZ(reg.dw1.bits.swizzle, w));
-   return reg;
-}
-
-
-static __inline struct brw_reg brw_swizzle1( struct brw_reg reg,
-					     unsigned x )
-{
-   return brw_swizzle(reg, x, x, x, x);
-}
-
-static __inline struct brw_reg brw_writemask( struct brw_reg reg,
-					      unsigned mask )
-{
-   reg.dw1.bits.writemask &= mask;
-   return reg;
-}
-
-static __inline struct brw_reg brw_set_writemask( struct brw_reg reg,
-						  unsigned mask )
-{
-   reg.dw1.bits.writemask = mask;
-   return reg;
-}
-
-static __inline struct brw_reg negate( struct brw_reg reg )
-{
-   reg.negate ^= 1;
-   return reg;
-}
-
-static __inline struct brw_reg brw_abs( struct brw_reg reg )
-{
-   reg.abs = 1;
-   return reg;
-}
-
-/***********************************************************************
- */
-static __inline struct brw_reg brw_vec4_indirect( unsigned subnr,
-						  int offset )
-{
-   struct brw_reg reg =  brw_vec4_grf(0, 0);
-   reg.subnr = subnr;
-   reg.address_mode = BRW_ADDRESS_REGISTER_INDIRECT_REGISTER;
-   reg.dw1.bits.indirect_offset = offset;
-   return reg;
-}
-
-static __inline struct brw_reg brw_vec1_indirect( unsigned subnr,
-						  int offset )
-{
-   struct brw_reg reg =  brw_vec1_grf(0, 0);
-   reg.subnr = subnr;
-   reg.address_mode = BRW_ADDRESS_REGISTER_INDIRECT_REGISTER;
-   reg.dw1.bits.indirect_offset = offset;
-   return reg;
-}
-
-static __inline struct brw_reg deref_4f(struct brw_indirect ptr, int offset)
-{
-   return brw_vec4_indirect(ptr.addr_subnr, ptr.addr_offset + offset);
-}
-
-static __inline struct brw_reg deref_1f(struct brw_indirect ptr, int offset)
-{
-   return brw_vec1_indirect(ptr.addr_subnr, ptr.addr_offset + offset);
-}
-
-static __inline struct brw_reg deref_4b(struct brw_indirect ptr, int offset)
-{
-   return retype(deref_4f(ptr, offset), BRW_REGISTER_TYPE_B);
-}
-
-static __inline struct brw_reg deref_1uw(struct brw_indirect ptr, int offset)
-{
-   return retype(deref_1f(ptr, offset), BRW_REGISTER_TYPE_UW);
-}
-
-static __inline struct brw_reg deref_1ud(struct brw_indirect ptr, int offset)
-{
-   return retype(deref_1f(ptr, offset), BRW_REGISTER_TYPE_UD);
-}
-
-static __inline struct brw_reg get_addr_reg(struct brw_indirect ptr)
-{
-   return brw_address_reg(ptr.addr_subnr);
-}
-
-static __inline struct brw_indirect brw_indirect_offset( struct brw_indirect ptr, int offset )
-{
-   ptr.addr_offset += offset;
-   return ptr;
-}
-
-static __inline struct brw_indirect brw_indirect( unsigned addr_subnr, int offset )
-{
-   struct brw_indirect ptr;
-   ptr.addr_subnr = addr_subnr;
-   ptr.addr_offset = offset;
-   ptr.pad = 0;
-   return ptr;
-}
-
-static __inline struct brw_instruction *current_insn( struct brw_compile *p)
-{
-	return &p->store[p->nr_insn];
-}
-
-void brw_pop_insn_state( struct brw_compile *p );
-void brw_push_insn_state( struct brw_compile *p );
-void brw_set_mask_control( struct brw_compile *p, unsigned value );
-void brw_set_saturate( struct brw_compile *p, unsigned value );
-void brw_set_access_mode( struct brw_compile *p, unsigned access_mode );
-void brw_set_compression_control( struct brw_compile *p, boolean control );
-void brw_set_predicate_control_flag_value( struct brw_compile *p, unsigned value );
-void brw_set_predicate_control( struct brw_compile *p, unsigned pc );
-void brw_set_conditionalmod( struct brw_compile *p, unsigned conditional );
-
-void brw_init_compile( struct brw_compile *p );
-const unsigned *brw_get_program( struct brw_compile *p, unsigned *sz );
-
-
-struct brw_instruction *brw_alu1( struct brw_compile *p,
-				  unsigned opcode,
-				  struct brw_reg dest,
-				  struct brw_reg src );
-
-struct brw_instruction *brw_alu2(struct brw_compile *p,
-				 unsigned opcode,
-				 struct brw_reg dest,
-				 struct brw_reg src0,
-				 struct brw_reg src1 );
-
-/* Helpers for regular instructions:
- */
-#define ALU1(OP)					\
-struct brw_instruction *brw_##OP(struct brw_compile *p,	\
-	      struct brw_reg dest,			\
-	      struct brw_reg src0);
-
-#define ALU2(OP)					\
-struct brw_instruction *brw_##OP(struct brw_compile *p,	\
-	      struct brw_reg dest,			\
-	      struct brw_reg src0,			\
-	      struct brw_reg src1);
-
-ALU1(MOV)
-ALU2(SEL)
-ALU1(NOT)
-ALU2(AND)
-ALU2(OR)
-ALU2(XOR)
-ALU2(SHR)
-ALU2(SHL)
-ALU2(RSR)
-ALU2(RSL)
-ALU2(ASR)
-ALU2(JMPI)
-ALU2(ADD)
-ALU2(MUL)
-ALU1(FRC)
-ALU1(RNDD)
-ALU2(MAC)
-ALU2(MACH)
-ALU1(LZD)
-ALU2(DP4)
-ALU2(DPH)
-ALU2(DP3)
-ALU2(DP2)
-ALU2(LINE)
-
-#undef ALU1
-#undef ALU2
-
-
-
-/* Helpers for SEND instruction:
- */
-void brw_urb_WRITE(struct brw_compile *p,
-		   struct brw_reg dest,
-		   unsigned msg_reg_nr,
-		   struct brw_reg src0,
-		   boolean allocate,
-		   boolean used,
-		   unsigned msg_length,
-		   unsigned response_length,
-		   boolean eot,
-		   boolean writes_complete,
-		   unsigned offset,
-		   unsigned swizzle);
-
-void brw_fb_WRITE(struct brw_compile *p,
-		   struct brw_reg dest,
-		   unsigned msg_reg_nr,
-		   struct brw_reg src0,
-		   unsigned binding_table_index,
-		   unsigned msg_length,
-		   unsigned response_length,
-		   boolean eot);
-
-void brw_SAMPLE(struct brw_compile *p,
-		struct brw_reg dest,
-		unsigned msg_reg_nr,
-		struct brw_reg src0,
-		unsigned binding_table_index,
-		unsigned sampler,
-		unsigned writemask,
-		unsigned msg_type,
-		unsigned response_length,
-		unsigned msg_length,
-		boolean eot);
-
-void brw_math_16( struct brw_compile *p,
-		  struct brw_reg dest,
-		  unsigned function,
-		  unsigned saturate,
-		  unsigned msg_reg_nr,
-		  struct brw_reg src,
-		  unsigned precision );
-
-void brw_math( struct brw_compile *p,
-	       struct brw_reg dest,
-	       unsigned function,
-	       unsigned saturate,
-	       unsigned msg_reg_nr,
-	       struct brw_reg src,
-	       unsigned data_type,
-	       unsigned precision );
-
-void brw_dp_READ_16( struct brw_compile *p,
-		     struct brw_reg dest,
-		     unsigned msg_reg_nr,
-		     unsigned scratch_offset );
-
-void brw_dp_WRITE_16( struct brw_compile *p,
-		      struct brw_reg src,
-		      unsigned msg_reg_nr,
-		      unsigned scratch_offset );
-
-/* If/else/endif.  Works by manipulating the execution flags on each
- * channel.
- */
-struct brw_instruction *brw_IF(struct brw_compile *p,
-			       unsigned execute_size);
-
-struct brw_instruction *brw_ELSE(struct brw_compile *p,
-				 struct brw_instruction *if_insn);
-
-void brw_ENDIF(struct brw_compile *p,
-	       struct brw_instruction *if_or_else_insn);
-
-
-/* DO/WHILE loops:
- */
-struct brw_instruction *brw_DO(struct brw_compile *p,
-			       unsigned execute_size);
-
-struct brw_instruction *brw_WHILE(struct brw_compile *p,
-	       struct brw_instruction *patch_insn);
-
-struct brw_instruction *brw_BREAK(struct brw_compile *p);
-struct brw_instruction *brw_CONT(struct brw_compile *p);
-/* Forward jumps:
- */
-void brw_land_fwd_jump(struct brw_compile *p,
-		       struct brw_instruction *jmp_insn);
-
-
-
-void brw_NOP(struct brw_compile *p);
-
-/* Special case: there is never a destination, execution size will be
- * taken from src0:
- */
-void brw_CMP(struct brw_compile *p,
-	     struct brw_reg dest,
-	     unsigned conditional,
-	     struct brw_reg src0,
-	     struct brw_reg src1);
-
-void brw_print_reg( struct brw_reg reg );
-
-
-/***********************************************************************
- * brw_eu_util.c:
- */
-
-void brw_copy_indirect_to_indirect(struct brw_compile *p,
-				   struct brw_indirect dst_ptr,
-				   struct brw_indirect src_ptr,
-				   unsigned count);
-
-void brw_copy_from_indirect(struct brw_compile *p,
-			    struct brw_reg dst,
-			    struct brw_indirect ptr,
-			    unsigned count);
-
-void brw_copy4(struct brw_compile *p,
-	       struct brw_reg dst,
-	       struct brw_reg src,
-	       unsigned count);
-
-void brw_copy8(struct brw_compile *p,
-	       struct brw_reg dst,
-	       struct brw_reg src,
-	       unsigned count);
-
-void brw_math_invert( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg src);
-
-void brw_set_src1( struct brw_instruction *insn,
-                          struct brw_reg reg );
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_eu_debug.c b/src/gallium/drivers/i965simple/brw_eu_debug.c
deleted file mode 100644
index 4adfb0c02f..0000000000
--- a/src/gallium/drivers/i965simple/brw_eu_debug.c
+++ /dev/null
@@ -1,90 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-    
-
-#include "util/u_debug.h"
-
-#include "brw_eu.h"
-
-void brw_print_reg( struct brw_reg hwreg )
-{
-   static const char *file[] = {
-      "arf",
-      "grf",
-      "msg",
-      "imm"
-   };
-
-   static const char *type[] = {
-      "ud",
-      "d",
-      "uw",
-      "w",
-      "ub",
-      "vf",
-      "hf",
-      "f"
-   };
-
-   debug_printf("%s%s", 
-		hwreg.abs ? "abs/" : "",
-		hwreg.negate ? "-" : "");
-     
-   if (hwreg.file == BRW_GENERAL_REGISTER_FILE &&
-       hwreg.nr % 2 == 0 &&
-       hwreg.subnr == 0 &&
-       hwreg.vstride == BRW_VERTICAL_STRIDE_8 &&
-       hwreg.width == BRW_WIDTH_8 &&
-       hwreg.hstride == BRW_HORIZONTAL_STRIDE_1 &&
-       hwreg.type == BRW_REGISTER_TYPE_F) {
-      debug_printf("vec%d", hwreg.nr);
-   }
-   else if (hwreg.file == BRW_GENERAL_REGISTER_FILE &&
-	    hwreg.vstride == BRW_VERTICAL_STRIDE_0 &&
-	    hwreg.width == BRW_WIDTH_1 &&
-	    hwreg.hstride == BRW_HORIZONTAL_STRIDE_0 &&
-	    hwreg.type == BRW_REGISTER_TYPE_F) {      
-      debug_printf("scl%d.%d", hwreg.nr, hwreg.subnr / 4);
-   }
-   else {
-      debug_printf("%s%d.%d<%d;%d,%d>:%s", 
-		   file[hwreg.file],
-		   hwreg.nr,
-		   hwreg.subnr / type_sz(hwreg.type),
-		   hwreg.vstride ? (1<<(hwreg.vstride-1)) : 0,
-		   1<<hwreg.width,
-		   hwreg.hstride ? (1<<(hwreg.hstride-1)) : 0,		
-		   type[hwreg.type]);
-   }
-}
-
-
-
diff --git a/src/gallium/drivers/i965simple/brw_eu_emit.c b/src/gallium/drivers/i965simple/brw_eu_emit.c
deleted file mode 100644
index 400a80b6fb..0000000000
--- a/src/gallium/drivers/i965simple/brw_eu_emit.c
+++ /dev/null
@@ -1,1080 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_eu.h"
-
-
-
-
-/***********************************************************************
- * Internal helper for constructing instructions
- */
-
-static void guess_execution_size( struct brw_instruction *insn,
-				  struct brw_reg reg )
-{
-   if (reg.width == BRW_WIDTH_8 &&
-       insn->header.compression_control == BRW_COMPRESSION_COMPRESSED)
-      insn->header.execution_size = BRW_EXECUTE_16;
-   else
-      insn->header.execution_size = reg.width;	/* note - definitions are compatible */
-}
-
-
-static void brw_set_dest( struct brw_instruction *insn,
-			  struct brw_reg dest )
-{
-   insn->bits1.da1.dest_reg_file = dest.file;
-   insn->bits1.da1.dest_reg_type = dest.type;
-   insn->bits1.da1.dest_address_mode = dest.address_mode;
-
-   if (dest.address_mode == BRW_ADDRESS_DIRECT) {
-      insn->bits1.da1.dest_reg_nr = dest.nr;
-
-      if (insn->header.access_mode == BRW_ALIGN_1) {
-	 insn->bits1.da1.dest_subreg_nr = dest.subnr;
-	 insn->bits1.da1.dest_horiz_stride = BRW_HORIZONTAL_STRIDE_1;
-      }
-      else {
-	 insn->bits1.da16.dest_subreg_nr = dest.subnr / 16;
-	 insn->bits1.da16.dest_writemask = dest.dw1.bits.writemask;
-      }
-   }
-   else {
-      insn->bits1.ia1.dest_subreg_nr = dest.subnr;
-
-      /* These are different sizes in align1 vs align16:
-       */
-      if (insn->header.access_mode == BRW_ALIGN_1) {
-	 insn->bits1.ia1.dest_indirect_offset = dest.dw1.bits.indirect_offset;
-	 insn->bits1.ia1.dest_horiz_stride = BRW_HORIZONTAL_STRIDE_1;
-      }
-      else {
-	 insn->bits1.ia16.dest_indirect_offset = dest.dw1.bits.indirect_offset;
-      }
-   }
-
-   /* NEW: Set the execution size based on dest.width and
-    * insn->compression_control:
-    */
-   guess_execution_size(insn, dest);
-}
-
-static void brw_set_src0( struct brw_instruction *insn,
-		      struct brw_reg reg )
-{
-   assert(reg.file != BRW_MESSAGE_REGISTER_FILE);
-
-   insn->bits1.da1.src0_reg_file = reg.file;
-   insn->bits1.da1.src0_reg_type = reg.type;
-   insn->bits2.da1.src0_abs = reg.abs;
-   insn->bits2.da1.src0_negate = reg.negate;
-   insn->bits2.da1.src0_address_mode = reg.address_mode;
-
-   if (reg.file == BRW_IMMEDIATE_VALUE) {
-      insn->bits3.ud = reg.dw1.ud;
-
-      /* Required to set some fields in src1 as well:
-       */
-      insn->bits1.da1.src1_reg_file = 0; /* arf */
-      insn->bits1.da1.src1_reg_type = reg.type;
-   }
-   else
-   {
-      if (reg.address_mode == BRW_ADDRESS_DIRECT) {
-	 if (insn->header.access_mode == BRW_ALIGN_1) {
-	    insn->bits2.da1.src0_subreg_nr = reg.subnr;
-	    insn->bits2.da1.src0_reg_nr = reg.nr;
-	 }
-	 else {
-	    insn->bits2.da16.src0_subreg_nr = reg.subnr / 16;
-	    insn->bits2.da16.src0_reg_nr = reg.nr;
-	 }
-      }
-      else {
-	 insn->bits2.ia1.src0_subreg_nr = reg.subnr;
-
-	 if (insn->header.access_mode == BRW_ALIGN_1) {
-	    insn->bits2.ia1.src0_indirect_offset = reg.dw1.bits.indirect_offset;
-	 }
-	 else {
-	    insn->bits2.ia16.src0_subreg_nr = reg.dw1.bits.indirect_offset;
-	 }
-      }
-
-      if (insn->header.access_mode == BRW_ALIGN_1) {
-	 if (reg.width == BRW_WIDTH_1 &&
-	     insn->header.execution_size == BRW_EXECUTE_1) {
-	    insn->bits2.da1.src0_horiz_stride = BRW_HORIZONTAL_STRIDE_0;
-	    insn->bits2.da1.src0_width = BRW_WIDTH_1;
-	    insn->bits2.da1.src0_vert_stride = BRW_VERTICAL_STRIDE_0;
-	 }
-	 else {
-	    insn->bits2.da1.src0_horiz_stride = reg.hstride;
-	    insn->bits2.da1.src0_width = reg.width;
-	    insn->bits2.da1.src0_vert_stride = reg.vstride;
-	 }
-      }
-      else {
-	 insn->bits2.da16.src0_swz_x = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_X);
-	 insn->bits2.da16.src0_swz_y = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_Y);
-	 insn->bits2.da16.src0_swz_z = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_Z);
-	 insn->bits2.da16.src0_swz_w = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_W);
-
-	 /* This is an oddity of the fact we're using the same
-	  * descriptions for registers in align_16 as align_1:
-	  */
-	 if (reg.vstride == BRW_VERTICAL_STRIDE_8)
-	    insn->bits2.da16.src0_vert_stride = BRW_VERTICAL_STRIDE_4;
-	 else
-	    insn->bits2.da16.src0_vert_stride = reg.vstride;
-      }
-   }
-}
-
-
-void brw_set_src1( struct brw_instruction *insn,
-			  struct brw_reg reg )
-{
-   assert(reg.file != BRW_MESSAGE_REGISTER_FILE);
-
-   insn->bits1.da1.src1_reg_file = reg.file;
-   insn->bits1.da1.src1_reg_type = reg.type;
-   insn->bits3.da1.src1_abs = reg.abs;
-   insn->bits3.da1.src1_negate = reg.negate;
-
-   /* Only src1 can be immediate in two-argument instructions.
-    */
-   assert(insn->bits1.da1.src0_reg_file != BRW_IMMEDIATE_VALUE);
-
-   if (reg.file == BRW_IMMEDIATE_VALUE) {
-      insn->bits3.ud = reg.dw1.ud;
-   }
-   else {
-      /* This is a hardware restriction, which may or may not be lifted
-       * in the future:
-       */
-      assert (reg.address_mode == BRW_ADDRESS_DIRECT);
-      //assert (reg.file == BRW_GENERAL_REGISTER_FILE);
-
-      if (insn->header.access_mode == BRW_ALIGN_1) {
-	 insn->bits3.da1.src1_subreg_nr = reg.subnr;
-	 insn->bits3.da1.src1_reg_nr = reg.nr;
-      }
-      else {
-	 insn->bits3.da16.src1_subreg_nr = reg.subnr / 16;
-	 insn->bits3.da16.src1_reg_nr = reg.nr;
-      }
-
-      if (insn->header.access_mode == BRW_ALIGN_1) {
-	 if (reg.width == BRW_WIDTH_1 &&
-	     insn->header.execution_size == BRW_EXECUTE_1) {
-	    insn->bits3.da1.src1_horiz_stride = BRW_HORIZONTAL_STRIDE_0;
-	    insn->bits3.da1.src1_width = BRW_WIDTH_1;
-	    insn->bits3.da1.src1_vert_stride = BRW_VERTICAL_STRIDE_0;
-	 }
-	 else {
-	    insn->bits3.da1.src1_horiz_stride = reg.hstride;
-	    insn->bits3.da1.src1_width = reg.width;
-	    insn->bits3.da1.src1_vert_stride = reg.vstride;
-	 }
-      }
-      else {
-	 insn->bits3.da16.src1_swz_x = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_X);
-	 insn->bits3.da16.src1_swz_y = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_Y);
-	 insn->bits3.da16.src1_swz_z = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_Z);
-	 insn->bits3.da16.src1_swz_w = BRW_GET_SWZ(reg.dw1.bits.swizzle, BRW_CHANNEL_W);
-
-	 /* This is an oddity of the fact we're using the same
-	  * descriptions for registers in align_16 as align_1:
-	  */
-	 if (reg.vstride == BRW_VERTICAL_STRIDE_8)
-	    insn->bits3.da16.src1_vert_stride = BRW_VERTICAL_STRIDE_4;
-	 else
-	    insn->bits3.da16.src1_vert_stride = reg.vstride;
-      }
-   }
-}
-
-
-
-static void brw_set_math_message( struct brw_instruction *insn,
-				  unsigned msg_length,
-				  unsigned response_length,
-				  unsigned function,
-				  unsigned integer_type,
-				  boolean low_precision,
-				  boolean saturate,
-				  unsigned dataType )
-{
-   brw_set_src1(insn, brw_imm_d(0));
-
-   insn->bits3.math.function = function;
-   insn->bits3.math.int_type = integer_type;
-   insn->bits3.math.precision = low_precision;
-   insn->bits3.math.saturate = saturate;
-   insn->bits3.math.data_type = dataType;
-   insn->bits3.math.response_length = response_length;
-   insn->bits3.math.msg_length = msg_length;
-   insn->bits3.math.msg_target = BRW_MESSAGE_TARGET_MATH;
-   insn->bits3.math.end_of_thread = 0;
-}
-
-static void brw_set_urb_message( struct brw_instruction *insn,
-				 boolean allocate,
-				 boolean used,
-				 unsigned msg_length,
-				 unsigned response_length,
-				 boolean end_of_thread,
-				 boolean complete,
-				 unsigned offset,
-				 unsigned swizzle_control )
-{
-   brw_set_src1(insn, brw_imm_d(0));
-
-   insn->bits3.urb.opcode = 0;	/* ? */
-   insn->bits3.urb.offset = offset;
-   insn->bits3.urb.swizzle_control = swizzle_control;
-   insn->bits3.urb.allocate = allocate;
-   insn->bits3.urb.used = used;	/* ? */
-   insn->bits3.urb.complete = complete;
-   insn->bits3.urb.response_length = response_length;
-   insn->bits3.urb.msg_length = msg_length;
-   insn->bits3.urb.msg_target = BRW_MESSAGE_TARGET_URB;
-   insn->bits3.urb.end_of_thread = end_of_thread;
-}
-
-static void brw_set_dp_write_message( struct brw_instruction *insn,
-				      unsigned binding_table_index,
-				      unsigned msg_control,
-				      unsigned msg_type,
-				      unsigned msg_length,
-				      unsigned pixel_scoreboard_clear,
-				      unsigned response_length,
-				      unsigned end_of_thread )
-{
-   brw_set_src1(insn, brw_imm_d(0));
-
-   insn->bits3.dp_write.binding_table_index = binding_table_index;
-   insn->bits3.dp_write.msg_control = msg_control;
-   insn->bits3.dp_write.pixel_scoreboard_clear = pixel_scoreboard_clear;
-   insn->bits3.dp_write.msg_type = msg_type;
-   insn->bits3.dp_write.send_commit_msg = 0;
-   insn->bits3.dp_write.response_length = response_length;
-   insn->bits3.dp_write.msg_length = msg_length;
-   insn->bits3.dp_write.msg_target = BRW_MESSAGE_TARGET_DATAPORT_WRITE;
-   insn->bits3.urb.end_of_thread = end_of_thread;
-}
-
-static void brw_set_dp_read_message( struct brw_instruction *insn,
-				      unsigned binding_table_index,
-				      unsigned msg_control,
-				      unsigned msg_type,
-				      unsigned target_cache,
-				      unsigned msg_length,
-				      unsigned response_length,
-				      unsigned end_of_thread )
-{
-   brw_set_src1(insn, brw_imm_d(0));
-
-   insn->bits3.dp_read.binding_table_index = binding_table_index;
-   insn->bits3.dp_read.msg_control = msg_control;
-   insn->bits3.dp_read.msg_type = msg_type;
-   insn->bits3.dp_read.target_cache = target_cache;
-   insn->bits3.dp_read.response_length = response_length;
-   insn->bits3.dp_read.msg_length = msg_length;
-   insn->bits3.dp_read.msg_target = BRW_MESSAGE_TARGET_DATAPORT_READ;
-   insn->bits3.dp_read.end_of_thread = end_of_thread;
-}
-
-static void brw_set_sampler_message( struct brw_instruction *insn,
-				     unsigned binding_table_index,
-				     unsigned sampler,
-				     unsigned msg_type,
-				     unsigned response_length,
-				     unsigned msg_length,
-				     boolean eot)
-{
-   brw_set_src1(insn, brw_imm_d(0));
-
-   insn->bits3.sampler.binding_table_index = binding_table_index;
-   insn->bits3.sampler.sampler = sampler;
-   insn->bits3.sampler.msg_type = msg_type;
-   insn->bits3.sampler.return_format = BRW_SAMPLER_RETURN_FORMAT_FLOAT32;
-   insn->bits3.sampler.response_length = response_length;
-   insn->bits3.sampler.msg_length = msg_length;
-   insn->bits3.sampler.end_of_thread = eot;
-   insn->bits3.sampler.msg_target = BRW_MESSAGE_TARGET_SAMPLER;
-}
-
-
-
-static struct brw_instruction *next_insn( struct brw_compile *p,
-					  unsigned opcode )
-{
-   struct brw_instruction *insn;
-
-   assert(p->nr_insn + 1 < BRW_EU_MAX_INSN);
-
-   insn = &p->store[p->nr_insn++];
-   memcpy(insn, p->current, sizeof(*insn));
-
-   /* Reset this one-shot flag:
-    */
-
-   if (p->current->header.destreg__conditonalmod) {
-      p->current->header.destreg__conditonalmod = 0;
-      p->current->header.predicate_control = BRW_PREDICATE_NORMAL;
-   }
-
-   insn->header.opcode = opcode;
-   return insn;
-}
-
-
-struct brw_instruction *brw_alu1( struct brw_compile *p,
-				  unsigned opcode,
-				  struct brw_reg dest,
-				  struct brw_reg src )
-{
-   struct brw_instruction *insn = next_insn(p, opcode);
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src);
-   return insn;
-}
-
-struct brw_instruction *brw_alu2(struct brw_compile *p,
-				 unsigned opcode,
-				 struct brw_reg dest,
-				 struct brw_reg src0,
-				 struct brw_reg src1 )
-{
-   struct brw_instruction *insn = next_insn(p, opcode);
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src0);
-   brw_set_src1(insn, src1);
-   return insn;
-}
-
-
-/***********************************************************************
- * Convenience routines.
- */
-#define ALU1(OP)					\
-struct brw_instruction *brw_##OP(struct brw_compile *p,			\
-	      struct brw_reg dest,			\
-	      struct brw_reg src0)   			\
-{							\
-   return brw_alu1(p, BRW_OPCODE_##OP, dest, src0);    	\
-}
-
-#define ALU2(OP)					\
-struct brw_instruction *brw_##OP(struct brw_compile *p,			\
-	      struct brw_reg dest,			\
-	      struct brw_reg src0,			\
-	      struct brw_reg src1)   			\
-{							\
-   return brw_alu2(p, BRW_OPCODE_##OP, dest, src0, src1);	\
-}
-
-
-ALU1(MOV)
-ALU2(SEL)
-ALU1(NOT)
-ALU2(AND)
-ALU2(OR)
-ALU2(XOR)
-ALU2(SHR)
-ALU2(SHL)
-ALU2(RSR)
-ALU2(RSL)
-ALU2(ASR)
-ALU2(ADD)
-ALU2(MUL)
-ALU1(FRC)
-ALU1(RNDD)
-ALU2(MAC)
-ALU2(MACH)
-ALU1(LZD)
-ALU2(DP4)
-ALU2(DPH)
-ALU2(DP3)
-ALU2(DP2)
-ALU2(LINE)
-
-
-
-
-void brw_NOP(struct brw_compile *p)
-{
-   struct brw_instruction *insn = next_insn(p, BRW_OPCODE_NOP);
-   brw_set_dest(insn, retype(brw_vec4_grf(0,0), BRW_REGISTER_TYPE_UD));
-   brw_set_src0(insn, retype(brw_vec4_grf(0,0), BRW_REGISTER_TYPE_UD));
-   brw_set_src1(insn, brw_imm_ud(0x0));
-}
-
-
-
-
-
-/***********************************************************************
- * Comparisons, if/else/endif
- */
-
-struct brw_instruction *brw_JMPI(struct brw_compile *p,
-	      struct brw_reg dest,
-	      struct brw_reg src0,
-	      struct brw_reg src1)
-{
-   struct brw_instruction *insn = brw_alu2(p, BRW_OPCODE_JMPI, dest, src0, src1);
-
-   p->current->header.predicate_control = BRW_PREDICATE_NONE;
-
-   return insn;
-}
-
-/* EU takes the value from the flag register and pushes it onto some
- * sort of a stack (presumably merging with any flag value already on
- * the stack).  Within an if block, the flags at the top of the stack
- * control execution on each channel of the unit, eg. on each of the
- * 16 pixel values in our wm programs.
- *
- * When the matching 'else' instruction is reached (presumably by
- * countdown of the instruction count patched in by our ELSE/ENDIF
- * functions), the relevent flags are inverted.
- *
- * When the matching 'endif' instruction is reached, the flags are
- * popped off.  If the stack is now empty, normal execution resumes.
- *
- * No attempt is made to deal with stack overflow (14 elements?).
- */
-struct brw_instruction *brw_IF(struct brw_compile *p, unsigned execute_size)
-{
-   struct brw_instruction *insn;
-
-   if (p->single_program_flow) {
-      assert(execute_size == BRW_EXECUTE_1);
-
-      insn = next_insn(p, BRW_OPCODE_ADD);
-      insn->header.predicate_inverse = 1;
-   } else {
-      insn = next_insn(p, BRW_OPCODE_IF);
-   }
-
-   /* Override the defaults for this instruction:
-    */
-   brw_set_dest(insn, brw_ip_reg());
-   brw_set_src0(insn, brw_ip_reg());
-   brw_set_src1(insn, brw_imm_d(0x0));
-
-   insn->header.execution_size = execute_size;
-   insn->header.compression_control = BRW_COMPRESSION_NONE;
-   insn->header.predicate_control = BRW_PREDICATE_NORMAL;
-   insn->header.mask_control = BRW_MASK_ENABLE;
-
-   p->current->header.predicate_control = BRW_PREDICATE_NONE;
-
-   return insn;
-}
-
-
-struct brw_instruction *brw_ELSE(struct brw_compile *p,
-				 struct brw_instruction *if_insn)
-{
-   struct brw_instruction *insn;
-
-   if (p->single_program_flow) {
-      insn = next_insn(p, BRW_OPCODE_ADD);
-   } else {
-      insn = next_insn(p, BRW_OPCODE_ELSE);
-   }
-
-   brw_set_dest(insn, brw_ip_reg());
-   brw_set_src0(insn, brw_ip_reg());
-   brw_set_src1(insn, brw_imm_d(0x0));
-
-   insn->header.compression_control = BRW_COMPRESSION_NONE;
-   insn->header.execution_size = if_insn->header.execution_size;
-   insn->header.mask_control = BRW_MASK_ENABLE;
-
-   /* Patch the if instruction to point at this instruction.
-    */
-   if (p->single_program_flow) {
-      assert(if_insn->header.opcode == BRW_OPCODE_ADD);
-
-      if_insn->bits3.ud = (insn - if_insn + 1) * 16;
-   } else {
-      assert(if_insn->header.opcode == BRW_OPCODE_IF);
-
-      if_insn->bits3.if_else.jump_count = insn - if_insn;
-      if_insn->bits3.if_else.pop_count = 1;
-      if_insn->bits3.if_else.pad0 = 0;
-   }
-
-   return insn;
-}
-
-void brw_ENDIF(struct brw_compile *p,
-	       struct brw_instruction *patch_insn)
-{
-   if (p->single_program_flow) {
-      /* In single program flow mode, there's no need to execute an ENDIF,
-       * since we don't need to do any stack operations, and if we're executing
-       * currently, we want to just continue executing.
-       */
-      struct brw_instruction *next = &p->store[p->nr_insn];
-
-      assert(patch_insn->header.opcode == BRW_OPCODE_ADD);
-
-      patch_insn->bits3.ud = (next - patch_insn) * 16;
-   } else {
-      struct brw_instruction *insn = next_insn(p, BRW_OPCODE_ENDIF);
-
-      brw_set_dest(insn, retype(brw_vec4_grf(0,0), BRW_REGISTER_TYPE_UD));
-      brw_set_src0(insn, retype(brw_vec4_grf(0,0), BRW_REGISTER_TYPE_UD));
-      brw_set_src1(insn, brw_imm_d(0x0));
-
-      insn->header.compression_control = BRW_COMPRESSION_NONE;
-      insn->header.execution_size = patch_insn->header.execution_size;
-      insn->header.mask_control = BRW_MASK_ENABLE;
-
-      assert(patch_insn->bits3.if_else.jump_count == 0);
-
-      /* Patch the if or else instructions to point at this or the next
-       * instruction respectively.
-       */
-      if (patch_insn->header.opcode == BRW_OPCODE_IF) {
-	 /* Automagically turn it into an IFF:
-	  */
-	 patch_insn->header.opcode = BRW_OPCODE_IFF;
-	 patch_insn->bits3.if_else.jump_count = insn - patch_insn + 1;
-	 patch_insn->bits3.if_else.pop_count = 0;
-	 patch_insn->bits3.if_else.pad0 = 0;
-      } else if (patch_insn->header.opcode == BRW_OPCODE_ELSE) {
-	 patch_insn->bits3.if_else.jump_count = insn - patch_insn + 1;
-	 patch_insn->bits3.if_else.pop_count = 1;
-	 patch_insn->bits3.if_else.pad0 = 0;
-      } else {
-	 assert(0);
-      }
-
-      /* Also pop item off the stack in the endif instruction:
-       */
-      insn->bits3.if_else.jump_count = 0;
-      insn->bits3.if_else.pop_count = 1;
-      insn->bits3.if_else.pad0 = 0;
-   }
-}
-
-struct brw_instruction *brw_BREAK(struct brw_compile *p)
-{
-   struct brw_instruction *insn;
-   insn = next_insn(p, BRW_OPCODE_BREAK);
-   brw_set_dest(insn, brw_ip_reg());
-   brw_set_src0(insn, brw_ip_reg());
-   brw_set_src1(insn, brw_imm_d(0x0));
-   insn->header.compression_control = BRW_COMPRESSION_NONE;
-   insn->header.execution_size = BRW_EXECUTE_8;
-   insn->header.mask_control = BRW_MASK_DISABLE;
-   insn->bits3.if_else.pad0 = 0;
-   return insn;
-}
-
-struct brw_instruction *brw_CONT(struct brw_compile *p)
-{
-   struct brw_instruction *insn;
-   insn = next_insn(p, BRW_OPCODE_CONTINUE);
-   brw_set_dest(insn, brw_ip_reg());
-   brw_set_src0(insn, brw_ip_reg());
-   brw_set_src1(insn, brw_imm_d(0x0));
-   insn->header.compression_control = BRW_COMPRESSION_NONE;
-   insn->header.execution_size = BRW_EXECUTE_8;
-   insn->header.mask_control = BRW_MASK_DISABLE;
-   insn->bits3.if_else.pad0 = 0;
-   return insn;
-}
-
-/* DO/WHILE loop:
- */
-struct brw_instruction *brw_DO(struct brw_compile *p, unsigned execute_size)
-{
-   if (p->single_program_flow) {
-      return &p->store[p->nr_insn];
-   } else {
-      struct brw_instruction *insn = next_insn(p, BRW_OPCODE_DO);
-
-      /* Override the defaults for this instruction:
-       */
-      brw_set_dest(insn, brw_null_reg());
-      brw_set_src0(insn, brw_null_reg());
-      brw_set_src1(insn, brw_null_reg());
-
-      insn->header.compression_control = BRW_COMPRESSION_NONE;
-      insn->header.execution_size = execute_size;
-      insn->header.predicate_control = BRW_PREDICATE_NONE;
-      /* insn->header.mask_control = BRW_MASK_ENABLE; */
-      insn->header.mask_control = BRW_MASK_DISABLE;
-
-      return insn;
-   }
-}
-
-
-
-struct brw_instruction *brw_WHILE(struct brw_compile *p,
-	       struct brw_instruction *do_insn)
-{
-   struct brw_instruction *insn;
-
-   if (p->single_program_flow)
-      insn = next_insn(p, BRW_OPCODE_ADD);
-   else
-      insn = next_insn(p, BRW_OPCODE_WHILE);
-
-   brw_set_dest(insn, brw_ip_reg());
-   brw_set_src0(insn, brw_ip_reg());
-   brw_set_src1(insn, brw_imm_d(0x0));
-
-   insn->header.compression_control = BRW_COMPRESSION_NONE;
-
-   if (p->single_program_flow) {
-      insn->header.execution_size = BRW_EXECUTE_1;
-
-      insn->bits3.d = (do_insn - insn) * 16;
-   } else {
-      insn->header.execution_size = do_insn->header.execution_size;
-
-      assert(do_insn->header.opcode == BRW_OPCODE_DO);
-      insn->bits3.if_else.jump_count = do_insn - insn;
-      insn->bits3.if_else.pop_count = 0;
-      insn->bits3.if_else.pad0 = 0;
-   }
-
-/*    insn->header.mask_control = BRW_MASK_ENABLE; */
-
-   insn->header.mask_control = BRW_MASK_DISABLE;
-   p->current->header.predicate_control = BRW_PREDICATE_NONE;
-   return insn;
-}
-
-
-/* FORWARD JUMPS:
- */
-void brw_land_fwd_jump(struct brw_compile *p,
-		       struct brw_instruction *jmp_insn)
-{
-   struct brw_instruction *landing = &p->store[p->nr_insn];
-
-   assert(jmp_insn->header.opcode == BRW_OPCODE_JMPI);
-   assert(jmp_insn->bits1.da1.src1_reg_file = BRW_IMMEDIATE_VALUE);
-
-   jmp_insn->bits3.ud = (landing - jmp_insn) - 1;
-}
-
-
-
-/* To integrate with the above, it makes sense that the comparison
- * instruction should populate the flag register.  It might be simpler
- * just to use the flag reg for most WM tasks?
- */
-void brw_CMP(struct brw_compile *p,
-	     struct brw_reg dest,
-	     unsigned conditional,
-	     struct brw_reg src0,
-	     struct brw_reg src1)
-{
-   struct brw_instruction *insn = next_insn(p, BRW_OPCODE_CMP);
-
-   insn->header.destreg__conditonalmod = conditional;
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src0);
-   brw_set_src1(insn, src1);
-
-/*    guess_execution_size(insn, src0); */
-
-
-   /* Make it so that future instructions will use the computed flag
-    * value until brw_set_predicate_control_flag_value() is called
-    * again.
-    */
-   if (dest.file == BRW_ARCHITECTURE_REGISTER_FILE &&
-       dest.nr == 0) {
-      p->current->header.predicate_control = BRW_PREDICATE_NORMAL;
-      p->flag_value = 0xff;
-   }
-}
-
-
-
-/***********************************************************************
- * Helpers for the various SEND message types:
- */
-
-/* Invert 8 values
- */
-void brw_math( struct brw_compile *p,
-	       struct brw_reg dest,
-	       unsigned function,
-	       unsigned saturate,
-	       unsigned msg_reg_nr,
-	       struct brw_reg src,
-	       unsigned data_type,
-	       unsigned precision )
-{
-   struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND);
-   unsigned msg_length = (function == BRW_MATH_FUNCTION_POW) ? 2 : 1;
-   unsigned response_length = (function == BRW_MATH_FUNCTION_SINCOS) ? 2 : 1;
-
-   /* Example code doesn't set predicate_control for send
-    * instructions.
-    */
-   insn->header.predicate_control = 0;
-   insn->header.destreg__conditonalmod = msg_reg_nr;
-
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src);
-   brw_set_math_message(insn,
-			msg_length, response_length,
-			function,
-			BRW_MATH_INTEGER_UNSIGNED,
-			precision,
-			saturate,
-			data_type);
-}
-
-/* Use 2 send instructions to invert 16 elements
- */
-void brw_math_16( struct brw_compile *p,
-		  struct brw_reg dest,
-		  unsigned function,
-		  unsigned saturate,
-		  unsigned msg_reg_nr,
-		  struct brw_reg src,
-		  unsigned precision )
-{
-   struct brw_instruction *insn;
-   unsigned msg_length = (function == BRW_MATH_FUNCTION_POW) ? 2 : 1;
-   unsigned response_length = (function == BRW_MATH_FUNCTION_SINCOS) ? 2 : 1;
-
-   /* First instruction:
-    */
-   brw_push_insn_state(p);
-   brw_set_predicate_control_flag_value(p, 0xff);
-   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-
-   insn = next_insn(p, BRW_OPCODE_SEND);
-   insn->header.destreg__conditonalmod = msg_reg_nr;
-
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src);
-   brw_set_math_message(insn,
-			msg_length, response_length,
-			function,
-			BRW_MATH_INTEGER_UNSIGNED,
-			precision,
-			saturate,
-			BRW_MATH_DATA_VECTOR);
-
-   /* Second instruction:
-    */
-   insn = next_insn(p, BRW_OPCODE_SEND);
-   insn->header.compression_control = BRW_COMPRESSION_2NDHALF;
-   insn->header.destreg__conditonalmod = msg_reg_nr+1;
-
-   brw_set_dest(insn, offset(dest,1));
-   brw_set_src0(insn, src);
-   brw_set_math_message(insn,
-			msg_length, response_length,
-			function,
-			BRW_MATH_INTEGER_UNSIGNED,
-			precision,
-			saturate,
-			BRW_MATH_DATA_VECTOR);
-
-   brw_pop_insn_state(p);
-}
-
-
-
-
-void brw_dp_WRITE_16( struct brw_compile *p,
-		      struct brw_reg src,
-		      unsigned msg_reg_nr,
-		      unsigned scratch_offset )
-{
-   {
-      brw_push_insn_state(p);
-      brw_set_mask_control(p, BRW_MASK_DISABLE);
-      brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-
-      brw_MOV(p,
-	      retype(brw_vec1_grf(0, 2), BRW_REGISTER_TYPE_D),
-	      brw_imm_d(scratch_offset));
-
-      brw_pop_insn_state(p);
-   }
-
-   {
-      unsigned msg_length = 3;
-      struct brw_reg dest = retype(brw_null_reg(), BRW_REGISTER_TYPE_UW);
-      struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND);
-
-      insn->header.predicate_control = 0; /* XXX */
-      insn->header.compression_control = BRW_COMPRESSION_NONE;
-      insn->header.destreg__conditonalmod = msg_reg_nr;
-
-      brw_set_dest(insn, dest);
-      brw_set_src0(insn, src);
-
-      brw_set_dp_write_message(insn,
-			       255, /* bti */
-			       BRW_DATAPORT_OWORD_BLOCK_4_OWORDS, /* msg_control */
-			       BRW_DATAPORT_WRITE_MESSAGE_OWORD_BLOCK_WRITE, /* msg_type */
-			       msg_length,
-			       0, /* pixel scoreboard */
-			       0, /* response_length */
-			       0); /* eot */
-   }
-
-}
-
-
-void brw_dp_READ_16( struct brw_compile *p,
-		      struct brw_reg dest,
-		      unsigned msg_reg_nr,
-		      unsigned scratch_offset )
-{
-   {
-      brw_push_insn_state(p);
-      brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-      brw_set_mask_control(p, BRW_MASK_DISABLE);
-
-      brw_MOV(p,
-	      retype(brw_vec1_grf(0, 2), BRW_REGISTER_TYPE_D),
-	      brw_imm_d(scratch_offset));
-
-      brw_pop_insn_state(p);
-   }
-
-   {
-      struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND);
-
-      insn->header.predicate_control = 0; /* XXX */
-      insn->header.compression_control = BRW_COMPRESSION_NONE;
-      insn->header.destreg__conditonalmod = msg_reg_nr;
-
-      brw_set_dest(insn, dest);	/* UW? */
-      brw_set_src0(insn, retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW));
-
-      brw_set_dp_read_message(insn,
-			      255, /* bti */
-			      3,  /* msg_control */
-			      BRW_DATAPORT_READ_MESSAGE_OWORD_BLOCK_READ, /* msg_type */
-			      1, /* target cache */
-			      1, /* msg_length */
-			      2, /* response_length */
-			      0); /* eot */
-   }
-}
-
-
-void brw_fb_WRITE(struct brw_compile *p,
-		   struct brw_reg dest,
-		   unsigned msg_reg_nr,
-		   struct brw_reg src0,
-		   unsigned binding_table_index,
-		   unsigned msg_length,
-		   unsigned response_length,
-		   boolean eot)
-{
-   struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND);
-
-   insn->header.predicate_control = 0; /* XXX */
-   insn->header.compression_control = BRW_COMPRESSION_NONE;
-   insn->header.destreg__conditonalmod = msg_reg_nr;
-
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src0);
-   brw_set_dp_write_message(insn,
-			    binding_table_index,
-			    BRW_DATAPORT_RENDER_TARGET_WRITE_SIMD16_SINGLE_SOURCE, /* msg_control */
-			    BRW_DATAPORT_WRITE_MESSAGE_RENDER_TARGET_WRITE, /* msg_type */
-			    msg_length,
-			    1,	/* pixel scoreboard */
-			    response_length,
-			    eot);
-}
-
-
-
-void brw_SAMPLE(struct brw_compile *p,
-		struct brw_reg dest,
-		unsigned msg_reg_nr,
-		struct brw_reg src0,
-		unsigned binding_table_index,
-		unsigned sampler,
-		unsigned writemask,
-		unsigned msg_type,
-		unsigned response_length,
-		unsigned msg_length,
-		boolean eot)
-{
-   boolean need_stall = 0;
-
-   if(writemask == 0) {
-/*       debug_printf("%s: zero writemask??\n", __FUNCTION__); */
-      return;
-   }
-
-   /* Hardware doesn't do destination dependency checking on send
-    * instructions properly.  Add a workaround which generates the
-    * dependency by other means.  In practice it seems like this bug
-    * only crops up for texture samples, and only where registers are
-    * written by the send and then written again later without being
-    * read in between.  Luckily for us, we already track that
-    * information and use it to modify the writemask for the
-    * instruction, so that is a guide for whether a workaround is
-    * needed.
-    */
-   if (writemask != TGSI_WRITEMASK_XYZW) {
-      unsigned dst_offset = 0;
-      unsigned i, newmask = 0, len = 0;
-
-      for (i = 0; i < 4; i++) {
-	 if (writemask & (1<<i))
-	    break;
-	 dst_offset += 2;
-      }
-      for (; i < 4; i++) {
-	 if (!(writemask & (1<<i)))
-	    break;
-	 newmask |= 1<<i;
-	 len++;
-      }
-
-      if (newmask != writemask) {
-	 need_stall = 1;
-/* 	 debug_printf("need stall %x %x\n", newmask , writemask); */
-      }
-      else {
-	 struct brw_reg m1 = brw_message_reg(msg_reg_nr);
-
-	 newmask = ~newmask & TGSI_WRITEMASK_XYZW;
-
-	 brw_push_insn_state(p);
-
-	 brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-	 brw_set_mask_control(p, BRW_MASK_DISABLE);
-
-	 brw_MOV(p, m1, brw_vec8_grf(0,0));
-  	 brw_MOV(p, get_element_ud(m1, 2), brw_imm_ud(newmask << 12));
-
-	 brw_pop_insn_state(p);
-
-  	 src0 = retype(brw_null_reg(), BRW_REGISTER_TYPE_UW);
-	 dest = offset(dest, dst_offset);
-	 response_length = len * 2;
-      }
-   }
-
-   {
-      struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND);
-
-      insn->header.predicate_control = 0; /* XXX */
-      insn->header.compression_control = BRW_COMPRESSION_NONE;
-      insn->header.destreg__conditonalmod = msg_reg_nr;
-
-      brw_set_dest(insn, dest);
-      brw_set_src0(insn, src0);
-      brw_set_sampler_message(insn,
-			      binding_table_index,
-			      sampler,
-			      msg_type,
-			      response_length,
-			      msg_length,
-			      eot);
-   }
-
-   if (need_stall)
-   {
-      struct brw_reg reg = vec8(offset(dest, response_length-1));
-
-      /*  mov (8) r9.0<1>:f    r9.0<8;8,1>:f    { Align1 }
-       */
-      brw_push_insn_state(p);
-      brw_set_compression_control(p, FALSE);
-      brw_MOV(p, reg, reg);
-      brw_pop_insn_state(p);
-   }
-
-}
-
-/* All these variables are pretty confusing - we might be better off
- * using bitmasks and macros for this, in the old style.  Or perhaps
- * just having the caller instantiate the fields in dword3 itself.
- */
-void brw_urb_WRITE(struct brw_compile *p,
-		   struct brw_reg dest,
-		   unsigned msg_reg_nr,
-		   struct brw_reg src0,
-		   boolean allocate,
-		   boolean used,
-		   unsigned msg_length,
-		   unsigned response_length,
-		   boolean eot,
-		   boolean writes_complete,
-		   unsigned offset,
-		   unsigned swizzle)
-{
-   struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND);
-
-   assert(msg_length < 16);
-
-   brw_set_dest(insn, dest);
-   brw_set_src0(insn, src0);
-   brw_set_src1(insn, brw_imm_d(0));
-
-   insn->header.destreg__conditonalmod = msg_reg_nr;
-
-   brw_set_urb_message(insn,
-		       allocate,
-		       used,
-		       msg_length,
-		       response_length,
-		       eot,
-		       writes_complete,
-		       offset,
-		       swizzle);
-}
-
diff --git a/src/gallium/drivers/i965simple/brw_eu_util.c b/src/gallium/drivers/i965simple/brw_eu_util.c
deleted file mode 100644
index 3a65b141f0..0000000000
--- a/src/gallium/drivers/i965simple/brw_eu_util.c
+++ /dev/null
@@ -1,126 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-      
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_eu.h"
-
-
-void brw_math_invert( struct brw_compile *p, 
-			     struct brw_reg dst,
-			     struct brw_reg src)
-{
-   brw_math( p, 
-	     dst,
-	     BRW_MATH_FUNCTION_INV, 
-	     BRW_MATH_SATURATE_NONE,
-	     0,
-	     src,
-	     BRW_MATH_PRECISION_FULL, 
-	     BRW_MATH_DATA_VECTOR );
-}
-
-
-
-void brw_copy4(struct brw_compile *p,
-	       struct brw_reg dst,
-	       struct brw_reg src,
-	       unsigned count)
-{
-   unsigned i;
-
-   dst = vec4(dst);
-   src = vec4(src);
-
-   for (i = 0; i < count; i++)
-   {
-      unsigned delta = i*32;
-      brw_MOV(p, byte_offset(dst, delta),    byte_offset(src, delta));
-      brw_MOV(p, byte_offset(dst, delta+16), byte_offset(src, delta+16));
-   }
-}
-
-
-void brw_copy8(struct brw_compile *p,
-	       struct brw_reg dst,
-	       struct brw_reg src,
-	       unsigned count)
-{
-   unsigned i;
-
-   dst = vec8(dst);
-   src = vec8(src);
-
-   for (i = 0; i < count; i++)
-   {
-      unsigned delta = i*32;
-      brw_MOV(p, byte_offset(dst, delta),    byte_offset(src, delta));
-   }
-}
-
-
-void brw_copy_indirect_to_indirect(struct brw_compile *p,
-				   struct brw_indirect dst_ptr,
-				   struct brw_indirect src_ptr,
-				   unsigned count)
-{
-   unsigned i;
-
-   for (i = 0; i < count; i++)
-   {
-      unsigned delta = i*32;
-      brw_MOV(p, deref_4f(dst_ptr, delta),    deref_4f(src_ptr, delta));
-      brw_MOV(p, deref_4f(dst_ptr, delta+16), deref_4f(src_ptr, delta+16));
-   }
-}
-
-
-void brw_copy_from_indirect(struct brw_compile *p,
-			    struct brw_reg dst,
-			    struct brw_indirect ptr,
-			    unsigned count)
-{
-   unsigned i;
-
-   dst = vec4(dst);
-
-   for (i = 0; i < count; i++)
-   {
-      unsigned delta = i*32;
-      brw_MOV(p, byte_offset(dst, delta),    deref_4f(ptr, delta));
-      brw_MOV(p, byte_offset(dst, delta+16), deref_4f(ptr, delta+16));
-   }
-}
-
-
-
-
diff --git a/src/gallium/drivers/i965simple/brw_gs.c b/src/gallium/drivers/i965simple/brw_gs.c
deleted file mode 100644
index de60868ccc..0000000000
--- a/src/gallium/drivers/i965simple/brw_gs.c
+++ /dev/null
@@ -1,196 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_state.h"
-#include "brw_gs.h"
-
-
-
-static void compile_gs_prog( struct brw_context *brw,
-			     struct brw_gs_prog_key *key )
-{
-   struct brw_gs_compile c;
-   const unsigned *program;
-   unsigned program_size;
-
-   memset(&c, 0, sizeof(c));
-
-   c.key = *key;
-
-   /* Need to locate the two positions present in vertex + header.
-    * These are currently hardcoded:
-    */
-   c.nr_attrs = brw_count_bits(c.key.attrs);
-   c.nr_regs = (c.nr_attrs + 1) / 2 + 1;  /* are vertices packed, or reg-aligned? */
-   c.nr_bytes = c.nr_regs * REG_SIZE;
-
-
-   /* Begin the compilation:
-    */
-   brw_init_compile(&c.func);
-
-   c.func.single_program_flow = 1;
-
-   /* For some reason the thread is spawned with only 4 channels
-    * unmasked.
-    */
-   brw_set_mask_control(&c.func, BRW_MASK_DISABLE);
-
-
-   /* Note that primitives which don't require a GS program have
-    * already been weeded out by this stage:
-    */
-   switch (key->primitive) {
-   case PIPE_PRIM_QUADS:
-      brw_gs_quads( &c );
-      break;
-   case PIPE_PRIM_QUAD_STRIP:
-      brw_gs_quad_strip( &c );
-      break;
-   case PIPE_PRIM_LINE_LOOP:
-      brw_gs_lines( &c );
-      break;
-   case PIPE_PRIM_LINES:
-      if (key->hint_gs_always)
-	 brw_gs_lines( &c );
-      else {
-	 return;
-      }
-      break;
-   case PIPE_PRIM_TRIANGLES:
-      if (key->hint_gs_always)
-	 brw_gs_tris( &c );
-      else {
-	 return;
-      }
-      break;
-   case PIPE_PRIM_POINTS:
-      if (key->hint_gs_always)
-	 brw_gs_points( &c );
-      else {
-	 return;
-      }
-      break;
-   default:
-      return;
-   }
-
-   /* get the program
-    */
-   program = brw_get_program(&c.func, &program_size);
-
-   /* Upload
-    */
-   brw->gs.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_GS_PROG],
-					      &c.key,
-					      sizeof(c.key),
-					      program,
-					      program_size,
-					      &c.prog_data,
-					      &brw->gs.prog_data );
-}
-
-
-static boolean search_cache( struct brw_context *brw,
-			       struct brw_gs_prog_key *key )
-{
-   return brw_search_cache(&brw->cache[BRW_GS_PROG],
-			   key, sizeof(*key),
-			   &brw->gs.prog_data,
-			   &brw->gs.prog_gs_offset);
-}
-
-
-static const int gs_prim[PIPE_PRIM_POLYGON+1] = {
-   PIPE_PRIM_POINTS,
-   PIPE_PRIM_LINES,
-   PIPE_PRIM_LINE_LOOP,
-   PIPE_PRIM_LINES,
-   PIPE_PRIM_TRIANGLES,
-   PIPE_PRIM_TRIANGLES,
-   PIPE_PRIM_TRIANGLES,
-   PIPE_PRIM_QUADS,
-   PIPE_PRIM_QUAD_STRIP,
-   PIPE_PRIM_TRIANGLES
-};
-
-static void populate_key( struct brw_context *brw,
-			  struct brw_gs_prog_key *key )
-{
-   memset(key, 0, sizeof(*key));
-
-   /* CACHE_NEW_VS_PROG */
-   key->attrs = brw->vs.prog_data->outputs_written;
-
-   /* BRW_NEW_PRIMITIVE */
-   key->primitive = gs_prim[brw->primitive];
-
-   key->hint_gs_always = 0;	/* debug code? */
-
-   key->need_gs_prog = (key->hint_gs_always ||
-			brw->primitive == PIPE_PRIM_QUADS ||
-			brw->primitive == PIPE_PRIM_QUAD_STRIP ||
-			brw->primitive == PIPE_PRIM_LINE_LOOP);
-}
-
-/* Calculate interpolants for triangle and line rasterization.
- */
-static void upload_gs_prog( struct brw_context *brw )
-{
-   struct brw_gs_prog_key key;
-
-   /* Populate the key:
-    */
-   populate_key(brw, &key);
-
-   if (brw->gs.prog_active != key.need_gs_prog) {
-      brw->state.dirty.cache |= CACHE_NEW_GS_PROG;
-      brw->gs.prog_active = key.need_gs_prog;
-   }
-
-   if (brw->gs.prog_active) {
-      if (!search_cache(brw, &key))
-	 compile_gs_prog( brw, &key );
-   }
-}
-
-
-const struct brw_tracked_state brw_gs_prog = {
-   .dirty = {
-      .brw   = BRW_NEW_PRIMITIVE,
-      .cache = CACHE_NEW_VS_PROG
-   },
-   .update = upload_gs_prog
-};
diff --git a/src/gallium/drivers/i965simple/brw_gs.h b/src/gallium/drivers/i965simple/brw_gs.h
deleted file mode 100644
index f09141c6aa..0000000000
--- a/src/gallium/drivers/i965simple/brw_gs.h
+++ /dev/null
@@ -1,75 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
- 
-
-#ifndef BRW_GS_H
-#define BRW_GS_H
-
-
-#include "brw_context.h"
-#include "brw_eu.h"
-
-#define MAX_GS_VERTS (4)	     
-
-struct brw_gs_prog_key {
-   unsigned attrs:32;
-   unsigned primitive:4;
-   unsigned hint_gs_always:1;
-   unsigned need_gs_prog:1;
-   unsigned pad:26;
-};
-
-struct brw_gs_compile {
-   struct brw_compile func;
-   struct brw_gs_prog_key key;
-   struct brw_gs_prog_data prog_data;
-   
-   struct {
-      struct brw_reg R0;
-      struct brw_reg vertex[MAX_GS_VERTS];
-   } reg;
-
-   /* 3 different ways of expressing vertex size:
-    */
-   unsigned nr_attrs;
-   unsigned nr_regs;
-   unsigned nr_bytes;
-};
-
-#define ATTR_SIZE  (4*4)
-
-void brw_gs_quads( struct brw_gs_compile *c );
-void brw_gs_quad_strip( struct brw_gs_compile *c );
-void brw_gs_tris( struct brw_gs_compile *c );
-void brw_gs_lines( struct brw_gs_compile *c );
-void brw_gs_points( struct brw_gs_compile *c );
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_gs_emit.c b/src/gallium/drivers/i965simple/brw_gs_emit.c
deleted file mode 100644
index c3cc90b10f..0000000000
--- a/src/gallium/drivers/i965simple/brw_gs_emit.c
+++ /dev/null
@@ -1,148 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_gs.h"
-
-static void brw_gs_alloc_regs( struct brw_gs_compile *c,
-			       unsigned nr_verts )
-{
-   unsigned i = 0,j;
-
-   /* Register usage is static, precompute here:
-    */
-   c->reg.R0 = retype(brw_vec8_grf(i, 0), BRW_REGISTER_TYPE_UD); i++;
-
-   /* Payload vertices plus space for more generated vertices:
-    */
-   for (j = 0; j < nr_verts; j++) {
-      c->reg.vertex[j] = brw_vec4_grf(i, 0);
-      i += c->nr_regs;
-   }
-
-   c->prog_data.urb_read_length = c->nr_regs; 
-   c->prog_data.total_grf = i;
-}
-
-
-static void brw_gs_emit_vue(struct brw_gs_compile *c, 
-			    struct brw_reg vert,
-			    boolean last,
-			    unsigned header)
-{
-   struct brw_compile *p = &c->func;
-   boolean allocate = !last;
-
-   /* Overwrite PrimType and PrimStart in the message header, for
-    * each vertex in turn:
-    */
-   brw_MOV(p, get_element_ud(c->reg.R0, 2), brw_imm_ud(header));
-
-   /* Copy the vertex from vertn into m1..mN+1:
-    */
-   brw_copy8(p, brw_message_reg(1), vert, c->nr_regs);
-
-   /* Send each vertex as a seperate write to the urb.  This is
-    * different to the concept in brw_sf_emit.c, where subsequent
-    * writes are used to build up a single urb entry.  Each of these
-    * writes instantiates a seperate urb entry, and a new one must be
-    * allocated each time.
-    */
-   brw_urb_WRITE(p, 
-		 allocate ? c->reg.R0 : retype(brw_null_reg(), BRW_REGISTER_TYPE_UD),
-		 0,
-		 c->reg.R0,
-		 allocate,
-		 1,		/* used */
-		 c->nr_regs + 1, /* msg length */
-		 allocate ? 1 : 0, /* response length */
-		 allocate ? 0 : 1, /* eot */
-		 1,		/* writes_complete */
-		 0,		/* urb offset */
-		 BRW_URB_SWIZZLE_NONE);
-}
-
-
-
-void brw_gs_quads( struct brw_gs_compile *c )
-{
-   brw_gs_alloc_regs(c, 4);
-   
-   /* Use polygons for correct edgeflag behaviour. Note that vertex 3
-    * is the PV for quads, but vertex 0 for polygons:
-    */
-   brw_gs_emit_vue(c, c->reg.vertex[3], 0, ((_3DPRIM_POLYGON << 2) | R02_PRIM_START));
-   brw_gs_emit_vue(c, c->reg.vertex[0], 0, (_3DPRIM_POLYGON << 2));
-   brw_gs_emit_vue(c, c->reg.vertex[1], 0, (_3DPRIM_POLYGON << 2)); 
-   brw_gs_emit_vue(c, c->reg.vertex[2], 1, ((_3DPRIM_POLYGON << 2) | R02_PRIM_END));
-}
-
-void brw_gs_quad_strip( struct brw_gs_compile *c )
-{
-   brw_gs_alloc_regs(c, 4);
-   
-   brw_gs_emit_vue(c, c->reg.vertex[2], 0, ((_3DPRIM_POLYGON << 2) | R02_PRIM_START));
-   brw_gs_emit_vue(c, c->reg.vertex[3], 0, (_3DPRIM_POLYGON << 2));
-   brw_gs_emit_vue(c, c->reg.vertex[0], 0, (_3DPRIM_POLYGON << 2)); 
-   brw_gs_emit_vue(c, c->reg.vertex[1], 1, ((_3DPRIM_POLYGON << 2) | R02_PRIM_END));
-}
-
-void brw_gs_tris( struct brw_gs_compile *c )
-{
-   brw_gs_alloc_regs(c, 3);
-   brw_gs_emit_vue(c, c->reg.vertex[0], 0, ((_3DPRIM_TRILIST << 2) | R02_PRIM_START));
-   brw_gs_emit_vue(c, c->reg.vertex[1], 0, (_3DPRIM_TRILIST << 2));
-   brw_gs_emit_vue(c, c->reg.vertex[2], 1, ((_3DPRIM_TRILIST << 2) | R02_PRIM_END));
-}
-
-void brw_gs_lines( struct brw_gs_compile *c )
-{
-   brw_gs_alloc_regs(c, 2);
-   brw_gs_emit_vue(c, c->reg.vertex[0], 0, ((_3DPRIM_LINESTRIP << 2) | R02_PRIM_START));
-   brw_gs_emit_vue(c, c->reg.vertex[1], 1, ((_3DPRIM_LINESTRIP << 2) | R02_PRIM_END));
-}
-
-void brw_gs_points( struct brw_gs_compile *c )
-{
-   brw_gs_alloc_regs(c, 1);
-   brw_gs_emit_vue(c, c->reg.vertex[0], 1, ((_3DPRIM_POINTLIST << 2) | R02_PRIM_START | R02_PRIM_END));
-}
-
-
-
-
-
-
-
-
diff --git a/src/gallium/drivers/i965simple/brw_gs_state.c b/src/gallium/drivers/i965simple/brw_gs_state.c
deleted file mode 100644
index 5b8016b2e9..0000000000
--- a/src/gallium/drivers/i965simple/brw_gs_state.c
+++ /dev/null
@@ -1,90 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-
-
-static void upload_gs_unit( struct brw_context *brw )
-{
-   struct brw_gs_unit_state gs;
-
-   memset(&gs, 0, sizeof(gs));
-
-   /* CACHE_NEW_GS_PROG */
-   if (brw->gs.prog_active) {
-      gs.thread0.grf_reg_count =
-	 align(brw->gs.prog_data->total_grf, 16) / 16 - 1;
-      gs.thread0.kernel_start_pointer = brw->gs.prog_gs_offset >> 6;
-      gs.thread3.urb_entry_read_length = brw->gs.prog_data->urb_read_length;
-   }
-   else {
-      gs.thread0.grf_reg_count = 0;
-      gs.thread0.kernel_start_pointer = 0;
-      gs.thread3.urb_entry_read_length = 1;
-   }
-
-   /* BRW_NEW_URB_FENCE */
-   gs.thread4.nr_urb_entries = brw->urb.nr_gs_entries;
-   gs.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
-
-   gs.thread4.max_threads = 0; /* Hardware requirement */
-
-   if (BRW_DEBUG & DEBUG_STATS)
-      gs.thread4.stats_enable = 1;
-
-   /* CONSTANT */
-   gs.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
-   gs.thread1.single_program_flow = 1;
-   gs.thread3.dispatch_grf_start_reg = 1;
-   gs.thread3.const_urb_entry_read_offset = 0;
-   gs.thread3.const_urb_entry_read_length = 0;
-   gs.thread3.urb_entry_read_offset = 0;
-
-
-   brw->gs.state_gs_offset = brw_cache_data( &brw->cache[BRW_GS_UNIT], &gs );
-}
-
-
-const struct brw_tracked_state brw_gs_unit = {
-   .dirty = {
-      .brw   = (BRW_NEW_CURBE_OFFSETS |
-		BRW_NEW_URB_FENCE),
-      .cache = CACHE_NEW_GS_PROG
-   },
-   .update = upload_gs_unit
-};
diff --git a/src/gallium/drivers/i965simple/brw_misc_state.c b/src/gallium/drivers/i965simple/brw_misc_state.c
deleted file mode 100644
index 99ff4403a5..0000000000
--- a/src/gallium/drivers/i965simple/brw_misc_state.c
+++ /dev/null
@@ -1,488 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_batch.h"
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-
-
-
-
-
-/***********************************************************************
- * Blend color
- */
-
-static void upload_blend_constant_color(struct brw_context *brw)
-{
-   struct brw_blend_constant_color bcc;
-
-   memset(&bcc, 0, sizeof(bcc));
-   bcc.header.opcode = CMD_BLEND_CONSTANT_COLOR;
-   bcc.header.length = sizeof(bcc)/4-2;
-   bcc.blend_constant_color[0] = brw->attribs.BlendColor.color[0];
-   bcc.blend_constant_color[1] = brw->attribs.BlendColor.color[1];
-   bcc.blend_constant_color[2] = brw->attribs.BlendColor.color[2];
-   bcc.blend_constant_color[3] = brw->attribs.BlendColor.color[3];
-
-   BRW_CACHED_BATCH_STRUCT(brw, &bcc);
-}
-
-
-const struct brw_tracked_state brw_blend_constant_color = {
-   .dirty = {
-      .brw = BRW_NEW_BLEND,
-      .cache = 0
-   },
-   .update = upload_blend_constant_color
-};
-
-
-/***********************************************************************
- * Drawing rectangle 
- */
-static void upload_drawing_rect(struct brw_context *brw)
-{
-   struct brw_drawrect bdr;
-
-   memset(&bdr, 0, sizeof(bdr));
-   bdr.header.opcode = CMD_DRAW_RECT;
-   bdr.header.length = sizeof(bdr)/4 - 2;
-   bdr.xmin = 0;
-   bdr.ymin = 0;
-   bdr.xmax = brw->attribs.FrameBuffer.cbufs[0]->width;
-   bdr.ymax = brw->attribs.FrameBuffer.cbufs[0]->height;
-   bdr.xorg = 0;
-   bdr.yorg = 0;
-
-   /* Can't use BRW_CACHED_BATCH_STRUCT because this is also emitted
-    * uncached in brw_draw.c:
-    */
-   BRW_BATCH_STRUCT(brw, &bdr);
-}
-
-const struct brw_tracked_state brw_drawing_rect = {
-   .dirty = {
-      .brw = BRW_NEW_SCENE,
-      .cache = 0
-   },
-   .update = upload_drawing_rect
-};
-
-/**
- * Upload the binding table pointers, which point each stage's array of surface
- * state pointers.
- *
- * The binding table pointers are relative to the surface state base address,
- * which is the BRW_SS_POOL cache buffer.
- */
-static void upload_binding_table_pointers(struct brw_context *brw)
-{
-   struct brw_binding_table_pointers btp;
-   memset(&btp, 0, sizeof(btp));
-
-   btp.header.opcode = CMD_BINDING_TABLE_PTRS;
-   btp.header.length = sizeof(btp)/4 - 2;
-   btp.vs = 0;
-   btp.gs = 0;
-   btp.clp = 0;
-   btp.sf = 0;
-   btp.wm = brw->wm.bind_ss_offset;
-
-   BRW_CACHED_BATCH_STRUCT(brw, &btp);
-}
-
-const struct brw_tracked_state brw_binding_table_pointers = {
-   .dirty = {
-      .brw = 0,
-      .cache = CACHE_NEW_SURF_BIND
-   },
-   .update = upload_binding_table_pointers,
-};
-
-
-/**
- * Upload pointers to the per-stage state.
- *
- * The state pointers in this packet are all relative to the general state
- * base address set by CMD_STATE_BASE_ADDRESS, which is the BRW_GS_POOL buffer.
- */
-static void upload_pipelined_state_pointers(struct brw_context *brw )
-{
-   struct brw_pipelined_state_pointers psp;
-   memset(&psp, 0, sizeof(psp));
-
-   psp.header.opcode = CMD_PIPELINED_STATE_POINTERS;
-   psp.header.length = sizeof(psp)/4 - 2;
-
-   psp.vs.offset = brw->vs.state_gs_offset >> 5;
-   psp.sf.offset = brw->sf.state_gs_offset >> 5;
-   psp.wm.offset = brw->wm.state_gs_offset >> 5;
-   psp.cc.offset = brw->cc.state_gs_offset >> 5;
-
-   /* GS gets turned on and off regularly.  Need to re-emit URB fence
-    * after this occurs.
-    */
-   if (brw->gs.prog_active) {
-      psp.gs.offset = brw->gs.state_gs_offset >> 5;
-      psp.gs.enable = 1;
-   }
-
-   if (0) {
-      psp.clp.offset = brw->clip.state_gs_offset >> 5;
-      psp.clp.enable = 1;
-   }
-
-
-   if (BRW_CACHED_BATCH_STRUCT(brw, &psp))
-      brw->state.dirty.brw |= BRW_NEW_PSP;
-}
-
-const struct brw_tracked_state brw_pipelined_state_pointers = {
-   .dirty = {
-      .brw = 0,
-      .cache = (CACHE_NEW_VS_UNIT |
-		CACHE_NEW_GS_UNIT |
-		CACHE_NEW_GS_PROG |
-		CACHE_NEW_CLIP_UNIT |
-		CACHE_NEW_SF_UNIT |
-		CACHE_NEW_WM_UNIT |
-		CACHE_NEW_CC_UNIT)
-   },
-   .update = upload_pipelined_state_pointers
-};
-
-static void upload_psp_urb_cbs(struct brw_context *brw )
-{
-   upload_pipelined_state_pointers(brw);
-   brw_upload_urb_fence(brw);
-   brw_upload_constant_buffer_state(brw);
-}
-
-
-const struct brw_tracked_state brw_psp_urb_cbs = {
-   .dirty = {
-      .brw = BRW_NEW_URB_FENCE,
-      .cache = (CACHE_NEW_VS_UNIT |
-		CACHE_NEW_GS_UNIT |
-		CACHE_NEW_GS_PROG |
-		CACHE_NEW_CLIP_UNIT |
-		CACHE_NEW_SF_UNIT |
-		CACHE_NEW_WM_UNIT |
-		CACHE_NEW_CC_UNIT)
-   },
-   .update = upload_psp_urb_cbs
-};
-
-/**
- * Upload the depthbuffer offset and format.
- *
- * We have to do this per state validation as we need to emit the relocation
- * in the batch buffer.
- */
-static void upload_depthbuffer(struct brw_context *brw)
-{
-   struct pipe_surface *depth_surface = brw->attribs.FrameBuffer.zsbuf;
-
-   BEGIN_BATCH(5, INTEL_BATCH_NO_CLIPRECTS);
-   OUT_BATCH(CMD_DEPTH_BUFFER << 16 | (5 - 2));
-   if (depth_surface == NULL) {
-      OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) |
-		(BRW_SURFACE_NULL << 29));
-      OUT_BATCH(0);
-      OUT_BATCH(0);
-      OUT_BATCH(0);
-   } else {
-      unsigned int format;
-      struct brw_texture *tex = (struct brw_texture *)depth_surface->texture;
-      assert(depth_surface->block.width == 1);
-      assert(depth_surface->block.height == 1);
-      switch (depth_surface->block.size) {
-      case 2:
-	 format = BRW_DEPTHFORMAT_D16_UNORM;
-	 break;
-      case 4:
-	 if (depth_surface->format == PIPE_FORMAT_Z32_FLOAT)
-	    format = BRW_DEPTHFORMAT_D32_FLOAT;
-	 else
-	    format = BRW_DEPTHFORMAT_D24_UNORM_S8_UINT;
-	 break;
-      default:
-	 assert(0);
-	 return;
-      }
-
-      OUT_BATCH((depth_surface->stride - 1) |
-		(format << 18) |
-		(BRW_TILEWALK_YMAJOR << 26) |
-//		(depth_surface->region->tiled << 27) |
-		(BRW_SURFACE_2D << 29));
-      OUT_RELOC(tex->buffer,
-		PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE, 0);
-      OUT_BATCH((BRW_SURFACE_MIPMAPLAYOUT_BELOW << 1) |
-		((depth_surface->stride/depth_surface->block.size - 1) << 6) |
-		((depth_surface->height - 1) << 19));
-      OUT_BATCH(0);
-   }
-   ADVANCE_BATCH();
-}
-
-const struct brw_tracked_state brw_depthbuffer = {
-   .dirty = {
-      .brw = BRW_NEW_SCENE,
-      .cache = 0
-   },
-   .update = upload_depthbuffer,
-};
-
-
-
-
-/***********************************************************************
- * Polygon stipple packet
- */
-
-static void upload_polygon_stipple(struct brw_context *brw)
-{
-   struct brw_polygon_stipple bps;
-   unsigned i;
-
-   memset(&bps, 0, sizeof(bps));
-   bps.header.opcode = CMD_POLY_STIPPLE_PATTERN;
-   bps.header.length = sizeof(bps)/4-2;
-
-   /* XXX: state tracker should send *all* state down initially!
-    */
-   if (brw->attribs.PolygonStipple)
-      for (i = 0; i < 32; i++)
-	 bps.stipple[i] = brw->attribs.PolygonStipple->stipple[31 - i]; /* invert */
-
-   BRW_CACHED_BATCH_STRUCT(brw, &bps);
-}
-
-const struct brw_tracked_state brw_polygon_stipple = {
-   .dirty = {
-      .brw = BRW_NEW_STIPPLE,
-      .cache = 0
-   },
-   .update = upload_polygon_stipple
-};
-
-
-/***********************************************************************
- * Line stipple packet
- */
-
-static void upload_line_stipple(struct brw_context *brw)
-{
-   struct brw_line_stipple bls;
-   float tmp;
-   int tmpi;
-
-   memset(&bls, 0, sizeof(bls));
-   bls.header.opcode = CMD_LINE_STIPPLE_PATTERN;
-   bls.header.length = sizeof(bls)/4 - 2;
-
-   bls.bits0.pattern = brw->attribs.Raster->line_stipple_pattern;
-   bls.bits1.repeat_count = brw->attribs.Raster->line_stipple_factor;
-
-   tmp = 1.0 / (float) brw->attribs.Raster->line_stipple_factor;
-   tmpi = tmp * (1<<13);
-
-
-   bls.bits1.inverse_repeat_count = tmpi;
-
-   BRW_CACHED_BATCH_STRUCT(brw, &bls);
-}
-
-const struct brw_tracked_state brw_line_stipple = {
-   .dirty = {
-      .brw = BRW_NEW_STIPPLE,
-      .cache = 0
-   },
-   .update = upload_line_stipple
-};
-
-
-/***********************************************************************
- * Misc constant state packets
- */
-
-static void upload_pipe_control(struct brw_context *brw)
-{
-   struct brw_pipe_control pc;
-
-   return;
-
-   memset(&pc, 0, sizeof(pc));
-
-   pc.header.opcode = CMD_PIPE_CONTROL;
-   pc.header.length = sizeof(pc)/4 - 2;
-   pc.header.post_sync_operation = PIPE_CONTROL_NOWRITE;
-
-   pc.header.instruction_state_cache_flush_enable = 1;
-
-   pc.bits1.dest_addr_type = PIPE_CONTROL_GTTWRITE_GLOBAL;
-
-   BRW_BATCH_STRUCT(brw, &pc);
-}
-
-const struct brw_tracked_state brw_pipe_control = {
-   .dirty = {
-      .brw = BRW_NEW_SCENE,
-      .cache = 0
-   },
-   .update = upload_pipe_control
-};
-
-
-/***********************************************************************
- * Misc invarient state packets
- */
-
-static void upload_invarient_state( struct brw_context *brw )
-{
-   {
-      struct brw_mi_flush flush;
-
-      memset(&flush, 0, sizeof(flush));      
-      flush.opcode = CMD_MI_FLUSH;
-      flush.flags = BRW_FLUSH_STATE_CACHE | BRW_FLUSH_READ_CACHE;
-      BRW_BATCH_STRUCT(brw, &flush);
-   }
-
-   {
-      /* 0x61040000  Pipeline Select */
-      /*     PipelineSelect            : 0 */
-      struct brw_pipeline_select ps;
-
-      memset(&ps, 0, sizeof(ps));
-      ps.header.opcode = CMD_PIPELINE_SELECT;
-      ps.header.pipeline_select = 0;
-      BRW_BATCH_STRUCT(brw, &ps);
-   }
-
-   {
-      struct brw_global_depth_offset_clamp gdo;
-      memset(&gdo, 0, sizeof(gdo));
-
-      /* Disable depth offset clamping.
-       */
-      gdo.header.opcode = CMD_GLOBAL_DEPTH_OFFSET_CLAMP;
-      gdo.header.length = sizeof(gdo)/4 - 2;
-      gdo.depth_offset_clamp = 0.0;
-
-      BRW_BATCH_STRUCT(brw, &gdo);
-   }
-
-
-   /* 0x61020000  State Instruction Pointer */
-   {
-      struct brw_system_instruction_pointer sip;
-      memset(&sip, 0, sizeof(sip));
-
-      sip.header.opcode = CMD_STATE_INSN_POINTER;
-      sip.header.length = 0;
-      sip.bits0.pad = 0;
-      sip.bits0.system_instruction_pointer = 0;
-      BRW_BATCH_STRUCT(brw, &sip);
-   }
-
-
-   {
-      struct brw_vf_statistics vfs;
-      memset(&vfs, 0, sizeof(vfs));
-
-      vfs.opcode = CMD_VF_STATISTICS;
-      if (BRW_DEBUG & DEBUG_STATS)
-	 vfs.statistics_enable = 1;
-
-      BRW_BATCH_STRUCT(brw, &vfs);
-   }
-
-   
-   {
-      struct brw_polygon_stipple_offset bpso;
-      
-      memset(&bpso, 0, sizeof(bpso));
-      bpso.header.opcode = CMD_POLY_STIPPLE_OFFSET;
-      bpso.header.length = sizeof(bpso)/4-2;      
-      bpso.bits0.x_offset = 0;
-      bpso.bits0.y_offset = 0;
-
-      BRW_BATCH_STRUCT(brw, &bpso);
-   }
-}
-
-const struct brw_tracked_state brw_invarient_state = {
-   .dirty = {
-      .brw = BRW_NEW_SCENE,
-      .cache = 0
-   },
-   .update = upload_invarient_state
-};
-
-/**
- * Define the base addresses which some state is referenced from.
- *
- * This allows us to avoid having to emit relocations in many places for
- * cached state, and instead emit pointers inside of large, mostly-static
- * state pools.  This comes at the expense of memory, and more expensive cache
- * misses.
- */
-static void upload_state_base_address( struct brw_context *brw )
-{
-   /* Output the structure (brw_state_base_address) directly to the
-    * batchbuffer, so we can emit relocations inline.
-    */
-   BEGIN_BATCH(6, INTEL_BATCH_NO_CLIPRECTS);
-   OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (6 - 2));
-   OUT_RELOC(brw->pool[BRW_GS_POOL].buffer,
-	     PIPE_BUFFER_USAGE_GPU_READ,
-	     1); /* General state base address */
-   OUT_RELOC(brw->pool[BRW_SS_POOL].buffer,
-	     PIPE_BUFFER_USAGE_GPU_READ,
-	     1); /* Surface state base address */
-   OUT_BATCH(1); /* Indirect object base address */
-   OUT_BATCH(1); /* General state upper bound */
-   OUT_BATCH(1); /* Indirect object upper bound */
-   ADVANCE_BATCH();
-}
-
-
-const struct brw_tracked_state brw_state_base_address = {
-   .dirty = {
-      .brw = BRW_NEW_SCENE,
-      .cache = 0
-   },
-   .update = upload_state_base_address
-};
diff --git a/src/gallium/drivers/i965simple/brw_reg.h b/src/gallium/drivers/i965simple/brw_reg.h
deleted file mode 100644
index 9e885c3b3b..0000000000
--- a/src/gallium/drivers/i965simple/brw_reg.h
+++ /dev/null
@@ -1,76 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#define CMD_MI				(0x0 << 29)
-#define CMD_2D				(0x2 << 29)
-#define CMD_3D				(0x3 << 29)
-
-#define MI_BATCH_BUFFER_END		(CMD_MI | 0xA << 23)
-
-/* Stalls command execution waiting for the given events to have occurred. */
-#define MI_WAIT_FOR_EVENT               (CMD_MI | (0x3 << 23))
-#define MI_WAIT_FOR_PLANE_B_FLIP        (1<<6)
-#define MI_WAIT_FOR_PLANE_A_FLIP        (1<<2)
-
-/* Primitive dispatch on 830-945 */
-#define _3DPRIMITIVE			(CMD_3D | (0x1f << 24))
-#define PRIM_INDIRECT            (1<<23)
-#define PRIM_INLINE              (0<<23)
-#define PRIM_INDIRECT_SEQUENTIAL (0<<17)
-#define PRIM_INDIRECT_ELTS       (1<<17)
-
-#define PRIM3D_TRILIST		(0x0<<18)
-#define PRIM3D_TRISTRIP 	(0x1<<18)
-#define PRIM3D_TRISTRIP_RVRSE	(0x2<<18)
-#define PRIM3D_TRIFAN		(0x3<<18)
-#define PRIM3D_POLY		(0x4<<18)
-#define PRIM3D_LINELIST 	(0x5<<18)
-#define PRIM3D_LINESTRIP	(0x6<<18)
-#define PRIM3D_RECTLIST 	(0x7<<18)
-#define PRIM3D_POINTLIST	(0x8<<18)
-#define PRIM3D_DIB		(0x9<<18)
-#define PRIM3D_MASK		(0x1f<<18)
-
-#define XY_SETUP_BLT_CMD		(CMD_2D | (0x01 << 22) | 6)
-
-#define XY_COLOR_BLT_CMD		(CMD_2D | (0x50 << 22) | 4)
-
-#define XY_SRC_COPY_BLT_CMD             (CMD_2D | (0x53 << 22) | 6)
-
-/* BR00 */
-#define XY_BLT_WRITE_ALPHA	(1 << 21)
-#define XY_BLT_WRITE_RGB	(1 << 20)
-#define XY_SRC_TILED		(1 << 15)
-#define XY_DST_TILED		(1 << 11)
-
-/* BR13 */
-#define BR13_565		(0x1 << 24)
-#define BR13_8888		(0x3 << 24)
-
-#define FENCE_LINEAR 0
-#define FENCE_XMAJOR 1
-#define FENCE_YMAJOR 2
diff --git a/src/gallium/drivers/i965simple/brw_screen.c b/src/gallium/drivers/i965simple/brw_screen.c
deleted file mode 100644
index b22e105f10..0000000000
--- a/src/gallium/drivers/i965simple/brw_screen.c
+++ /dev/null
@@ -1,246 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "util/u_memory.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_string.h"
-#include "util/u_simple_screen.h"
-
-#include "brw_context.h"
-#include "brw_screen.h"
-#include "brw_tex_layout.h"
-
-
-static const char *
-brw_get_vendor( struct pipe_screen *screen )
-{
-   return "Tungsten Graphics, Inc.";
-}
-
-
-static const char *
-brw_get_name( struct pipe_screen *screen )
-{
-   static char buffer[128];
-   const char *chipset;
-
-   switch (brw_screen(screen)->pci_id) {
-   case PCI_CHIP_I965_Q:
-      chipset = "Intel(R) 965Q";
-      break;
-   case PCI_CHIP_I965_G:
-   case PCI_CHIP_I965_G_1:
-      chipset = "Intel(R) 965G";
-      break;
-   case PCI_CHIP_I965_GM:
-      chipset = "Intel(R) 965GM";
-      break;
-   case PCI_CHIP_I965_GME:
-      chipset = "Intel(R) 965GME/GLE";
-      break;
-   default:
-      chipset = "unknown";
-      break;
-   }
-
-   util_snprintf(buffer, sizeof(buffer), "i965 (chipset: %s)", chipset);
-   return buffer;
-}
-
-
-static int
-brw_get_param(struct pipe_screen *screen, int param)
-{
-   switch (param) {
-   case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
-      return 8;
-   case PIPE_CAP_NPOT_TEXTURES:
-      return 1;
-   case PIPE_CAP_TWO_SIDED_STENCIL:
-      return 1;
-   case PIPE_CAP_GLSL:
-      return 0;
-   case PIPE_CAP_S3TC:
-      return 0;
-   case PIPE_CAP_ANISOTROPIC_FILTER:
-      return 0;
-   case PIPE_CAP_POINT_SPRITE:
-      return 0;
-   case PIPE_CAP_MAX_RENDER_TARGETS:
-      return 1;
-   case PIPE_CAP_OCCLUSION_QUERY:
-      return 0;
-   case PIPE_CAP_TEXTURE_SHADOW_MAP:
-      return 1;
-   case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
-      return 11; /* max 1024x1024 */
-   case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
-      return 8;  /* max 128x128x128 */
-   case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
-      return 11; /* max 1024x1024 */
-   default:
-      return 0;
-   }
-}
-
-
-static float
-brw_get_paramf(struct pipe_screen *screen, int param)
-{
-   switch (param) {
-   case PIPE_CAP_MAX_LINE_WIDTH:
-      /* fall-through */
-   case PIPE_CAP_MAX_LINE_WIDTH_AA:
-      return 7.5;
-
-   case PIPE_CAP_MAX_POINT_WIDTH:
-      /* fall-through */
-   case PIPE_CAP_MAX_POINT_WIDTH_AA:
-      return 255.0;
-
-   case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
-      return 4.0;
-
-   case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
-      return 16.0;
-
-   default:
-      return 0;
-   }
-}
-
-
-static boolean
-brw_is_format_supported( struct pipe_screen *screen,
-                         enum pipe_format format, 
-                         enum pipe_texture_target target,
-                         unsigned tex_usage, 
-                         unsigned geom_flags )
-{
-#if 0
-   /* XXX: This is broken -- rewrite if still needed. */
-   static const unsigned tex_supported[] = {
-      PIPE_FORMAT_R8G8B8A8_UNORM,
-      PIPE_FORMAT_A8R8G8B8_UNORM,
-      PIPE_FORMAT_R5G6B5_UNORM,
-      PIPE_FORMAT_L8_UNORM,
-      PIPE_FORMAT_A8_UNORM,
-      PIPE_FORMAT_I8_UNORM,
-      PIPE_FORMAT_L8A8_UNORM,
-      PIPE_FORMAT_YCBCR,
-      PIPE_FORMAT_YCBCR_REV,
-      PIPE_FORMAT_S8_Z24,
-   };
-
-
-   /* Actually a lot more than this - add later:
-    */
-   static const unsigned render_supported[] = {
-      PIPE_FORMAT_A8R8G8B8_UNORM,
-      PIPE_FORMAT_R5G6B5_UNORM,
-   };
-
-   /*
-    */
-   static const unsigned z_stencil_supported[] = {
-      PIPE_FORMAT_Z16_UNORM,
-      PIPE_FORMAT_Z32_UNORM,
-      PIPE_FORMAT_S8Z24_UNORM,
-   };
-
-   switch (type) {
-   case PIPE_RENDER_FORMAT:
-      *numFormats = Elements(render_supported);
-      return render_supported;
-
-   case PIPE_TEX_FORMAT:
-      *numFormats = Elements(tex_supported);
-      return render_supported;
-
-   case PIPE_Z_STENCIL_FORMAT:
-      *numFormats = Elements(render_supported);
-      return render_supported;
-
-   default:
-      *numFormats = 0;
-      return NULL;
-   }
-#else
-   switch (format) {
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-   case PIPE_FORMAT_R5G6B5_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
-      return TRUE;
-   default:
-      return FALSE;
-   };
-   return FALSE;
-#endif
-}
-
-
-static void
-brw_destroy_screen( struct pipe_screen *screen )
-{
-   struct pipe_winsys *winsys = screen->winsys;
-
-   if(winsys->destroy)
-      winsys->destroy(winsys);
-
-   FREE(screen);
-}
-
-
-/**
- * Create a new brw_screen object
- */
-struct pipe_screen *
-brw_create_screen(struct pipe_winsys *winsys, uint pci_id)
-{
-   struct brw_screen *brwscreen = CALLOC_STRUCT(brw_screen);
-
-   if (!brwscreen)
-      return NULL;
-
-   brwscreen->pci_id = pci_id;
-
-   brwscreen->screen.winsys = winsys;
-
-   brwscreen->screen.destroy = brw_destroy_screen;
-
-   brwscreen->screen.get_name = brw_get_name;
-   brwscreen->screen.get_vendor = brw_get_vendor;
-   brwscreen->screen.get_param = brw_get_param;
-   brwscreen->screen.get_paramf = brw_get_paramf;
-   brwscreen->screen.is_format_supported = brw_is_format_supported;
-
-   brw_init_screen_texture_funcs(&brwscreen->screen);
-   u_simple_screen_init(&brwscreen->screen);
-
-   return &brwscreen->screen;
-}
diff --git a/src/gallium/drivers/i965simple/brw_sf.c b/src/gallium/drivers/i965simple/brw_sf.c
deleted file mode 100644
index b82a2e143b..0000000000
--- a/src/gallium/drivers/i965simple/brw_sf.c
+++ /dev/null
@@ -1,351 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_sf.h"
-#include "brw_state.h"
-#include "tgsi/tgsi_parse.h"
-
-
-static void compile_sf_prog( struct brw_context *brw,
-			     struct brw_sf_prog_key *key )
-{
-   struct brw_sf_compile c;
-   const unsigned *program;
-   unsigned program_size;
-
-   memset(&c, 0, sizeof(c));
-
-   /* Begin the compilation:
-    */
-   brw_init_compile(&c.func);
-
-   c.key = *key;
-
-
-   c.nr_attrs = c.key.vp_output_count;
-   c.nr_attr_regs = (c.nr_attrs+1)/2;
-
-   c.nr_setup_attrs = c.key.fp_input_count + 1; /* +1 for position */
-   c.nr_setup_regs = (c.nr_setup_attrs+1)/2;
-
-   c.prog_data.urb_read_length = c.nr_attr_regs;
-   c.prog_data.urb_entry_size = c.nr_setup_regs * 2;
-
-
-   /* Which primitive?  Or all three?
-    */
-   switch (key->primitive) {
-   case SF_TRIANGLES:
-      c.nr_verts = 3;
-      brw_emit_tri_setup( &c );
-      break;
-   case SF_LINES:
-      c.nr_verts = 2;
-      brw_emit_line_setup( &c );
-      break;
-   case SF_POINTS:
-      c.nr_verts = 1;
-      brw_emit_point_setup( &c );
-      break;
-
-   case SF_UNFILLED_TRIS:
-   default:
-      assert(0);
-      return;
-   }
-
-
-
-   /* get the program
-    */
-   program = brw_get_program(&c.func, &program_size);
-
-   /* Upload
-    */
-   brw->sf.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_SF_PROG],
-					      &c.key,
-					      sizeof(c.key),
-					      program,
-					      program_size,
-					      &c.prog_data,
-					      &brw->sf.prog_data );
-}
-
-
-static boolean search_cache( struct brw_context *brw,
-			       struct brw_sf_prog_key *key )
-{
-   return brw_search_cache(&brw->cache[BRW_SF_PROG],
-			   key, sizeof(*key),
-			   &brw->sf.prog_data,
-			   &brw->sf.prog_gs_offset);
-}
-
-
-/* Calculate interpolants for triangle and line rasterization.
- */
-static void upload_sf_prog( struct brw_context *brw )
-{
-   const struct brw_fragment_program *fs = brw->attribs.FragmentProgram;
-   struct brw_sf_prog_key key;
-   struct tgsi_parse_context parse;
-   int i, done = 0;
-
-
-   memset(&key, 0, sizeof(key));
-
-   /* Populate the key, noting state dependencies:
-    */
-   /* CACHE_NEW_VS_PROG */
-   key.vp_output_count = brw->vs.prog_data->outputs_written;
-
-   /* BRW_NEW_FS */
-   key.fp_input_count = brw->attribs.FragmentProgram->info.file_max[TGSI_FILE_INPUT] + 1;
-
-
-   /* BRW_NEW_REDUCED_PRIMITIVE */
-   switch (brw->reduced_primitive) {
-   case PIPE_PRIM_TRIANGLES:
-//      if (key.attrs & (1<<VERT_RESULT_EDGE))
-//	 key.primitive = SF_UNFILLED_TRIS;
-//      else
-      key.primitive = SF_TRIANGLES;
-      break;
-   case PIPE_PRIM_LINES:
-      key.primitive = SF_LINES;
-      break;
-   case PIPE_PRIM_POINTS:
-      key.primitive = SF_POINTS;
-      break;
-   }
-
-
-
-   /* Scan fp inputs to figure out what interpolation modes are
-    * required for each incoming vp output.  There is an assumption
-    * that the state tracker makes sure there is a 1:1 linkage between
-    * these sets of attributes (XXX: position??)
-    */
-   tgsi_parse_init( &parse, fs->program.tokens );
-   while( !done &&
-	  !tgsi_parse_end_of_tokens( &parse ) ) 
-   {
-      tgsi_parse_token( &parse );
-
-      switch( parse.FullToken.Token.Type ) {
-      case TGSI_TOKEN_TYPE_DECLARATION:
-	 if (parse.FullToken.FullDeclaration.Declaration.File == TGSI_FILE_INPUT) 
-	 {
-	    int first = parse.FullToken.FullDeclaration.DeclarationRange.First;
-	    int last = parse.FullToken.FullDeclaration.DeclarationRange.Last;
-	    int interp_mode = parse.FullToken.FullDeclaration.Declaration.Interpolate;
-	    //int semantic = parse.FullToken.FullDeclaration.Semantic.SemanticName;
-	    //int semantic_index = parse.FullToken.FullDeclaration.Semantic.SemanticIndex;
-
-	    debug_printf("fs input %d..%d interp mode %d\n", first, last, interp_mode);
-	    
-	    switch (interp_mode) {
-	    case TGSI_INTERPOLATE_CONSTANT:
-	       for (i = first; i <= last; i++) 
-		  key.const_mask |= (1 << i);
-	       break;
-	    case TGSI_INTERPOLATE_LINEAR:
-	       for (i = first; i <= last; i++) 
-		  key.linear_mask |= (1 << i);
-	       break;
-	    case TGSI_INTERPOLATE_PERSPECTIVE:
-	       for (i = first; i <= last; i++) 
-		  key.persp_mask |= (1 << i);
-	       break;
-	    default:
-	       break;
-	    }
-
-	    /* Also need stuff for flat shading, twosided color.
-	     */
-
-	 }
-	 break;
-      default:
-	 done = 1;
-	 break;
-      }
-   }
-
-   /* Hack: Adjust for position.  Optimize away when not required (ie
-    * for perspective interpolation).
-    */
-   key.persp_mask <<= 1;
-   key.linear_mask <<= 1; 
-   key.linear_mask |= 1;
-   key.const_mask <<= 1;
-
-   debug_printf("key.persp_mask: %x\n", key.persp_mask);
-   debug_printf("key.linear_mask: %x\n", key.linear_mask);
-   debug_printf("key.const_mask: %x\n", key.const_mask);
-
-
-//   key.do_point_sprite = brw->attribs.Point->PointSprite;
-//   key.SpriteOrigin = brw->attribs.Point->SpriteOrigin;
-
-//   key.do_flat_shading = (brw->attribs.Raster->flatshade);
-//   key.do_twoside_color = (brw->attribs.Light->Enabled && brw->attribs.Light->Model.TwoSide);
-
-//   if (key.do_twoside_color)
-//      key.frontface_ccw = (brw->attribs.Polygon->FrontFace == GL_CCW);
-
-
-   if (!search_cache(brw, &key))
-      compile_sf_prog( brw, &key );
-}
-
-
-const struct brw_tracked_state brw_sf_prog = {
-   .dirty = {
-      .brw   = (BRW_NEW_RASTERIZER |
-		BRW_NEW_REDUCED_PRIMITIVE |
-		BRW_NEW_VS |
-		BRW_NEW_FS),
-      .cache = 0,
-   },
-   .update = upload_sf_prog
-};
-
-
-
-#if 0
-/* Build a struct like the one we'd like the state tracker to pass to
- * us.
- */
-static void update_sf_linkage( struct brw_context *brw )
-{
-   const struct brw_vertex_program *vs = brw->attribs.VertexProgram;
-   const struct brw_fragment_program *fs = brw->attribs.FragmentProgram;
-   struct pipe_setup_linkage state;
-   struct tgsi_parse_context parse;
-
-   int i, j;
-   int nr_vp_outputs = 0;
-   int done = 0;
-
-   struct { 
-      unsigned semantic:8;
-      unsigned semantic_index:16;
-   } fp_semantic[32], vp_semantic[32];
-
-   memset(&state, 0, sizeof(state));
-
-   state.fp_input_count = 0;
-
-
-
-   
-
-
-   assert(state.fp_input_count == fs->program.num_inputs);
-
-      
-   /* Then scan vp outputs
-    */
-   done = 0;
-   tgsi_parse_init( &parse, vs->program.tokens );
-   while( !done &&
-	  !tgsi_parse_end_of_tokens( &parse ) ) 
-   {
-      tgsi_parse_token( &parse );
-
-      switch( parse.FullToken.Token.Type ) {
-      case TGSI_TOKEN_TYPE_DECLARATION:
-	 if (parse.FullToken.FullDeclaration.Declaration.File == TGSI_FILE_INPUT) 
-	 {
-	    int first = parse.FullToken.FullDeclaration.DeclarationRange.First;
-	    int last = parse.FullToken.FullDeclaration.DeclarationRange.Last;
-
-	    for (i = first; i < last; i++) {
-	       vp_semantic[i].semantic = 
-		  parse.FullToken.FullDeclaration.Semantic.SemanticName;
-	       vp_semantic[i].semantic_index = 
-		  parse.FullToken.FullDeclaration.Semantic.SemanticIndex;
-	    }
-	    
-	    assert(last > nr_vp_outputs);
-	    nr_vp_outputs = last;
-	 }
-	 break;
-      default:
-	 done = 1;
-	 break;
-      }
-   }
-
-
-   /* Now match based on semantic information.
-    */
-   for (i = 0; i< state.fp_input_count; i++) {
-      for (j = 0; j < nr_vp_outputs; j++) {
-	 if (fp_semantic[i].semantic == vp_semantic[j].semantic &&
-	     fp_semantic[i].semantic_index == vp_semantic[j].semantic_index) {
-	    state.fp_input[i].vp_output = j;
-	 }
-      }
-      if (fp_semantic[i].semantic == TGSI_SEMANTIC_COLOR) {
-	 for (j = 0; j < nr_vp_outputs; j++) {
-	    if (TGSI_SEMANTIC_BCOLOR == vp_semantic[j].semantic &&
-		fp_semantic[i].semantic_index == vp_semantic[j].semantic_index) {
-	       state.fp_input[i].bf_vp_output = j;
-	    }
-	 }
-      }
-   }
-
-   if (memcmp(&brw->sf.linkage, &state, sizeof(state)) != 0) {
-      brw->sf.linkage = state;
-      brw->state.dirty.brw |= BRW_NEW_SF_LINKAGE;
-   }
-}
-
-
-const struct brw_tracked_state brw_sf_linkage = {
-   .dirty = {
-      .brw   = (BRW_NEW_VS |
-		BRW_NEW_FS),
-      .cache = 0,
-   },
-   .update = update_sf_linkage
-};
-
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_sf.h b/src/gallium/drivers/i965simple/brw_sf.h
deleted file mode 100644
index b7ada47560..0000000000
--- a/src/gallium/drivers/i965simple/brw_sf.h
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#ifndef BRW_SF_H
-#define BRW_SF_H
-
-#include "brw_context.h"
-#include "brw_eu.h"
-
-
-#define SF_POINTS    0
-#define SF_LINES     1
-#define SF_TRIANGLES 2
-#define SF_UNFILLED_TRIS   3
-
-
-
-struct brw_sf_prog_key {
-   unsigned vp_output_count:5;
-   unsigned fp_input_count:5;
-
-   unsigned primitive:2;
-   unsigned do_twoside_color:1;
-   unsigned do_flat_shading:1;
-   unsigned frontface_ccw:1;
-   unsigned do_point_sprite:1;
-
-   /* Interpolation masks;
-    */
-   unsigned linear_mask;
-   unsigned persp_mask;
-   unsigned const_mask;
-
-
-//   int SpriteOrigin;
-};
-
-struct brw_sf_point_tex {
-	boolean CoordReplace;
-};
-
-struct brw_sf_compile {
-   struct brw_compile func;
-   struct brw_sf_prog_key key;
-   struct brw_sf_prog_data prog_data;
-
-   struct brw_reg pv;
-   struct brw_reg det;
-   struct brw_reg dx0;
-   struct brw_reg dx2;
-   struct brw_reg dy0;
-   struct brw_reg dy2;
-
-   /* z and 1/w passed in seperately:
-    */
-   struct brw_reg z[3];
-   struct brw_reg inv_w[3];
-
-   /* The vertices:
-    */
-   struct brw_reg vert[3];
-
-    /* Temporaries, allocated after last vertex reg.
-    */
-   struct brw_reg inv_det;
-   struct brw_reg a1_sub_a0;
-   struct brw_reg a2_sub_a0;
-   struct brw_reg tmp;
-
-   struct brw_reg m1Cx;
-   struct brw_reg m2Cy;
-   struct brw_reg m3C0;
-
-   unsigned nr_verts;
-   unsigned nr_attrs;
-   unsigned nr_attr_regs;
-   unsigned nr_setup_attrs;
-   unsigned nr_setup_regs;
-#if 0
-   ubyte attr_to_idx[VERT_RESULT_MAX];
-   ubyte idx_to_attr[VERT_RESULT_MAX];
-   struct brw_sf_point_tex point_attrs[VERT_RESULT_MAX];
-#endif
-};
-
-
-void brw_emit_tri_setup( struct brw_sf_compile *c );
-void brw_emit_line_setup( struct brw_sf_compile *c );
-void brw_emit_point_setup( struct brw_sf_compile *c );
-void brw_emit_point_sprite_setup( struct brw_sf_compile *c );
-void brw_emit_anyprim_setup( struct brw_sf_compile *c );
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_sf_emit.c b/src/gallium/drivers/i965simple/brw_sf_emit.c
deleted file mode 100644
index 78d6fa5e9e..0000000000
--- a/src/gallium/drivers/i965simple/brw_sf_emit.c
+++ /dev/null
@@ -1,382 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_util.h"
-#include "brw_sf.h"
-
-
-
-/***********************************************************************
- * Triangle setup.
- */
-
-
-static void alloc_regs( struct brw_sf_compile *c )
-{
-   unsigned reg, i;
-
-   /* Values computed by fixed function unit:
-    */
-   c->pv  = retype(brw_vec1_grf(1, 1), BRW_REGISTER_TYPE_UD);
-   c->det = brw_vec1_grf(1, 2);
-   c->dx0 = brw_vec1_grf(1, 3);
-   c->dx2 = brw_vec1_grf(1, 4);
-   c->dy0 = brw_vec1_grf(1, 5);
-   c->dy2 = brw_vec1_grf(1, 6);
-
-   /* z and 1/w passed in seperately:
-    */
-   c->z[0]     = brw_vec1_grf(2, 0);
-   c->inv_w[0] = brw_vec1_grf(2, 1);
-   c->z[1]     = brw_vec1_grf(2, 2);
-   c->inv_w[1] = brw_vec1_grf(2, 3);
-   c->z[2]     = brw_vec1_grf(2, 4);
-   c->inv_w[2] = brw_vec1_grf(2, 5);
-
-   /* The vertices:
-    */
-   reg = 3;
-   for (i = 0; i < c->nr_verts; i++) {
-      c->vert[i] = brw_vec8_grf(reg, 0);
-      reg += c->nr_attr_regs;
-   }
-
-   /* Temporaries, allocated after last vertex reg.
-    */
-   c->inv_det = brw_vec1_grf(reg, 0);  reg++;
-   c->a1_sub_a0 = brw_vec8_grf(reg, 0);  reg++;
-   c->a2_sub_a0 = brw_vec8_grf(reg, 0);  reg++;
-   c->tmp = brw_vec8_grf(reg, 0);  reg++;
-
-   /* Note grf allocation:
-    */
-   c->prog_data.total_grf = reg;
-
-
-   /* Outputs of this program - interpolation coefficients for
-    * rasterization:
-    */
-   c->m1Cx = brw_vec8_reg(BRW_MESSAGE_REGISTER_FILE, 1, 0);
-   c->m2Cy = brw_vec8_reg(BRW_MESSAGE_REGISTER_FILE, 2, 0);
-   c->m3C0 = brw_vec8_reg(BRW_MESSAGE_REGISTER_FILE, 3, 0);
-}
-
-
-static void copy_z_inv_w( struct brw_sf_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   unsigned i;
-
-   brw_push_insn_state(p);
-
-   /* Copy both scalars with a single MOV:
-    */
-   for (i = 0; i < c->nr_verts; i++)
-      brw_MOV(p, vec2(suboffset(c->vert[i], 2)), vec2(c->z[i]));
-
-   brw_pop_insn_state(p);
-}
-
-
-static void invert_det( struct brw_sf_compile *c)
-{
-   brw_math(&c->func,
-	    c->inv_det,
-	    BRW_MATH_FUNCTION_INV,
-	    BRW_MATH_SATURATE_NONE,
-	    0,
-	    c->det,
-	    BRW_MATH_DATA_SCALAR,
-	    BRW_MATH_PRECISION_FULL);
-
-}
-
-#define NON_PERPECTIVE_ATTRS  (FRAG_BIT_WPOS | \
-                               FRAG_BIT_COL0 | \
-			       FRAG_BIT_COL1)
-
-static boolean calculate_masks( struct brw_sf_compile *c,
-				  unsigned reg,
-				  ushort *pc,
-				  ushort *pc_persp,
-				  ushort *pc_linear)
-{
-   boolean is_last_attr = (reg == c->nr_setup_regs - 1);
-   unsigned persp_mask = c->key.persp_mask;
-   unsigned linear_mask = c->key.linear_mask;
-
-   debug_printf("persp_mask: %x\n", persp_mask);
-   debug_printf("linear_mask: %x\n", linear_mask);
-
-   *pc_persp = 0;
-   *pc_linear = 0;
-   *pc = 0xf;
-
-   if (persp_mask & (1 << (reg*2)))
-      *pc_persp = 0xf;
-
-   if (linear_mask & (1 << (reg*2)))
-      *pc_linear = 0xf;
-
-   /* Maybe only processs one attribute on the final round:
-    */
-   if (reg*2+1 < c->nr_setup_attrs) {
-      *pc |= 0xf0;
-
-      if (persp_mask & (1 << (reg*2+1)))
-	 *pc_persp |= 0xf0;
-
-      if (linear_mask & (1 << (reg*2+1)))
-	 *pc_linear |= 0xf0;
-   }
-
-   debug_printf("pc: %x\n", *pc);
-   debug_printf("pc_persp: %x\n", *pc_persp);
-   debug_printf("pc_linear: %x\n", *pc_linear);
-   
-
-   return is_last_attr;
-}
-
-
-
-void brw_emit_tri_setup( struct brw_sf_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   unsigned i;
-
-   debug_printf("%s START ==============\n", __FUNCTION__);
-
-   c->nr_verts = 3;
-   alloc_regs(c);
-   invert_det(c);
-   copy_z_inv_w(c);
-
-
-   for (i = 0; i < c->nr_setup_regs; i++)
-   {
-      /* Pair of incoming attributes:
-       */
-      struct brw_reg a0 = offset(c->vert[0], i);
-      struct brw_reg a1 = offset(c->vert[1], i);
-      struct brw_reg a2 = offset(c->vert[2], i);
-      ushort pc = 0, pc_persp = 0, pc_linear = 0;
-      boolean last = calculate_masks(c, i, &pc, &pc_persp, &pc_linear);
-
-      if (pc_persp)
-      {
-	 brw_set_predicate_control_flag_value(p, pc_persp);
-	 brw_MUL(p, a0, a0, c->inv_w[0]);
-	 brw_MUL(p, a1, a1, c->inv_w[1]);
-	 brw_MUL(p, a2, a2, c->inv_w[2]);
-      }
-
-
-      /* Calculate coefficients for interpolated values:
-       */
-      if (pc_linear)
-      {
-	 brw_set_predicate_control_flag_value(p, pc_linear);
-
-	 brw_ADD(p, c->a1_sub_a0, a1, negate(a0));
-	 brw_ADD(p, c->a2_sub_a0, a2, negate(a0));
-
-	 /* calculate dA/dx
-	  */
-	 brw_MUL(p, brw_null_reg(), c->a1_sub_a0, c->dy2);
-	 brw_MAC(p, c->tmp, c->a2_sub_a0, negate(c->dy0));
-	 brw_MUL(p, c->m1Cx, c->tmp, c->inv_det);
-
-	 /* calculate dA/dy
-	  */
-	 brw_MUL(p, brw_null_reg(), c->a2_sub_a0, c->dx0);
-	 brw_MAC(p, c->tmp, c->a1_sub_a0, negate(c->dx2));
-	 brw_MUL(p, c->m2Cy, c->tmp, c->inv_det);
-      }
-
-      {
-	 brw_set_predicate_control_flag_value(p, pc);
-	 /* start point for interpolation
-	  */
-	 brw_MOV(p, c->m3C0, a0);
-
-	 /* Copy m0..m3 to URB.  m0 is implicitly copied from r0 in
-	  * the send instruction:
-	  */
-	 brw_urb_WRITE(p,
-		       brw_null_reg(),
-		       0,
-		       brw_vec8_grf(0, 0), /* r0, will be copied to m0 */
-		       0, 	/* allocate */
-		       1,	/* used */
-		       4, 	/* msg len */
-		       0,	/* response len */
-		       last,	/* eot */
-		       last, 	/* writes complete */
-		       i*4,	/* offset */
-		       BRW_URB_SWIZZLE_TRANSPOSE); /* XXX: Swizzle control "SF to windower" */
-      }
-   }
-
-   debug_printf("%s DONE ==============\n", __FUNCTION__);
-
-}
-
-
-
-void brw_emit_line_setup( struct brw_sf_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   unsigned i;
-
-
-   c->nr_verts = 2;
-   alloc_regs(c);
-   invert_det(c);
-   copy_z_inv_w(c);
-
-   for (i = 0; i < c->nr_setup_regs; i++)
-   {
-      /* Pair of incoming attributes:
-       */
-      struct brw_reg a0 = offset(c->vert[0], i);
-      struct brw_reg a1 = offset(c->vert[1], i);
-      ushort pc, pc_persp, pc_linear;
-      boolean last = calculate_masks(c, i, &pc, &pc_persp, &pc_linear);
-
-      if (pc_persp)
-      {
-	 brw_set_predicate_control_flag_value(p, pc_persp);
-	 brw_MUL(p, a0, a0, c->inv_w[0]);
-	 brw_MUL(p, a1, a1, c->inv_w[1]);
-      }
-
-      /* Calculate coefficients for position, color:
-       */
-      if (pc_linear) {
-	 brw_set_predicate_control_flag_value(p, pc_linear);
-
-	 brw_ADD(p, c->a1_sub_a0, a1, negate(a0));
-
- 	 brw_MUL(p, c->tmp, c->a1_sub_a0, c->dx0);
-	 brw_MUL(p, c->m1Cx, c->tmp, c->inv_det);
-
-	 brw_MUL(p, c->tmp, c->a1_sub_a0, c->dy0);
-	 brw_MUL(p, c->m2Cy, c->tmp, c->inv_det);
-      }
-
-      {
-	 brw_set_predicate_control_flag_value(p, pc);
-
-	 /* start point for interpolation
-	  */
-	 brw_MOV(p, c->m3C0, a0);
-
-	 /* Copy m0..m3 to URB.
-	  */
-	 brw_urb_WRITE(p,
-		       brw_null_reg(),
-		       0,
-		       brw_vec8_grf(0, 0),
-		       0, 	/* allocate */
-		       1, 	/* used */
-		       4, 	/* msg len */
-		       0,	/* response len */
-		       last, 	/* eot */
-		       last, 	/* writes complete */
-		       i*4,	/* urb destination offset */
-		       BRW_URB_SWIZZLE_TRANSPOSE);
-      }
-   }
-}
-
-
-/* Points setup - several simplifications as all attributes are
- * constant across the face of the point (point sprites excluded!)
- */
-void brw_emit_point_setup( struct brw_sf_compile *c )
-{
-   struct brw_compile *p = &c->func;
-   unsigned i;
-
-   c->nr_verts = 1;
-   alloc_regs(c);
-   copy_z_inv_w(c);
-
-   brw_MOV(p, c->m1Cx, brw_imm_ud(0)); /* zero - move out of loop */
-   brw_MOV(p, c->m2Cy, brw_imm_ud(0)); /* zero - move out of loop */
-
-   for (i = 0; i < c->nr_setup_regs; i++)
-   {
-      struct brw_reg a0 = offset(c->vert[0], i);
-      ushort pc, pc_persp, pc_linear;
-      boolean last = calculate_masks(c, i, &pc, &pc_persp, &pc_linear);
-
-      if (pc_persp)
-      {
-	 /* This seems odd as the values are all constant, but the
-	  * fragment shader will be expecting it:
-	  */
-	 brw_set_predicate_control_flag_value(p, pc_persp);
-	 brw_MUL(p, a0, a0, c->inv_w[0]);
-      }
-
-
-      /* The delta values are always zero, just send the starting
-       * coordinate.  Again, this is to fit in with the interpolation
-       * code in the fragment shader.
-       */
-      {
-	 brw_set_predicate_control_flag_value(p, pc);
-
-	 brw_MOV(p, c->m3C0, a0); /* constant value */
-
-	 /* Copy m0..m3 to URB.
-	  */
-	 brw_urb_WRITE(p,
-		       brw_null_reg(),
-		       0,
-		       brw_vec8_grf(0, 0),
-		       0, 	/* allocate */
-		       1,	/* used */
-		       4, 	/* msg len */
-		       0,	/* response len */
-		       last, 	/* eot */
-		       last, 	/* writes complete */
-		       i*4,	/* urb destination offset */
-		       BRW_URB_SWIZZLE_TRANSPOSE);
-      }
-   }
-}
diff --git a/src/gallium/drivers/i965simple/brw_sf_state.c b/src/gallium/drivers/i965simple/brw_sf_state.c
deleted file mode 100644
index 2a5de61c21..0000000000
--- a/src/gallium/drivers/i965simple/brw_sf_state.c
+++ /dev/null
@@ -1,181 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-
-static void upload_sf_vp(struct brw_context *brw)
-{
-   struct brw_sf_viewport sfv;
-
-   memset(&sfv, 0, sizeof(sfv));
-
-
-   /* BRW_NEW_VIEWPORT */
-   {
-      const float *scale = brw->attribs.Viewport.scale;
-      const float *trans = brw->attribs.Viewport.translate;
-
-      sfv.viewport.m00 = scale[0];
-      sfv.viewport.m11 = scale[1];
-      sfv.viewport.m22 = scale[2]; 
-      sfv.viewport.m30 = trans[0];
-      sfv.viewport.m31 = trans[1];
-      sfv.viewport.m32 = trans[2];
-   }
-
-   /* _NEW_SCISSOR */
-   sfv.scissor.xmin = brw->attribs.Scissor.minx;
-   sfv.scissor.xmax = brw->attribs.Scissor.maxx - 1;
-   sfv.scissor.ymin = brw->attribs.Scissor.miny;
-   sfv.scissor.ymax = brw->attribs.Scissor.maxy - 1;
-
-   brw->sf.vp_gs_offset = brw_cache_data( &brw->cache[BRW_SF_VP], &sfv );
-}
-
-const struct brw_tracked_state brw_sf_vp = {
-   .dirty = {
-      .brw   = (BRW_NEW_SCISSOR |
-		BRW_NEW_VIEWPORT),
-      .cache = 0
-   },
-   .update = upload_sf_vp
-};
-
-static void upload_sf_unit( struct brw_context *brw )
-{
-   struct brw_sf_unit_state sf;
-   memset(&sf, 0, sizeof(sf));
-
-   /* CACHE_NEW_SF_PROG */
-   sf.thread0.grf_reg_count = align(brw->sf.prog_data->total_grf, 16) / 16 - 1;
-   sf.thread0.kernel_start_pointer = brw->sf.prog_gs_offset >> 6;
-   sf.thread3.urb_entry_read_length = brw->sf.prog_data->urb_read_length;
-
-   sf.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
-   sf.thread3.dispatch_grf_start_reg = 3;
-   sf.thread3.urb_entry_read_offset = 1;
-
-   /* BRW_NEW_URB_FENCE */
-   sf.thread4.nr_urb_entries = brw->urb.nr_sf_entries;
-   sf.thread4.urb_entry_allocation_size = brw->urb.sfsize - 1;
-   sf.thread4.max_threads = MIN2(12, brw->urb.nr_sf_entries / 2) - 1;
-
-   if (BRW_DEBUG & DEBUG_SINGLE_THREAD)
-      sf.thread4.max_threads = 0;
-
-   if (BRW_DEBUG & DEBUG_STATS)
-      sf.thread4.stats_enable = 1;
-
-   /* CACHE_NEW_SF_VP */
-   sf.sf5.sf_viewport_state_offset = brw->sf.vp_gs_offset >> 5;
-   sf.sf5.viewport_transform = 1;
-
-   /* BRW_NEW_RASTER */
-   if (brw->attribs.Raster->scissor)
-      sf.sf6.scissor = 1;
-
-#if 0
-   if (brw->attribs.Polygon->FrontFace == GL_CCW)
-      sf.sf5.front_winding = BRW_FRONTWINDING_CCW;
-   else
-      sf.sf5.front_winding = BRW_FRONTWINDING_CW;
-
-
-   if (brw->attribs.Polygon->CullFlag) {
-      switch (brw->attribs.Polygon->CullFaceMode) {
-      case GL_FRONT:
-	 sf.sf6.cull_mode = BRW_CULLMODE_FRONT;
-	 break;
-      case GL_BACK:
-	 sf.sf6.cull_mode = BRW_CULLMODE_BACK;
-	 break;
-      case GL_FRONT_AND_BACK:
-	 sf.sf6.cull_mode = BRW_CULLMODE_BOTH;
-	 break;
-      default:
-	 assert(0);
-	 break;
-      }
-   }
-   else
-      sf.sf6.cull_mode = BRW_CULLMODE_NONE;
-#else
-   sf.sf5.front_winding = BRW_FRONTWINDING_CCW;
-   sf.sf6.cull_mode = BRW_CULLMODE_NONE;
-#endif
-
-   sf.sf6.line_width = CLAMP(brw->attribs.Raster->line_width, 1.0, 5.0) * (1<<1);
-
-   sf.sf6.line_endcap_aa_region_width = 1;
-   if (brw->attribs.Raster->line_smooth)
-      sf.sf6.aa_enable = 1;
-   else if (sf.sf6.line_width <= 0x2)
-       sf.sf6.line_width = 0;
-
-   sf.sf6.point_rast_rule = 1;	/* opengl conventions */
-
-   sf.sf7.sprite_point = brw->attribs.Raster->point_sprite;
-   sf.sf7.point_size = CLAMP(brw->attribs.Raster->line_width, 1.0, 255.0) * (1<<3);
-   sf.sf7.use_point_size_state = !brw->attribs.Raster->point_size_per_vertex;
-
-   /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
-    */
-   sf.sf7.trifan_pv = 2;
-   sf.sf7.linestrip_pv = 1;
-   sf.sf7.tristrip_pv = 2;
-   sf.sf7.line_last_pixel_enable = 0;
-
-   /* Set bias for OpenGL rasterization rules:
-    */
-   sf.sf6.dest_org_vbias = 0x8;
-   sf.sf6.dest_org_hbias = 0x8;
-
-   brw->sf.state_gs_offset = brw_cache_data( &brw->cache[BRW_SF_UNIT], &sf );
-}
-
-
-const struct brw_tracked_state brw_sf_unit = {
-   .dirty = {
-      .brw   = (BRW_NEW_RASTERIZER |
-		BRW_NEW_URB_FENCE),
-      .cache = (CACHE_NEW_SF_VP |
-		CACHE_NEW_SF_PROG)
-   },
-   .update = upload_sf_unit
-};
-
-
diff --git a/src/gallium/drivers/i965simple/brw_shader_info.c b/src/gallium/drivers/i965simple/brw_shader_info.c
deleted file mode 100644
index 86d877d7ef..0000000000
--- a/src/gallium/drivers/i965simple/brw_shader_info.c
+++ /dev/null
@@ -1,48 +0,0 @@
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "util/u_memory.h"
-#include "pipe/p_shader_tokens.h"
-#include "tgsi/tgsi_parse.h"
-
-
-/**
- * XXX this obsolete new and no longer compiled.
- */
-void brw_shader_info(const struct tgsi_token *tokens,
-		     struct brw_shader_info *info )
-{
-   struct tgsi_parse_context parse;
-   int done = 0;
-
-   tgsi_parse_init( &parse, tokens );
-
-   while( !done &&
-	  !tgsi_parse_end_of_tokens( &parse ) ) 
-   {
-      tgsi_parse_token( &parse );
-
-      switch( parse.FullToken.Token.Type ) {
-      case TGSI_TOKEN_TYPE_DECLARATION:
-      {
-	 const struct tgsi_full_declaration *decl = &parse.FullToken.FullDeclaration;
-	 unsigned last = decl->DeclarationRange.Last;
-      
-	 // Broken by crazy wpos init:
-	 //assert( info->nr_regs[decl->Declaration.File] <= last);
-
-	 info->nr_regs[decl->Declaration.File] = MAX2(info->nr_regs[decl->Declaration.File],
-						      last+1);
-	 break;
-      }
-      case TGSI_TOKEN_TYPE_IMMEDIATE:
-      case TGSI_TOKEN_TYPE_INSTRUCTION:
-      default:
-	 done = 1;
-	 break;
-      }
-   }
-
-   tgsi_parse_free (&parse);
-   
-}
diff --git a/src/gallium/drivers/i965simple/brw_state.c b/src/gallium/drivers/i965simple/brw_state.c
deleted file mode 100644
index b47f5373f3..0000000000
--- a/src/gallium/drivers/i965simple/brw_state.c
+++ /dev/null
@@ -1,469 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/* Authors:  Zack Rusin <zack@tungstengraphics.com>
- *           Keith Whitwell <keith@tungstengraphics.com>
- */
-
-
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_memory.h"
-#include "pipe/p_inlines.h"
-#include "pipe/p_shader_tokens.h"
-#include "tgsi/tgsi_dump.h"
-#include "tgsi/tgsi_parse.h"
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_state.h"
-#include "brw_draw.h"
-
-
-#define DUP( TYPE, VAL )                        \
-do {                                            \
-   struct TYPE *x = malloc(sizeof(*x));         \
-   memcpy(x, VAL, sizeof(*x) );                 \
-   return x;                                    \
-} while (0)
-
-/************************************************************************
- * Blend 
- */
-static void *
-brw_create_blend_state(struct pipe_context *pipe,
-                        const struct pipe_blend_state *blend)
-{   
-   DUP( pipe_blend_state, blend );
-}
-
-static void brw_bind_blend_state(struct pipe_context *pipe,
-                                 void *blend)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.Blend = (struct pipe_blend_state*)blend;
-   brw->state.dirty.brw |= BRW_NEW_BLEND;
-}
-
-
-static void brw_delete_blend_state(struct pipe_context *pipe, void *blend)
-{
-   free(blend);
-}
-
-static void brw_set_blend_color( struct pipe_context *pipe,
-			     const struct pipe_blend_color *blend_color )
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.BlendColor = *blend_color;
-
-   brw->state.dirty.brw |= BRW_NEW_BLEND;
-}
-
-/************************************************************************
- * Sampler 
- */
-
-static void *
-brw_create_sampler_state(struct pipe_context *pipe,
-                          const struct pipe_sampler_state *sampler)
-{
-   DUP( pipe_sampler_state, sampler );
-}
-
-static void brw_bind_sampler_states(struct pipe_context *pipe,
-                                    unsigned num, void **sampler)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   assert(num <= PIPE_MAX_SAMPLERS);
-
-   /* Check for no-op */
-   if (num == brw->num_samplers &&
-       !memcmp(brw->attribs.Samplers, sampler, num * sizeof(void *)))
-      return;
-
-   memcpy(brw->attribs.Samplers, sampler, num * sizeof(void *));
-   memset(&brw->attribs.Samplers[num], 0, (PIPE_MAX_SAMPLERS - num) *
-          sizeof(void *));
-
-   brw->num_samplers = num;
-
-   brw->state.dirty.brw |= BRW_NEW_SAMPLER;
-}
-
-static void brw_delete_sampler_state(struct pipe_context *pipe,
-                                      void *sampler)
-{
-   free(sampler);
-}
-
-
-/************************************************************************
- * Depth stencil 
- */
-
-static void *
-brw_create_depth_stencil_state(struct pipe_context *pipe,
-                           const struct pipe_depth_stencil_alpha_state *depth_stencil)
-{
-   DUP( pipe_depth_stencil_alpha_state, depth_stencil );
-}
-
-static void brw_bind_depth_stencil_state(struct pipe_context *pipe,
-                                         void *depth_stencil)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.DepthStencil = (const struct pipe_depth_stencil_alpha_state *)depth_stencil;
-
-   brw->state.dirty.brw |= BRW_NEW_DEPTH_STENCIL;
-}
-
-static void brw_delete_depth_stencil_state(struct pipe_context *pipe,
-                                           void *depth_stencil)
-{
-   free(depth_stencil);
-}
-
-/************************************************************************
- * Scissor
- */
-static void brw_set_scissor_state( struct pipe_context *pipe,
-                                 const struct pipe_scissor_state *scissor )
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   memcpy( &brw->attribs.Scissor, scissor, sizeof(*scissor) );
-   brw->state.dirty.brw |= BRW_NEW_SCISSOR;
-}
-
-
-/************************************************************************
- * Stipple
- */
-
-static void brw_set_polygon_stipple( struct pipe_context *pipe,
-                                   const struct pipe_poly_stipple *stipple )
-{
-}
-
-
-/************************************************************************
- * Fragment shader
- */
-
-static void * brw_create_fs_state(struct pipe_context *pipe,
-                                   const struct pipe_shader_state *shader)
-{
-   struct brw_fragment_program *brw_fp = CALLOC_STRUCT(brw_fragment_program);
-
-   brw_fp->program.tokens = tgsi_dup_tokens(shader->tokens);
-   brw_fp->id = brw_context(pipe)->program_id++;
-
-   tgsi_scan_shader(shader->tokens, &brw_fp->info);
-
-#if 0
-   brw_shader_info(shader->tokens,
-		   &brw_fp->info2);
-#endif
-
-   tgsi_dump(shader->tokens, 0);
-
-
-   return (void *)brw_fp;
-}
-
-static void brw_bind_fs_state(struct pipe_context *pipe, void *shader)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.FragmentProgram = (struct brw_fragment_program *)shader;
-   brw->state.dirty.brw |= BRW_NEW_FS;
-}
-
-static void brw_delete_fs_state(struct pipe_context *pipe, void *shader)
-{
-   struct brw_fragment_program *brw_fp = (struct brw_fragment_program *) shader;
-
-   FREE((void *) brw_fp->program.tokens);
-   FREE(brw_fp);
-}
-
-
-/************************************************************************
- * Vertex shader and other TNL state 
- */
-
-static void *brw_create_vs_state(struct pipe_context *pipe,
-                                 const struct pipe_shader_state *shader)
-{
-   struct brw_vertex_program *brw_vp = CALLOC_STRUCT(brw_vertex_program);
-
-   brw_vp->program.tokens = tgsi_dup_tokens(shader->tokens);
-   brw_vp->id = brw_context(pipe)->program_id++;
-
-   tgsi_scan_shader(shader->tokens, &brw_vp->info);
-
-#if 0
-   brw_shader_info(shader->tokens,
-		   &brw_vp->info2);
-#endif
-   tgsi_dump(shader->tokens, 0);
-
-   return (void *)brw_vp;
-}
-
-static void brw_bind_vs_state(struct pipe_context *pipe, void *vs)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.VertexProgram = (struct brw_vertex_program *)vs;
-   brw->state.dirty.brw |= BRW_NEW_VS;
-
-   debug_printf("YYYYYYYYYYYYY BINDING VERTEX SHADER\n");
-}
-
-static void brw_delete_vs_state(struct pipe_context *pipe, void *shader)
-{
-   struct brw_vertex_program *brw_vp = (struct brw_vertex_program *) shader;
-
-   FREE((void *) brw_vp->program.tokens);
-   FREE(brw_vp);
-}
-
-
-static void brw_set_clip_state( struct pipe_context *pipe,
-                                const struct pipe_clip_state *clip )
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.Clip = *clip;
-}
-
-
-static void brw_set_viewport_state( struct pipe_context *pipe,
-				     const struct pipe_viewport_state *viewport )
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.Viewport = *viewport; /* struct copy */
-   brw->state.dirty.brw |= BRW_NEW_VIEWPORT;
-
-   /* pass the viewport info to the draw module */
-   //draw_set_viewport_state(brw->draw, viewport);
-}
-
-
-static void brw_set_vertex_buffers(struct pipe_context *pipe,
-				   unsigned count,
-				   const struct pipe_vertex_buffer *buffers)
-{
-   struct brw_context *brw = brw_context(pipe);
-   memcpy(brw->vb.vbo_array, buffers, count * sizeof(buffers[0]));
-}
-
-static void brw_set_vertex_elements(struct pipe_context *pipe,
-                                    unsigned count,
-                                    const struct pipe_vertex_element *elements)
-{
-   /* flush ? */
-   struct brw_context *brw = brw_context(pipe);
-   uint i;
-
-   assert(count <= PIPE_MAX_ATTRIBS);
-
-   for (i = 0; i < count; i++) {
-      struct brw_vertex_element_state el;
-      memset(&el, 0, sizeof(el));
-
-      el.ve0.src_offset = elements[i].src_offset;
-      el.ve0.src_format = brw_translate_surface_format(elements[i].src_format);
-      el.ve0.valid = 1;
-      el.ve0.vertex_buffer_index = elements[i].vertex_buffer_index;
-
-      el.ve1.dst_offset   = i * 4;
-
-      el.ve1.vfcomponent3 = BRW_VFCOMPONENT_STORE_SRC;
-      el.ve1.vfcomponent2 = BRW_VFCOMPONENT_STORE_SRC;
-      el.ve1.vfcomponent1 = BRW_VFCOMPONENT_STORE_SRC;
-      el.ve1.vfcomponent0 = BRW_VFCOMPONENT_STORE_SRC;
-
-      switch (elements[i].nr_components) {
-      case 1: el.ve1.vfcomponent1 = BRW_VFCOMPONENT_STORE_0;
-      case 2: el.ve1.vfcomponent2 = BRW_VFCOMPONENT_STORE_0;
-      case 3: el.ve1.vfcomponent3 = BRW_VFCOMPONENT_STORE_1_FLT;
-         break;
-      }
-
-      brw->vb.inputs[i] = el;
-   }
-}
-
-
-
-/************************************************************************
- * Constant buffers
- */
-
-static void brw_set_constant_buffer(struct pipe_context *pipe,
-                                     uint shader, uint index,
-                                     const struct pipe_constant_buffer *buf)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   assert(buf == 0 || index == 0);
-
-   brw->attribs.Constants[shader] = buf;
-   brw->state.dirty.brw |= BRW_NEW_CONSTANTS;
-}
-
-
-/************************************************************************
- * Texture surfaces
- */
-
-
-static void brw_set_sampler_textures(struct pipe_context *pipe,
-                                     unsigned num,
-                                     struct pipe_texture **texture)
-{
-   struct brw_context *brw = brw_context(pipe);
-   uint i;
-
-   assert(num <= PIPE_MAX_SAMPLERS);
-
-   /* Check for no-op */
-   if (num == brw->num_textures &&
-       !memcmp(brw->attribs.Texture, texture, num *
-               sizeof(struct pipe_texture *)))
-      return;
-
-   for (i = 0; i < num; i++)
-      pipe_texture_reference((struct pipe_texture **) &brw->attribs.Texture[i],
-                             texture[i]);
-
-   for (i = num; i < brw->num_textures; i++)
-      pipe_texture_reference((struct pipe_texture **) &brw->attribs.Texture[i],
-                             NULL);
-
-   brw->num_textures = num;
-
-   brw->state.dirty.brw |= BRW_NEW_TEXTURE;
-}
-
-
-/************************************************************************
- * Render targets, etc
- */
-
-static void brw_set_framebuffer_state(struct pipe_context *pipe,
-				       const struct pipe_framebuffer_state *fb)
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.FrameBuffer = *fb; /* struct copy */
-
-   brw->state.dirty.brw |= BRW_NEW_FRAMEBUFFER;
-}
-
-
-
-/************************************************************************
- * Rasterizer state
- */
-
-static void *
-brw_create_rasterizer_state(struct pipe_context *pipe,
-                             const struct pipe_rasterizer_state *rasterizer)
-{
-   DUP(pipe_rasterizer_state, rasterizer);
-}
-
-static void brw_bind_rasterizer_state( struct pipe_context *pipe,
-                                        void *setup )
-{
-   struct brw_context *brw = brw_context(pipe);
-
-   brw->attribs.Raster = (struct pipe_rasterizer_state *)setup;
-
-   /* Also pass-through to draw module:
-    */
-   //draw_set_rasterizer_state(brw->draw, setup);
-
-   brw->state.dirty.brw |= BRW_NEW_RASTERIZER;
-}
-
-static void brw_delete_rasterizer_state(struct pipe_context *pipe,
-                                         void *setup)
-{
-   free(setup);
-}
-
-
-
-void
-brw_init_state_functions( struct brw_context *brw )
-{
-   brw->pipe.create_blend_state = brw_create_blend_state;
-   brw->pipe.bind_blend_state = brw_bind_blend_state;
-   brw->pipe.delete_blend_state = brw_delete_blend_state;
-
-   brw->pipe.create_sampler_state = brw_create_sampler_state;
-   brw->pipe.bind_sampler_states = brw_bind_sampler_states;
-   brw->pipe.delete_sampler_state = brw_delete_sampler_state;
-
-   brw->pipe.create_depth_stencil_alpha_state = brw_create_depth_stencil_state;
-   brw->pipe.bind_depth_stencil_alpha_state = brw_bind_depth_stencil_state;
-   brw->pipe.delete_depth_stencil_alpha_state = brw_delete_depth_stencil_state;
-
-   brw->pipe.create_rasterizer_state = brw_create_rasterizer_state;
-   brw->pipe.bind_rasterizer_state = brw_bind_rasterizer_state;
-   brw->pipe.delete_rasterizer_state = brw_delete_rasterizer_state;
-   brw->pipe.create_fs_state = brw_create_fs_state;
-   brw->pipe.bind_fs_state = brw_bind_fs_state;
-   brw->pipe.delete_fs_state = brw_delete_fs_state;
-   brw->pipe.create_vs_state = brw_create_vs_state;
-   brw->pipe.bind_vs_state = brw_bind_vs_state;
-   brw->pipe.delete_vs_state = brw_delete_vs_state;
-
-   brw->pipe.set_blend_color = brw_set_blend_color;
-   brw->pipe.set_clip_state = brw_set_clip_state;
-   brw->pipe.set_constant_buffer = brw_set_constant_buffer;
-   brw->pipe.set_framebuffer_state = brw_set_framebuffer_state;
-
-//   brw->pipe.set_feedback_state = brw_set_feedback_state;
-//   brw->pipe.set_feedback_buffer = brw_set_feedback_buffer;
-
-   brw->pipe.set_polygon_stipple = brw_set_polygon_stipple;
-   brw->pipe.set_scissor_state = brw_set_scissor_state;
-   brw->pipe.set_sampler_textures = brw_set_sampler_textures;
-   brw->pipe.set_viewport_state = brw_set_viewport_state;
-   brw->pipe.set_vertex_buffers = brw_set_vertex_buffers;
-   brw->pipe.set_vertex_elements = brw_set_vertex_elements;
-}
diff --git a/src/gallium/drivers/i965simple/brw_state.h b/src/gallium/drivers/i965simple/brw_state.h
deleted file mode 100644
index de0a6371b8..0000000000
--- a/src/gallium/drivers/i965simple/brw_state.h
+++ /dev/null
@@ -1,151 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-    
-
-#ifndef BRW_STATE_H
-#define BRW_STATE_H
-
-#include "brw_context.h"
-#include "brw_winsys.h"
-
-
-const struct brw_tracked_state brw_blend_constant_color;
-const struct brw_tracked_state brw_cc_unit;
-const struct brw_tracked_state brw_cc_vp;
-const struct brw_tracked_state brw_clip_prog;
-const struct brw_tracked_state brw_clip_unit;
-const struct brw_tracked_state brw_constant_buffer_state;
-const struct brw_tracked_state brw_constant_buffer;
-const struct brw_tracked_state brw_curbe_offsets;
-const struct brw_tracked_state brw_invarient_state;
-const struct brw_tracked_state brw_gs_prog;
-const struct brw_tracked_state brw_gs_unit;
-const struct brw_tracked_state brw_drawing_rect;
-const struct brw_tracked_state brw_line_stipple;
-const struct brw_tracked_state brw_pipelined_state_pointers;
-const struct brw_tracked_state brw_binding_table_pointers;
-const struct brw_tracked_state brw_depthbuffer;
-const struct brw_tracked_state brw_polygon_stipple_offset;
-const struct brw_tracked_state brw_polygon_stipple;
-const struct brw_tracked_state brw_program_parameters;
-const struct brw_tracked_state brw_recalculate_urb_fence;
-const struct brw_tracked_state brw_sf_prog;
-const struct brw_tracked_state brw_sf_unit;
-const struct brw_tracked_state brw_sf_vp;
-const struct brw_tracked_state brw_state_base_address;
-const struct brw_tracked_state brw_urb_fence;
-const struct brw_tracked_state brw_vertex_state;
-const struct brw_tracked_state brw_vs_prog;
-const struct brw_tracked_state brw_vs_unit;
-const struct brw_tracked_state brw_wm_prog;
-const struct brw_tracked_state brw_wm_samplers;
-const struct brw_tracked_state brw_wm_surfaces;
-const struct brw_tracked_state brw_wm_unit;
-
-const struct brw_tracked_state brw_psp_urb_cbs;
-
-const struct brw_tracked_state brw_active_vertprog;
-const struct brw_tracked_state brw_tnl_vertprog;
-const struct brw_tracked_state brw_pipe_control;
-
-const struct brw_tracked_state brw_clear_surface_cache;
-const struct brw_tracked_state brw_clear_batch_cache;
-
-/***********************************************************************
- * brw_state_cache.c
- */
-unsigned brw_cache_data(struct brw_cache *cache,
-		      const void *data );
-
-unsigned brw_cache_data_sz(struct brw_cache *cache,
-			 const void *data,
-			 unsigned data_sz);
-
-unsigned brw_upload_cache( struct brw_cache *cache,
-			 const void *key,
-			 unsigned key_sz,
-			 const void *data,
-			 unsigned data_sz,
-			 const void *aux,
-			 void *aux_return );
-
-boolean brw_search_cache( struct brw_cache *cache,
-			    const void *key,
-			    unsigned key_size,
-			    void *aux_return,
-			    unsigned *offset_return);
-
-void brw_init_caches( struct brw_context *brw );
-void brw_destroy_caches( struct brw_context *brw );
-
-static inline struct pipe_buffer *brw_cache_buffer(struct brw_context *brw,
-                                                          enum brw_cache_id id)
-{
-   return brw->cache[id].pool->buffer;
-}
-
-/***********************************************************************
- * brw_state_batch.c
- */
-#define BRW_CACHED_BATCH_STRUCT(brw, s) brw_cached_batch_struct( brw, (s), sizeof(*(s)) )
-
-boolean brw_cached_batch_struct( struct brw_context *brw,
-				   const void *data,
-				   unsigned sz );
-
-void brw_destroy_batch_cache( struct brw_context *brw );
-
-
-/***********************************************************************
- * brw_state_pool.c
- */
-void brw_init_pools( struct brw_context *brw );
-void brw_destroy_pools( struct brw_context *brw );
-
-boolean brw_pool_alloc( struct brw_mem_pool *pool,
-			  unsigned size,
-			  unsigned alignment,
-			  unsigned *offset_return);
-
-void brw_pool_fence( struct brw_context *brw,
-		     struct brw_mem_pool *pool,
-		     unsigned fence );
-
-
-void brw_pool_check_wrap( struct brw_context *brw,
-			  struct brw_mem_pool *pool );
-
-void brw_clear_all_caches( struct brw_context *brw );
-void brw_invalidate_pools( struct brw_context *brw );
-void brw_clear_batch_cache_flush( struct brw_context *brw );
-
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_state_batch.c b/src/gallium/drivers/i965simple/brw_state_batch.c
deleted file mode 100644
index 43a1c89fc4..0000000000
--- a/src/gallium/drivers/i965simple/brw_state_batch.c
+++ /dev/null
@@ -1,113 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_state.h"
-#include "brw_winsys.h"
-
-#include "util/u_memory.h"
-
-/* A facility similar to the data caching code above, which aims to
- * prevent identical commands being issued repeatedly.
- */
-boolean brw_cached_batch_struct( struct brw_context *brw,
-                                 const void *data,
-                                 unsigned sz )
-{
-   struct brw_cached_batch_item *item = brw->cached_batch_items;
-   struct header *newheader = (struct header *)data;
-
-   if (brw->emit_state_always) {
-      brw_batchbuffer_data(brw->winsys, data, sz);
-      return TRUE;
-   }
-
-   while (item) {
-      if (item->header->opcode == newheader->opcode) {
-	 if (item->sz == sz && memcmp(item->header, newheader, sz) == 0)
-	    return FALSE;
-	 if (item->sz != sz) {
-	    FREE(item->header);
-	    item->header = MALLOC(sz);
-	    item->sz = sz;
-	 }
-	 goto emit;
-      }
-      item = item->next;
-   }
-
-   assert(!item);
-   item = CALLOC_STRUCT(brw_cached_batch_item);
-   item->header = MALLOC(sz);
-   item->sz = sz;
-   item->next = brw->cached_batch_items;
-   brw->cached_batch_items = item;
-
-emit:
-   memcpy(item->header, newheader, sz);
-   brw_batchbuffer_data(brw->winsys, data, sz);
-   return TRUE;
-}
-
-static void clear_batch_cache( struct brw_context *brw )
-{
-   struct brw_cached_batch_item *item = brw->cached_batch_items;
-
-   while (item) {
-      struct brw_cached_batch_item *next = item->next;
-      free((void *)item->header);
-      free(item);
-      item = next;
-   }
-
-   brw->cached_batch_items = NULL;
-
-
-   brw_clear_all_caches(brw);
-
-   brw_invalidate_pools(brw);
-}
-
-void brw_clear_batch_cache_flush( struct brw_context *brw )
-{
-   clear_batch_cache(brw);
-
-/*    brw_do_flush(brw, BRW_FLUSH_STATE_CACHE|BRW_FLUSH_READ_CACHE); */
-
-   brw->state.dirty.brw |= ~0;
-   brw->state.dirty.cache |= ~0;
-}
-
-
-
-void brw_destroy_batch_cache( struct brw_context *brw )
-{
-   clear_batch_cache(brw);
-}
diff --git a/src/gallium/drivers/i965simple/brw_state_cache.c b/src/gallium/drivers/i965simple/brw_state_cache.c
deleted file mode 100644
index 094248fa69..0000000000
--- a/src/gallium/drivers/i965simple/brw_state_cache.c
+++ /dev/null
@@ -1,443 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_state.h"
-
-#include "brw_wm.h"
-#include "brw_vs.h"
-#include "brw_clip.h"
-#include "brw_sf.h"
-#include "brw_gs.h"
-
-#include "util/u_memory.h"
-
-
-
-/***********************************************************************
- * Check cache for uploaded version of struct, else upload new one.
- * Fail when memory is exhausted.
- *
- * XXX: FIXME: Currently search is so slow it would be quicker to
- * regenerate the data every time...
- */
-
-static unsigned hash_key( const void *key, unsigned key_size )
-{
-   unsigned *ikey = (unsigned *)key;
-   unsigned hash = 0, i;
-
-   assert(key_size % 4 == 0);
-
-   /* I'm sure this can be improved on:
-    */
-   for (i = 0; i < key_size/4; i++)
-      hash ^= ikey[i];
-
-   return hash;
-}
-
-static struct brw_cache_item *search_cache( struct brw_cache *cache,
-					     unsigned hash,
-					     const void *key,
-					     unsigned key_size)
-{
-   struct brw_cache_item *c;
-
-   for (c = cache->items[hash % cache->size]; c; c = c->next) {
-      if (c->hash == hash &&
-	  c->key_size == key_size &&
-	  memcmp(c->key, key, key_size) == 0)
-	 return c;
-   }
-
-   return NULL;
-}
-
-
-static void rehash( struct brw_cache *cache )
-{
-   struct brw_cache_item **items;
-   struct brw_cache_item *c, *next;
-   unsigned size, i;
-
-   size = cache->size * 3;
-   items = (struct brw_cache_item**) MALLOC(size * sizeof(*items));
-   memset(items, 0, size * sizeof(*items));
-
-   for (i = 0; i < cache->size; i++)
-      for (c = cache->items[i]; c; c = next) {
-	 next = c->next;
-	 c->next = items[c->hash % size];
-	 items[c->hash % size] = c;
-      }
-
-   FREE(cache->items);
-   cache->items = items;
-   cache->size = size;
-}
-
-
-boolean brw_search_cache( struct brw_cache *cache,
-			    const void *key,
-			    unsigned key_size,
-			    void *aux_return,
-			    unsigned *offset_return)
-{
-   struct brw_cache_item *item;
-   unsigned addr = 0;
-   unsigned hash = hash_key(key, key_size);
-
-   item = search_cache(cache, hash, key, key_size);
-
-   if (item) {
-      if (aux_return)
-	 *(void **)aux_return = (void *)((char *)item->key + item->key_size);
-
-      *offset_return = addr = item->offset;
-   }
-
-   if (item == NULL || addr != cache->last_addr) {
-      cache->brw->state.dirty.cache |= 1<<cache->id;
-      cache->last_addr = addr;
-   }
-
-   return item != NULL;
-}
-
-unsigned brw_upload_cache( struct brw_cache *cache,
-			 const void *key,
-			 unsigned key_size,
-			 const void *data,
-			 unsigned data_size,
-			 const void *aux,
-			 void *aux_return )
-{
-   unsigned offset;
-   struct brw_cache_item *item = CALLOC_STRUCT(brw_cache_item);
-   unsigned hash = hash_key(key, key_size);
-   void *tmp = MALLOC(key_size + cache->aux_size);
-
-   if (!brw_pool_alloc(cache->pool, data_size, 1 << 6, &offset)) {
-      /* Should not be possible:
-       */
-      debug_printf("brw_pool_alloc failed\n");
-      exit(1);
-   }
-
-   memcpy(tmp, key, key_size);
-
-   if (cache->aux_size)
-      memcpy(tmp+key_size, aux, cache->aux_size);
-
-   item->key = tmp;
-   item->hash = hash;
-   item->key_size = key_size;
-   item->offset = offset;
-   item->data_size = data_size;
-
-   if (++cache->n_items > cache->size * 1.5)
-      rehash(cache);
-
-   hash %= cache->size;
-   item->next = cache->items[hash];
-   cache->items[hash] = item;
-
-   if (aux_return) {
-      assert(cache->aux_size);
-      *(void **)aux_return = (void *)((char *)item->key + item->key_size);
-   }
-
-   if (BRW_DEBUG & DEBUG_STATE)
-      debug_printf("upload %s: %d bytes to pool buffer %p offset %x\n",
-             cache->name, 
-	     data_size,
-             (void*)cache->pool->buffer,
-             offset);
-
-   /* Copy data to the buffer:
-    */
-   cache->brw->winsys->buffer_subdata_typed(cache->brw->winsys,
-					    cache->pool->buffer, 
-					    offset, 
-					    data_size, 
-					    data,
-					    cache->id);
-
-   cache->brw->state.dirty.cache |= 1<<cache->id;
-   cache->last_addr = offset;
-
-   return offset;
-}
-
-/* This doesn't really work with aux data.  Use search/upload instead
- */
-unsigned brw_cache_data_sz(struct brw_cache *cache,
-			 const void *data,
-			 unsigned data_size)
-{
-   unsigned addr;
-
-   if (!brw_search_cache(cache, data, data_size, NULL, &addr)) {
-      addr = brw_upload_cache(cache,
-			      data, data_size,
-			      data, data_size,
-			      NULL, NULL);
-   }
-
-   return addr;
-}
-
-unsigned brw_cache_data(struct brw_cache *cache,
-		      const void *data)
-{
-   return brw_cache_data_sz(cache, data, cache->key_size);
-}
-
-enum pool_type {
-   DW_SURFACE_STATE,
-   DW_GENERAL_STATE
-};
-
-static void brw_init_cache( struct brw_context *brw,
-			    const char *name,
-			    unsigned id,
-			    unsigned key_size,
-			    unsigned aux_size,
-			    enum pool_type pool_type)
-{
-   struct brw_cache *cache = &brw->cache[id];
-   cache->brw = brw;
-   cache->id = id;
-   cache->name = name;
-   cache->items = NULL;
-
-   cache->size = 7;
-   cache->n_items = 0;
-   cache->items = (struct brw_cache_item **)
-      CALLOC(cache->size, sizeof(struct brw_cache_item));
-
-
-   cache->key_size = key_size;
-   cache->aux_size = aux_size;
-   switch (pool_type) {
-   case DW_GENERAL_STATE: cache->pool = &brw->pool[BRW_GS_POOL]; break;
-   case DW_SURFACE_STATE: cache->pool = &brw->pool[BRW_SS_POOL]; break;
-   default: assert(0); break;
-   }
-}
-
-void brw_init_caches( struct brw_context *brw )
-{
-
-   brw_init_cache(brw,
-		  "CC_VP",
-		  BRW_CC_VP,
-		  sizeof(struct brw_cc_viewport),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "CC_UNIT",
-		  BRW_CC_UNIT,
-		  sizeof(struct brw_cc_unit_state),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "WM_PROG",
-		  BRW_WM_PROG,
-		  sizeof(struct brw_wm_prog_key),
-		  sizeof(struct brw_wm_prog_data),
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "SAMPLER_DEFAULT_COLOR",
-		  BRW_SAMPLER_DEFAULT_COLOR,
-		  sizeof(struct brw_sampler_default_color),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "SAMPLER",
-		  BRW_SAMPLER,
-		  0,		/* variable key/data size */
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "WM_UNIT",
-		  BRW_WM_UNIT,
-		  sizeof(struct brw_wm_unit_state),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "SF_PROG",
-		  BRW_SF_PROG,
-		  sizeof(struct brw_sf_prog_key),
-		  sizeof(struct brw_sf_prog_data),
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "SF_VP",
-		  BRW_SF_VP,
-		  sizeof(struct brw_sf_viewport),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "SF_UNIT",
-		  BRW_SF_UNIT,
-		  sizeof(struct brw_sf_unit_state),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "VS_UNIT",
-		  BRW_VS_UNIT,
-		  sizeof(struct brw_vs_unit_state),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "VS_PROG",
-		  BRW_VS_PROG,
-		  sizeof(struct brw_vs_prog_key),
-		  sizeof(struct brw_vs_prog_data),
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "CLIP_UNIT",
-		  BRW_CLIP_UNIT,
-		  sizeof(struct brw_clip_unit_state),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "CLIP_PROG",
-		  BRW_CLIP_PROG,
-		  sizeof(struct brw_clip_prog_key),
-		  sizeof(struct brw_clip_prog_data),
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "GS_UNIT",
-		  BRW_GS_UNIT,
-		  sizeof(struct brw_gs_unit_state),
-		  0,
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "GS_PROG",
-		  BRW_GS_PROG,
-		  sizeof(struct brw_gs_prog_key),
-		  sizeof(struct brw_gs_prog_data),
-		  DW_GENERAL_STATE);
-
-   brw_init_cache(brw,
-		  "SS_SURFACE",
-		  BRW_SS_SURFACE,
-		  sizeof(struct brw_surface_state),
-		  0,
-		  DW_SURFACE_STATE);
-
-   brw_init_cache(brw,
-		  "SS_SURF_BIND",
-		  BRW_SS_SURF_BIND,
-		  sizeof(struct brw_surface_binding_table),
-		  0,
-		  DW_SURFACE_STATE);
-}
-
-
-/* When we lose hardware context, need to invalidate the surface cache
- * as these structs must be explicitly re-uploaded.  They are subject
- * to fixup by the memory manager as they contain absolute agp
- * offsets, so we need to ensure there is a fresh version of the
- * struct available to receive the fixup.
- *
- * XXX: Need to ensure that there aren't two versions of a surface or
- * bufferobj with different backing data active in the same buffer at
- * once?  Otherwise the cache could confuse them.  Maybe better not to
- * cache at all?
- *
- * --> Isn't this the same as saying need to ensure batch is flushed
- *         before new data is uploaded to an existing buffer?  We
- *         already try to make sure of that.
- */
-static void clear_cache( struct brw_cache *cache )
-{
-   struct brw_cache_item *c, *next;
-   unsigned i;
-
-   for (i = 0; i < cache->size; i++) {
-      for (c = cache->items[i]; c; c = next) {
-	 next = c->next;
-	 free((void *)c->key);
-	 free(c);
-      }
-      cache->items[i] = NULL;
-   }
-
-   cache->n_items = 0;
-}
-
-void brw_clear_all_caches( struct brw_context *brw )
-{
-   int i;
-
-   if (BRW_DEBUG & DEBUG_STATE)
-      debug_printf("%s\n", __FUNCTION__);
-
-   for (i = 0; i < BRW_MAX_CACHE; i++)
-      clear_cache(&brw->cache[i]);
-
-   if (brw->curbe.last_buf) {
-      FREE(brw->curbe.last_buf);
-      brw->curbe.last_buf = NULL;
-   }
-
-   brw->state.dirty.brw |= ~0;
-   brw->state.dirty.cache |= ~0;
-}
-
-
-
-
-
-void brw_destroy_caches( struct brw_context *brw )
-{
-   unsigned i;
-
-   for (i = 0; i < BRW_MAX_CACHE; i++)
-      clear_cache(&brw->cache[i]);
-}
diff --git a/src/gallium/drivers/i965simple/brw_state_pool.c b/src/gallium/drivers/i965simple/brw_state_pool.c
deleted file mode 100644
index e91263cb1f..0000000000
--- a/src/gallium/drivers/i965simple/brw_state_pool.c
+++ /dev/null
@@ -1,138 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-/** @file brw_state_pool.c
- * Implements the state pool allocator.
- *
- * For the 965, we create two state pools for state cache entries.  Objects
- * will be allocated into the pools depending on which state base address
- * their pointer is relative to in other 965 state.
- *
- * The state pools are relatively simple: As objects are allocated, increment
- * the offset to allocate space.  When the pool is "full" (rather, close to
- * full), we reset the pool and reset the state cache entries that point into
- * the pool.
- */
-
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "pipe/p_inlines.h"
-#include "brw_context.h"
-#include "brw_state.h"
-
-boolean brw_pool_alloc( struct brw_mem_pool *pool,
-			  unsigned size,
-			  unsigned alignment,
-			  unsigned *offset_return)
-{
-   unsigned fixup = align(pool->offset, alignment) - pool->offset;
-
-   size = align(size, 4);
-
-   if (pool->offset + fixup + size >= pool->size) {
-      debug_printf("%s failed\n", __FUNCTION__);
-      assert(0);
-      exit(0);
-   }
-
-   pool->offset += fixup;
-   *offset_return = pool->offset;
-   pool->offset += size;
-
-   return TRUE;
-}
-
-static
-void brw_invalidate_pool( struct brw_mem_pool *pool )
-{
-   if (BRW_DEBUG & DEBUG_STATE)
-      debug_printf("\n\n\n %s \n\n\n", __FUNCTION__);
-
-   pool->offset = 0;
-
-   brw_clear_all_caches(pool->brw);
-}
-
-
-static void brw_init_pool( struct brw_context *brw,
-			   unsigned pool_id,
-			   unsigned size )
-{
-   struct brw_mem_pool *pool = &brw->pool[pool_id];
-
-   pool->size = size;
-   pool->brw = brw;
-
-   pool->buffer = pipe_buffer_create(brw->pipe.screen,
-                                     4096,
-                                     0 /*  DRM_BO_FLAG_MEM_TT */,
-                                     size);
-}
-
-static void brw_destroy_pool( struct brw_context *brw,
-			      unsigned pool_id )
-{
-   struct brw_mem_pool *pool = &brw->pool[pool_id];
-
-   pipe_buffer_reference( pool->brw->pipe.screen,
-			  &pool->buffer,
-			  NULL );
-}
-
-
-void brw_pool_check_wrap( struct brw_context *brw,
-			  struct brw_mem_pool *pool )
-{
-   if (pool->offset > (pool->size * 3) / 4) {
-      brw->state.dirty.brw |= BRW_NEW_SCENE;
-   }
-
-}
-
-void brw_init_pools( struct brw_context *brw )
-{
-   brw_init_pool(brw, BRW_GS_POOL, 0x80000);
-   brw_init_pool(brw, BRW_SS_POOL, 0x80000);
-}
-
-void brw_destroy_pools( struct brw_context *brw )
-{
-   brw_destroy_pool(brw, BRW_GS_POOL);
-   brw_destroy_pool(brw, BRW_SS_POOL);
-}
-
-
-void brw_invalidate_pools( struct brw_context *brw )
-{
-   brw_invalidate_pool(&brw->pool[BRW_GS_POOL]);
-   brw_invalidate_pool(&brw->pool[BRW_SS_POOL]);
-}
diff --git a/src/gallium/drivers/i965simple/brw_state_upload.c b/src/gallium/drivers/i965simple/brw_state_upload.c
deleted file mode 100644
index bac9161b5f..0000000000
--- a/src/gallium/drivers/i965simple/brw_state_upload.c
+++ /dev/null
@@ -1,202 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_state.h"
-
-#include "util/u_memory.h"
-
-/* This is used to initialize brw->state.atoms[].  We could use this
- * list directly except for a single atom, brw_constant_buffer, which
- * has a .dirty value which changes according to the parameters of the
- * current fragment and vertex programs, and so cannot be a static
- * value.
- */
-const struct brw_tracked_state *atoms[] =
-{
-   &brw_vs_prog,
-   &brw_gs_prog,
-   &brw_clip_prog,
-   &brw_sf_prog,
-   &brw_wm_prog,
-
-   /* Once all the programs are done, we know how large urb entry
-    * sizes need to be and can decide if we need to change the urb
-    * layout.
-    */
-   &brw_curbe_offsets,
-   &brw_recalculate_urb_fence,
-
-
-   &brw_cc_vp,
-   &brw_cc_unit,
-
-   &brw_wm_surfaces,		/* must do before samplers */
-   &brw_wm_samplers,
-
-   &brw_wm_unit,
-   &brw_sf_vp,
-   &brw_sf_unit,
-   &brw_vs_unit,		/* always required, enabled or not */
-   &brw_clip_unit,
-   &brw_gs_unit,
-
-   /* Command packets:
-    */
-   &brw_invarient_state,
-   &brw_state_base_address,
-   &brw_pipe_control,
-
-   &brw_binding_table_pointers,
-   &brw_blend_constant_color,
-
-   &brw_drawing_rect,
-   &brw_depthbuffer,
-
-   &brw_polygon_stipple,
-   &brw_line_stipple,
-
-   &brw_psp_urb_cbs,
-
-   &brw_constant_buffer
-};
-
-
-void brw_init_state( struct brw_context *brw )
-{
-   brw_init_pools(brw);
-   brw_init_caches(brw);
-
-   brw->state.dirty.brw = ~0;
-   brw->emit_state_always = 0;
-}
-
-
-void brw_destroy_state( struct brw_context *brw )
-{
-   brw_destroy_caches(brw);
-   brw_destroy_batch_cache(brw);
-   brw_destroy_pools(brw);
-}
-
-/***********************************************************************
- */
-
-static boolean check_state( const struct brw_state_flags *a,
-			      const struct brw_state_flags *b )
-{
-   return ((a->brw & b->brw) ||
-	   (a->cache & b->cache));
-}
-
-static void accumulate_state( struct brw_state_flags *a,
-			      const struct brw_state_flags *b )
-{
-   a->brw |= b->brw;
-   a->cache |= b->cache;
-}
-
-
-static void xor_states( struct brw_state_flags *result,
-			     const struct brw_state_flags *a,
-			      const struct brw_state_flags *b )
-{
-   result->brw = a->brw ^ b->brw;
-   result->cache = a->cache ^ b->cache;
-}
-
-
-/***********************************************************************
- * Emit all state:
- */
-void brw_validate_state( struct brw_context *brw )
-{
-   struct brw_state_flags *state = &brw->state.dirty;
-   unsigned i;
-
-   if (brw->emit_state_always) 
-      state->brw |= ~0;
-
-   if (state->cache == 0 &&
-       state->brw == 0)
-      return;
-
-   if (brw->state.dirty.brw & BRW_NEW_SCENE)
-      brw_clear_batch_cache_flush(brw);
-
-   if (BRW_DEBUG) {
-      /* Debug version which enforces various sanity checks on the
-       * state flags which are generated and checked to help ensure
-       * state atoms are ordered correctly in the list.
-       */
-      struct brw_state_flags examined, prev;
-      memset(&examined, 0, sizeof(examined));
-      prev = *state;
-
-      for (i = 0; i < Elements(atoms); i++) {
-	 const struct brw_tracked_state *atom = atoms[i];
-	 struct brw_state_flags generated;
-
-	 assert(atom->dirty.brw ||
-		atom->dirty.cache);
-	 assert(atom->update);
-
-	 if (check_state(state, &atom->dirty)) {
-	    atom->update( brw );
-	 }
-
-	 accumulate_state(&examined, &atom->dirty);
-
-	 /* generated = (prev ^ state)
-	  * if (examined & generated)
-	  *     fail;
-	  */
-	 xor_states(&generated, &prev, state);
-	 assert(!check_state(&examined, &generated));
-	 prev = *state;
-      }
-   }
-   else {
-      for (i = 0; i < Elements(atoms); i++) {
-	 const struct brw_tracked_state *atom = atoms[i];
-
-	 assert(atom->dirty.brw ||
-		atom->dirty.cache);
-	 assert(atom->update);
-
-	 if (check_state(state, &atom->dirty))
-	    atom->update( brw );
-      }
-   }
-
-   memset(state, 0, sizeof(*state));
-}
diff --git a/src/gallium/drivers/i965simple/brw_structs.h b/src/gallium/drivers/i965simple/brw_structs.h
deleted file mode 100644
index bbb087e95d..0000000000
--- a/src/gallium/drivers/i965simple/brw_structs.h
+++ /dev/null
@@ -1,1348 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#ifndef BRW_STRUCTS_H
-#define BRW_STRUCTS_H
-
-#include "pipe/p_compiler.h"
-
-/* Command packets:
- */
-struct header
-{
-   unsigned length:16;
-   unsigned opcode:16;
-};
-
-
-union header_union
-{
-   struct header bits;
-   unsigned dword;
-};
-
-struct brw_3d_control
-{
-   struct
-   {
-      unsigned length:8;
-      unsigned notify_enable:1;
-      unsigned pad:3;
-      unsigned wc_flush_enable:1;
-      unsigned depth_stall_enable:1;
-      unsigned operation:2;
-      unsigned opcode:16;
-   } header;
-
-   struct
-   {
-      unsigned pad:2;
-      unsigned dest_addr_type:1;
-      unsigned dest_addr:29;
-   } dest;
-
-   unsigned dword2;
-   unsigned dword3;
-};
-
-
-struct brw_3d_primitive
-{
-   struct
-   {
-      unsigned length:8;
-      unsigned pad:2;
-      unsigned topology:5;
-      unsigned indexed:1;
-      unsigned opcode:16;
-   } header;
-
-   unsigned verts_per_instance;
-   unsigned start_vert_location;
-   unsigned instance_count;
-   unsigned start_instance_location;
-   unsigned base_vert_location;
-};
-
-/* These seem to be passed around as function args, so it works out
- * better to keep them as #defines:
- */
-#define BRW_FLUSH_READ_CACHE           0x1
-#define BRW_FLUSH_STATE_CACHE          0x2
-#define BRW_INHIBIT_FLUSH_RENDER_CACHE 0x4
-#define BRW_FLUSH_SNAPSHOT_COUNTERS    0x8
-
-struct brw_mi_flush
-{
-   unsigned flags:4;
-   unsigned pad:12;
-   unsigned opcode:16;
-};
-
-struct brw_vf_statistics
-{
-   unsigned statistics_enable:1;
-   unsigned pad:15;
-   unsigned opcode:16;
-};
-
-
-
-struct brw_binding_table_pointers
-{
-   struct header header;
-   unsigned vs;
-   unsigned gs;
-   unsigned clp;
-   unsigned sf;
-   unsigned wm;
-};
-
-
-struct brw_blend_constant_color
-{
-   struct header header;
-   float blend_constant_color[4];
-};
-
-
-struct brw_depthbuffer
-{
-   union header_union header;
-
-   union {
-      struct {
-	 unsigned pitch:18;
-	 unsigned format:3;
-	 unsigned pad:4;
-	 unsigned depth_offset_disable:1;
-	 unsigned tile_walk:1;
-	 unsigned tiled_surface:1;
-	 unsigned pad2:1;
-	 unsigned surface_type:3;
-      } bits;
-      unsigned dword;
-   } dword1;
-
-   unsigned dword2_base_addr;
-
-   union {
-      struct {
-	 unsigned pad:1;
-	 unsigned mipmap_layout:1;
-	 unsigned lod:4;
-	 unsigned width:13;
-	 unsigned height:13;
-      } bits;
-      unsigned dword;
-   } dword3;
-
-   union {
-      struct {
-	 unsigned pad:12;
-	 unsigned min_array_element:9;
-	 unsigned depth:11;
-      } bits;
-      unsigned dword;
-   } dword4;
-};
-
-struct brw_drawrect
-{
-   struct header header;
-   unsigned xmin:16;
-   unsigned ymin:16;
-   unsigned xmax:16;
-   unsigned ymax:16;
-   unsigned xorg:16;
-   unsigned yorg:16;
-};
-
-
-
-
-struct brw_global_depth_offset_clamp
-{
-   struct header header;
-   float depth_offset_clamp;
-};
-
-struct brw_indexbuffer
-{
-   union {
-      struct
-      {
-	 unsigned length:8;
-	 unsigned index_format:2;
-	 unsigned cut_index_enable:1;
-	 unsigned pad:5;
-	 unsigned opcode:16;
-      } bits;
-      unsigned dword;
-
-   } header;
-
-   unsigned buffer_start;
-   unsigned buffer_end;
-};
-
-
-struct brw_line_stipple
-{
-   struct header header;
-
-   struct
-   {
-      unsigned pattern:16;
-      unsigned pad:16;
-   } bits0;
-
-   struct
-   {
-      unsigned repeat_count:9;
-      unsigned pad:7;
-      unsigned inverse_repeat_count:16;
-   } bits1;
-};
-
-
-struct brw_pipelined_state_pointers
-{
-   struct header header;
-
-   struct {
-      unsigned pad:5;
-      unsigned offset:27;
-   } vs;
-
-   struct
-   {
-      unsigned enable:1;
-      unsigned pad:4;
-      unsigned offset:27;
-   } gs;
-
-   struct
-   {
-      unsigned enable:1;
-      unsigned pad:4;
-      unsigned offset:27;
-   } clp;
-
-   struct
-   {
-      unsigned pad:5;
-      unsigned offset:27;
-   } sf;
-
-   struct
-   {
-      unsigned pad:5;
-      unsigned offset:27;
-   } wm;
-
-   struct
-   {
-      unsigned pad:5;
-      unsigned offset:27; /* KW: check me! */
-   } cc;
-};
-
-
-struct brw_polygon_stipple_offset
-{
-   struct header header;
-
-   struct {
-      unsigned y_offset:5;
-      unsigned pad:3;
-      unsigned x_offset:5;
-      unsigned pad0:19;
-   } bits0;
-};
-
-
-
-struct brw_polygon_stipple
-{
-   struct header header;
-   unsigned stipple[32];
-};
-
-
-
-struct brw_pipeline_select
-{
-   struct
-   {
-      unsigned pipeline_select:1;
-      unsigned pad:15;
-      unsigned opcode:16;
-   } header;
-};
-
-
-struct brw_pipe_control
-{
-   struct
-   {
-      unsigned length:8;
-      unsigned notify_enable:1;
-      unsigned pad:2;
-      unsigned instruction_state_cache_flush_enable:1;
-      unsigned write_cache_flush_enable:1;
-      unsigned depth_stall_enable:1;
-      unsigned post_sync_operation:2;
-
-      unsigned opcode:16;
-   } header;
-
-   struct
-   {
-      unsigned pad:2;
-      unsigned dest_addr_type:1;
-      unsigned dest_addr:29;
-   } bits1;
-
-   unsigned data0;
-   unsigned data1;
-};
-
-
-struct brw_urb_fence
-{
-   struct
-   {
-      unsigned length:8;
-      unsigned vs_realloc:1;
-      unsigned gs_realloc:1;
-      unsigned clp_realloc:1;
-      unsigned sf_realloc:1;
-      unsigned vfe_realloc:1;
-      unsigned cs_realloc:1;
-      unsigned pad:2;
-      unsigned opcode:16;
-   } header;
-
-   struct
-   {
-      unsigned vs_fence:10;
-      unsigned gs_fence:10;
-      unsigned clp_fence:10;
-      unsigned pad:2;
-   } bits0;
-
-   struct
-   {
-      unsigned sf_fence:10;
-      unsigned vf_fence:10;
-      unsigned cs_fence:10;
-      unsigned pad:2;
-   } bits1;
-};
-
-struct brw_constant_buffer_state /* previously brw_command_streamer */
-{
-   struct header header;
-
-   struct
-   {
-      unsigned nr_urb_entries:3;
-      unsigned pad:1;
-      unsigned urb_entry_size:5;
-      unsigned pad0:23;
-   } bits0;
-};
-
-struct brw_constant_buffer
-{
-   struct
-   {
-      unsigned length:8;
-      unsigned valid:1;
-      unsigned pad:7;
-      unsigned opcode:16;
-   } header;
-
-   struct
-   {
-      unsigned buffer_length:6;
-      unsigned buffer_address:26;
-   } bits0;
-};
-
-struct brw_state_base_address
-{
-   struct header header;
-
-   struct
-   {
-      unsigned modify_enable:1;
-      unsigned pad:4;
-      unsigned general_state_address:27;
-   } bits0;
-
-   struct
-   {
-      unsigned modify_enable:1;
-      unsigned pad:4;
-      unsigned surface_state_address:27;
-   } bits1;
-
-   struct
-   {
-      unsigned modify_enable:1;
-      unsigned pad:4;
-      unsigned indirect_object_state_address:27;
-   } bits2;
-
-   struct
-   {
-      unsigned modify_enable:1;
-      unsigned pad:11;
-      unsigned general_state_upper_bound:20;
-   } bits3;
-
-   struct
-   {
-      unsigned modify_enable:1;
-      unsigned pad:11;
-      unsigned indirect_object_state_upper_bound:20;
-   } bits4;
-};
-
-struct brw_state_prefetch
-{
-   struct header header;
-
-   struct
-   {
-      unsigned prefetch_count:3;
-      unsigned pad:3;
-      unsigned prefetch_pointer:26;
-   } bits0;
-};
-
-struct brw_system_instruction_pointer
-{
-   struct header header;
-
-   struct
-   {
-      unsigned pad:4;
-      unsigned system_instruction_pointer:28;
-   } bits0;
-};
-
-
-
-
-/* State structs for the various fixed function units:
- */
-
-
-struct thread0
-{
-   unsigned pad0:1;
-   unsigned grf_reg_count:3;
-   unsigned pad1:2;
-   unsigned kernel_start_pointer:26;
-};
-
-struct thread1
-{
-   unsigned ext_halt_exception_enable:1;
-   unsigned sw_exception_enable:1;
-   unsigned mask_stack_exception_enable:1;
-   unsigned timeout_exception_enable:1;
-   unsigned illegal_op_exception_enable:1;
-   unsigned pad0:3;
-   unsigned depth_coef_urb_read_offset:6;	/* WM only */
-   unsigned pad1:2;
-   unsigned floating_point_mode:1;
-   unsigned thread_priority:1;
-   unsigned binding_table_entry_count:8;
-   unsigned pad3:5;
-   unsigned single_program_flow:1;
-};
-
-struct thread2
-{
-   unsigned per_thread_scratch_space:4;
-   unsigned pad0:6;
-   unsigned scratch_space_base_pointer:22;
-};
-
-
-struct thread3
-{
-   unsigned dispatch_grf_start_reg:4;
-   unsigned urb_entry_read_offset:6;
-   unsigned pad0:1;
-   unsigned urb_entry_read_length:6;
-   unsigned pad1:1;
-   unsigned const_urb_entry_read_offset:6;
-   unsigned pad2:1;
-   unsigned const_urb_entry_read_length:6;
-   unsigned pad3:1;
-};
-
-
-
-struct brw_clip_unit_state
-{
-   struct thread0 thread0;
-   struct
-   {
-      unsigned pad0:7;
-      unsigned sw_exception_enable:1;
-      unsigned pad1:3;
-      unsigned mask_stack_exception_enable:1;
-      unsigned pad2:1;
-      unsigned illegal_op_exception_enable:1;
-      unsigned pad3:2;
-      unsigned floating_point_mode:1;
-      unsigned thread_priority:1;
-      unsigned binding_table_entry_count:8;
-      unsigned pad4:5;
-      unsigned single_program_flow:1;
-   } thread1;
-
-   struct thread2 thread2;
-   struct thread3 thread3;
-
-   struct
-   {
-      unsigned pad0:9;
-      unsigned gs_output_stats:1; /* not always */
-      unsigned stats_enable:1;
-      unsigned nr_urb_entries:7;
-      unsigned pad1:1;
-      unsigned urb_entry_allocation_size:5;
-      unsigned pad2:1;
-      unsigned max_threads:1; 	/* may be less */
-      unsigned pad3:6;
-   } thread4;
-
-   struct
-   {
-      unsigned pad0:13;
-      unsigned clip_mode:3;
-      unsigned userclip_enable_flags:8;
-      unsigned userclip_must_clip:1;
-      unsigned pad1:1;
-      unsigned guard_band_enable:1;
-      unsigned viewport_z_clip_enable:1;
-      unsigned viewport_xy_clip_enable:1;
-      unsigned vertex_position_space:1;
-      unsigned api_mode:1;
-      unsigned pad2:1;
-   } clip5;
-
-   struct
-   {
-      unsigned pad0:5;
-      unsigned clipper_viewport_state_ptr:27;
-   } clip6;
-
-
-   float viewport_xmin;
-   float viewport_xmax;
-   float viewport_ymin;
-   float viewport_ymax;
-};
-
-
-
-struct brw_cc_unit_state
-{
-   struct
-   {
-      unsigned pad0:3;
-      unsigned bf_stencil_pass_depth_pass_op:3;
-      unsigned bf_stencil_pass_depth_fail_op:3;
-      unsigned bf_stencil_fail_op:3;
-      unsigned bf_stencil_func:3;
-      unsigned bf_stencil_enable:1;
-      unsigned pad1:2;
-      unsigned stencil_write_enable:1;
-      unsigned stencil_pass_depth_pass_op:3;
-      unsigned stencil_pass_depth_fail_op:3;
-      unsigned stencil_fail_op:3;
-      unsigned stencil_func:3;
-      unsigned stencil_enable:1;
-   } cc0;
-
-
-   struct
-   {
-      unsigned bf_stencil_ref:8;
-      unsigned stencil_write_mask:8;
-      unsigned stencil_test_mask:8;
-      unsigned stencil_ref:8;
-   } cc1;
-
-
-   struct
-   {
-      unsigned logicop_enable:1;
-      unsigned pad0:10;
-      unsigned depth_write_enable:1;
-      unsigned depth_test_function:3;
-      unsigned depth_test:1;
-      unsigned bf_stencil_write_mask:8;
-      unsigned bf_stencil_test_mask:8;
-   } cc2;
-
-
-   struct
-   {
-      unsigned pad0:8;
-      unsigned alpha_test_func:3;
-      unsigned alpha_test:1;
-      unsigned blend_enable:1;
-      unsigned ia_blend_enable:1;
-      unsigned pad1:1;
-      unsigned alpha_test_format:1;
-      unsigned pad2:16;
-   } cc3;
-
-   struct
-   {
-      unsigned pad0:5;
-      unsigned cc_viewport_state_offset:27;
-   } cc4;
-
-   struct
-   {
-      unsigned pad0:2;
-      unsigned ia_dest_blend_factor:5;
-      unsigned ia_src_blend_factor:5;
-      unsigned ia_blend_function:3;
-      unsigned statistics_enable:1;
-      unsigned logicop_func:4;
-      unsigned pad1:11;
-      unsigned dither_enable:1;
-   } cc5;
-
-   struct
-   {
-      unsigned clamp_post_alpha_blend:1;
-      unsigned clamp_pre_alpha_blend:1;
-      unsigned clamp_range:2;
-      unsigned pad0:11;
-      unsigned y_dither_offset:2;
-      unsigned x_dither_offset:2;
-      unsigned dest_blend_factor:5;
-      unsigned src_blend_factor:5;
-      unsigned blend_function:3;
-   } cc6;
-
-   struct {
-      union {
-	 float f;
-	 ubyte ub[4];
-      } alpha_ref;
-   } cc7;
-};
-
-
-
-struct brw_sf_unit_state
-{
-   struct thread0 thread0;
-   struct thread1 thread1;
-   struct thread2 thread2;
-   struct thread3 thread3;
-
-   struct
-   {
-      unsigned pad0:10;
-      unsigned stats_enable:1;
-      unsigned nr_urb_entries:7;
-      unsigned pad1:1;
-      unsigned urb_entry_allocation_size:5;
-      unsigned pad2:1;
-      unsigned max_threads:6;
-      unsigned pad3:1;
-   } thread4;
-
-   struct
-   {
-      unsigned front_winding:1;
-      unsigned viewport_transform:1;
-      unsigned pad0:3;
-      unsigned sf_viewport_state_offset:27;
-   } sf5;
-
-   struct
-   {
-      unsigned pad0:9;
-      unsigned dest_org_vbias:4;
-      unsigned dest_org_hbias:4;
-      unsigned scissor:1;
-      unsigned disable_2x2_trifilter:1;
-      unsigned disable_zero_pix_trifilter:1;
-      unsigned point_rast_rule:2;
-      unsigned line_endcap_aa_region_width:2;
-      unsigned line_width:4;
-      unsigned fast_scissor_disable:1;
-      unsigned cull_mode:2;
-      unsigned aa_enable:1;
-   } sf6;
-
-   struct
-   {
-      unsigned point_size:11;
-      unsigned use_point_size_state:1;
-      unsigned subpixel_precision:1;
-      unsigned sprite_point:1;
-      unsigned pad0:11;
-      unsigned trifan_pv:2;
-      unsigned linestrip_pv:2;
-      unsigned tristrip_pv:2;
-      unsigned line_last_pixel_enable:1;
-   } sf7;
-
-};
-
-
-struct brw_gs_unit_state
-{
-   struct thread0 thread0;
-   struct thread1 thread1;
-   struct thread2 thread2;
-   struct thread3 thread3;
-
-   struct
-   {
-      unsigned pad0:10;
-      unsigned stats_enable:1;
-      unsigned nr_urb_entries:7;
-      unsigned pad1:1;
-      unsigned urb_entry_allocation_size:5;
-      unsigned pad2:1;
-      unsigned max_threads:1;
-      unsigned pad3:6;
-   } thread4;
-
-   struct
-   {
-      unsigned sampler_count:3;
-      unsigned pad0:2;
-      unsigned sampler_state_pointer:27;
-   } gs5;
-
-
-   struct
-   {
-      unsigned max_vp_index:4;
-      unsigned pad0:26;
-      unsigned reorder_enable:1;
-      unsigned pad1:1;
-   } gs6;
-};
-
-
-struct brw_vs_unit_state
-{
-   struct thread0 thread0;
-   struct thread1 thread1;
-   struct thread2 thread2;
-   struct thread3 thread3;
-
-   struct
-   {
-      unsigned pad0:10;
-      unsigned stats_enable:1;
-      unsigned nr_urb_entries:7;
-      unsigned pad1:1;
-      unsigned urb_entry_allocation_size:5;
-      unsigned pad2:1;
-      unsigned max_threads:4;
-      unsigned pad3:3;
-   } thread4;
-
-   struct
-   {
-      unsigned sampler_count:3;
-      unsigned pad0:2;
-      unsigned sampler_state_pointer:27;
-   } vs5;
-
-   struct
-   {
-      unsigned vs_enable:1;
-      unsigned vert_cache_disable:1;
-      unsigned pad0:30;
-   } vs6;
-};
-
-
-struct brw_wm_unit_state
-{
-   struct thread0 thread0;
-   struct thread1 thread1;
-   struct thread2 thread2;
-   struct thread3 thread3;
-
-   struct {
-      unsigned stats_enable:1;
-      unsigned pad0:1;
-      unsigned sampler_count:3;
-      unsigned sampler_state_pointer:27;
-   } wm4;
-
-   struct
-   {
-      unsigned enable_8_pix:1;
-      unsigned enable_16_pix:1;
-      unsigned enable_32_pix:1;
-      unsigned pad0:7;
-      unsigned legacy_global_depth_bias:1;
-      unsigned line_stipple:1;
-      unsigned depth_offset:1;
-      unsigned polygon_stipple:1;
-      unsigned line_aa_region_width:2;
-      unsigned line_endcap_aa_region_width:2;
-      unsigned early_depth_test:1;
-      unsigned thread_dispatch_enable:1;
-      unsigned program_uses_depth:1;
-      unsigned program_computes_depth:1;
-      unsigned program_uses_killpixel:1;
-      unsigned legacy_line_rast: 1;
-      unsigned pad1:1;
-      unsigned max_threads:6;
-      unsigned pad2:1;
-   } wm5;
-
-   float global_depth_offset_constant;
-   float global_depth_offset_scale;
-};
-
-struct brw_sampler_default_color {
-   float color[4];
-};
-
-struct brw_sampler_state
-{
-
-   struct
-   {
-      unsigned shadow_function:3;
-      unsigned lod_bias:11;
-      unsigned min_filter:3;
-      unsigned mag_filter:3;
-      unsigned mip_filter:2;
-      unsigned base_level:5;
-      unsigned pad:1;
-      unsigned lod_preclamp:1;
-      unsigned default_color_mode:1;
-      unsigned pad0:1;
-      unsigned disable:1;
-   } ss0;
-
-   struct
-   {
-      unsigned r_wrap_mode:3;
-      unsigned t_wrap_mode:3;
-      unsigned s_wrap_mode:3;
-      unsigned pad:3;
-      unsigned max_lod:10;
-      unsigned min_lod:10;
-   } ss1;
-
-
-   struct
-   {
-      unsigned pad:5;
-      unsigned default_color_pointer:27;
-   } ss2;
-
-   struct
-   {
-      unsigned pad:19;
-      unsigned max_aniso:3;
-      unsigned chroma_key_mode:1;
-      unsigned chroma_key_index:2;
-      unsigned chroma_key_enable:1;
-      unsigned monochrome_filter_width:3;
-      unsigned monochrome_filter_height:3;
-   } ss3;
-};
-
-
-struct brw_clipper_viewport
-{
-   float xmin;
-   float xmax;
-   float ymin;
-   float ymax;
-};
-
-struct brw_cc_viewport
-{
-   float min_depth;
-   float max_depth;
-};
-
-struct brw_sf_viewport
-{
-   struct {
-      float m00;
-      float m11;
-      float m22;
-      float m30;
-      float m31;
-      float m32;
-   } viewport;
-
-   struct {
-      short xmin;
-      short ymin;
-      short xmax;
-      short ymax;
-   } scissor;
-};
-
-/* Documented in the subsystem/shared-functions/sampler chapter...
- */
-struct brw_surface_state
-{
-   struct {
-      unsigned cube_pos_z:1;
-      unsigned cube_neg_z:1;
-      unsigned cube_pos_y:1;
-      unsigned cube_neg_y:1;
-      unsigned cube_pos_x:1;
-      unsigned cube_neg_x:1;
-      unsigned pad:4;
-      unsigned mipmap_layout_mode:1;
-      unsigned vert_line_stride_ofs:1;
-      unsigned vert_line_stride:1;
-      unsigned color_blend:1;
-      unsigned writedisable_blue:1;
-      unsigned writedisable_green:1;
-      unsigned writedisable_red:1;
-      unsigned writedisable_alpha:1;
-      unsigned surface_format:9;
-      unsigned data_return_format:1;
-      unsigned pad0:1;
-      unsigned surface_type:3;
-   } ss0;
-
-   struct {
-      unsigned base_addr;
-   } ss1;
-
-   struct {
-      unsigned pad:2;
-      unsigned mip_count:4;
-      unsigned width:13;
-      unsigned height:13;
-   } ss2;
-
-   struct {
-      unsigned tile_walk:1;
-      unsigned tiled_surface:1;
-      unsigned pad:1;
-      unsigned pitch:18;
-      unsigned depth:11;
-   } ss3;
-
-   struct {
-      unsigned pad:19;
-      unsigned min_array_elt:9;
-      unsigned min_lod:4;
-   } ss4;
-};
-
-
-
-struct brw_vertex_buffer_state
-{
-   struct {
-      unsigned pitch:11;
-      unsigned pad:15;
-      unsigned access_type:1;
-      unsigned vb_index:5;
-   } vb0;
-
-   unsigned start_addr;
-   unsigned max_index;
-#if 1
-   unsigned instance_data_step_rate; /* not included for sequential/random vertices? */
-#endif
-};
-
-#define BRW_VBP_MAX 17
-
-struct brw_vb_array_state {
-   struct header header;
-   struct brw_vertex_buffer_state vb[BRW_VBP_MAX];
-};
-
-
-struct brw_vertex_element_state
-{
-   struct
-   {
-      unsigned src_offset:11;
-      unsigned pad:5;
-      unsigned src_format:9;
-      unsigned pad0:1;
-      unsigned valid:1;
-      unsigned vertex_buffer_index:5;
-   } ve0;
-
-   struct
-   {
-      unsigned dst_offset:8;
-      unsigned pad:8;
-      unsigned vfcomponent3:4;
-      unsigned vfcomponent2:4;
-      unsigned vfcomponent1:4;
-      unsigned vfcomponent0:4;
-   } ve1;
-};
-
-#define BRW_VEP_MAX 18
-
-struct brw_vertex_element_packet {
-   struct header header;
-   struct brw_vertex_element_state ve[BRW_VEP_MAX]; /* note: less than _TNL_ATTRIB_MAX */
-};
-
-
-struct brw_urb_immediate {
-   unsigned opcode:4;
-   unsigned offset:6;
-   unsigned swizzle_control:2;
-   unsigned pad:1;
-   unsigned allocate:1;
-   unsigned used:1;
-   unsigned complete:1;
-   unsigned response_length:4;
-   unsigned msg_length:4;
-   unsigned msg_target:4;
-   unsigned pad1:3;
-   unsigned end_of_thread:1;
-};
-
-/* Instruction format for the execution units:
- */
-
-struct brw_instruction
-{
-   struct
-   {
-      unsigned opcode:7;
-      unsigned pad:1;
-      unsigned access_mode:1;
-      unsigned mask_control:1;
-      unsigned dependency_control:2;
-      unsigned compression_control:2;
-      unsigned thread_control:2;
-      unsigned predicate_control:4;
-      unsigned predicate_inverse:1;
-      unsigned execution_size:3;
-      unsigned destreg__conditonalmod:4; /* destreg - send, conditionalmod - others */
-      unsigned pad0:2;
-      unsigned debug_control:1;
-      unsigned saturate:1;
-   } header;
-
-   union {
-      struct
-      {
-	 unsigned dest_reg_file:2;
-	 unsigned dest_reg_type:3;
-	 unsigned src0_reg_file:2;
-	 unsigned src0_reg_type:3;
-	 unsigned src1_reg_file:2;
-	 unsigned src1_reg_type:3;
-	 unsigned pad:1;
-	 unsigned dest_subreg_nr:5;
-	 unsigned dest_reg_nr:8;
-	 unsigned dest_horiz_stride:2;
-	 unsigned dest_address_mode:1;
-      } da1;
-
-      struct
-      {
-	 unsigned dest_reg_file:2;
-	 unsigned dest_reg_type:3;
-	 unsigned src0_reg_file:2;
-	 unsigned src0_reg_type:3;
-	 unsigned pad:6;
-	 int dest_indirect_offset:10;	/* offset against the deref'd address reg */
-	 unsigned dest_subreg_nr:3; /* subnr for the address reg a0.x */
-	 unsigned dest_horiz_stride:2;
-	 unsigned dest_address_mode:1;
-      } ia1;
-
-      struct
-      {
-	 unsigned dest_reg_file:2;
-	 unsigned dest_reg_type:3;
-	 unsigned src0_reg_file:2;
-	 unsigned src0_reg_type:3;
-	 unsigned src1_reg_file:2;
-	 unsigned src1_reg_type:3;
-	 unsigned pad0:1;
-	 unsigned dest_writemask:4;
-	 unsigned dest_subreg_nr:1;
-	 unsigned dest_reg_nr:8;
-	 unsigned pad1:2;
-	 unsigned dest_address_mode:1;
-      } da16;
-
-      struct
-      {
-	 unsigned dest_reg_file:2;
-	 unsigned dest_reg_type:3;
-	 unsigned src0_reg_file:2;
-	 unsigned src0_reg_type:3;
-	 unsigned pad0:6;
-	 unsigned dest_writemask:4;
-	 int dest_indirect_offset:6;
-	 unsigned dest_subreg_nr:3;
-	 unsigned pad1:2;
-	 unsigned dest_address_mode:1;
-      } ia16;
-   } bits1;
-
-
-   union {
-      struct
-      {
-	 unsigned src0_subreg_nr:5;
-	 unsigned src0_reg_nr:8;
-	 unsigned src0_abs:1;
-	 unsigned src0_negate:1;
-	 unsigned src0_address_mode:1;
-	 unsigned src0_horiz_stride:2;
-	 unsigned src0_width:3;
-	 unsigned src0_vert_stride:4;
-	 unsigned flag_reg_nr:1;
-	 unsigned pad:6;
-      } da1;
-
-      struct
-      {
-	 int src0_indirect_offset:10;
-	 unsigned src0_subreg_nr:3;
-	 unsigned src0_abs:1;
-	 unsigned src0_negate:1;
-	 unsigned src0_address_mode:1;
-	 unsigned src0_horiz_stride:2;
-	 unsigned src0_width:3;
-	 unsigned src0_vert_stride:4;
-	 unsigned flag_reg_nr:1;
-	 unsigned pad:6;
-      } ia1;
-
-      struct
-      {
-	 unsigned src0_swz_x:2;
-	 unsigned src0_swz_y:2;
-	 unsigned src0_subreg_nr:1;
-	 unsigned src0_reg_nr:8;
-	 unsigned src0_abs:1;
-	 unsigned src0_negate:1;
-	 unsigned src0_address_mode:1;
-	 unsigned src0_swz_z:2;
-	 unsigned src0_swz_w:2;
-	 unsigned pad0:1;
-	 unsigned src0_vert_stride:4;
-	 unsigned flag_reg_nr:1;
-	 unsigned pad1:6;
-      } da16;
-
-      struct
-      {
-	 unsigned src0_swz_x:2;
-	 unsigned src0_swz_y:2;
-	 int src0_indirect_offset:6;
-	 unsigned src0_subreg_nr:3;
-	 unsigned src0_abs:1;
-	 unsigned src0_negate:1;
-	 unsigned src0_address_mode:1;
-	 unsigned src0_swz_z:2;
-	 unsigned src0_swz_w:2;
-	 unsigned pad0:1;
-	 unsigned src0_vert_stride:4;
-	 unsigned flag_reg_nr:1;
-	 unsigned pad1:6;
-      } ia16;
-
-   } bits2;
-
-   union
-   {
-      struct
-      {
-	 unsigned src1_subreg_nr:5;
-	 unsigned src1_reg_nr:8;
-	 unsigned src1_abs:1;
-	 unsigned src1_negate:1;
-	 unsigned pad:1;
-	 unsigned src1_horiz_stride:2;
-	 unsigned src1_width:3;
-	 unsigned src1_vert_stride:4;
-	 unsigned pad0:7;
-      } da1;
-
-      struct
-      {
-	 unsigned src1_swz_x:2;
-	 unsigned src1_swz_y:2;
-	 unsigned src1_subreg_nr:1;
-	 unsigned src1_reg_nr:8;
-	 unsigned src1_abs:1;
-	 unsigned src1_negate:1;
-	 unsigned pad0:1;
-	 unsigned src1_swz_z:2;
-	 unsigned src1_swz_w:2;
-	 unsigned pad1:1;
-	 unsigned src1_vert_stride:4;
-	 unsigned pad2:7;
-      } da16;
-
-      struct
-      {
-	 int  src1_indirect_offset:10;
-	 unsigned src1_subreg_nr:3;
-	 unsigned src1_abs:1;
-	 unsigned src1_negate:1;
-	 unsigned pad0:1;
-	 unsigned src1_horiz_stride:2;
-	 unsigned src1_width:3;
-	 unsigned src1_vert_stride:4;
-	 unsigned flag_reg_nr:1;
-	 unsigned pad1:6;
-      } ia1;
-
-      struct
-      {
-	 unsigned src1_swz_x:2;
-	 unsigned src1_swz_y:2;
-	 int  src1_indirect_offset:6;
-	 unsigned src1_subreg_nr:3;
-	 unsigned src1_abs:1;
-	 unsigned src1_negate:1;
-	 unsigned pad0:1;
-	 unsigned src1_swz_z:2;
-	 unsigned src1_swz_w:2;
-	 unsigned pad1:1;
-	 unsigned src1_vert_stride:4;
-	 unsigned flag_reg_nr:1;
-	 unsigned pad2:6;
-      } ia16;
-
-
-      struct
-      {
-	 int  jump_count:16;	/* note: signed */
-	 unsigned  pop_count:4;
-	 unsigned  pad0:12;
-      } if_else;
-
-      struct {
-	 unsigned function:4;
-	 unsigned int_type:1;
-	 unsigned precision:1;
-	 unsigned saturate:1;
-	 unsigned data_type:1;
-	 unsigned pad0:8;
-	 unsigned response_length:4;
-	 unsigned msg_length:4;
-	 unsigned msg_target:4;
-	 unsigned pad1:3;
-	 unsigned end_of_thread:1;
-      } math;
-
-      struct {
-	 unsigned binding_table_index:8;
-	 unsigned sampler:4;
-	 unsigned return_format:2;
-	 unsigned msg_type:2;
-	 unsigned response_length:4;
-	 unsigned msg_length:4;
-	 unsigned msg_target:4;
-	 unsigned pad1:3;
-	 unsigned end_of_thread:1;
-      } sampler;
-
-      struct brw_urb_immediate urb;
-
-      struct {
-	 unsigned binding_table_index:8;
-	 unsigned msg_control:4;
-	 unsigned msg_type:2;
-	 unsigned target_cache:2;
-	 unsigned response_length:4;
-	 unsigned msg_length:4;
-	 unsigned msg_target:4;
-	 unsigned pad1:3;
-	 unsigned end_of_thread:1;
-      } dp_read;
-
-      struct {
-	 unsigned binding_table_index:8;
-	 unsigned msg_control:3;
-	 unsigned pixel_scoreboard_clear:1;
-	 unsigned msg_type:3;
-	 unsigned send_commit_msg:1;
-	 unsigned response_length:4;
-	 unsigned msg_length:4;
-	 unsigned msg_target:4;
-	 unsigned pad1:3;
-	 unsigned end_of_thread:1;
-      } dp_write;
-
-      struct {
-	 unsigned pad:16;
-	 unsigned response_length:4;
-	 unsigned msg_length:4;
-	 unsigned msg_target:4;
-	 unsigned pad1:3;
-	 unsigned end_of_thread:1;
-      } generic;
-
-      int d;
-      unsigned ud;
-   } bits3;
-};
-
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_surface.c b/src/gallium/drivers/i965simple/brw_surface.c
deleted file mode 100644
index 724a69b2ee..0000000000
--- a/src/gallium/drivers/i965simple/brw_surface.c
+++ /dev/null
@@ -1,126 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "brw_blit.h"
-#include "brw_context.h"
-#include "brw_state.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_tile.h"
-#include "util/u_rect.h"
-
-
-
-/* Assumes all values are within bounds -- no checking at this level -
- * do it higher up if required.
- */
-static void
-brw_surface_copy(struct pipe_context *pipe,
-                 struct pipe_surface *dst,
-                 unsigned dstx, unsigned dsty,
-                 struct pipe_surface *src,
-                 unsigned srcx, unsigned srcy, unsigned width, unsigned height)
-{
-   assert( dst != src );
-   assert( dst->block.size == src->block.size );
-   assert( dst->block.width == src->block.height );
-   assert( dst->block.height == src->block.height );
-
-   if (0) {
-      void *dst_map = pipe->screen->surface_map( pipe->screen,
-                                                 dst,
-                                                 PIPE_BUFFER_USAGE_CPU_WRITE );
-      
-      const void *src_map = pipe->screen->surface_map( pipe->screen,
-                                                       src,
-                                                       PIPE_BUFFER_USAGE_CPU_READ );
-      
-      util_copy_rect(dst_map,
-                     &dst->block,
-                     dst->stride,
-                     dstx, dsty,
-                     width, height,
-                     src_map,
-                     src->stride,
-                     srcx, srcy);
-
-      pipe->screen->surface_unmap(pipe->screen, src);
-      pipe->screen->surface_unmap(pipe->screen, dst);
-   }
-   else {
-      struct brw_texture *dst_tex = (struct brw_texture *)dst->texture;
-      struct brw_texture *src_tex = (struct brw_texture *)src->texture;
-      assert(dst->block.width == 1);
-      assert(dst->block.height == 1);
-      brw_copy_blit(brw_context(pipe),
-                    FALSE,
-                    dst->block.size,
-                    (short) src->stride/src->block.size, src_tex->buffer, src->offset, FALSE,
-                    (short) dst->stride/dst->block.size, dst_tex->buffer, dst->offset, FALSE,
-                    (short) srcx, (short) srcy, (short) dstx, (short) dsty,
-                    (short) width, (short) height, PIPE_LOGICOP_COPY);
-   }
-}
-
-
-static void
-brw_surface_fill(struct pipe_context *pipe,
-                 struct pipe_surface *dst,
-                 unsigned dstx, unsigned dsty,
-                 unsigned width, unsigned height, unsigned value)
-{
-   if (0) {
-      void *dst_map = pipe->screen->surface_map( pipe->screen,
-                                                 dst,
-                                                 PIPE_BUFFER_USAGE_CPU_WRITE );
-
-      util_fill_rect(dst_map, &dst->block, dst->stride, dstx, dsty, width, height, value);
-
-      pipe->screen->surface_unmap(pipe->screen, dst);
-   }
-   else {
-      struct brw_texture *tex = (struct brw_texture *)dst->texture;
-      assert(dst->block.width == 1);
-      assert(dst->block.height == 1);
-      brw_fill_blit(brw_context(pipe),
-                    dst->block.size,
-                    (short) dst->stride/dst->block.size, 
-                    tex->buffer, dst->offset, FALSE,
-                    (short) dstx, (short) dsty,
-                    (short) width, (short) height,
-                    value);
-   }
-}
-
-
-void
-brw_init_surface_functions(struct brw_context *brw)
-{
-   brw->pipe.surface_copy  = brw_surface_copy;
-   brw->pipe.surface_fill  = brw_surface_fill;
-}
diff --git a/src/gallium/drivers/i965simple/brw_tex_layout.c b/src/gallium/drivers/i965simple/brw_tex_layout.c
deleted file mode 100644
index 998ffaeac4..0000000000
--- a/src/gallium/drivers/i965simple/brw_tex_layout.c
+++ /dev/null
@@ -1,380 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-/* Code to layout images in a mipmap tree for i965.
- */
-
-#include "pipe/p_state.h"
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "brw_context.h"
-#include "brw_tex_layout.h"
-
-
-#define FILE_DEBUG_FLAG DEBUG_TEXTURE
-
-#if 0
-unsigned intel_compressed_alignment(unsigned internalFormat)
-{
-    unsigned alignment = 4;
-
-    switch (internalFormat) {
-    case GL_COMPRESSED_RGB_FXT1_3DFX:
-    case GL_COMPRESSED_RGBA_FXT1_3DFX:
-        alignment = 8;
-        break;
-
-    default:
-        break;
-    }
-
-    return alignment;
-}
-#endif
-
-
-static void intel_miptree_set_image_offset(struct brw_texture *tex,
-                                           unsigned level,
-                                           unsigned img,
-                                           unsigned x, unsigned y)
-{
-   struct pipe_texture *pt = &tex->base;
-   if (img == 0 && level == 0)
-      assert(x == 0 && y == 0);
-   assert(img < tex->nr_images[level]);
-
-   tex->image_offset[level][img] = y * tex->stride + x * pt->block.size;
-}
-
-static void intel_miptree_set_level_info(struct brw_texture *tex,
-                                         unsigned level,
-                                         unsigned nr_images,
-                                         unsigned x, unsigned y,
-                                         unsigned w, unsigned h, unsigned d)
-{
-   struct pipe_texture *pt = &tex->base;
-
-   assert(level < PIPE_MAX_TEXTURE_LEVELS);
-
-   pt->width[level] = w;
-   pt->height[level] = h;
-   pt->depth[level] = d;
-   
-   pt->nblocksx[level] = pf_get_nblocksx(&pt->block, w);
-   pt->nblocksy[level] = pf_get_nblocksy(&pt->block, h);
-
-   tex->level_offset[level] = y * tex->stride + x * tex->base.block.size;
-   tex->nr_images[level] = nr_images;
-
-   /*
-   DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__,
-       level, w, h, d, x, y, tex->level_offset[level]);
-   */
-
-   /* Not sure when this would happen, but anyway: 
-    */
-   if (tex->image_offset[level]) {
-      FREE(tex->image_offset[level]);
-      tex->image_offset[level] = NULL;
-   }
-
-   assert(nr_images);
-   assert(!tex->image_offset[level]);
-
-   tex->image_offset[level] = (unsigned *) MALLOC(nr_images * sizeof(unsigned));
-   tex->image_offset[level][0] = 0;
-}
-
-static void i945_miptree_layout_2d(struct brw_texture *tex)
-{
-   struct pipe_texture *pt = &tex->base;
-   const int align_x = 2, align_y = 4;
-   unsigned level;
-   unsigned x = 0;
-   unsigned y = 0;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned nblocksx = pt->nblocksx[0];
-   unsigned nblocksy = pt->nblocksy[0];
-
-   tex->stride = align(pt->nblocksx[0] * pt->block.size, 4);
-
-   /* May need to adjust pitch to accomodate the placement of
-    * the 2nd mipmap level.  This occurs when the alignment
-    * constraints of mipmap placement push the right edge of the
-    * 2nd mipmap level out past the width of its parent.
-    */
-   if (pt->last_level > 0) {
-      unsigned mip1_nblocksx 
-	 = align(pf_get_nblocksx(&pt->block, minify(width)), align_x)
-         + pf_get_nblocksx(&pt->block, minify(minify(width)));
-
-      if (mip1_nblocksx > nblocksx)
-	 tex->stride = mip1_nblocksx * pt->block.size;
-   }
-
-   /* Pitch must be a whole number of dwords
-    */
-   tex->stride = align(tex->stride, 64);
-   tex->total_nblocksy = 0;
-
-   for (level = 0; level <= pt->last_level; level++) {
-      intel_miptree_set_level_info(tex, level, 1, x, y, width,
-				   height, 1);
-
-      nblocksy = align(nblocksy, align_y);
-
-      /* Because the images are packed better, the final offset
-       * might not be the maximal one:
-       */
-      tex->total_nblocksy = MAX2(tex->total_nblocksy, y + nblocksy);
-
-      /* Layout_below: step right after second mipmap level.
-       */
-      if (level == 1) {
-	 x += align(nblocksx, align_x);
-      }
-      else {
-	 y += nblocksy;
-      }
-
-      width  = minify(width);
-      height = minify(height);
-      nblocksx = pf_get_nblocksx(&pt->block, width);
-      nblocksy = pf_get_nblocksy(&pt->block, height);
-   }
-}
-
-static boolean brw_miptree_layout(struct brw_texture *tex)
-{
-   struct pipe_texture *pt = &tex->base;
-   /* XXX: these vary depending on image format:
-    */
-/*    int align_w = 4; */
-
-   switch (pt->target) {
-   case PIPE_TEXTURE_CUBE:
-   case PIPE_TEXTURE_3D: {
-      unsigned width  = pt->width[0];
-      unsigned height = pt->height[0];
-      unsigned depth = pt->depth[0];
-      unsigned nblocksx = pt->nblocksx[0];
-      unsigned nblocksy = pt->nblocksy[0];
-      unsigned pack_x_pitch, pack_x_nr;
-      unsigned pack_y_pitch;
-      unsigned level;
-      unsigned align_h = 2;
-      unsigned align_w = 4;
-
-      tex->total_nblocksy = 0;
-
-      tex->stride = align(pt->nblocksx[0], 4);
-      pack_y_pitch = align(pt->nblocksy[0], align_h);
-
-      pack_x_pitch = tex->stride / pt->block.size;
-      pack_x_nr = 1;
-
-      for (level = 0; level <= pt->last_level; level++) {
-	 unsigned nr_images = pt->target == PIPE_TEXTURE_3D ? depth : 6;
-	 int x = 0;
-	 int y = 0;
-	 uint q, j;
-
-	 intel_miptree_set_level_info(tex, level, nr_images,
-				      0, tex->total_nblocksy,
-				      width, height, depth);
-
-	 for (q = 0; q < nr_images;) {
-	    for (j = 0; j < pack_x_nr && q < nr_images; j++, q++) {
-	       intel_miptree_set_image_offset(tex, level, q, x, y);
-	       x += pack_x_pitch;
-	    }
-
-	    x = 0;
-	    y += pack_y_pitch;
-	 }
-
-
-	 tex->total_nblocksy += y;
-	 width  = minify(width);
-	 height = minify(height);
-	 depth  = minify(depth);
-         nblocksx = pf_get_nblocksx(&pt->block, width);
-         nblocksy = pf_get_nblocksy(&pt->block, height);
-
-         if (pf_is_compressed(pt->format)) {
-            pack_y_pitch = (height + 3) / 4;
-
-            if (pack_x_pitch > align(width, align_w)) {
-               pack_x_pitch = align(width, align_w);
-               pack_x_nr <<= 1;
-            }
-         } else {
-            if (pack_x_pitch > 4) {
-               pack_x_pitch >>= 1;
-               pack_x_nr <<= 1;
-               assert(pack_x_pitch * pack_x_nr * pt->block.size <= tex->stride);
-            }
-
-            if (pack_y_pitch > 2) {
-               pack_y_pitch >>= 1;
-               pack_y_pitch = align(pack_y_pitch, align_h);
-            }
-         }
-
-      }
-      break;
-   }
-
-   default:
-      i945_miptree_layout_2d(tex);
-      break;
-   }
-#if 0
-   PRINT("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__,
-       pt->pitch,
-       pt->total_nblocksy,
-       pt->block.size,
-       pt->stride * pt->total_nblocksy );
-#endif
-
-   return TRUE;
-}
-
-
-static struct pipe_texture *
-brw_texture_create_screen(struct pipe_screen *screen,
-                          const struct pipe_texture *templat)
-{
-   struct brw_texture *tex = CALLOC_STRUCT(brw_texture);
-
-   if (tex) {
-      tex->base = *templat;
-      pipe_reference_init(&tex->base.reference, 1);
-
-      tex->base.nblocksx[0] = pf_get_nblocksx(&tex->base.block, tex->base.width[0]);
-      tex->base.nblocksy[0] = pf_get_nblocksy(&tex->base.block, tex->base.height[0]);
-   
-      if (brw_miptree_layout(tex))
-	 tex->buffer = screen->buffer_create(screen, 64,
-                                          PIPE_BUFFER_USAGE_PIXEL,
-                                          tex->stride *
-                                          tex->total_nblocksy);
-
-      if (!tex->buffer) {
-	 FREE(tex);
-         return NULL;
-      }
-   }
-
-   return &tex->base;
-}
-
-
-static void
-brw_texture_destroy_screen(struct pipe_texture *pt)
-{
-   struct brw_texture *tex = (struct brw_texture *)pt;
-   uint i;
-
-   /*
-     DBG("%s deleting %p\n", __FUNCTION__, (void *) tex);
-   */
-
-   pipe_buffer_reference(&tex->buffer, NULL);
-
-   for (i = 0; i < PIPE_MAX_TEXTURE_LEVELS; i++)
-      if (tex->image_offset[i])
-         free(tex->image_offset[i]);
-
-   free(tex);
-}
-
-
-static struct pipe_surface *
-brw_get_tex_surface_screen(struct pipe_screen *screen,
-                           struct pipe_texture *pt,
-                           unsigned face, unsigned level, unsigned zslice)
-{
-   struct brw_texture *tex = (struct brw_texture *)pt;
-   struct pipe_surface *ps;
-   unsigned offset;  /* in bytes */
-
-   offset = tex->level_offset[level];
-
-   if (pt->target == PIPE_TEXTURE_CUBE) {
-      offset += tex->image_offset[level][face];
-   }
-   else if (pt->target == PIPE_TEXTURE_3D) {
-      offset += tex->image_offset[level][zslice];
-   }
-   else {
-      assert(face == 0);
-      assert(zslice == 0);
-   }
-
-   ps = CALLOC_STRUCT(pipe_surface);
-   if (ps) {
-      pipe_reference_init(&ps->reference, 1);
-      pipe_texture_reference(&ps->texture, pt);
-      ps->format = pt->format;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
-      ps->block = pt->block;
-      ps->nblocksx = pt->nblocksx[level];
-      ps->nblocksy = pt->nblocksy[level];
-      ps->stride = tex->stride;
-      ps->offset = offset;
-   }
-   return ps;
-}
-
-
-void
-brw_init_texture_functions(struct brw_context *brw)
-{
-//   brw->pipe.texture_update = brw_texture_update;
-}
-
-
-void
-brw_init_screen_texture_funcs(struct pipe_screen *screen)
-{
-   screen->texture_create  = brw_texture_create_screen;
-   screen->texture_destroy = brw_texture_destroy_screen;
-   screen->get_tex_surface = brw_get_tex_surface_screen;
-}
-
diff --git a/src/gallium/drivers/i965simple/brw_tex_layout.h b/src/gallium/drivers/i965simple/brw_tex_layout.h
deleted file mode 100644
index a6b6ba8146..0000000000
--- a/src/gallium/drivers/i965simple/brw_tex_layout.h
+++ /dev/null
@@ -1,44 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
-
-
-#ifndef BRW_TEX_LAYOUT_H
-#define BRW_TEX_LAYOUT_H
-
-
-struct brw_context;
-struct pipe_screen;
-
-
-extern void
-brw_init_texture_functions(struct brw_context *brw);
-
-extern void
-brw_init_screen_texture_funcs(struct pipe_screen *screen);
-
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_urb.c b/src/gallium/drivers/i965simple/brw_urb.c
deleted file mode 100644
index 101a4367b9..0000000000
--- a/src/gallium/drivers/i965simple/brw_urb.c
+++ /dev/null
@@ -1,186 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-//#include "brw_state.h"
-#include "brw_batch.h"
-#include "brw_defines.h"
-
-#define VS 0
-#define GS 1
-#define CLP 2
-#define SF 3
-#define CS 4
-
-/* XXX: Are the min_entry_size numbers useful?
- * XXX: Verify min_nr_entries, esp for VS.
- * XXX: Verify SF min_entry_size.
- */
-static const struct {
-   unsigned min_nr_entries;
-   unsigned preferred_nr_entries;
-   unsigned min_entry_size;
-   unsigned max_entry_size;
-} limits[CS+1] = {
-   { 8, 32, 1, 5 },			/* vs */
-   { 4, 8,  1, 5 },			/* gs */
-   { 6, 8,  1, 5 },			/* clp */
-   { 1, 8,  1, 12 },		        /* sf */
-   { 1, 4,  1, 32 }			/* cs */
-};
-
-
-static boolean check_urb_layout( struct brw_context *brw )
-{
-   brw->urb.vs_start = 0;
-   brw->urb.gs_start = brw->urb.nr_vs_entries * brw->urb.vsize;
-   brw->urb.clip_start = brw->urb.gs_start + brw->urb.nr_gs_entries * brw->urb.vsize;
-   brw->urb.sf_start = brw->urb.clip_start + brw->urb.nr_clip_entries * brw->urb.vsize;
-   brw->urb.cs_start = brw->urb.sf_start + brw->urb.nr_sf_entries * brw->urb.sfsize;
-
-   return brw->urb.cs_start + brw->urb.nr_cs_entries * brw->urb.csize <= 256;
-}
-
-/* Most minimal update, forces re-emit of URB fence packet after GS
- * unit turned on/off.
- */
-static void recalculate_urb_fence( struct brw_context *brw )
-{
-   unsigned csize = brw->curbe.total_size;
-   unsigned vsize = brw->vs.prog_data->urb_entry_size;
-   unsigned sfsize = brw->sf.prog_data->urb_entry_size;
-
-   if (csize < limits[CS].min_entry_size)
-      csize = limits[CS].min_entry_size;
-
-   if (vsize < limits[VS].min_entry_size)
-      vsize = limits[VS].min_entry_size;
-
-   if (sfsize < limits[SF].min_entry_size)
-      sfsize = limits[SF].min_entry_size;
-
-   if (brw->urb.vsize < vsize ||
-       brw->urb.sfsize < sfsize ||
-       brw->urb.csize < csize ||
-       (brw->urb.constrained && (brw->urb.vsize > brw->urb.vsize ||
-				 brw->urb.sfsize > brw->urb.sfsize ||
-				 brw->urb.csize > brw->urb.csize))) {
-
-
-      brw->urb.csize = csize;
-      brw->urb.sfsize = sfsize;
-      brw->urb.vsize = vsize;
-
-      brw->urb.nr_vs_entries = limits[VS].preferred_nr_entries;
-      brw->urb.nr_gs_entries = limits[GS].preferred_nr_entries;
-      brw->urb.nr_clip_entries = limits[CLP].preferred_nr_entries;
-      brw->urb.nr_sf_entries = limits[SF].preferred_nr_entries;
-      brw->urb.nr_cs_entries = limits[CS].preferred_nr_entries;
-
-      if (!check_urb_layout(brw)) {
-	 brw->urb.nr_vs_entries = limits[VS].min_nr_entries;
-	 brw->urb.nr_gs_entries = limits[GS].min_nr_entries;
-	 brw->urb.nr_clip_entries = limits[CLP].min_nr_entries;
-	 brw->urb.nr_sf_entries = limits[SF].min_nr_entries;
-	 brw->urb.nr_cs_entries = limits[CS].min_nr_entries;
-
-	 brw->urb.constrained = 1;
-
-	 if (!check_urb_layout(brw)) {
-	    /* This is impossible, given the maximal sizes of urb
-	     * entries and the values for minimum nr of entries
-	     * provided above.
-	     */
-	    debug_printf("couldn't calculate URB layout!\n");
-	    exit(1);
-	 }
-
-	 if (BRW_DEBUG & (DEBUG_URB|DEBUG_FALLBACKS))
-	    debug_printf("URB CONSTRAINED\n");
-      }
-      else
-	 brw->urb.constrained = 0;
-
-      if (BRW_DEBUG & DEBUG_URB)
-	 debug_printf("URB fence: %d ..VS.. %d ..GS.. %d ..CLP.. %d ..SF.. %d ..CS.. %d\n",
-		      brw->urb.vs_start,
-		      brw->urb.gs_start,
-		      brw->urb.clip_start,
-		      brw->urb.sf_start,
-		      brw->urb.cs_start,
-		      256);
-
-      brw->state.dirty.brw |= BRW_NEW_URB_FENCE;
-   }
-}
-
-
-const struct brw_tracked_state brw_recalculate_urb_fence = {
-   .dirty = {
-      .brw = BRW_NEW_CURBE_OFFSETS,
-      .cache = (CACHE_NEW_VS_PROG |
-		CACHE_NEW_SF_PROG)
-   },
-   .update = recalculate_urb_fence
-};
-
-
-
-
-
-void brw_upload_urb_fence(struct brw_context *brw)
-{
-   struct brw_urb_fence uf;
-   memset(&uf, 0, sizeof(uf));
-
-   uf.header.opcode = CMD_URB_FENCE;
-   uf.header.length = sizeof(uf)/4-2;
-   uf.header.vs_realloc = 1;
-   uf.header.gs_realloc = 1;
-   uf.header.clp_realloc = 1;
-   uf.header.sf_realloc = 1;
-   uf.header.vfe_realloc = 1;
-   uf.header.cs_realloc = 1;
-
-   /* The ordering below is correct, not the layout in the
-    * instruction.
-    *
-    * There are 256 urb reg pairs in total.
-    */
-   uf.bits0.vs_fence  = brw->urb.gs_start;
-   uf.bits0.gs_fence  = brw->urb.clip_start;
-   uf.bits0.clp_fence = brw->urb.sf_start;
-   uf.bits1.sf_fence  = brw->urb.cs_start;
-   uf.bits1.cs_fence  = 256;
-
-   BRW_BATCH_STRUCT(brw, &uf);
-}
diff --git a/src/gallium/drivers/i965simple/brw_util.c b/src/gallium/drivers/i965simple/brw_util.c
deleted file mode 100644
index 42391d7c8c..0000000000
--- a/src/gallium/drivers/i965simple/brw_util.c
+++ /dev/null
@@ -1,104 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_util.h"
-#include "brw_defines.h"
-
-#include "pipe/p_defines.h"
-
-unsigned brw_count_bits( unsigned val )
-{
-   unsigned i;
-   for (i = 0; val ; val >>= 1)
-      if (val & 1)
-	 i++;
-   return i;
-}
-
-
-unsigned brw_translate_blend_equation( int mode )
-{
-   switch (mode) {
-   case PIPE_BLEND_ADD:
-      return BRW_BLENDFUNCTION_ADD;
-   case PIPE_BLEND_MIN:
-      return BRW_BLENDFUNCTION_MIN;
-   case PIPE_BLEND_MAX:
-      return BRW_BLENDFUNCTION_MAX;
-   case PIPE_BLEND_SUBTRACT:
-      return BRW_BLENDFUNCTION_SUBTRACT;
-   case PIPE_BLEND_REVERSE_SUBTRACT:
-      return BRW_BLENDFUNCTION_REVERSE_SUBTRACT;
-   default:
-      assert(0);
-      return BRW_BLENDFUNCTION_ADD;
-   }
-}
-
-unsigned brw_translate_blend_factor( int factor )
-{
-   switch(factor) {
-   case PIPE_BLENDFACTOR_ZERO:
-      return BRW_BLENDFACTOR_ZERO;
-   case PIPE_BLENDFACTOR_SRC_ALPHA:
-      return BRW_BLENDFACTOR_SRC_ALPHA;
-   case PIPE_BLENDFACTOR_ONE:
-      return BRW_BLENDFACTOR_ONE;
-   case PIPE_BLENDFACTOR_SRC_COLOR:
-      return BRW_BLENDFACTOR_SRC_COLOR;
-   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
-      return BRW_BLENDFACTOR_INV_SRC_COLOR;
-   case PIPE_BLENDFACTOR_DST_COLOR:
-      return BRW_BLENDFACTOR_DST_COLOR;
-   case PIPE_BLENDFACTOR_INV_DST_COLOR:
-      return BRW_BLENDFACTOR_INV_DST_COLOR;
-   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-      return BRW_BLENDFACTOR_INV_SRC_ALPHA;
-   case PIPE_BLENDFACTOR_DST_ALPHA:
-      return BRW_BLENDFACTOR_DST_ALPHA;
-   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
-      return BRW_BLENDFACTOR_INV_DST_ALPHA;
-   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
-      return BRW_BLENDFACTOR_SRC_ALPHA_SATURATE;
-   case PIPE_BLENDFACTOR_CONST_COLOR:
-      return BRW_BLENDFACTOR_CONST_COLOR;
-   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
-      return BRW_BLENDFACTOR_INV_CONST_COLOR;
-   case PIPE_BLENDFACTOR_CONST_ALPHA:
-      return BRW_BLENDFACTOR_CONST_ALPHA;
-   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
-      return BRW_BLENDFACTOR_INV_CONST_ALPHA;
-   default:
-      assert(0);
-      return BRW_BLENDFACTOR_ZERO;
-   }
-}
diff --git a/src/gallium/drivers/i965simple/brw_util.h b/src/gallium/drivers/i965simple/brw_util.h
deleted file mode 100644
index d60e5934db..0000000000
--- a/src/gallium/drivers/i965simple/brw_util.h
+++ /dev/null
@@ -1,43 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-          
-
-#ifndef BRW_UTIL_H
-#define BRW_UTIL_H
-
-#include "pipe/p_state.h"
-
-extern unsigned brw_count_bits( unsigned val );
-extern unsigned brw_translate_blend_factor( int factor );
-extern unsigned brw_translate_blend_equation( int mode );
-
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_vs.c b/src/gallium/drivers/i965simple/brw_vs.c
deleted file mode 100644
index 92327e896d..0000000000
--- a/src/gallium/drivers/i965simple/brw_vs.c
+++ /dev/null
@@ -1,120 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_vs.h"
-#include "brw_util.h"
-#include "brw_state.h"
-
-
-static void do_vs_prog( struct brw_context *brw,
-			const struct brw_vertex_program *vp,
-			struct brw_vs_prog_key *key )
-{
-   unsigned program_size;
-   const unsigned *program;
-   struct brw_vs_compile c;
-
-   memset(&c, 0, sizeof(c));
-   memcpy(&c.key, key, sizeof(*key));
-
-   brw_init_compile(&c.func);
-   c.vp = vp;
-
-   c.prog_data.outputs_written = vp->info.num_outputs;
-   c.prog_data.inputs_read = vp->info.num_inputs;
-
-#if 0
-   if (c.key.copy_edgeflag) {
-      c.prog_data.outputs_written |= 1<<VERT_RESULT_EDGE;
-      c.prog_data.inputs_read |= 1<<VERT_ATTRIB_EDGEFLAG;
-   }
-#endif
-
-   /* Emit GEN4 code.
-    */
-   brw_vs_emit(&c);
-
-   /* get the program
-    */
-   program = brw_get_program(&c.func, &program_size);
-
-   /*
-    */
-   brw->vs.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_VS_PROG],
-					      &c.key,
-					      sizeof(c.key),
-					      program,
-					      program_size,
-					      &c.prog_data,
-					      &brw->vs.prog_data);
-}
-
-
-static void brw_upload_vs_prog( struct brw_context *brw )
-{
-   struct brw_vs_prog_key key;
-   const struct brw_vertex_program *vp = brw->attribs.VertexProgram;
-
-   assert(vp);
-
-   memset(&key, 0, sizeof(key));
-
-   /* Just upload the program verbatim for now.  Always send it all
-    * the inputs it asks for, whether they are varying or not.
-    */
-   key.program_string_id = vp->id;
-   key.nr_userclip = brw->attribs.Clip.nr;
-   key.copy_edgeflag = (brw->attribs.Raster->fill_cw != PIPE_POLYGON_MODE_FILL ||
-			brw->attribs.Raster->fill_ccw != PIPE_POLYGON_MODE_FILL);
-
-   /* Make an early check for the key.
-    */
-   if (brw_search_cache(&brw->cache[BRW_VS_PROG],
-			&key, sizeof(key),
-			&brw->vs.prog_data,
-			&brw->vs.prog_gs_offset))
-       return;
-
-   do_vs_prog(brw, vp, &key);
-}
-
-
-/* See brw_vs.c:
- */
-const struct brw_tracked_state brw_vs_prog = {
-   .dirty = {
-      .brw   = BRW_NEW_VS,
-      .cache = 0
-   },
-   .update = brw_upload_vs_prog
-};
diff --git a/src/gallium/drivers/i965simple/brw_vs.h b/src/gallium/drivers/i965simple/brw_vs.h
deleted file mode 100644
index 070f9dfcae..0000000000
--- a/src/gallium/drivers/i965simple/brw_vs.h
+++ /dev/null
@@ -1,82 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#ifndef BRW_VS_H
-#define BRW_VS_H
-
-
-#include "brw_context.h"
-#include "brw_eu.h"
-
-
-struct brw_vs_prog_key {
-   unsigned program_string_id;
-   unsigned nr_userclip:4;
-   unsigned copy_edgeflag:1;
-   unsigned know_w_is_one:1;
-   unsigned pad:26;
-};
-
-
-struct brw_vs_compile {
-   struct brw_compile func;
-   struct brw_vs_prog_key key;
-   struct brw_vs_prog_data prog_data;
-
-   const struct brw_vertex_program *vp;
-
-   unsigned nr_inputs;
-
-   unsigned first_output;
-   unsigned nr_outputs;
-
-   unsigned first_tmp;
-   unsigned last_tmp;
-
-   struct brw_reg r0;
-   struct brw_reg r1;
-   struct brw_reg regs[12][128];
-   struct brw_reg tmp;
-   struct brw_reg stack;
-
-   struct {
-       boolean used_in_src;
-       struct brw_reg reg;
-   } output_regs[128];
-
-   struct brw_reg userplane[6];
-
-};
-
-void brw_vs_emit( struct brw_vs_compile *c );
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_vs_emit.c b/src/gallium/drivers/i965simple/brw_vs_emit.c
deleted file mode 100644
index 3ee82d95b3..0000000000
--- a/src/gallium/drivers/i965simple/brw_vs_emit.c
+++ /dev/null
@@ -1,1330 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_context.h"
-#include "brw_vs.h"
-
-#include "pipe/p_shader_tokens.h"
-#include "tgsi/tgsi_parse.h"
-
-struct brw_prog_info {
-   unsigned num_temps;
-   unsigned num_addrs;
-   unsigned num_consts;
-
-   unsigned writes_psize;
-
-   unsigned pos_idx;
-   unsigned result_edge_idx;
-   unsigned edge_flag_idx;
-   unsigned psize_idx;
-};
-
-/* Do things as simply as possible.  Allocate and populate all regs
- * ahead of time.
- */
-static void brw_vs_alloc_regs( struct brw_vs_compile *c,
-                               struct brw_prog_info *info )
-{
-   unsigned i, reg = 0, mrf;
-   unsigned nr_params;
-
-   /* r0 -- reserved as usual
-    */
-   c->r0 = brw_vec8_grf(reg, 0); reg++;
-
-   /* User clip planes from curbe:
-    */
-   if (c->key.nr_userclip) {
-      for (i = 0; i < c->key.nr_userclip; i++) {
-	 c->userplane[i] = stride( brw_vec4_grf(reg+3+i/2, (i%2) * 4), 0, 4, 1);
-      }
-
-      /* Deal with curbe alignment:
-       */
-      reg += ((6+c->key.nr_userclip+3)/4)*2;
-   }
-
-   /* Vertex program parameters from curbe:
-    */
-   nr_params = c->prog_data.max_const;
-   for (i = 0; i < nr_params; i++) {
-      c->regs[TGSI_FILE_CONSTANT][i] = stride(brw_vec4_grf(reg+i/2, (i%2) * 4), 0, 4, 1);
-   }
-   reg += (nr_params+1)/2;
-   c->prog_data.curb_read_length = reg - 1;
-
-
-
-   /* Allocate input regs:
-    */
-   c->nr_inputs = c->vp->info.num_inputs;
-   for (i = 0; i < c->nr_inputs; i++) {
-	 c->regs[TGSI_FILE_INPUT][i] = brw_vec8_grf(reg, 0);
-	 reg++;
-   }
-
-
-   /* Allocate outputs: TODO: could organize the non-position outputs
-    * to go straight into message regs.
-    */
-   c->nr_outputs = 0;
-   c->first_output = reg;
-   mrf = 4;
-   for (i = 0; i < c->vp->info.num_outputs; i++) {
-      c->nr_outputs++;
-#if 0
-      if (i == VERT_RESULT_HPOS) {
-         c->regs[TGSI_FILE_OUTPUT][i] = brw_vec8_grf(reg, 0);
-         reg++;
-      }
-      else if (i == VERT_RESULT_PSIZ) {
-         c->regs[TGSI_FILE_OUTPUT][i] = brw_vec8_grf(reg, 0);
-         reg++;
-         mrf++;		/* just a placeholder?  XXX fix later stages & remove this */
-      }
-      else {
-         c->regs[TGSI_FILE_OUTPUT][i] = brw_message_reg(mrf);
-         mrf++;
-      }
-#else
-      /*treat pos differently for now */
-      if (i == info->pos_idx) {
-         c->regs[TGSI_FILE_OUTPUT][i] = brw_vec8_grf(reg, 0);
-         reg++;
-      } else {
-         c->regs[TGSI_FILE_OUTPUT][i] = brw_message_reg(mrf);
-         mrf++;
-      }
-#endif
-   }
-
-   /* Allocate program temporaries:
-    */
-   for (i = 0; i < info->num_temps; i++) {
-      c->regs[TGSI_FILE_TEMPORARY][i] = brw_vec8_grf(reg, 0);
-      reg++;
-   }
-
-   /* Address reg(s).  Don't try to use the internal address reg until
-    * deref time.
-    */
-   for (i = 0; i < info->num_addrs; i++) {
-      c->regs[TGSI_FILE_ADDRESS][i] =  brw_reg(BRW_GENERAL_REGISTER_FILE,
-                                               reg,
-                                               0,
-                                               BRW_REGISTER_TYPE_D,
-                                               BRW_VERTICAL_STRIDE_8,
-                                               BRW_WIDTH_8,
-                                               BRW_HORIZONTAL_STRIDE_1,
-                                               BRW_SWIZZLE_XXXX,
-                                               TGSI_WRITEMASK_X);
-      reg++;
-   }
-
-   for (i = 0; i < 128; i++) {
-      if (c->output_regs[i].used_in_src) {
-         c->output_regs[i].reg = brw_vec8_grf(reg, 0);
-         reg++;
-      }
-   }
-
-   c->stack =  brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, reg, 0);
-   reg += 2;
-
-
-   /* Some opcodes need an internal temporary:
-    */
-   c->first_tmp = reg;
-   c->last_tmp = reg;		/* for allocation purposes */
-
-   /* Each input reg holds data from two vertices.  The
-    * urb_read_length is the number of registers read from *each*
-    * vertex urb, so is half the amount:
-    */
-   c->prog_data.urb_read_length = (c->nr_inputs+1)/2;
-
-   c->prog_data.urb_entry_size = (c->nr_outputs+2+3)/4;
-   c->prog_data.total_grf = reg;
-}
-
-
-static struct brw_reg get_tmp( struct brw_vs_compile *c )
-{
-   struct brw_reg tmp = brw_vec8_grf(c->last_tmp, 0);
-
-   if (++c->last_tmp > c->prog_data.total_grf)
-      c->prog_data.total_grf = c->last_tmp;
-
-   return tmp;
-}
-
-static void release_tmp( struct brw_vs_compile *c, struct brw_reg tmp )
-{
-   if (tmp.nr == c->last_tmp-1)
-      c->last_tmp--;
-}
-
-static void release_tmps( struct brw_vs_compile *c )
-{
-   c->last_tmp = c->first_tmp;
-}
-
-
-static void unalias1( struct brw_vs_compile *c,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      void (*func)( struct brw_vs_compile *,
-				    struct brw_reg,
-				    struct brw_reg ))
-{
-   if (dst.file == arg0.file && dst.nr == arg0.nr) {
-      struct brw_compile *p = &c->func;
-      struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
-      func(c, tmp, arg0);
-      brw_MOV(p, dst, tmp);
-   }
-   else {
-      func(c, dst, arg0);
-   }
-}
-
-static void unalias2( struct brw_vs_compile *c,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1,
-		      void (*func)( struct brw_vs_compile *,
-				    struct brw_reg,
-				    struct brw_reg,
-				    struct brw_reg ))
-{
-   if ((dst.file == arg0.file && dst.nr == arg0.nr) ||
-       (dst.file == arg1.file && dst.nr == arg1.nr)) {
-      struct brw_compile *p = &c->func;
-      struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
-      func(c, tmp, arg0, arg1);
-      brw_MOV(p, dst, tmp);
-   }
-   else {
-      func(c, dst, arg0, arg1);
-   }
-}
-
-static void emit_sop( struct brw_compile *p,
-                      struct brw_reg dst,
-                      struct brw_reg arg0,
-                      struct brw_reg arg1,
-		      unsigned cond)
-{
-   brw_push_insn_state(p);
-   brw_CMP(p, brw_null_reg(), cond, arg0, arg1);
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-   brw_MOV(p, dst, brw_imm_f(1.0f));
-   brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-   brw_MOV(p, dst, brw_imm_f(0.0f));
-   brw_pop_insn_state(p);
-}
-
-static void emit_seq( struct brw_compile *p,
-                      struct brw_reg dst,
-                      struct brw_reg arg0,
-                      struct brw_reg arg1 )
-{
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_EQ);
-}
-
-static void emit_sne( struct brw_compile *p,
-                      struct brw_reg dst,
-                      struct brw_reg arg0,
-                      struct brw_reg arg1 )
-{
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_NEQ);
-}
-static void emit_slt( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1 )
-{
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_L);
-}
-
-static void emit_sle( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1 )
-{
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_LE);
-}
-
-static void emit_sgt( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1 )
-{
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_G);
-}
-
-static void emit_sge( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1 )
-{
-  emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_GE);
-}
-
-static void emit_max( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1 )
-{
-   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0, arg1);
-   brw_SEL(p, dst, arg1, arg0);
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-}
-
-static void emit_min( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg arg0,
-		      struct brw_reg arg1 )
-{
-   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0, arg1);
-   brw_SEL(p, dst, arg0, arg1);
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-}
-
-
-static void emit_math1( struct brw_vs_compile *c,
-			unsigned function,
-			struct brw_reg dst,
-			struct brw_reg arg0,
-			unsigned precision)
-{
-   /* There are various odd behaviours with SEND on the simulator.  In
-    * addition there are documented issues with the fact that the GEN4
-    * processor doesn't do dependency control properly on SEND
-    * results.  So, on balance, this kludge to get around failures
-    * with writemasked math results looks like it might be necessary
-    * whether that turns out to be a simulator bug or not:
-    */
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = dst;
-   boolean need_tmp = (dst.dw1.bits.writemask != 0xf ||
-			 dst.file != BRW_GENERAL_REGISTER_FILE);
-
-   if (need_tmp)
-      tmp = get_tmp(c);
-
-   brw_math(p,
-	    tmp,
-	    function,
-	    BRW_MATH_SATURATE_NONE,
-	    2,
-	    arg0,
-	    BRW_MATH_DATA_SCALAR,
-	    precision);
-
-   if (need_tmp) {
-      brw_MOV(p, dst, tmp);
-      release_tmp(c, tmp);
-   }
-}
-
-static void emit_math2( struct brw_vs_compile *c,
-			unsigned function,
-			struct brw_reg dst,
-			struct brw_reg arg0,
-			struct brw_reg arg1,
-			unsigned precision)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = dst;
-   boolean need_tmp = (dst.dw1.bits.writemask != 0xf ||
-			 dst.file != BRW_GENERAL_REGISTER_FILE);
-
-   if (need_tmp)
-      tmp = get_tmp(c);
-
-   brw_MOV(p, brw_message_reg(3), arg1);
-
-   brw_math(p,
-	    tmp,
-	    function,
-	    BRW_MATH_SATURATE_NONE,
-	    2,
- 	    arg0,
-	    BRW_MATH_DATA_SCALAR,
-	    precision);
-
-   if (need_tmp) {
-      brw_MOV(p, dst, tmp);
-      release_tmp(c, tmp);
-   }
-}
-
-
-
-static void emit_exp_noalias( struct brw_vs_compile *c,
-			      struct brw_reg dst,
-			      struct brw_reg arg0 )
-{
-   struct brw_compile *p = &c->func;
-
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_X) {
-      struct brw_reg tmp = get_tmp(c);
-      struct brw_reg tmp_d = retype(tmp, BRW_REGISTER_TYPE_D);
-
-      /* tmp_d = floor(arg0.x) */
-      brw_RNDD(p, tmp_d, brw_swizzle1(arg0, 0));
-
-      /* result[0] = 2.0 ^ tmp */
-
-      /* Adjust exponent for floating point:
-       * exp += 127
-       */
-      brw_ADD(p, brw_writemask(tmp_d, TGSI_WRITEMASK_X), tmp_d, brw_imm_d(127));
-
-      /* Install exponent and sign.
-       * Excess drops off the edge:
-       */
-      brw_SHL(p, brw_writemask(retype(dst, BRW_REGISTER_TYPE_D), TGSI_WRITEMASK_X),
-	      tmp_d, brw_imm_d(23));
-
-      release_tmp(c, tmp);
-   }
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_Y) {
-      /* result[1] = arg0.x - floor(arg0.x) */
-      brw_FRC(p, brw_writemask(dst, TGSI_WRITEMASK_Y), brw_swizzle1(arg0, 0));
-   }
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_Z) {
-      /* As with the LOG instruction, we might be better off just
-       * doing a taylor expansion here, seeing as we have to do all
-       * the prep work.
-       *
-       * If mathbox partial precision is too low, consider also:
-       * result[3] = result[0] * EXP(result[1])
-       */
-      emit_math1(c,
-		 BRW_MATH_FUNCTION_EXP,
-		 brw_writemask(dst, TGSI_WRITEMASK_Z),
-		 brw_swizzle1(arg0, 0),
-		 BRW_MATH_PRECISION_PARTIAL);
-   }
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_W) {
-      /* result[3] = 1.0; */
-      brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_W), brw_imm_f(1));
-   }
-}
-
-
-static void emit_log_noalias( struct brw_vs_compile *c,
-			      struct brw_reg dst,
-			      struct brw_reg arg0 )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = dst;
-   struct brw_reg tmp_ud = retype(tmp, BRW_REGISTER_TYPE_UD);
-   struct brw_reg arg0_ud = retype(arg0, BRW_REGISTER_TYPE_UD);
-   boolean need_tmp = (dst.dw1.bits.writemask != 0xf ||
-			 dst.file != BRW_GENERAL_REGISTER_FILE);
-
-   if (need_tmp) {
-      tmp = get_tmp(c);
-      tmp_ud = retype(tmp, BRW_REGISTER_TYPE_UD);
-   }
-
-   /* Perform mant = frexpf(fabsf(x), &exp), adjust exp and mnt
-    * according to spec:
-    *
-    * These almost look likey they could be joined up, but not really
-    * practical:
-    *
-    * result[0].f = (x.i & ((1<<31)-1) >> 23) - 127
-    * result[1].i = (x.i & ((1<<23)-1)        + (127<<23)
-    */
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_XZ) {
-      brw_AND(p,
-	      brw_writemask(tmp_ud, TGSI_WRITEMASK_X),
-	      brw_swizzle1(arg0_ud, 0),
-	      brw_imm_ud((1U<<31)-1));
-
-      brw_SHR(p,
-	      brw_writemask(tmp_ud, TGSI_WRITEMASK_X),
-	      tmp_ud,
-	      brw_imm_ud(23));
-
-      brw_ADD(p,
-	      brw_writemask(tmp, TGSI_WRITEMASK_X),
-	      retype(tmp_ud, BRW_REGISTER_TYPE_D),	/* does it matter? */
-	      brw_imm_d(-127));
-   }
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_YZ) {
-      brw_AND(p,
-	      brw_writemask(tmp_ud, TGSI_WRITEMASK_Y),
-	      brw_swizzle1(arg0_ud, 0),
-	      brw_imm_ud((1<<23)-1));
-
-      brw_OR(p,
-	     brw_writemask(tmp_ud, TGSI_WRITEMASK_Y),
-	     tmp_ud,
-	     brw_imm_ud(127<<23));
-   }
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_Z) {
-      /* result[2] = result[0] + LOG2(result[1]); */
-
-      /* Why bother?  The above is just a hint how to do this with a
-       * taylor series.  Maybe we *should* use a taylor series as by
-       * the time all the above has been done it's almost certainly
-       * quicker than calling the mathbox, even with low precision.
-       *
-       * Options are:
-       *    - result[0] + mathbox.LOG2(result[1])
-       *    - mathbox.LOG2(arg0.x)
-       *    - result[0] + inline_taylor_approx(result[1])
-       */
-      emit_math1(c,
-		 BRW_MATH_FUNCTION_LOG,
-		 brw_writemask(tmp, TGSI_WRITEMASK_Z),
-		 brw_swizzle1(tmp, 1),
-		 BRW_MATH_PRECISION_FULL);
-
-      brw_ADD(p,
-	      brw_writemask(tmp, TGSI_WRITEMASK_Z),
-	      brw_swizzle1(tmp, 2),
-	      brw_swizzle1(tmp, 0));
-   }
-
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_W) {
-      /* result[3] = 1.0; */
-      brw_MOV(p, brw_writemask(tmp, TGSI_WRITEMASK_W), brw_imm_f(1));
-   }
-
-   if (need_tmp) {
-      brw_MOV(p, dst, tmp);
-      release_tmp(c, tmp);
-   }
-}
-
-
-
-
-/* Need to unalias - consider swizzles:   r0 = DST r0.xxxx r1
- */
-static void emit_dst_noalias( struct brw_vs_compile *c,
-			      struct brw_reg dst,
-			      struct brw_reg arg0,
-			      struct brw_reg arg1)
-{
-   struct brw_compile *p = &c->func;
-
-   /* There must be a better way to do this:
-    */
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_X)
-      brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_X), brw_imm_f(1.0));
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_Y)
-      brw_MUL(p, brw_writemask(dst, TGSI_WRITEMASK_Y), arg0, arg1);
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_Z)
-      brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_Z), arg0);
-   if (dst.dw1.bits.writemask & TGSI_WRITEMASK_W)
-      brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_W), arg1);
-}
-
-static void emit_xpd( struct brw_compile *p,
-		      struct brw_reg dst,
-		      struct brw_reg t,
-		      struct brw_reg u)
-{
-   brw_MUL(p, brw_null_reg(), brw_swizzle(t, 1,2,0,3),  brw_swizzle(u,2,0,1,3));
-   brw_MAC(p, dst,     negate(brw_swizzle(t, 2,0,1,3)), brw_swizzle(u,1,2,0,3));
-}
-
-
-
-static void emit_lit_noalias( struct brw_vs_compile *c,
-			      struct brw_reg dst,
-			      struct brw_reg arg0 )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *if_insn;
-   struct brw_reg tmp = dst;
-   boolean need_tmp = (dst.file != BRW_GENERAL_REGISTER_FILE);
-
-   if (need_tmp)
-      tmp = get_tmp(c);
-
-   brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_YZ), brw_imm_f(0));
-   brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_XW), brw_imm_f(1));
-
-   /* Need to use BRW_EXECUTE_8 and also do an 8-wide compare in order
-    * to get all channels active inside the IF.  In the clipping code
-    * we run with NoMask, so it's not an option and we can use
-    * BRW_EXECUTE_1 for all comparisions.
-    */
-   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_G, brw_swizzle1(arg0,0), brw_imm_f(0));
-   if_insn = brw_IF(p, BRW_EXECUTE_8);
-   {
-      brw_MOV(p, brw_writemask(dst, TGSI_WRITEMASK_Y), brw_swizzle1(arg0,0));
-
-      brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_G, brw_swizzle1(arg0,1), brw_imm_f(0));
-      brw_MOV(p, brw_writemask(tmp, TGSI_WRITEMASK_Z),  brw_swizzle1(arg0,1));
-      brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
-      emit_math2(c,
-		 BRW_MATH_FUNCTION_POW,
-		 brw_writemask(dst, TGSI_WRITEMASK_Z),
-		 brw_swizzle1(tmp, 2),
-		 brw_swizzle1(arg0, 3),
-		 BRW_MATH_PRECISION_PARTIAL);
-   }
-
-   brw_ENDIF(p, if_insn);
-}
-
-
-
-
-
-/* TODO: relative addressing!
- */
-static struct brw_reg get_reg( struct brw_vs_compile *c,
-			       unsigned file,
-			       unsigned index )
-{
-   switch (file) {
-   case TGSI_FILE_TEMPORARY:
-   case TGSI_FILE_INPUT:
-   case TGSI_FILE_OUTPUT:
-      assert(c->regs[file][index].nr != 0);
-      return c->regs[file][index];
-   case TGSI_FILE_CONSTANT:
-      assert(c->regs[TGSI_FILE_CONSTANT][index + c->prog_data.num_imm].nr != 0);
-      return c->regs[TGSI_FILE_CONSTANT][index + c->prog_data.num_imm];
-   case TGSI_FILE_IMMEDIATE:
-      assert(c->regs[TGSI_FILE_CONSTANT][index].nr != 0);
-      return c->regs[TGSI_FILE_CONSTANT][index];
-   case TGSI_FILE_ADDRESS:
-      assert(index == 0);
-      return c->regs[file][index];
-
-   case TGSI_FILE_NULL:			/* undef values */
-      return brw_null_reg();
-
-   default:
-      assert(0);
-      return brw_null_reg();
-   }
-}
-
-
-
-static struct brw_reg deref( struct brw_vs_compile *c,
-			     struct brw_reg arg,
-			     int offset)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = vec4(get_tmp(c));
-   struct brw_reg vp_address = retype(vec1(get_reg(c, TGSI_FILE_ADDRESS, 0)), BRW_REGISTER_TYPE_UW);
-   unsigned byte_offset = arg.nr * 32 + arg.subnr + offset * 16;
-   struct brw_reg indirect = brw_vec4_indirect(0,0);
-
-   {
-      brw_push_insn_state(p);
-      brw_set_access_mode(p, BRW_ALIGN_1);
-
-      /* This is pretty clunky - load the address register twice and
-       * fetch each 4-dword value in turn.  There must be a way to do
-       * this in a single pass, but I couldn't get it to work.
-       */
-      brw_ADD(p, brw_address_reg(0), vp_address, brw_imm_d(byte_offset));
-      brw_MOV(p, tmp, indirect);
-
-      brw_ADD(p, brw_address_reg(0), suboffset(vp_address, 8), brw_imm_d(byte_offset));
-      brw_MOV(p, suboffset(tmp, 4), indirect);
-
-      brw_pop_insn_state(p);
-   }
-
-   return vec8(tmp);
-}
-
-
-static void emit_arl( struct brw_vs_compile *c,
-		      struct brw_reg dst,
-		      struct brw_reg arg0 )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg tmp = dst;
-   boolean need_tmp = (dst.file != BRW_GENERAL_REGISTER_FILE);
-
-   if (need_tmp)
-      tmp = get_tmp(c);
-
-   brw_RNDD(p, tmp, arg0);
-   brw_MUL(p, dst, tmp, brw_imm_d(16));
-
-   if (need_tmp)
-      release_tmp(c, tmp);
-}
-
-
-/* Will return mangled results for SWZ op.  The emit_swz() function
- * ignores this result and recalculates taking extended swizzles into
- * account.
- */
-static struct brw_reg get_arg( struct brw_vs_compile *c,
-			       struct tgsi_src_register *src )
-{
-   struct brw_reg reg;
-
-   if (src->File == TGSI_FILE_NULL)
-      return brw_null_reg();
-
-#if 0
-   if (src->RelAddr)
-      reg = deref(c, c->regs[PROGRAM_STATE_VAR][0], src->Index);
-   else
-#endif
-      reg = get_reg(c, src->File, src->Index);
-
-   /* Convert 3-bit swizzle to 2-bit.
-    */
-   reg.dw1.bits.swizzle = BRW_SWIZZLE4(src->SwizzleX,
-				       src->SwizzleY,
-				       src->SwizzleZ,
-				       src->SwizzleW);
-
-   /* Note this is ok for non-swizzle instructions:
-    */
-   reg.negate = src->Negate ? 1 : 0;
-
-   return reg;
-}
-
-
-static struct brw_reg get_dst( struct brw_vs_compile *c,
-			       const struct tgsi_dst_register *dst )
-{
-   struct brw_reg reg = get_reg(c, dst->File, dst->Index);
-
-   reg.dw1.bits.writemask = dst->WriteMask;
-
-   return reg;
-}
-
-
-
-
-static void emit_swz( struct brw_vs_compile *c,
-		      struct brw_reg dst,
-		      struct tgsi_src_register src )
-{
-   struct brw_compile *p = &c->func;
-   unsigned zeros_mask = 0;
-   unsigned ones_mask = 0;
-   unsigned src_mask = 0;
-   ubyte src_swz[4];
-   boolean need_tmp = (src.Negate &&
-			 dst.file != BRW_GENERAL_REGISTER_FILE);
-   struct brw_reg tmp = dst;
-   unsigned i;
-
-   if (need_tmp)
-      tmp = get_tmp(c);
-
-   for (i = 0; i < 4; i++) {
-      if (dst.dw1.bits.writemask & (1<<i)) {
-	 ubyte s = 0;
-         switch(i) {
-         case 0:
-            s = src.SwizzleX;
-            break;
-            s = src.SwizzleY;
-         case 1:
-            break;
-            s = src.SwizzleZ;
-         case 2:
-            break;
-            s = src.SwizzleW;
-         case 3:
-            break;
-         }
-	 switch (s) {
-	 case TGSI_SWIZZLE_X:
-	 case TGSI_SWIZZLE_Y:
-	 case TGSI_SWIZZLE_Z:
-	 case TGSI_SWIZZLE_W:
-	    src_mask |= 1<<i;
-	    src_swz[i] = s;
-	    break;
-	 case TGSI_EXTSWIZZLE_ZERO:
-	    zeros_mask |= 1<<i;
-	    break;
-	 case TGSI_EXTSWIZZLE_ONE:
-	    ones_mask |= 1<<i;
-	    break;
-	 }
-      }
-   }
-
-   /* Do src first, in case dst aliases src:
-    */
-   if (src_mask) {
-      struct brw_reg arg0;
-
-#if 0
-      if (src.RelAddr)
-	 arg0 = deref(c, c->regs[PROGRAM_STATE_VAR][0], src.Index);
-      else
-#endif
-	 arg0 = get_reg(c, src.File, src.Index);
-
-      arg0 = brw_swizzle(arg0,
-			 src_swz[0], src_swz[1],
-			 src_swz[2], src_swz[3]);
-
-      brw_MOV(p, brw_writemask(tmp, src_mask), arg0);
-   }
-
-   if (zeros_mask)
-      brw_MOV(p, brw_writemask(tmp, zeros_mask), brw_imm_f(0));
-
-   if (ones_mask)
-      brw_MOV(p, brw_writemask(tmp, ones_mask), brw_imm_f(1));
-
-   if (src.Negate)
-      brw_MOV(p, brw_writemask(tmp, src.Negate), negate(tmp));
-
-   if (need_tmp) {
-      brw_MOV(p, dst, tmp);
-      release_tmp(c, tmp);
-   }
-}
-
-
-
-/* Post-vertex-program processing.  Send the results to the URB.
- */
-static void emit_vertex_write( struct brw_vs_compile *c, struct brw_prog_info *info)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg m0 = brw_message_reg(0);
-   struct brw_reg pos = c->regs[TGSI_FILE_OUTPUT][info->pos_idx];
-   struct brw_reg ndc;
-
-   if (c->key.copy_edgeflag) {
-      brw_MOV(p,
-	      get_reg(c, TGSI_FILE_OUTPUT, info->result_edge_idx),
-	      get_reg(c, TGSI_FILE_INPUT, info->edge_flag_idx));
-   }
-
-
-   /* Build ndc coords?   TODO: Shortcircuit when w is known to be one.
-    */
-   if (!c->key.know_w_is_one) {
-      ndc = get_tmp(c);
-      emit_math1(c, BRW_MATH_FUNCTION_INV, ndc, brw_swizzle1(pos, 3), BRW_MATH_PRECISION_FULL);
-      brw_MUL(p, brw_writemask(ndc, TGSI_WRITEMASK_XYZ), pos, ndc);
-   }
-   else {
-      ndc = pos;
-   }
-
-   /* This includes the workaround for -ve rhw, so is no longer an
-    * optional step:
-    */
-   if (info->writes_psize ||
-       c->key.nr_userclip ||
-       !c->key.know_w_is_one)
-   {
-      struct brw_reg header1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
-      unsigned i;
-
-      brw_MOV(p, header1, brw_imm_ud(0));
-
-      brw_set_access_mode(p, BRW_ALIGN_16);
-
-      if (info->writes_psize) {
-	 struct brw_reg psiz = c->regs[TGSI_FILE_OUTPUT][info->psize_idx];
-	 brw_MUL(p, brw_writemask(header1, TGSI_WRITEMASK_W),
-                 brw_swizzle1(psiz, 0), brw_imm_f(1<<11));
-	 brw_AND(p, brw_writemask(header1, TGSI_WRITEMASK_W), header1,
-                 brw_imm_ud(0x7ff<<8));
-      }
-
-
-      for (i = 0; i < c->key.nr_userclip; i++) {
-	 brw_set_conditionalmod(p, BRW_CONDITIONAL_L);
-	 brw_DP4(p, brw_null_reg(), pos, c->userplane[i]);
-	 brw_OR(p, brw_writemask(header1, TGSI_WRITEMASK_W), header1, brw_imm_ud(1<<i));
-	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-      }
-
-
-      /* i965 clipping workaround:
-       * 1) Test for -ve rhw
-       * 2) If set,
-       *      set ndc = (0,0,0,0)
-       *      set ucp[6] = 1
-       *
-       * Later, clipping will detect ucp[6] and ensure the primitive is
-       * clipped against all fixed planes.
-       */
-      if (!c->key.know_w_is_one) {
-	 brw_CMP(p,
-		 vec8(brw_null_reg()),
-		 BRW_CONDITIONAL_L,
-		 brw_swizzle1(ndc, 3),
-		 brw_imm_f(0));
-
-	 brw_OR(p, brw_writemask(header1, TGSI_WRITEMASK_W), header1, brw_imm_ud(1<<6));
-	 brw_MOV(p, ndc, brw_imm_f(0));
-	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-      }
-
-      brw_set_access_mode(p, BRW_ALIGN_1);	/* why? */
-      brw_MOV(p, retype(brw_message_reg(1), BRW_REGISTER_TYPE_UD), header1);
-      brw_set_access_mode(p, BRW_ALIGN_16);
-
-      release_tmp(c, header1);
-   }
-   else {
-      brw_MOV(p, retype(brw_message_reg(1), BRW_REGISTER_TYPE_UD), brw_imm_ud(0));
-   }
-
-
-   /* Emit the (interleaved) headers for the two vertices - an 8-reg
-    * of zeros followed by two sets of NDC coordinates:
-    */
-   brw_set_access_mode(p, BRW_ALIGN_1);
-   brw_MOV(p, offset(m0, 2), ndc);
-   brw_MOV(p, offset(m0, 3), pos);
-
-
-   brw_urb_WRITE(p,
-		 brw_null_reg(), /* dest */
-		 0,		/* starting mrf reg nr */
-		 c->r0,		/* src */
-		 0,		/* allocate */
-		 1,		/* used */
-		 c->nr_outputs + 3, /* msg len */
-		 0,		/* response len */
-		 1, 		/* eot */
-		 1, 		/* writes complete */
-		 0, 		/* urb destination offset */
-		 BRW_URB_SWIZZLE_INTERLEAVE);
-
-}
-
-static void
-post_vs_emit( struct brw_vs_compile *c, struct brw_instruction *end_inst )
-{
-   struct tgsi_parse_context parse;
-   const struct tgsi_token *tokens = c->vp->program.tokens;
-   tgsi_parse_init(&parse, tokens);
-   while (!tgsi_parse_end_of_tokens(&parse)) {
-      tgsi_parse_token(&parse);
-      if (parse.FullToken.Token.Type == TGSI_TOKEN_TYPE_INSTRUCTION) {
-#if 0
-         struct brw_instruction *brw_inst1, *brw_inst2;
-         const struct tgsi_full_instruction *inst1, *inst2;
-         int offset;
-         inst1 = &parse.FullToken.FullInstruction;
-         brw_inst1 = inst1->Data;
-         switch (inst1->Opcode) {
-	 case TGSI_OPCODE_CAL:
-	 case TGSI_OPCODE_BRA:
-	    target_insn = inst1->BranchTarget;
-	    inst2 = &c->vp->program.Base.Instructions[target_insn];
-	    brw_inst2 = inst2->Data;
-	    offset = brw_inst2 - brw_inst1;
-	    brw_set_src1(brw_inst1, brw_imm_d(offset*16));
-	    break;
-	 case TGSI_OPCODE_END:
-	    offset = end_inst - brw_inst1;
-	    brw_set_src1(brw_inst1, brw_imm_d(offset*16));
-	    break;
-	 default:
-	    break;
-         }
-#endif
-      }
-   }
-   tgsi_parse_free(&parse);
-}
-
-static void process_declaration(const struct tgsi_full_declaration *decl,
-                                struct brw_prog_info *info)
-{
-   int first = decl->DeclarationRange.First;
-   int last = decl->DeclarationRange.Last;
-   
-   switch(decl->Declaration.File) {
-   case TGSI_FILE_CONSTANT: 
-      info->num_consts += last - first + 1;
-      break;
-   case TGSI_FILE_INPUT: {
-   }
-      break;
-   case TGSI_FILE_OUTPUT: {
-      assert(last == first);	/* for now */
-      if (decl->Declaration.Semantic) {
-         switch (decl->Semantic.SemanticName) {
-         case TGSI_SEMANTIC_POSITION: {
-            info->pos_idx = first;
-         }
-            break;
-         case TGSI_SEMANTIC_COLOR:
-            break;
-         case TGSI_SEMANTIC_BCOLOR:
-            break;
-         case TGSI_SEMANTIC_FOG:
-            break;
-         case TGSI_SEMANTIC_PSIZE: {
-            info->writes_psize = TRUE;
-            info->psize_idx = first;
-         }
-            break;
-         case TGSI_SEMANTIC_GENERIC:
-            break;
-         }
-      }
-   }
-      break;
-   case TGSI_FILE_TEMPORARY: {
-      info->num_temps += (last - first) + 1;
-   }
-      break;
-   case TGSI_FILE_SAMPLER: {
-   }
-      break;
-   case TGSI_FILE_ADDRESS: {
-      info->num_addrs += (last - first) + 1;
-   }
-      break;
-   case TGSI_FILE_IMMEDIATE: {
-   }
-      break;
-   case TGSI_FILE_NULL: {
-   }
-      break;
-   }
-}
-
-static void process_instruction(struct brw_vs_compile *c,
-                                struct tgsi_full_instruction *inst,
-                                struct brw_prog_info *info)
-{
-   struct brw_reg args[3], dst;
-   struct brw_compile *p = &c->func;
-   /*struct brw_indirect stack_index = brw_indirect(0, 0);*/
-   unsigned i;
-   unsigned index;
-   unsigned file;
-   /*FIXME: might not be the only one*/
-   const struct tgsi_dst_register *dst_reg = &inst->FullDstRegisters[0].DstRegister;
-   /*
-   struct brw_instruction *if_inst[MAX_IFSN];
-   unsigned insn, if_insn = 0;
-   */
-
-   for (i = 0; i < 3; i++) {
-      struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i];
-      index = src->SrcRegister.Index;
-      file = src->SrcRegister.File;
-      if (file == TGSI_FILE_OUTPUT && c->output_regs[index].used_in_src)
-         args[i] = c->output_regs[index].reg;
-      else
-         args[i] = get_arg(c, &src->SrcRegister);
-   }
-
-   /* Get dest regs.  Note that it is possible for a reg to be both
-    * dst and arg, given the static allocation of registers.  So
-    * care needs to be taken emitting multi-operation instructions.
-    */
-   index = dst_reg->Index;
-   file = dst_reg->File;
-   if (file == TGSI_FILE_OUTPUT && c->output_regs[index].used_in_src)
-      dst = c->output_regs[index].reg;
-   else
-      dst = get_dst(c, dst_reg);
-
-   switch (inst->Instruction.Opcode) {
-   case TGSI_OPCODE_ABS:
-      brw_MOV(p, dst, brw_abs(args[0]));
-      break;
-   case TGSI_OPCODE_ADD:
-      brw_ADD(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_DP3:
-      brw_DP3(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_DP4:
-      brw_DP4(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_DPH:
-      brw_DPH(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_DST:
-      unalias2(c, dst, args[0], args[1], emit_dst_noalias);
-      break;
-   case TGSI_OPCODE_EXP:
-      unalias1(c, dst, args[0], emit_exp_noalias);
-      break;
-   case TGSI_OPCODE_EX2:
-      emit_math1(c, BRW_MATH_FUNCTION_EXP, dst, args[0], BRW_MATH_PRECISION_FULL);
-      break;
-   case TGSI_OPCODE_ARL:
-      emit_arl(c, dst, args[0]);
-      break;
-   case TGSI_OPCODE_FLR:
-      brw_RNDD(p, dst, args[0]);
-      break;
-   case TGSI_OPCODE_FRC:
-      brw_FRC(p, dst, args[0]);
-      break;
-   case TGSI_OPCODE_LOG:
-      unalias1(c, dst, args[0], emit_log_noalias);
-      break;
-   case TGSI_OPCODE_LG2:
-      emit_math1(c, BRW_MATH_FUNCTION_LOG, dst, args[0], BRW_MATH_PRECISION_FULL);
-      break;
-   case TGSI_OPCODE_LIT:
-      unalias1(c, dst, args[0], emit_lit_noalias);
-      break;
-   case TGSI_OPCODE_MAD:
-      brw_MOV(p, brw_acc_reg(), args[2]);
-      brw_MAC(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_MAX:
-      emit_max(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_MIN:
-      emit_min(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_MOV:
-   case TGSI_OPCODE_SWZ:
-#if 0
-      /* The args[0] value can't be used here as it won't have
-       * correctly encoded the full swizzle:
-       */
-      emit_swz(c, dst, inst->SrcReg[0] );
-#endif
-      brw_MOV(p, dst, args[0]);
-      break;
-   case TGSI_OPCODE_MUL:
-      brw_MUL(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_POW:
-      emit_math2(c, BRW_MATH_FUNCTION_POW, dst, args[0], args[1], BRW_MATH_PRECISION_FULL);
-      break;
-   case TGSI_OPCODE_RCP:
-      emit_math1(c, BRW_MATH_FUNCTION_INV, dst, args[0], BRW_MATH_PRECISION_FULL);
-      break;
-   case TGSI_OPCODE_RSQ:
-      emit_math1(c, BRW_MATH_FUNCTION_RSQ, dst, args[0], BRW_MATH_PRECISION_FULL);
-      break;
-
-   case TGSI_OPCODE_SEQ:
-      emit_seq(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_SNE:
-      emit_sne(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_SGE:
-      emit_sge(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_SGT:
-      emit_sgt(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_SLT:
-      emit_slt(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_SLE:
-      emit_sle(p, dst, args[0], args[1]);
-      break;
-   case TGSI_OPCODE_SUB:
-      brw_ADD(p, dst, args[0], negate(args[1]));
-      break;
-   case TGSI_OPCODE_XPD:
-      emit_xpd(p, dst, args[0], args[1]);
-      break;
-#if 0
-   case TGSI_OPCODE_IF:
-      assert(if_insn < MAX_IFSN);
-      if_inst[if_insn++] = brw_IF(p, BRW_EXECUTE_8);
-      break;
-   case TGSI_OPCODE_ELSE:
-      if_inst[if_insn-1] = brw_ELSE(p, if_inst[if_insn-1]);
-      break;
-   case TGSI_OPCODE_ENDIF:
-      assert(if_insn > 0);
-      brw_ENDIF(p, if_inst[--if_insn]);
-      break;
-   case TGSI_OPCODE_BRA:
-      brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-      brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-      brw_set_predicate_control_flag_value(p, 0xff);
-      break;
-   case TGSI_OPCODE_CAL:
-      brw_set_access_mode(p, BRW_ALIGN_1);
-      brw_ADD(p, deref_1uw(stack_index, 0), brw_ip_reg(), brw_imm_d(3*16));
-      brw_set_access_mode(p, BRW_ALIGN_16);
-      brw_ADD(p, get_addr_reg(stack_index),
-              get_addr_reg(stack_index), brw_imm_d(4));
-      inst->Data = &p->store[p->nr_insn];
-      brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-      break;
-#endif
-   case TGSI_OPCODE_RET:
-#if 0
-      brw_ADD(p, get_addr_reg(stack_index),
-              get_addr_reg(stack_index), brw_imm_d(-4));
-      brw_set_access_mode(p, BRW_ALIGN_1);
-      brw_MOV(p, brw_ip_reg(), deref_1uw(stack_index, 0));
-      brw_set_access_mode(p, BRW_ALIGN_16);
-#else
-      /*brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));*/
-#endif
-      break;
-   case TGSI_OPCODE_END:
-      brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-      break;
-   case TGSI_OPCODE_BGNSUB:
-   case TGSI_OPCODE_ENDSUB:
-      break;
-   default:
-      debug_printf("Unsupport opcode %d in vertex shader\n", inst->Instruction.Opcode);
-      break;
-   }
-
-   if (dst_reg->File == TGSI_FILE_OUTPUT
-       && dst_reg->Index != info->pos_idx
-       && c->output_regs[dst_reg->Index].used_in_src)
-      brw_MOV(p, get_dst(c, dst_reg), dst);
-
-   release_tmps(c);
-}
-
-/* Emit the fragment program instructions here.
- */
-void brw_vs_emit(struct brw_vs_compile *c)
-{
-#define MAX_IFSN 32
-   struct brw_compile *p = &c->func;
-   struct brw_instruction *end_inst;
-   struct tgsi_parse_context parse;
-   struct brw_indirect stack_index = brw_indirect(0, 0);
-   const struct tgsi_token *tokens = c->vp->program.tokens;
-   struct brw_prog_info prog_info;
-   unsigned allocated_registers = 0;
-   memset(&prog_info, 0, sizeof(struct brw_prog_info));
-
-   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-   brw_set_access_mode(p, BRW_ALIGN_16);
-
-   tgsi_parse_init(&parse, tokens);
-   /* Message registers can't be read, so copy the output into GRF register
-      if they are used in source registers */
-   while (!tgsi_parse_end_of_tokens(&parse)) {
-      tgsi_parse_token(&parse);
-      unsigned i;
-      switch (parse.FullToken.Token.Type) {
-      case TGSI_TOKEN_TYPE_INSTRUCTION: {
-         const struct tgsi_full_instruction *inst = &parse.FullToken.FullInstruction;
-         for (i = 0; i < 3; ++i) {
-            const struct tgsi_src_register *src = &inst->FullSrcRegisters[i].SrcRegister;
-            unsigned index = src->Index;
-            unsigned file = src->File;
-            if (file == TGSI_FILE_OUTPUT)
-               c->output_regs[index].used_in_src = TRUE;
-         }
-      }
-         break;
-      default:
-         /* nothing */
-         break;
-      }
-   }
-   tgsi_parse_free(&parse);
-
-   tgsi_parse_init(&parse, tokens);
-
-   while (!tgsi_parse_end_of_tokens(&parse)) {
-      tgsi_parse_token(&parse);
-
-      switch (parse.FullToken.Token.Type) {
-      case TGSI_TOKEN_TYPE_DECLARATION: {
-         struct tgsi_full_declaration *decl = &parse.FullToken.FullDeclaration;
-         process_declaration(decl, &prog_info);
-      }
-         break;
-      case TGSI_TOKEN_TYPE_IMMEDIATE: {
-         struct tgsi_full_immediate *imm = &parse.FullToken.FullImmediate;
-         assert(imm->Immediate.NrTokens == 4 + 1);
-         c->prog_data.imm_buf[c->prog_data.num_imm][0] = imm->u[0].Float;
-         c->prog_data.imm_buf[c->prog_data.num_imm][1] = imm->u[1].Float;
-         c->prog_data.imm_buf[c->prog_data.num_imm][2] = imm->u[2].Float;
-         c->prog_data.imm_buf[c->prog_data.num_imm][3] = imm->u[3].Float;
-         c->prog_data.num_imm++;
-      }
-         break;
-      case TGSI_TOKEN_TYPE_INSTRUCTION: {
-         struct tgsi_full_instruction *inst = &parse.FullToken.FullInstruction;
-         if (!allocated_registers) {
-            /* first instruction (declerations finished).
-             * now that we know what vars are being used allocate
-             * registers for them.*/
-            c->prog_data.num_consts = prog_info.num_consts;
-            c->prog_data.max_const = prog_info.num_consts + c->prog_data.num_imm;
-            brw_vs_alloc_regs(c, &prog_info);
-
-	    brw_set_access_mode(p, BRW_ALIGN_1);
-            brw_MOV(p, get_addr_reg(stack_index), brw_address(c->stack));
-	    brw_set_access_mode(p, BRW_ALIGN_16);
-            allocated_registers = 1;
-         }
-         process_instruction(c, inst, &prog_info);
-      }
-         break;
-      }
-   }
-
-   end_inst = &p->store[p->nr_insn];
-   emit_vertex_write(c, &prog_info);
-   post_vs_emit(c, end_inst);
-   tgsi_parse_free(&parse);
-
-}
diff --git a/src/gallium/drivers/i965simple/brw_vs_state.c b/src/gallium/drivers/i965simple/brw_vs_state.c
deleted file mode 100644
index 1eaff87892..0000000000
--- a/src/gallium/drivers/i965simple/brw_vs_state.c
+++ /dev/null
@@ -1,103 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-static void upload_vs_unit( struct brw_context *brw )
-{
-   struct brw_vs_unit_state vs;
-
-   memset(&vs, 0, sizeof(vs));
-
-   /* CACHE_NEW_VS_PROG */
-   vs.thread0.kernel_start_pointer = brw->vs.prog_gs_offset >> 6;
-   vs.thread0.grf_reg_count = align(brw->vs.prog_data->total_grf, 16) / 16 - 1;
-   vs.thread3.urb_entry_read_length = brw->vs.prog_data->urb_read_length;
-   vs.thread3.const_urb_entry_read_length = brw->vs.prog_data->curb_read_length;
-   vs.thread3.dispatch_grf_start_reg = 1;
-
-
-   /* BRW_NEW_URB_FENCE  */
-   vs.thread4.nr_urb_entries = brw->urb.nr_vs_entries;
-   vs.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
-   vs.thread4.max_threads = MIN2(
-      MAX2(0, (brw->urb.nr_vs_entries - 6) / 2 - 1),
-      15);
-
-
-
-   if (BRW_DEBUG & DEBUG_SINGLE_THREAD)
-      vs.thread4.max_threads = 0;
-
-   /* BRW_NEW_CURBE_OFFSETS, _NEW_TRANSFORM */
-   if (0 /*brw->attribs.Clip->ClipPlanesEnabled*/) {
-      /* Note that we read in the userclip planes as well, hence
-       * clip_start:
-       */
-      vs.thread3.const_urb_entry_read_offset = brw->curbe.clip_start * 2;
-   }
-   else {
-      vs.thread3.const_urb_entry_read_offset = brw->curbe.vs_start * 2;
-   }
-
-   vs.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
-   vs.thread3.urb_entry_read_offset = 0;
-
-   /* No samplers for ARB_vp programs:
-    */
-   vs.vs5.sampler_count = 0;
-
-   if (BRW_DEBUG & DEBUG_STATS)
-      vs.thread4.stats_enable = 1;
-
-   /* Vertex program always enabled:
-    */
-   vs.vs6.vs_enable = 1;
-
-   brw->vs.state_gs_offset = brw_cache_data( &brw->cache[BRW_VS_UNIT], &vs );
-}
-
-
-const struct brw_tracked_state brw_vs_unit = {
-   .dirty = {
-      .brw   = (BRW_NEW_CLIP |
-		BRW_NEW_CURBE_OFFSETS |
-		BRW_NEW_URB_FENCE),
-      .cache = CACHE_NEW_VS_PROG
-   },
-   .update = upload_vs_unit
-};
diff --git a/src/gallium/drivers/i965simple/brw_winsys.h b/src/gallium/drivers/i965simple/brw_winsys.h
deleted file mode 100644
index ec1e400418..0000000000
--- a/src/gallium/drivers/i965simple/brw_winsys.h
+++ /dev/null
@@ -1,209 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/**
- * \file
- * This is the interface that i965simple requires any window system
- * hosting it to implement.  This is the only include file in i965simple
- * which is public.
- *
- */
-
-#ifndef BRW_WINSYS_H
-#define BRW_WINSYS_H
-
-
-#include "pipe/p_defines.h"
-
-
-/* Pipe drivers are (meant to be!) independent of both GL and the
- * window system.  The window system provides a buffer manager and a
- * set of additional hooks for things like command buffer submission,
- * etc.
- *
- * There clearly has to be some agreement between the window system
- * driver and the hardware driver about the format of command buffers,
- * etc.
- */
-
-struct pipe_buffer;
-struct pipe_fence_handle;
-struct pipe_winsys;
-struct pipe_screen;
-
-
-/* The pipe driver currently understands the following chipsets:
- */
-#define PCI_CHIP_I965_G			0x29A2
-#define PCI_CHIP_I965_Q			0x2992
-#define PCI_CHIP_I965_G_1		0x2982
-#define PCI_CHIP_I965_GM                0x2A02
-#define PCI_CHIP_I965_GME               0x2A12
-
-
-/* These are the names of all the state caches managed by the driver.
- * 
- * When data is uploaded to a buffer with buffer_subdata, we use the
- * special version of that function below so that information about
- * what type of data this is can be passed to the winsys backend.
- * That in turn allows the correct flags to be set in the aub file
- * dump to allow human-readable file dumps later on.
- */
-
-enum brw_cache_id {
-   BRW_CC_VP,
-   BRW_CC_UNIT,
-   BRW_WM_PROG,
-   BRW_SAMPLER_DEFAULT_COLOR,
-   BRW_SAMPLER,
-   BRW_WM_UNIT,
-   BRW_SF_PROG,
-   BRW_SF_VP,
-   BRW_SF_UNIT,
-   BRW_VS_UNIT,
-   BRW_VS_PROG,
-   BRW_GS_UNIT,
-   BRW_GS_PROG,
-   BRW_CLIP_VP,
-   BRW_CLIP_UNIT,
-   BRW_CLIP_PROG,
-   BRW_SS_SURFACE,
-   BRW_SS_SURF_BIND,
-
-   BRW_MAX_CACHE
-};
-
-#define BRW_CONSTANT_BUFFER BRW_MAX_CACHE
-
-/**
- * Additional winsys interface for i965simple.
- *
- * It is an over-simple batchbuffer mechanism.  Will want to improve the
- * performance of this, perhaps based on the cmdstream stuff.  It
- * would be pretty impossible to implement swz on top of this
- * interface.
- *
- * Will also need additions/changes to implement static/dynamic
- * indirect state.
- */
-struct brw_winsys {
-
-   void (*destroy)(struct brw_winsys *);
-   
-   /**
-    * Reserve space on batch buffer.
-    *
-    * Returns a null pointer if there is insufficient space in the batch buffer
-    * to hold the requested number of dwords and relocations.
-    *
-    * The number of dwords should also include the number of relocations.
-    */
-   unsigned *(*batch_start)(struct brw_winsys *sws,
-                            unsigned dwords,
-                            unsigned relocs);
-
-   void (*batch_dword)(struct brw_winsys *sws,
-                       unsigned dword);
-
-   /**
-    * Emit a relocation to a buffer.
-    *
-    * Used not only when the buffer addresses are not pinned, but also to
-    * ensure refered buffers will not be destroyed until the current batch
-    * buffer execution is finished.
-    *
-    * The access flags is a combination of I915_BUFFER_ACCESS_WRITE and
-    * I915_BUFFER_ACCESS_READ macros.
-    */
-   void (*batch_reloc)(struct brw_winsys *sws,
-                       struct pipe_buffer *buf,
-                       unsigned access_flags,
-                       unsigned delta);
-
-
-   /* Not used yet, but really want this:
-    */
-   void (*batch_end)( struct brw_winsys *sws );
-
-   /**
-    * Flush the batch buffer.
-    *
-    * Fence argument must point to NULL or to a previous fence, and the caller
-    * must call fence_reference when done with the fence.
-    */
-   void (*batch_flush)(struct brw_winsys *sws,
-                       struct pipe_fence_handle **fence);
-
-
-   /* A version of buffer_subdata that includes information for the
-    * simulator:
-    */
-   void (*buffer_subdata_typed)(struct brw_winsys *sws, 
-				struct pipe_buffer *buf,
-				unsigned long offset, 
-				unsigned long size, 
-				const void *data,
-				unsigned data_type);
-   
-
-   /* A cheat so we don't have to think about relocations in a couple
-    * of places yet:
-    */
-   unsigned (*get_buffer_offset)( struct brw_winsys *sws,
-				  struct pipe_buffer *buf,
-				  unsigned flags );
-
-};
-
-#define BRW_BUFFER_ACCESS_WRITE   0x1
-#define BRW_BUFFER_ACCESS_READ    0x2
-
-#define BRW_BUFFER_USAGE_LIT_VERTEX  (PIPE_BUFFER_USAGE_CUSTOM << 0)
-
-
-struct pipe_context *brw_create(struct pipe_screen *,
-                                struct brw_winsys *,
-                                unsigned pci_id);
-
-static inline boolean brw_batchbuffer_data(struct brw_winsys *winsys,
-                                           const void *data,
-                                           unsigned bytes)
-{
-   static const unsigned incr = sizeof(unsigned);
-   uint i;
-   const unsigned *udata = (const unsigned*)(data);
-   unsigned size = bytes/incr;
-
-   winsys->batch_start(winsys, size, 0);
-   for (i = 0; i < size; ++i) {
-      winsys->batch_dword(winsys, udata[i]);
-   }
-   winsys->batch_end(winsys);
-
-   return (i == size);
-}
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_wm.c b/src/gallium/drivers/i965simple/brw_wm.c
deleted file mode 100644
index 10161f2d2f..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm.c
+++ /dev/null
@@ -1,209 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_util.h"
-#include "brw_wm.h"
-#include "brw_eu.h"
-#include "brw_state.h"
-#include "util/u_memory.h"
-
-
-
-static void do_wm_prog( struct brw_context *brw,
-			struct brw_fragment_program *fp,
-			struct brw_wm_prog_key *key)
-{
-   struct brw_wm_compile *c = CALLOC_STRUCT(brw_wm_compile);
-   const unsigned *program;
-   unsigned program_size;
-
-   c->key = *key;
-   c->fp = fp;
-   
-   c->delta_xy[0] = brw_null_reg();
-   c->delta_xy[1] = brw_null_reg();
-   c->pixel_xy[0] = brw_null_reg();
-   c->pixel_xy[1] = brw_null_reg();
-   c->pixel_w = brw_null_reg();
-
-
-   debug_printf("XXXXXXXX FP\n");
-   
-   brw_wm_glsl_emit(c);
-
-   /* get the program
-    */
-   program = brw_get_program(&c->func, &program_size);
-
-   /*
-    */
-   brw->wm.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_WM_PROG],
-					      &c->key,
-					      sizeof(c->key),
-					      program,
-					      program_size,
-					      &c->prog_data,
-					      &brw->wm.prog_data );
-
-   FREE(c);
-}
-
-
-
-static void brw_wm_populate_key( struct brw_context *brw,
-				 struct brw_wm_prog_key *key )
-{
-   /* BRW_NEW_FRAGMENT_PROGRAM */
-   struct brw_fragment_program *fp =
-      (struct brw_fragment_program *)brw->attribs.FragmentProgram;
-   unsigned lookup = 0;
-   unsigned line_aa;
-   
-   memset(key, 0, sizeof(*key));
-
-   /* Build the index for table lookup
-    */
-   /* BRW_NEW_DEPTH_STENCIL */
-   if (fp->info.uses_kill ||
-       brw->attribs.DepthStencil->alpha.enabled)
-      lookup |= IZ_PS_KILL_ALPHATEST_BIT;
-
-   if (fp->info.writes_z)
-      lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
-
-   if (brw->attribs.DepthStencil->depth.enabled)
-      lookup |= IZ_DEPTH_TEST_ENABLE_BIT;
-
-   if (brw->attribs.DepthStencil->depth.enabled &&
-       brw->attribs.DepthStencil->depth.writemask) /* ?? */
-      lookup |= IZ_DEPTH_WRITE_ENABLE_BIT;
-
-   if (brw->attribs.DepthStencil->stencil[0].enabled) {
-      lookup |= IZ_STENCIL_TEST_ENABLE_BIT;
-
-      if (brw->attribs.DepthStencil->stencil[0].writemask ||
-	  brw->attribs.DepthStencil->stencil[1].writemask)
-	 lookup |= IZ_STENCIL_WRITE_ENABLE_BIT;
-   }
-
-   /* XXX: when should this be disabled?
-    */
-   if (1)
-      lookup |= IZ_EARLY_DEPTH_TEST_BIT;
-
-
-   line_aa = AA_NEVER;
-
-   /* _NEW_LINE, _NEW_POLYGON, BRW_NEW_REDUCED_PRIMITIVE */
-   if (brw->attribs.Raster->line_smooth) {
-      if (brw->reduced_primitive == PIPE_PRIM_LINES) {
-	 line_aa = AA_ALWAYS;
-      }
-      else if (brw->reduced_primitive == PIPE_PRIM_TRIANGLES) {
-	 if (brw->attribs.Raster->fill_ccw == PIPE_POLYGON_MODE_LINE) {
-	    line_aa = AA_SOMETIMES;
-
-	    if (brw->attribs.Raster->fill_cw == PIPE_POLYGON_MODE_LINE ||
-		(brw->attribs.Raster->cull_mode == PIPE_WINDING_CW))
-	       line_aa = AA_ALWAYS;
-	 }
-	 else if (brw->attribs.Raster->fill_cw == PIPE_POLYGON_MODE_LINE) {
-	    line_aa = AA_SOMETIMES;
-
-	    if (brw->attribs.Raster->cull_mode == PIPE_WINDING_CCW)
-	       line_aa = AA_ALWAYS;
-	 }
-      }
-   }
-
-   brw_wm_lookup_iz(line_aa,
-		    lookup,
-		    key);
-
-
-#if 0
-   /* BRW_NEW_SAMPLER 
-    *
-    * Not doing any of this at the moment:
-    */
-   for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-      const struct pipe_sampler_state *unit = brw->attribs.Samplers[i];
-
-      if (unit) {
-
-	 if (unit->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-	    key->shadowtex_mask |= 1<<i;
-	 }
-	 if (t->Image[0][t->BaseLevel]->InternalFormat == GL_YCBCR_MESA)
-	    key->yuvtex_mask |= 1<<i;
-      }
-   }
-#endif
-
-
-   /* Extra info:
-    */
-   key->program_string_id = fp->id;
-
-}
-
-
-static void brw_upload_wm_prog( struct brw_context *brw )
-{
-   struct brw_wm_prog_key key;
-   struct brw_fragment_program *fp = (struct brw_fragment_program *)
-      brw->attribs.FragmentProgram;
-
-   brw_wm_populate_key(brw, &key);
-
-   /* Make an early check for the key.
-    */
-   if (brw_search_cache(&brw->cache[BRW_WM_PROG],
-			&key, sizeof(key),
-			&brw->wm.prog_data,
-			&brw->wm.prog_gs_offset))
-      return;
-
-   do_wm_prog(brw, fp, &key);
-}
-
-
-const struct brw_tracked_state brw_wm_prog = {
-   .dirty = {
-      .brw   = (BRW_NEW_FS |
-		BRW_NEW_REDUCED_PRIMITIVE),
-      .cache = 0
-   },
-   .update = brw_upload_wm_prog
-};
-
diff --git a/src/gallium/drivers/i965simple/brw_wm.h b/src/gallium/drivers/i965simple/brw_wm.h
deleted file mode 100644
index b29c4393f0..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm.h
+++ /dev/null
@@ -1,142 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-              
-
-#ifndef BRW_WM_H
-#define BRW_WM_H
-
-
-#include "brw_context.h"
-#include "brw_eu.h"
-
-/* A big lookup table is used to figure out which and how many
- * additional regs will inserted before the main payload in the WM
- * program execution.  These mainly relate to depth and stencil
- * processing and the early-depth-test optimization.
- */
-#define IZ_PS_KILL_ALPHATEST_BIT    0x1
-#define IZ_PS_COMPUTES_DEPTH_BIT    0x2
-#define IZ_DEPTH_WRITE_ENABLE_BIT   0x4
-#define IZ_DEPTH_TEST_ENABLE_BIT    0x8
-#define IZ_STENCIL_WRITE_ENABLE_BIT 0x10
-#define IZ_STENCIL_TEST_ENABLE_BIT  0x20
-#define IZ_EARLY_DEPTH_TEST_BIT     0x40
-#define IZ_BIT_MAX                  0x80
-
-#define AA_NEVER     0
-#define AA_SOMETIMES 1
-#define AA_ALWAYS    2
-
-struct brw_wm_prog_key {
-   unsigned source_depth_reg:3;
-   unsigned aa_dest_stencil_reg:3;
-   unsigned dest_depth_reg:3;
-   unsigned nr_depth_regs:3;
-   unsigned shadowtex_mask:8;
-   unsigned computes_depth:1;	/* could be derived from program string */
-   unsigned source_depth_to_render_target:1;
-   unsigned runtime_check_aads_emit:1;
-
-   unsigned yuvtex_mask:8;
-
-   unsigned program_string_id;
-};
-
-
-
-
-
-#define PROGRAM_INTERNAL_PARAM
-#define MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS 1024 /* 72 for GL_ARB_f_p */
-#define BRW_WM_MAX_INSN  (MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS*3 + PIPE_MAX_ATTRIBS + 3)
-#define BRW_WM_MAX_GRF   128		/* hardware limit */
-#define BRW_WM_MAX_VREG  (BRW_WM_MAX_INSN * 4)
-#define BRW_WM_MAX_REF   (BRW_WM_MAX_INSN * 12)
-#define BRW_WM_MAX_PARAM 256
-#define BRW_WM_MAX_CONST 256
-#define BRW_WM_MAX_KILLS MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS
-
-#define PAYLOAD_DEPTH     (PIPE_MAX_ATTRIBS)
-
-#define MAX_IFSN 32
-#define MAX_LOOP_DEPTH 32
-
-struct brw_wm_compile {
-   struct brw_compile func;
-   struct brw_wm_prog_key key;
-   struct brw_wm_prog_data prog_data; /* result */
-
-   struct brw_fragment_program *fp;
-
-   unsigned grf_limit;
-   unsigned max_wm_grf;
-
-
-   struct brw_reg pixel_xy[2];
-   struct brw_reg delta_xy[2];
-   struct brw_reg pixel_w;
-
-
-   struct brw_reg wm_regs[8][32][4];
-
-   struct brw_reg payload_depth[4];
-   struct brw_reg payload_coef[16];
-
-   struct brw_reg emit_mask_reg;
-
-   struct brw_instruction *if_inst[MAX_IFSN];
-   int if_insn;
-
-   struct brw_instruction *loop_inst[MAX_LOOP_DEPTH];
-   int loop_insn;
-
-   struct brw_instruction *inst0;
-   struct brw_instruction *inst1;
-
-   struct brw_reg stack;
-   struct brw_indirect stack_index;
-
-   unsigned reg_index;
-
-   unsigned tmp_start;
-   unsigned tmp_index;
-};
-
-
-
-void brw_wm_lookup_iz( unsigned line_aa,
-		       unsigned lookup,
-		       struct brw_wm_prog_key *key );
-
-void brw_wm_glsl_emit(struct brw_wm_compile *c);
-void brw_wm_emit_decls(struct brw_wm_compile *c);
-
-#endif
diff --git a/src/gallium/drivers/i965simple/brw_wm_decl.c b/src/gallium/drivers/i965simple/brw_wm_decl.c
deleted file mode 100644
index d50e66f613..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm_decl.c
+++ /dev/null
@@ -1,392 +0,0 @@
-
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_wm.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "pipe/p_shader_tokens.h"
-#include "tgsi/tgsi_parse.h"
-
-static struct brw_reg alloc_tmp(struct brw_wm_compile *c)
-{
-   c->tmp_index++;
-   c->reg_index = MAX2(c->reg_index, c->tmp_start + c->tmp_index);
-   return brw_vec8_grf(c->tmp_start + c->tmp_index, 0);
-}
-
-static void release_tmps(struct brw_wm_compile *c)
-{
-   c->tmp_index = 0;
-}
-
-
-
-static int is_null( struct brw_reg reg )
-{
-   return (reg.file == BRW_ARCHITECTURE_REGISTER_FILE &&
-	   reg.nr == BRW_ARF_NULL);
-}
-
-static void emit_pixel_xy( struct brw_wm_compile *c )
-{
-   if (is_null(c->pixel_xy[0])) {
-
-      struct brw_compile *p = &c->func;
-      struct brw_reg r1_uw = retype(brw_vec1_grf(1, 0), BRW_REGISTER_TYPE_UW);
-
-      c->pixel_xy[0] = vec8(retype(alloc_tmp(c), BRW_REGISTER_TYPE_UW));
-      c->pixel_xy[1] = vec8(retype(alloc_tmp(c), BRW_REGISTER_TYPE_UW));
-
-      /* Calculate pixel centers by adding 1 or 0 to each of the
-       * micro-tile coordinates passed in r1.
-       */
-      brw_ADD(p,
-	      c->pixel_xy[0],
-	      stride(suboffset(r1_uw, 4), 2, 4, 0),
-	      brw_imm_v(0x10101010));
-
-      brw_ADD(p,
-	      c->pixel_xy[1],
-	      stride(suboffset(r1_uw, 5), 2, 4, 0),
-	      brw_imm_v(0x11001100));
-   }
-}
-
-
-
-
-
-
-static void emit_delta_xy( struct brw_wm_compile *c )
-{
-   if (is_null(c->delta_xy[0])) {
-      struct brw_compile *p = &c->func;
-      struct brw_reg r1 = brw_vec1_grf(1, 0);
-
-      emit_pixel_xy(c);
-
-      c->delta_xy[0] = alloc_tmp(c);
-      c->delta_xy[1] = alloc_tmp(c);
-
-      /* Calc delta X,Y by subtracting origin in r1 from the pixel
-       * centers.
-       */
-      brw_ADD(p,
-	      c->delta_xy[0],
-	      retype(c->pixel_xy[0], BRW_REGISTER_TYPE_UW),
-	      negate(r1));
-
-      brw_ADD(p,
-	      c->delta_xy[1],
-	      retype(c->pixel_xy[1], BRW_REGISTER_TYPE_UW),
-	      negate(suboffset(r1,1)));
-   }
-}
-
-
-
-#if 0
-static void emit_pixel_w( struct brw_wm_compile *c )
-{
-   if (is_null(c->pixel_w)) {
-      struct brw_compile *p = &c->func;
-
-      struct brw_reg interp_wpos = c->coef_wpos;
-      
-      c->pixel_w = alloc_tmp(c);
-
-      emit_delta_xy(c);
-
-      /* Calc 1/w - just linterp wpos[3] optimized by putting the
-       * result straight into a message reg.
-       */
-      struct brw_reg interp3 = brw_vec1_grf(interp_wpos.nr+1, 4);
-      brw_LINE(p, brw_null_reg(), interp3, c->delta_xy[0]);
-      brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), c->delta_xy[1]);
-
-      /* Calc w */
-      brw_math_16( p, 
-		   c->pixel_w,
-		   BRW_MATH_FUNCTION_INV,
-		   BRW_MATH_SATURATE_NONE,
-		   2, 
-		   brw_null_reg(),
-		   BRW_MATH_PRECISION_FULL);
-   }
-}
-#endif
-
-
-static void emit_cinterp(struct brw_wm_compile *c,
-			 int idx,
-			 int mask )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg interp[4];
-   struct brw_reg coef = c->payload_coef[idx];
-   int i;
-
-   interp[0] = brw_vec1_grf(coef.nr, 0);
-   interp[1] = brw_vec1_grf(coef.nr, 4);
-   interp[2] = brw_vec1_grf(coef.nr+1, 0);
-   interp[3] = brw_vec1_grf(coef.nr+1, 4);
-
-   for(i = 0; i < 4; i++ ) {
-      if (mask & (1<<i)) {
-	 struct brw_reg dst = c->wm_regs[TGSI_FILE_INPUT][idx][i];
-	 brw_MOV(p, dst, suboffset(interp[i],3));
-      }
-   }
-}
-
-static void emit_linterp(struct brw_wm_compile *c,
-			 int idx,
-			 int mask )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg interp[4];
-   struct brw_reg coef = c->payload_coef[idx];
-   int i;
-
-   emit_delta_xy(c);
-
-   interp[0] = brw_vec1_grf(coef.nr, 0);
-   interp[1] = brw_vec1_grf(coef.nr, 4);
-   interp[2] = brw_vec1_grf(coef.nr+1, 0);
-   interp[3] = brw_vec1_grf(coef.nr+1, 4);
-
-   for(i = 0; i < 4; i++ ) {
-      if (mask & (1<<i)) {
-	 struct brw_reg dst = c->wm_regs[TGSI_FILE_INPUT][idx][i];
-	 brw_LINE(p, brw_null_reg(), interp[i], c->delta_xy[0]);
-	 brw_MAC(p, dst, suboffset(interp[i],1), c->delta_xy[1]);
-      }
-   }
-}
-
-#if 0
-static void emit_pinterp(struct brw_wm_compile *c,
-			 int idx,
-			 int mask )
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg interp[4];
-   struct brw_reg coef = c->payload_coef[idx];
-   int i;
-
-   get_delta_xy(c);
-   get_pixel_w(c);
-
-   interp[0] = brw_vec1_grf(coef.nr, 0);
-   interp[1] = brw_vec1_grf(coef.nr, 4);
-   interp[2] = brw_vec1_grf(coef.nr+1, 0);
-   interp[3] = brw_vec1_grf(coef.nr+1, 4);
-
-   for(i = 0; i < 4; i++ ) {
-      if (mask & (1<<i)) {
-	 struct brw_reg dst = allocate_reg(c, TGSI_FILE_INPUT, idx, i);
-	 brw_LINE(p, brw_null_reg(), interp[i], c->delta_xy[0]);
-	 brw_MAC(p, dst, suboffset(interp[i],1), c->delta_xy[1]);
-	 brw_MUL(p, dst, dst, c->pixel_w);
-      }
-   }
-}
-#endif
-
-
-
-#if 0
-static void emit_wpos( )
-{ 
-   struct prog_dst_register dst = dst_reg(PROGRAM_INPUT, idx);
-   struct tgsi_full_src_register interp = src_reg(PROGRAM_PAYLOAD, idx);
-   struct tgsi_full_src_register deltas = get_delta_xy(c);
-   struct tgsi_full_src_register arg2;
-   unsigned opcode;
-
-   opcode = WM_LINTERP;
-   arg2 = src_undef();
-
-   /* Have to treat wpos.xy specially:
-    */
-   emit_op(c,
-	   WM_WPOSXY,
-	   dst_mask(dst, WRITEMASK_XY),
-	   0, 0, 0,
-	   get_pixel_xy(c),
-	   src_undef(),
-	   src_undef());
-      
-   dst = dst_mask(dst, WRITEMASK_ZW);
-
-   /* PROGRAM_INPUT.attr.xyzw = INTERP payload.interp[attr].x, deltas.xyw
-    */
-   emit_op(c,
-	   WM_LINTERP,
-	   dst,
-	   0, 0, 0,
-	   interp,
-	   deltas,
-	   arg2);
-}
-#endif
-
-
-
-
-/* Perform register allocation:
- * 
- *  -- r0???
- *  -- passthrough depth regs (and stencil/aa??)
- *  -- curbe ??
- *  -- inputs (coefficients)
- *
- * Use a totally static register allocation.  This will perform poorly
- * but is an easy way to get started (again).
- */
-static void prealloc_reg(struct brw_wm_compile *c)
-{
-   int i, j;
-   int nr_curbe_regs = 0;
-
-   /* R0, then some depth related regs:
-    */
-   for (i = 0; i < c->key.nr_depth_regs; i++) {
-      c->payload_depth[i] =  brw_vec8_grf(i*2, 0);
-      c->reg_index += 2;
-   }
-
-
-   /* Then a copy of our part of the CURBE entry:
-    */
-   {
-      int nr_constants = c->fp->info.file_max[TGSI_FILE_CONSTANT] + 1;
-      int index = 0;
-
-      /* XXX number of constants, or highest numbered constant? */
-      assert(nr_constants == c->fp->info.file_count[TGSI_FILE_CONSTANT]);
-
-      c->prog_data.max_const = 4*nr_constants;
-      for (i = 0; i < nr_constants; i++) {
-	 for (j = 0; j < 4; j++, index++) 
-	    c->wm_regs[TGSI_FILE_CONSTANT][i][j] = brw_vec1_grf(c->reg_index + index/8,
-								index%8);
-      }
-
-      nr_curbe_regs = 2*((4*nr_constants+15)/16);
-      c->reg_index += nr_curbe_regs;
-   }
-
-   /* Adjust for parameter coefficients for position, which are
-    * currently always provided.
-    */
-//   c->position_coef[i] = brw_vec8_grf(c->reg_index, 0);
-   c->reg_index += 2;
-
-   /* Next we receive the plane coefficients for parameter
-    * interpolation:
-    */
-   assert(c->fp->info.file_max[TGSI_FILE_INPUT] == c->fp->info.num_inputs);
-   for (i = 0; i < c->fp->info.file_max[TGSI_FILE_INPUT] + 1; i++) {
-      c->payload_coef[i] = brw_vec8_grf(c->reg_index, 0);
-      c->reg_index += 2;
-   }
-
-   c->prog_data.first_curbe_grf = c->key.nr_depth_regs * 2;
-   c->prog_data.urb_read_length = (c->fp->info.num_inputs + 1) * 2;
-   c->prog_data.curb_read_length = nr_curbe_regs;
-
-   /* That's the end of the payload, now we can start allocating registers.
-    */
-   c->emit_mask_reg = brw_uw1_reg(BRW_GENERAL_REGISTER_FILE, c->reg_index, 0);
-   c->reg_index++;
-
-   c->stack = brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, c->reg_index, 0);
-   c->reg_index += 2;
-
-   /* Now allocate room for the interpolated inputs and staging
-    * registers for the outputs:
-    */
-   /* XXX do we want to loop over the _number_ of inputs/outputs or loop
-    * to the highest input/output index that's used?
-    *  Probably the same, actually.
-    */
-   assert(c->fp->info.file_max[TGSI_FILE_INPUT] + 1 == c->fp->info.num_inputs);
-   assert(c->fp->info.file_max[TGSI_FILE_OUTPUT] + 1 == c->fp->info.num_outputs);
-   for (i = 0; i < c->fp->info.file_max[TGSI_FILE_INPUT] + 1; i++) 
-      for (j = 0; j < 4; j++)
-	 c->wm_regs[TGSI_FILE_INPUT][i][j] = brw_vec8_grf( c->reg_index++, 0 );
-
-   for (i = 0; i < c->fp->info.file_max[TGSI_FILE_OUTPUT] + 1; i++) 
-      for (j = 0; j < 4; j++)
-	 c->wm_regs[TGSI_FILE_OUTPUT][i][j] = brw_vec8_grf( c->reg_index++, 0 );
-
-   /* Beyond this we should only need registers for internal temporaries:
-    */
-   c->tmp_start = c->reg_index;
-}
-
-
-
-
-
-/* Need to interpolate fragment program inputs in as a preamble to the
- * shader.  A more sophisticated compiler would do this on demand, but
- * we'll do it up front:
- */
-void brw_wm_emit_decls(struct brw_wm_compile *c)
-{
-   struct tgsi_parse_context parse;
-   int done = 0;
-
-   prealloc_reg(c);
-
-   tgsi_parse_init( &parse, c->fp->program.tokens );
-
-   while( !done &&
-	  !tgsi_parse_end_of_tokens( &parse ) ) 
-   {
-      tgsi_parse_token( &parse );
-
-      switch( parse.FullToken.Token.Type ) {
-      case TGSI_TOKEN_TYPE_DECLARATION:
-      {
-	 const struct tgsi_full_declaration *decl = &parse.FullToken.FullDeclaration;
-	 unsigned first = decl->DeclarationRange.First;
-	 unsigned last = decl->DeclarationRange.Last;
-	 unsigned mask = decl->Declaration.UsageMask; /* ? */
-	 unsigned i;
-
-	 if (decl->Declaration.File != TGSI_FILE_INPUT)
-	    break;
-
-	 for( i = first; i <= last; i++ ) {
-	    switch (decl->Declaration.Interpolate) {
-	    case TGSI_INTERPOLATE_CONSTANT:
-	       emit_cinterp(c, i, mask);
-	       break;
-
-	    case TGSI_INTERPOLATE_LINEAR:
-	       emit_linterp(c, i, mask);
-	       break;
-
-	    case TGSI_INTERPOLATE_PERSPECTIVE:
-	       //emit_pinterp(c, i, mask);
-	       emit_linterp(c, i, mask);
-	       break;
-	    }
-	 }
-	 break;
-      }
-      case TGSI_TOKEN_TYPE_IMMEDIATE:
-      case TGSI_TOKEN_TYPE_INSTRUCTION:
-      default:
-         done = 1;
-	 break;
-      }
-   }
-
-   tgsi_parse_free (&parse);
-   
-   release_tmps(c);
-}
diff --git a/src/gallium/drivers/i965simple/brw_wm_glsl.c b/src/gallium/drivers/i965simple/brw_wm_glsl.c
deleted file mode 100644
index db75963932..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm_glsl.c
+++ /dev/null
@@ -1,1076 +0,0 @@
-
-#include "brw_context.h"
-#include "brw_eu.h"
-#include "brw_wm.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "pipe/p_shader_tokens.h"
-#include "tgsi/tgsi_parse.h"
-
-
-
-static int get_scalar_dst_index(struct tgsi_full_instruction *inst)
-{
-   struct tgsi_dst_register dst = inst->FullDstRegisters[0].DstRegister;
-   int i;
-   for (i = 0; i < 4; i++)
-      if (dst.WriteMask & (1<<i))
-	 break;
-   return i;
-}
-
-static struct brw_reg alloc_tmp(struct brw_wm_compile *c)
-{
-   c->tmp_index++;
-   c->reg_index = MAX2(c->reg_index, c->tmp_index);
-   return brw_vec8_grf(c->tmp_start + c->tmp_index, 0);
-}
-
-static void release_tmps(struct brw_wm_compile *c)
-{
-   c->tmp_index = 0;
-}
-
-
-static struct brw_reg
-get_reg(struct brw_wm_compile *c, int file, int index, int component )
-{
-   switch (file) {
-   case TGSI_FILE_NULL:
-      return brw_null_reg();
-
-   case TGSI_FILE_SAMPLER:
-      /* Should never get here:
-       */
-      assert (0);	       
-      return brw_null_reg();
-
-   case TGSI_FILE_IMMEDIATE:
-      /* These need a different path:
-       */
-      assert(0);
-      return brw_null_reg();
-
-       
-   case TGSI_FILE_CONSTANT:
-   case TGSI_FILE_INPUT:
-   case TGSI_FILE_OUTPUT:
-   case TGSI_FILE_TEMPORARY:
-   case TGSI_FILE_ADDRESS:
-      return c->wm_regs[file][index][component];
-
-   default:
-      assert(0);
-      return brw_null_reg();
-   }
-}
-
-
-static struct brw_reg get_dst_reg(struct brw_wm_compile *c,
-				  struct tgsi_full_instruction *inst, 
-				  int component)
-{
-   return get_reg(c, 
-		  inst->FullDstRegisters[0].DstRegister.File, 
-		  inst->FullDstRegisters[0].DstRegister.Index,
-		  component);
-}
-
-static int get_swz( struct tgsi_src_register src, int index )
-{
-   switch (index & 3) {
-   case 0: return src.SwizzleX;
-   case 1: return src.SwizzleY;
-   case 2: return src.SwizzleZ;
-   case 3: return src.SwizzleW;
-   default: return 0;
-   }
-}
-
-static int get_ext_swz( struct tgsi_src_register_ext_swz src, int index )
-{
-   switch (index & 3) {
-   case 0: return src.ExtSwizzleX;
-   case 1: return src.ExtSwizzleY;
-   case 2: return src.ExtSwizzleZ;
-   case 3: return src.ExtSwizzleW;
-   default: return 0;
-   }
-}
-
-static struct brw_reg get_src_reg(struct brw_wm_compile *c,
-				  struct tgsi_full_src_register *src, 
-				  int index)
-{
-   struct brw_reg reg;
-   int component = index;
-   int neg = 0;
-   int abs = 0;
-
-   if (src->SrcRegister.Negate)
-      neg = 1;
-
-   component = get_swz(src->SrcRegister, component);
-
-   /* Yes, there are multiple negates:
-    */
-   switch (component & 3) {
-   case 0: neg ^= src->SrcRegisterExtSwz.NegateX; break;
-   case 1: neg ^= src->SrcRegisterExtSwz.NegateY; break;
-   case 2: neg ^= src->SrcRegisterExtSwz.NegateZ; break;
-   case 3: neg ^= src->SrcRegisterExtSwz.NegateW; break;
-   }
-
-   /* And multiple swizzles, fun isn't it:
-    */
-   component = get_ext_swz(src->SrcRegisterExtSwz, component);
-
-   /* Not handling indirect lookups yet:
-    */
-   assert(src->SrcRegister.Indirect == 0);
-
-   /* Don't know what dimension means:
-    */
-   assert(src->SrcRegister.Dimension == 0);
-
-   /* Will never handle any of this stuff: 
-    */
-   assert(src->SrcRegisterExtMod.Complement == 0);
-   assert(src->SrcRegisterExtMod.Bias == 0);
-   assert(src->SrcRegisterExtMod.Scale2X == 0);
-
-   if (src->SrcRegisterExtMod.Absolute)
-      abs = 1;
-
-   /* Another negate!  This is a post-absolute negate, which we
-    * can't do.  Need to clean the crap out of tgsi somehow.
-    */
-   assert(src->SrcRegisterExtMod.Negate == 0);
-
-   switch( component ) {
-   case TGSI_EXTSWIZZLE_X:
-   case TGSI_EXTSWIZZLE_Y:
-   case TGSI_EXTSWIZZLE_Z:
-   case TGSI_EXTSWIZZLE_W:
-      reg = get_reg(c, 
-		    src->SrcRegister.File, 
-		    src->SrcRegister.Index, 
-		    component );
-
-      if (neg) 
-	 reg = negate(reg);
-   
-      if (abs)
-	 reg = brw_abs(reg);
-
-      break;
-
-      /* XXX: this won't really work in the general case, but we know
-       * that the extended swizzle is only allowed in the SWZ
-       * instruction (right??), in which case using an immediate
-       * directly will work.
-       */
-   case TGSI_EXTSWIZZLE_ZERO:
-      reg = brw_imm_f(0);
-      break;
-
-   case TGSI_EXTSWIZZLE_ONE:
-      if (neg && !abs)
-	 reg = brw_imm_f(-1.0);
-      else
-	 reg = brw_imm_f(1.0);
-      break;
-
-   default:
-      assert(0);
-      break;
-   }
-
-    
-   return reg;
-}
-
-static void emit_abs( struct brw_wm_compile *c,
-		      struct tgsi_full_instruction *inst)
-{
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-
-   int i;
-   struct brw_compile *p = &c->func;
-   brw_set_saturate(p, inst->Instruction.Saturate != TGSI_SAT_NONE);
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 struct brw_reg src, dst;
-	 dst = get_dst_reg(c, inst, i);
-	 src = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 brw_MOV(p, dst, brw_abs(src)); /* NOTE */
-      }
-   }
-   brw_set_saturate(p, 0);
-}
-
-
-static void emit_xpd(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   int i;
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   for (i = 0; i < 4; i++) {
-      unsigned i2 = (i+2)%3;
-      unsigned i1 = (i+1)%3;
-      if (mask & (1<<i)) {
-	 struct brw_reg src0, src1, dst;
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = negate(get_src_reg(c, &inst->FullSrcRegisters[0], i2));
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i1);
-	 brw_MUL(p, brw_null_reg(), src0, src1);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i1);
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i2);
-	 brw_set_saturate(p, inst->Instruction.Saturate != TGSI_SAT_NONE);
-	 brw_MAC(p, dst, src0, src1);
-	 brw_set_saturate(p, 0);
-      }
-   }
-   brw_set_saturate(p, 0);
-}
-
-static void emit_dp3(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_reg src0[3], src1[3], dst;
-   int i;
-   struct brw_compile *p = &c->func;
-   for (i = 0; i < 3; i++) {
-      src0[i] = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-      src1[i] = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-   }
-
-   dst = get_dst_reg(c, inst, get_scalar_dst_index(inst));
-   brw_MUL(p, brw_null_reg(), src0[0], src1[0]);
-   brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
-   brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-   brw_MAC(p, dst, src0[2], src1[2]);
-   brw_set_saturate(p, 0);
-}
-
-static void emit_dp4(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_reg src0[4], src1[4], dst;
-   int i;
-   struct brw_compile *p = &c->func;
-   for (i = 0; i < 4; i++) {
-      src0[i] = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-      src1[i] = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-   }
-   dst = get_dst_reg(c, inst, get_scalar_dst_index(inst));
-   brw_MUL(p, brw_null_reg(), src0[0], src1[0]);
-   brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
-   brw_MAC(p, brw_null_reg(), src0[2], src1[2]);
-   brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-   brw_MAC(p, dst, src0[3], src1[3]);
-   brw_set_saturate(p, 0);
-}
-
-static void emit_dph(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_reg src0[4], src1[4], dst;
-   int i;
-   struct brw_compile *p = &c->func;
-   for (i = 0; i < 4; i++) {
-      src0[i] = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-      src1[i] = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-   }
-   dst = get_dst_reg(c, inst, get_scalar_dst_index(inst));
-   brw_MUL(p, brw_null_reg(), src0[0], src1[0]);
-   brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
-   brw_MAC(p, dst, src0[2], src1[2]);
-   brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-   brw_ADD(p, dst, src0[3], src1[3]);
-   brw_set_saturate(p, 0);
-}
-
-static void emit_math1(struct brw_wm_compile *c,
-		       struct tgsi_full_instruction *inst, unsigned func)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg src0, dst;
-
-   src0 = get_src_reg(c, &inst->FullSrcRegisters[0], 0);
-   dst = get_dst_reg(c, inst, get_scalar_dst_index(inst));
-   brw_MOV(p, brw_message_reg(2), src0);
-   brw_math(p,
-	    dst,
-	    func,
-	    ((inst->Instruction.Saturate != TGSI_SAT_NONE) 
-	     ? BRW_MATH_SATURATE_SATURATE 
-	     : BRW_MATH_SATURATE_NONE),
-	    2,
-	    brw_null_reg(),
-	    BRW_MATH_DATA_VECTOR,
-	    BRW_MATH_PRECISION_FULL);
-}
-
-
-static void emit_alu2(struct brw_wm_compile *c,		      
-		      struct tgsi_full_instruction *inst,
-		      unsigned opcode)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg src0, src1, dst;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   int i;
-   brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-   for (i = 0 ; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-	 brw_alu2(p, opcode, dst, src0, src1);
-      }
-   }
-   brw_set_saturate(p, 0);
-}
-
-
-static void emit_alu1(struct brw_wm_compile *c,
-		      struct tgsi_full_instruction *inst,
-		      unsigned opcode)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg src0, dst;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   int i;
-   brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-   for (i = 0 ; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 brw_alu1(p, opcode, dst, src0);
-      }
-   }
-   if (inst->Instruction.Saturate != TGSI_SAT_NONE)
-      brw_set_saturate(p, 0);
-}
-
-
-static void emit_max(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg src0, src1, dst;
-   int i;
-   brw_push_insn_state(p);
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-	 brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-	 brw_MOV(p, dst, src0);
-	 brw_set_saturate(p, 0);
-
-	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, src0, src1);
-	 brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-	 brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-	 brw_MOV(p, dst, src1);
-	 brw_set_saturate(p, 0);
-	 brw_set_predicate_control_flag_value(p, 0xff);
-      }
-   }
-   brw_pop_insn_state(p);
-}
-
-static void emit_min(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg src0, src1, dst;
-   int i;
-   brw_push_insn_state(p);
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-	 brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-	 brw_MOV(p, dst, src0);
-	 brw_set_saturate(p, 0);
-
-	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, src1, src0);
-	 brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-	 brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-	 brw_MOV(p, dst, src1);
-	 brw_set_saturate(p, 0);
-	 brw_set_predicate_control_flag_value(p, 0xff);
-      }
-   }
-   brw_pop_insn_state(p);
-}
-
-static void emit_pow(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg dst, src0, src1;
-   dst = get_dst_reg(c, inst, get_scalar_dst_index(inst));
-   src0 = get_src_reg(c, &inst->FullSrcRegisters[0], 0);
-   src1 = get_src_reg(c, &inst->FullSrcRegisters[1], 0);
-
-   brw_MOV(p, brw_message_reg(2), src0);
-   brw_MOV(p, brw_message_reg(3), src1);
-
-   brw_math(p,
-	    dst,
-	    BRW_MATH_FUNCTION_POW,
-	    (inst->Instruction.Saturate != TGSI_SAT_NONE 
-	     ? BRW_MATH_SATURATE_SATURATE 
-	     : BRW_MATH_SATURATE_NONE),
-	    2,
-	    brw_null_reg(),
-	    BRW_MATH_DATA_VECTOR,
-	    BRW_MATH_PRECISION_FULL);
-}
-
-static void emit_lrp(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg dst, tmp1, tmp2, src0, src1, src2;
-   int i;
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-
-	 if (src1.nr == dst.nr) {
-	    tmp1 = alloc_tmp(c);
-	    brw_MOV(p, tmp1, src1);
-	 } else
-	    tmp1 = src1;
-
-	 src2 = get_src_reg(c, &inst->FullSrcRegisters[2], i);
-	 if (src2.nr == dst.nr) {
-	    tmp2 = alloc_tmp(c);
-	    brw_MOV(p, tmp2, src2);
-	 } else
-	    tmp2 = src2;
-
-	 brw_ADD(p, dst, negate(src0), brw_imm_f(1.0));
-	 brw_MUL(p, brw_null_reg(), dst, tmp2);
-	 brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-	 brw_MAC(p, dst, src0, tmp1);
-	 brw_set_saturate(p, 0);
-      }
-      release_tmps(c);
-   }
-}
-
-static void emit_kil(struct brw_wm_compile *c)
-{
-   struct brw_compile *p = &c->func;
-   struct brw_reg depth = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
-   brw_push_insn_state(p);
-   brw_set_mask_control(p, BRW_MASK_DISABLE);
-   brw_NOT(p, c->emit_mask_reg, brw_mask_reg(1)); //IMASK
-   brw_AND(p, depth, c->emit_mask_reg, depth);
-   brw_pop_insn_state(p);
-}
-
-static void emit_mad(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg dst, src0, src1, src2;
-   int i;
-
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-	 src2 = get_src_reg(c, &inst->FullSrcRegisters[2], i);
-	 brw_MUL(p, dst, src0, src1);
-
-	 brw_set_saturate(p, (inst->Instruction.Saturate != TGSI_SAT_NONE) ? 1 : 0);
-	 brw_ADD(p, dst, dst, src2);
-	 brw_set_saturate(p, 0);
-      }
-   }
-}
-
-static void emit_sop(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst, unsigned cond)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg dst, src0, src1;
-   int i;
-
-   brw_push_insn_state(p);
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 src0 = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-	 src1 = get_src_reg(c, &inst->FullSrcRegisters[1], i);
-	 brw_CMP(p, brw_null_reg(), cond, src0, src1);
-	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-	 brw_MOV(p, dst, brw_imm_f(0.0));
-	 brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-	 brw_MOV(p, dst, brw_imm_f(1.0));
-      }
-   }
-   brw_pop_insn_state(p);
-}
-
-
-static void emit_ddx(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg interp[4];
-   struct brw_reg dst;
-   struct brw_reg src0, w;
-   unsigned nr, i;
-   src0 = get_src_reg(c, &inst->FullSrcRegisters[0], 0);
-   w = get_src_reg(c, &inst->FullSrcRegisters[1], 3);
-   nr = src0.nr;
-   interp[0] = brw_vec1_grf(nr, 0);
-   interp[1] = brw_vec1_grf(nr, 4);
-   interp[2] = brw_vec1_grf(nr+1, 0);
-   interp[3] = brw_vec1_grf(nr+1, 4);
-   brw_set_saturate(p, inst->Instruction.Saturate != TGSI_SAT_NONE);
-   for(i = 0; i < 4; i++ ) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 brw_MOV(p, dst, interp[i]);
-	 brw_MUL(p, dst, dst, w);
-      }
-   }
-   brw_set_saturate(p, 0);
-}
-
-static void emit_ddy(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   unsigned mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
-   struct brw_reg interp[4];
-   struct brw_reg dst;
-   struct brw_reg src0, w;
-   unsigned nr, i;
-
-   src0 = get_src_reg(c, &inst->FullSrcRegisters[0], 0);
-   nr = src0.nr;
-   w = get_src_reg(c, &inst->FullSrcRegisters[1], 3);
-   interp[0] = brw_vec1_grf(nr, 0);
-   interp[1] = brw_vec1_grf(nr, 4);
-   interp[2] = brw_vec1_grf(nr+1, 0);
-   interp[3] = brw_vec1_grf(nr+1, 4);
-   brw_set_saturate(p, inst->Instruction.Saturate != TGSI_SAT_NONE);
-   for(i = 0; i < 4; i++ ) {
-      if (mask & (1<<i)) {
-	 dst = get_dst_reg(c, inst, i);
-	 brw_MOV(p, dst, suboffset(interp[i], 1));
-	 brw_MUL(p, dst, dst, w);
-      }
-   }
-   brw_set_saturate(p, 0);
-}
-
-/* TODO
-   BIAS on SIMD8 not workind yet...
-*/
-static void emit_txb(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-#if 0
-   struct brw_compile *p = &c->func;
-   struct brw_reg payload_reg = c->payload_depth[0];
-   struct brw_reg dst[4], src[4];
-   unsigned i;
-   for (i = 0; i < 4; i++)
-      dst[i] = get_dst_reg(c, inst, i);
-   for (i = 0; i < 4; i++)
-      src[i] = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-
-#if 0
-   switch (inst->TexSrcTarget) {
-   case TEXTURE_1D_INDEX:
-      brw_MOV(p, brw_message_reg(2), src[0]);
-      brw_MOV(p, brw_message_reg(3), brw_imm_f(0));
-      brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
-      break;
-   case TEXTURE_2D_INDEX:
-   case TEXTURE_RECT_INDEX:
-      brw_MOV(p, brw_message_reg(2), src[0]);
-      brw_MOV(p, brw_message_reg(3), src[1]);
-      brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
-      break;
-   default:
-      brw_MOV(p, brw_message_reg(2), src[0]);
-      brw_MOV(p, brw_message_reg(3), src[1]);
-      brw_MOV(p, brw_message_reg(4), src[2]);
-      break;
-   }
-#else
-   brw_MOV(p, brw_message_reg(2), src[0]);
-   brw_MOV(p, brw_message_reg(3), src[1]);
-   brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
-#endif
-
-   brw_MOV(p, brw_message_reg(5), src[3]);
-   brw_MOV(p, brw_message_reg(6), brw_imm_f(0));
-   brw_SAMPLE(p,
-	      retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW),
-	      1,
-	      retype(payload_reg, BRW_REGISTER_TYPE_UW),
-	      inst->TexSrcUnit + 1, /* surface */
-	      inst->TexSrcUnit,     /* sampler */
-	      inst->FullDstRegisters[0].DstRegister.WriteMask,
-	      BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS,
-	      4,
-	      4,
-	      0);
-#endif
-}
-
-static void emit_tex(struct brw_wm_compile *c,
-		     struct tgsi_full_instruction *inst)
-{
-#if 0
-   struct brw_compile *p = &c->func;
-   struct brw_reg payload_reg = c->payload_depth[0];
-   struct brw_reg dst[4], src[4];
-   unsigned msg_len;
-   unsigned i, nr;
-   unsigned emit;
-   boolean shadow = (c->key.shadowtex_mask & (1<<inst->TexSrcUnit)) ? 1 : 0;
-
-   for (i = 0; i < 4; i++)
-      dst[i] = get_dst_reg(c, inst, i);
-   for (i = 0; i < 4; i++)
-      src[i] = get_src_reg(c, &inst->FullSrcRegisters[0], i);
-
-#if 0
-   switch (inst->TexSrcTarget) {
-   case TEXTURE_1D_INDEX:
-      emit = WRITEMASK_X;
-      nr = 1;
-      break;
-   case TEXTURE_2D_INDEX:
-   case TEXTURE_RECT_INDEX:
-      emit = WRITEMASK_XY;
-      nr = 2;
-      break;
-   default:
-      emit = WRITEMASK_XYZ;
-      nr = 3;
-      break;
-   }
-#else
-   emit = WRITEMASK_XY;
-   nr = 2;
-#endif
-
-   msg_len = 1;
-
-   for (i = 0; i < nr; i++) {
-      static const unsigned swz[4] = {0,1,2,2};
-      if (emit & (1<<i))
-	 brw_MOV(p, brw_message_reg(msg_len+1), src[swz[i]]);
-      else
-	 brw_MOV(p, brw_message_reg(msg_len+1), brw_imm_f(0));
-      msg_len += 1;
-   }
-
-   if (shadow) {
-      brw_MOV(p, brw_message_reg(5), brw_imm_f(0));
-      brw_MOV(p, brw_message_reg(6), src[2]);
-   }
-
-   brw_SAMPLE(p,
-	      retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW),
-	      1,
-	      retype(payload_reg, BRW_REGISTER_TYPE_UW),
-	      inst->TexSrcUnit + 1, /* surface */
-	      inst->TexSrcUnit,     /* sampler */
-	      inst->FullDstRegisters[0].DstRegister.WriteMask,
-	      BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE,
-	      4,
-	      shadow ? 6 : 4,
-	      0);
-
-   if (shadow)
-      brw_MOV(p, dst[3], brw_imm_f(1.0));
-#endif
-}
-
-
-
-
-
-
-
-
-static void emit_fb_write(struct brw_wm_compile *c,
-			  struct tgsi_full_instruction *inst)
-{
-   struct brw_compile *p = &c->func;
-   int nr = 2;
-   int channel;
-   int base_reg = 0;
-
-   // src0 = output color
-   // src1 = payload_depth[0]
-   // src2 = output depth
-   // dst = ???
-
-
-
-   /* Reserve a space for AA - may not be needed:
-    */
-   if (c->key.aa_dest_stencil_reg)
-      nr += 1;
-
-   {
-      brw_push_insn_state(p);
-      for (channel = 0; channel < 4; channel++) {
-	 struct brw_reg src0 = c->wm_regs[TGSI_FILE_OUTPUT][0][channel];
-
-	 /*  mov (8) m2.0<1>:ud   r28.0<8;8,1>:ud  { Align1 } */
-	 /*  mov (8) m6.0<1>:ud   r29.0<8;8,1>:ud  { Align1 SecHalf } */
-	 brw_MOV(p, brw_message_reg(nr + channel), src0);
-      }
-      /* skip over the regs populated above: */
-      nr += 8;
-      brw_pop_insn_state(p);
-   }
-    
-
-   /* Pass through control information:
-    */
-   /*  mov (8) m1.0<1>:ud   r1.0<8;8,1>:ud   { Align1 NoMask } */
-   {
-      brw_push_insn_state(p);
-      brw_set_mask_control(p, BRW_MASK_DISABLE); /* ? */
-      brw_MOV(p,
-	      brw_message_reg(base_reg + 1),
-	      brw_vec8_grf(1, 0));
-      brw_pop_insn_state(p);
-   }
-
-   /* Send framebuffer write message: */
-   brw_fb_WRITE(p,
-		retype(vec8(brw_null_reg()), BRW_REGISTER_TYPE_UW),
-		base_reg,
-		retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW),
-		0,              /* render surface always 0 */
-		nr,
-		0,
-		1);
-
-}
-
-
-static void brw_wm_emit_instruction( struct brw_wm_compile *c,
-				     struct tgsi_full_instruction *inst )
-{
-   struct brw_compile *p = &c->func;
-
-#if 0   
-   if (inst->CondUpdate)
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-   else
-      brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE);
-#else
-   brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE);
-#endif
-
-   switch (inst->Instruction.Opcode) {
-   case TGSI_OPCODE_ABS:
-      emit_abs(c, inst);
-      break;
-   case TGSI_OPCODE_ADD:
-      emit_alu2(c, inst, BRW_OPCODE_ADD);
-      break;
-   case TGSI_OPCODE_SUB:
-      assert(0);
-//      emit_alu2(c, inst, BRW_OPCODE_SUB);
-      break;
-   case TGSI_OPCODE_FRC:
-      emit_alu1(c, inst, BRW_OPCODE_FRC);
-      break;
-   case TGSI_OPCODE_FLR:
-      assert(0);
-//      emit_alu1(c, inst, BRW_OPCODE_FLR);
-      break;
-   case TGSI_OPCODE_LRP:
-      emit_lrp(c, inst);
-      break;
-   case TGSI_OPCODE_INT:
-      emit_alu1(c, inst, BRW_OPCODE_RNDD);
-      break;
-   case TGSI_OPCODE_MOV:
-      emit_alu1(c, inst, BRW_OPCODE_MOV);
-      break;
-   case TGSI_OPCODE_DP3:
-      emit_dp3(c, inst);
-      break;
-   case TGSI_OPCODE_DP4:
-      emit_dp4(c, inst);
-      break;
-   case TGSI_OPCODE_XPD:
-      emit_xpd(c, inst);
-      break;
-   case TGSI_OPCODE_DPH:
-      emit_dph(c, inst);
-      break;
-   case TGSI_OPCODE_RCP:
-      emit_math1(c, inst, BRW_MATH_FUNCTION_INV);
-      break;
-   case TGSI_OPCODE_RSQ:
-      emit_math1(c, inst, BRW_MATH_FUNCTION_RSQ);
-      break;
-   case TGSI_OPCODE_SIN:
-      emit_math1(c, inst, BRW_MATH_FUNCTION_SIN);
-      break;
-   case TGSI_OPCODE_COS:
-      emit_math1(c, inst, BRW_MATH_FUNCTION_COS);
-      break;
-   case TGSI_OPCODE_EX2:
-      emit_math1(c, inst, BRW_MATH_FUNCTION_EXP);
-      break;
-   case TGSI_OPCODE_LG2:
-      emit_math1(c, inst, BRW_MATH_FUNCTION_LOG);
-      break;
-   case TGSI_OPCODE_MAX:
-      emit_max(c, inst);
-      break;
-   case TGSI_OPCODE_MIN:
-      emit_min(c, inst);
-      break;
-   case TGSI_OPCODE_DDX:
-      emit_ddx(c, inst);
-      break;
-   case TGSI_OPCODE_DDY:
-      emit_ddy(c, inst);
-      break;
-   case TGSI_OPCODE_SLT:
-      emit_sop(c, inst, BRW_CONDITIONAL_L);
-      break;
-   case TGSI_OPCODE_SLE:
-      emit_sop(c, inst, BRW_CONDITIONAL_LE);
-      break;
-   case TGSI_OPCODE_SGT:
-      emit_sop(c, inst, BRW_CONDITIONAL_G);
-      break;
-   case TGSI_OPCODE_SGE:
-      emit_sop(c, inst, BRW_CONDITIONAL_GE);
-      break;
-   case TGSI_OPCODE_SEQ:
-      emit_sop(c, inst, BRW_CONDITIONAL_EQ);
-      break;
-   case TGSI_OPCODE_SNE:
-      emit_sop(c, inst, BRW_CONDITIONAL_NEQ);
-      break;
-   case TGSI_OPCODE_MUL:
-      emit_alu2(c, inst, BRW_OPCODE_MUL);
-      break;
-   case TGSI_OPCODE_POW:
-      emit_pow(c, inst);
-      break;
-   case TGSI_OPCODE_MAD:
-      emit_mad(c, inst);
-      break;
-   case TGSI_OPCODE_TEX:
-      emit_tex(c, inst);
-      break;
-   case TGSI_OPCODE_TXB:
-      emit_txb(c, inst);
-      break;
-   case TGSI_OPCODE_TEXKILL:
-      emit_kil(c);
-      break;
-   case TGSI_OPCODE_IF:
-      assert(c->if_insn < MAX_IFSN);
-      c->if_inst[c->if_insn++] = brw_IF(p, BRW_EXECUTE_8);
-      break;
-   case TGSI_OPCODE_ELSE:
-      c->if_inst[c->if_insn-1]  = brw_ELSE(p, c->if_inst[c->if_insn-1]);
-      break;
-   case TGSI_OPCODE_ENDIF:
-      assert(c->if_insn > 0);
-      brw_ENDIF(p, c->if_inst[--c->if_insn]);
-      break;
-   case TGSI_OPCODE_BGNSUB:
-   case TGSI_OPCODE_ENDSUB:
-      break;
-   case TGSI_OPCODE_CAL:
-      brw_push_insn_state(p);
-      brw_set_mask_control(p, BRW_MASK_DISABLE);
-      brw_set_access_mode(p, BRW_ALIGN_1);
-      brw_ADD(p, deref_1ud(c->stack_index, 0), brw_ip_reg(), brw_imm_d(3*16));
-      brw_set_access_mode(p, BRW_ALIGN_16);
-      brw_ADD(p, 
-	      get_addr_reg(c->stack_index),
-	      get_addr_reg(c->stack_index), brw_imm_d(4));
-//      orig_inst = inst->Data;
-//      orig_inst->Data = &p->store[p->nr_insn];
-      assert(0);
-      brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-      brw_pop_insn_state(p);
-      break;
-
-   case TGSI_OPCODE_RET:
-#if 0
-      brw_push_insn_state(p);
-      brw_set_mask_control(p, BRW_MASK_DISABLE);
-      brw_ADD(p, 
-	      get_addr_reg(c->stack_index),
-	      get_addr_reg(c->stack_index), brw_imm_d(-4));
-      brw_set_access_mode(p, BRW_ALIGN_1);
-      brw_MOV(p, brw_ip_reg(), deref_1ud(c->stack_index, 0));
-      brw_set_access_mode(p, BRW_ALIGN_16);
-      brw_pop_insn_state(p);
-#else
-      emit_fb_write(c, inst);
-#endif
-
-      break;
-   case TGSI_OPCODE_BGNFOR:
-      c->loop_inst[c->loop_insn++] = brw_DO(p, BRW_EXECUTE_8);
-      break;
-   case TGSI_OPCODE_BRK:
-      brw_BREAK(p);
-      brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-      break;
-   case TGSI_OPCODE_CONT:
-      brw_CONT(p);
-      brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-      break;
-   case TGSI_OPCODE_ENDFOR:
-      c->loop_insn--;
-      c->inst0 = c->inst1 = brw_WHILE(p, c->loop_inst[c->loop_insn]);
-      /* patch all the BREAK instructions from
-         last BGNFOR */
-      while (c->inst0 > c->loop_inst[c->loop_insn]) {
-	 c->inst0--;
-	 if (c->inst0->header.opcode == BRW_OPCODE_BREAK) {
-	    c->inst0->bits3.if_else.jump_count = c->inst1 - c->inst0 + 1;
-	    c->inst0->bits3.if_else.pop_count = 0;
-	 } else if (c->inst0->header.opcode == BRW_OPCODE_CONTINUE) {
-	    c->inst0->bits3.if_else.jump_count = c->inst1 - c->inst0;
-	    c->inst0->bits3.if_else.pop_count = 0;
-	 }
-      }
-      break;
-   case TGSI_OPCODE_END:
-      emit_fb_write(c, inst);
-      break;
-
-   default:
-      debug_printf("unsupported IR in fragment shader %d\n",
-		   inst->Instruction.Opcode);
-   }
-#if 0
-   if (inst->CondUpdate)
-      brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-   else
-      brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-#endif
-}
-
-
-
-
-
-
-void brw_wm_glsl_emit(struct brw_wm_compile *c)
-{
-   struct tgsi_parse_context parse;
-   struct brw_compile *p = &c->func;
-
-   brw_init_compile(&c->func);
-   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-
-   c->reg_index = 0;
-   c->if_insn = 0;
-   c->loop_insn = 0;
-   c->stack_index = brw_indirect(0,0);
-
-   /* Do static register allocation and parameter interpolation:
-    */
-   brw_wm_emit_decls( c );
-
-   /* Emit the actual program.  All done with very direct translation,
-    * hopefully we can improve on this shortly...
-    */
-   brw_MOV(p, get_addr_reg(c->stack_index), brw_address(c->stack));
-
-   tgsi_parse_init( &parse, c->fp->program.tokens );
-
-   while( !tgsi_parse_end_of_tokens( &parse ) ) 
-   {
-      tgsi_parse_token( &parse );
-
-      switch( parse.FullToken.Token.Type ) {
-      case TGSI_TOKEN_TYPE_DECLARATION:
-	 /* already done */
-	 break;
-
-      case TGSI_TOKEN_TYPE_IMMEDIATE:
-         /* not handled yet */
-	 assert(0);
-         break;
-
-      case TGSI_TOKEN_TYPE_INSTRUCTION:
-         brw_wm_emit_instruction(c, &parse.FullToken.FullInstruction);
-         break;
-
-      default:
-         assert( 0 );
-      }
-   }
-
-   tgsi_parse_free (&parse);
-   
-   /* Fix up call targets:
-    */
-#if 0
-   {
-      unsigned nr_insns = c->fp->program.Base.NumInstructions;
-      unsigned insn, target_insn;
-      struct tgsi_full_instruction *inst1, *inst2;
-      struct brw_instruction *brw_inst1, *brw_inst2;
-      int offset;
-      for (insn = 0; insn < nr_insns; insn++) {
-	 inst1 = &c->fp->program.Base.Instructions[insn];
-	 brw_inst1 = inst1->Data;
-	 switch (inst1->Opcode) {
-	 case TGSI_OPCODE_CAL:
-	    target_insn = inst1->BranchTarget;
-	    inst2 = &c->fp->program.Base.Instructions[target_insn];
-	    brw_inst2 = inst2->Data;
-	    offset = brw_inst2 - brw_inst1;
-	    brw_set_src1(brw_inst1, brw_imm_d(offset*16));
-	    break;
-	 default:
-	    break;
-	 }
-      }
-   }
-#endif
-
-   c->prog_data.total_grf = c->reg_index;
-   c->prog_data.total_scratch = 0;
-}
diff --git a/src/gallium/drivers/i965simple/brw_wm_iz.c b/src/gallium/drivers/i965simple/brw_wm_iz.c
deleted file mode 100644
index 6c5f25bf39..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm_iz.c
+++ /dev/null
@@ -1,214 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_wm.h"
-
-
-#undef P			/* prompted depth */
-#undef C			/* computed */
-#undef N			/* non-promoted? */
-
-#define P 0
-#define C 1
-#define N 2
-
-const struct {
-   unsigned mode:2;
-   unsigned sd_present:1;
-   unsigned sd_to_rt:1;
-   unsigned dd_present:1;
-   unsigned ds_present:1;
-} wm_iz_table[IZ_BIT_MAX] =
-{
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { N, 1, 1, 0, 0 }, 
- { N, 0, 1, 0, 0 }, 
- { N, 0, 1, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { N, 1, 1, 0, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { N, 1, 1, 0, 0 }, 
- { N, 0, 1, 0, 0 }, 
- { N, 0, 1, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { N, 1, 1, 0, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { N, 1, 1, 0, 1 }, 
- { N, 0, 1, 0, 1 }, 
- { N, 0, 1, 0, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { N, 1, 1, 0, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 0, 0, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 } 
-};
-
-void brw_wm_lookup_iz( unsigned line_aa,
-		       unsigned lookup,
-		       struct brw_wm_prog_key *key )
-{
-   unsigned reg = 2;
-
-   assert (lookup < IZ_BIT_MAX);
-      
-   if (lookup & IZ_PS_COMPUTES_DEPTH_BIT)
-      key->computes_depth = 1;
-
-   if (wm_iz_table[lookup].sd_present) {
-      key->source_depth_reg = reg;
-      reg += 2;
-   }
-
-   if (wm_iz_table[lookup].sd_to_rt)
-      key->source_depth_to_render_target = 1;
-
-   if (wm_iz_table[lookup].ds_present || line_aa != AA_NEVER) {
-      key->aa_dest_stencil_reg = reg;
-      key->runtime_check_aads_emit = (!wm_iz_table[lookup].ds_present &&
-				      line_aa == AA_SOMETIMES);
-      reg++;
-   }
-
-   if (wm_iz_table[lookup].dd_present) {
-      key->dest_depth_reg = reg;
-      reg+=2;
-   }
-
-   key->nr_depth_regs = (reg+1)/2;
-}
-
diff --git a/src/gallium/drivers/i965simple/brw_wm_sampler_state.c b/src/gallium/drivers/i965simple/brw_wm_sampler_state.c
deleted file mode 100644
index 52b2909a65..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm_sampler_state.c
+++ /dev/null
@@ -1,275 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-
-#define COMPAREFUNC_ALWAYS		0
-#define COMPAREFUNC_NEVER		0x1
-#define COMPAREFUNC_LESS		0x2
-#define COMPAREFUNC_EQUAL		0x3
-#define COMPAREFUNC_LEQUAL		0x4
-#define COMPAREFUNC_GREATER		0x5
-#define COMPAREFUNC_NOTEQUAL		0x6
-#define COMPAREFUNC_GEQUAL		0x7
-
-/* Samplers aren't strictly wm state from the hardware's perspective,
- * but that is the only situation in which we use them in this driver.
- */
-
-static int intel_translate_shadow_compare_func(unsigned func)
-{
-   switch(func) {
-   case PIPE_FUNC_NEVER:
-       return COMPAREFUNC_ALWAYS;
-   case PIPE_FUNC_LESS:
-       return COMPAREFUNC_LEQUAL;
-   case PIPE_FUNC_LEQUAL:
-       return COMPAREFUNC_LESS;
-   case PIPE_FUNC_GREATER:
-       return COMPAREFUNC_GEQUAL;
-   case PIPE_FUNC_GEQUAL:
-      return COMPAREFUNC_GREATER;
-   case PIPE_FUNC_NOTEQUAL:
-      return COMPAREFUNC_EQUAL;
-   case PIPE_FUNC_EQUAL:
-      return COMPAREFUNC_NOTEQUAL;
-   case PIPE_FUNC_ALWAYS:
-       return COMPAREFUNC_NEVER;
-   }
-
-   debug_printf("Unknown value in %s: %x\n", __FUNCTION__, func);
-   return COMPAREFUNC_NEVER;
-}
-
-/* The brw (and related graphics cores) do not support GL_CLAMP.  The
- * Intel drivers for "other operating systems" implement GL_CLAMP as
- * GL_CLAMP_TO_EDGE, so the same is done here.
- */
-static unsigned translate_wrap_mode( int wrap )
-{
-   switch( wrap ) {
-   case PIPE_TEX_WRAP_REPEAT:
-      return BRW_TEXCOORDMODE_WRAP;
-   case PIPE_TEX_WRAP_CLAMP:
-      return BRW_TEXCOORDMODE_CLAMP;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      return BRW_TEXCOORDMODE_CLAMP; /* conform likes it this way */
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      return BRW_TEXCOORDMODE_CLAMP_BORDER;
-   case PIPE_TEX_WRAP_MIRROR_REPEAT:
-      return BRW_TEXCOORDMODE_MIRROR;
-   default:
-      return BRW_TEXCOORDMODE_WRAP;
-   }
-}
-
-
-static unsigned U_FIXED(float value, unsigned frac_bits)
-{
-   value *= (1<<frac_bits);
-   return value < 0 ? 0 : value;
-}
-
-static int S_FIXED(float value, unsigned frac_bits)
-{
-   return value * (1<<frac_bits);
-}
-
-
-static unsigned upload_default_color( struct brw_context *brw,
-                                      const float *color )
-{
-   struct brw_sampler_default_color sdc;
-
-   COPY_4V(sdc.color, color);
-
-   return brw_cache_data( &brw->cache[BRW_SAMPLER_DEFAULT_COLOR], &sdc );
-}
-
-
-/*
- */
-static void brw_update_sampler_state( const struct pipe_sampler_state *pipe_sampler,
-				      unsigned sdc_gs_offset,
-				      struct brw_sampler_state *sampler)
-{
-   memset(sampler, 0, sizeof(*sampler));
-
-   switch (pipe_sampler->min_mip_filter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      sampler->ss0.min_filter = BRW_MAPFILTER_NEAREST;
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-      sampler->ss0.min_filter = BRW_MAPFILTER_LINEAR;
-      break;
-   case PIPE_TEX_FILTER_ANISO:
-      sampler->ss0.min_filter = BRW_MAPFILTER_ANISOTROPIC;
-      break;
-   default:
-      break;
-   }
-
-   switch (pipe_sampler->min_mip_filter) {
-   case PIPE_TEX_MIPFILTER_NEAREST:
-      sampler->ss0.mip_filter = BRW_MIPFILTER_NEAREST;
-      break;
-   case PIPE_TEX_MIPFILTER_LINEAR:
-      sampler->ss0.mip_filter = BRW_MIPFILTER_LINEAR;
-      break;
-   case PIPE_TEX_MIPFILTER_NONE:
-      sampler->ss0.mip_filter = BRW_MIPFILTER_NONE;
-      break;
-   default:
-      break;
-   }
-   /* Set Anisotropy:
-    */
-   switch (pipe_sampler->mag_img_filter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      sampler->ss0.mag_filter = BRW_MAPFILTER_NEAREST;
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-      sampler->ss0.mag_filter = BRW_MAPFILTER_LINEAR;
-      break;
-   case PIPE_TEX_FILTER_ANISO:
-      sampler->ss0.mag_filter = BRW_MAPFILTER_LINEAR;
-      break;
-   default:
-      break;
-   }
-
-   if (pipe_sampler->max_anisotropy > 2.0) {
-      sampler->ss3.max_aniso = MAX2((pipe_sampler->max_anisotropy - 2) / 2,
-                                    BRW_ANISORATIO_16);
-   }
-
-   sampler->ss1.s_wrap_mode = translate_wrap_mode(pipe_sampler->wrap_s);
-   sampler->ss1.r_wrap_mode = translate_wrap_mode(pipe_sampler->wrap_r);
-   sampler->ss1.t_wrap_mode = translate_wrap_mode(pipe_sampler->wrap_t);
-
-   /* Fulsim complains if I don't do this.  Hardware doesn't mind:
-    */
-#if 0
-   if (texObj->Target == GL_TEXTURE_CUBE_MAP_ARB) {
-      sampler->ss1.r_wrap_mode = BRW_TEXCOORDMODE_CUBE;
-      sampler->ss1.s_wrap_mode = BRW_TEXCOORDMODE_CUBE;
-      sampler->ss1.t_wrap_mode = BRW_TEXCOORDMODE_CUBE;
-   }
-#endif
-
-   /* Set shadow function:
-    */
-   if (pipe_sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-      /* Shadowing is "enabled" by emitting a particular sampler
-       * message (sample_c).  So need to recompile WM program when
-       * shadow comparison is enabled on each/any texture unit.
-       */
-      sampler->ss0.shadow_function = intel_translate_shadow_compare_func(pipe_sampler->compare_func);
-   }
-
-   /* Set LOD bias:
-    */
-   sampler->ss0.lod_bias = S_FIXED(CLAMP(pipe_sampler->lod_bias, -16, 15), 6);
-
-   sampler->ss0.lod_preclamp = 1; /* OpenGL mode */
-   sampler->ss0.default_color_mode = 0; /* OpenGL/DX10 mode */
-
-   /* Set BaseMipLevel, MaxLOD, MinLOD:
-    *
-    * XXX: I don't think that using firstLevel, lastLevel works,
-    * because we always setup the surface state as if firstLevel ==
-    * level zero.  Probably have to subtract firstLevel from each of
-    * these:
-    */
-   sampler->ss0.base_level = U_FIXED(0, 1);
-
-   sampler->ss1.max_lod = U_FIXED(MIN2(MAX2(pipe_sampler->max_lod, 0), 13), 6);
-   sampler->ss1.min_lod = U_FIXED(MIN2(MAX2(pipe_sampler->min_lod, 0), 13), 6);
-
-   sampler->ss2.default_color_pointer = sdc_gs_offset >> 5;
-}
-
-
-
-/* All samplers must be uploaded in a single contiguous array, which
- * complicates various things.  However, this is still too confusing -
- * FIXME: simplify all the different new texture state flags.
- */
-static void upload_wm_samplers(struct brw_context *brw)
-{
-   unsigned unit;
-   unsigned sampler_count = 0;
-
-   /* BRW_NEW_SAMPLER */
-   for (unit = 0; unit < brw->num_textures && unit < brw->num_samplers;
-        unit++) {
-      /* determine unit enable/disable by looking for a bound texture */
-      if (brw->attribs.Texture[unit]) {
-         const struct pipe_sampler_state *sampler = brw->attribs.Samplers[unit];
-	 unsigned sdc_gs_offset = upload_default_color(brw, sampler->border_color);
-
-	 brw_update_sampler_state(sampler,
-				  sdc_gs_offset,
-				  &brw->wm.sampler[unit]);
-
-	 sampler_count = unit + 1;
-      }
-   }
-
-   if (brw->wm.sampler_count != sampler_count) {
-      brw->wm.sampler_count = sampler_count;
-      brw->state.dirty.cache |= CACHE_NEW_SAMPLER;
-   }
-
-   brw->wm.sampler_gs_offset = 0;
-
-   if (brw->wm.sampler_count)
-      brw->wm.sampler_gs_offset =
-	 brw_cache_data_sz(&brw->cache[BRW_SAMPLER],
-			   brw->wm.sampler,
-			   sizeof(struct brw_sampler_state) * brw->wm.sampler_count);
-}
-
-const struct brw_tracked_state brw_wm_samplers = {
-   .dirty = {
-      .brw = BRW_NEW_SAMPLER,
-      .cache = 0
-   },
-   .update = upload_wm_samplers
-};
-
diff --git a/src/gallium/drivers/i965simple/brw_wm_state.c b/src/gallium/drivers/i965simple/brw_wm_state.c
deleted file mode 100644
index 37a9bf919c..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm_state.c
+++ /dev/null
@@ -1,195 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-#include "brw_wm.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-/***********************************************************************
- * WM unit - fragment programs and rasterization
- */
-static void upload_wm_unit(struct brw_context *brw )
-{
-   struct brw_wm_unit_state wm;
-   unsigned max_threads;
-   unsigned per_thread;
-
-   if (BRW_DEBUG & DEBUG_SINGLE_THREAD)
-      max_threads = 0;
-   else
-      max_threads = 31;
-
-
-   memset(&wm, 0, sizeof(wm));
-
-   /* CACHE_NEW_WM_PROG */
-   wm.thread0.grf_reg_count = align(brw->wm.prog_data->total_grf, 16) / 16 - 1;
-   wm.thread0.kernel_start_pointer = brw->wm.prog_gs_offset >> 6;
-   wm.thread3.dispatch_grf_start_reg = brw->wm.prog_data->first_curbe_grf;
-   wm.thread3.urb_entry_read_length = brw->wm.prog_data->urb_read_length;
-   wm.thread3.const_urb_entry_read_length = brw->wm.prog_data->curb_read_length;
-
-   wm.wm5.max_threads = max_threads;
-
-   per_thread = align(brw->wm.prog_data->total_scratch, 1024);
-   assert(per_thread <= 12 * 1024);
-
-#if 0
-   if (brw->wm.prog_data->total_scratch) {
-      unsigned total = per_thread * (max_threads + 1);
-
-      /* Scratch space -- just have to make sure there is sufficient
-       * allocated for the active program and current number of threads.
-       */
-      brw->wm.scratch_buffer_size = total;
-      if (brw->wm.scratch_buffer &&
-	  brw->wm.scratch_buffer_size > brw->wm.scratch_buffer->size) {
-	 dri_bo_unreference(brw->wm.scratch_buffer);
-	 brw->wm.scratch_buffer = NULL;
-      }
-      if (!brw->wm.scratch_buffer) {
-	 brw->wm.scratch_buffer = dri_bo_alloc(intel->intelScreen->bufmgr,
-					       "wm scratch",
-					       brw->wm.scratch_buffer_size,
-					       4096, DRM_BO_FLAG_MEM_TT);
-      }
-   }
-   /* XXX: Scratch buffers are not implemented correectly.
-    *
-    * The scratch offset to be programmed into wm is relative to the general
-    * state base address.  However, using dri_bo_alloc/dri_bo_emit_reloc (or
-    * the previous bmGenBuffers scheme), we get an offset relative to the
-    * start of framebuffer.  Even before then, it was broken in other ways,
-    * so just fail for now if we hit that path.
-    */
-   assert(brw->wm.prog_data->total_scratch == 0);
-#endif
-
-   /* CACHE_NEW_SURFACE */
-   wm.thread1.binding_table_entry_count = brw->wm.nr_surfaces;
-
-   /* BRW_NEW_CURBE_OFFSETS */
-   wm.thread3.const_urb_entry_read_offset = brw->curbe.wm_start * 2;
-
-   wm.thread3.urb_entry_read_offset = 0;
-   wm.thread1.depth_coef_urb_read_offset = 1;
-   wm.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
-
-   /* CACHE_NEW_SAMPLER */
-   wm.wm4.sampler_count = (brw->wm.sampler_count + 1) / 4;
-   wm.wm4.sampler_state_pointer = brw->wm.sampler_gs_offset >> 5;
-
-   /* BRW_NEW_FRAGMENT_PROGRAM */
-   {
-      const struct brw_fragment_program *fp = brw->attribs.FragmentProgram;
-
-      if (fp->UsesDepth)
-	 wm.wm5.program_uses_depth = 1; /* as far as we can tell */
-
-      if (fp->info.writes_z)
-	 wm.wm5.program_computes_depth = 1;
-
-      /* BRW_NEW_ALPHA_TEST */
-      if (fp->info.uses_kill ||
-	  brw->attribs.DepthStencil->alpha.enabled)
-	 wm.wm5.program_uses_killpixel = 1;
-
-      wm.wm5.enable_8_pix = 1;
-   }
-
-   wm.wm5.thread_dispatch_enable = 1;	/* AKA: color_write */
-   wm.wm5.legacy_line_rast = 0;
-   wm.wm5.legacy_global_depth_bias = 0;
-   wm.wm5.early_depth_test = 1;	        /* never need to disable */
-   wm.wm5.line_aa_region_width = 0;
-   wm.wm5.line_endcap_aa_region_width = 1;
-
-   /* BRW_NEW_RASTERIZER */
-   if (brw->attribs.Raster->poly_stipple_enable)
-      wm.wm5.polygon_stipple = 1;
-
-#if 0
-   if (brw->attribs.Polygon->OffsetFill) {
-      wm.wm5.depth_offset = 1;
-      /* Something wierd going on with legacy_global_depth_bias,
-       * offset_constant, scaling and MRD.  This value passes glean
-       * but gives some odd results elsewere (eg. the
-       * quad-offset-units test).
-       */
-      wm.global_depth_offset_constant = brw->attribs.Polygon->OffsetUnits * 2;
-
-      /* This is the only value that passes glean:
-       */
-      wm.global_depth_offset_scale = brw->attribs.Polygon->OffsetFactor;
-   }
-#endif
-
-   if (brw->attribs.Raster->line_stipple_enable) {
-      wm.wm5.line_stipple = 1;
-   }
-
-   if (BRW_DEBUG & DEBUG_STATS)
-      wm.wm4.stats_enable = 1;
-
-   brw->wm.state_gs_offset = brw_cache_data( &brw->cache[BRW_WM_UNIT], &wm );
-
-   if (brw->wm.prog_data->total_scratch) {
-      /*
-      dri_emit_reloc(brw->cache[BRW_WM_UNIT].pool->buffer,
-		     DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE,
-		     (per_thread / 1024) - 1,
-		     brw->wm.state_gs_offset +
-		     ((char *)&wm.thread2 - (char *)&wm),
-		     brw->wm.scratch_buffer);
-      */
-   } else {
-      wm.thread2.scratch_space_base_pointer = 0;
-   }
-}
-
-const struct brw_tracked_state brw_wm_unit = {
-   .dirty = {
-      .brw = (BRW_NEW_RASTERIZER |
-	      BRW_NEW_ALPHA_TEST |
-	      BRW_NEW_FS |
-	      BRW_NEW_CURBE_OFFSETS),
-
-      .cache = (CACHE_NEW_SURFACE |
-		CACHE_NEW_WM_PROG |
-		CACHE_NEW_SAMPLER)
-   },
-   .update = upload_wm_unit
-};
-
diff --git a/src/gallium/drivers/i965simple/brw_wm_surface_state.c b/src/gallium/drivers/i965simple/brw_wm_surface_state.c
deleted file mode 100644
index b5b9e0e702..0000000000
--- a/src/gallium/drivers/i965simple/brw_wm_surface_state.c
+++ /dev/null
@@ -1,305 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
-
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
-
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-
-static unsigned translate_tex_target( enum pipe_texture_target target )
-{
-   switch (target) {
-   case PIPE_TEXTURE_1D:
-      return BRW_SURFACE_1D;
-
-   case PIPE_TEXTURE_2D:
-      return BRW_SURFACE_2D;
-
-   case PIPE_TEXTURE_3D:
-      return BRW_SURFACE_3D;
-
-   case PIPE_TEXTURE_CUBE:
-      return BRW_SURFACE_CUBE;
-
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-static unsigned translate_tex_format( enum pipe_format pipe_format )
-{
-   switch( pipe_format ) {
-   case PIPE_FORMAT_L8_UNORM:
-      return BRW_SURFACEFORMAT_L8_UNORM;
-
-   case PIPE_FORMAT_I8_UNORM:
-      return BRW_SURFACEFORMAT_I8_UNORM;
-
-   case PIPE_FORMAT_A8_UNORM:
-      return BRW_SURFACEFORMAT_A8_UNORM;
-
-   case PIPE_FORMAT_A8L8_UNORM:
-      return BRW_SURFACEFORMAT_L8A8_UNORM;
-
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      assert(0);		/* not supported for sampling */
-      return BRW_SURFACEFORMAT_R8G8B8_UNORM;
-
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      return BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-
-   case PIPE_FORMAT_R8G8B8A8_UNORM:
-      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM;
-
-   case PIPE_FORMAT_R5G6B5_UNORM:
-      return BRW_SURFACEFORMAT_B5G6R5_UNORM;
-
-   case PIPE_FORMAT_A1R5G5B5_UNORM:
-      return BRW_SURFACEFORMAT_B5G5R5A1_UNORM;
-
-   case PIPE_FORMAT_A4R4G4B4_UNORM:
-      return BRW_SURFACEFORMAT_B4G4R4A4_UNORM;
-
-   case PIPE_FORMAT_YCBCR_REV:
-      return BRW_SURFACEFORMAT_YCRCB_NORMAL;
-
-   case PIPE_FORMAT_YCBCR:
-      return BRW_SURFACEFORMAT_YCRCB_SWAPUVY;
-#if 0
-   case PIPE_FORMAT_RGB_FXT1:
-   case PIPE_FORMAT_RGBA_FXT1:
-      return BRW_SURFACEFORMAT_FXT1;
-#endif
-
-   case PIPE_FORMAT_Z16_UNORM:
-      return BRW_SURFACEFORMAT_I16_UNORM;
-#if 0
-   case PIPE_FORMAT_RGB_DXT1:
-       return BRW_SURFACEFORMAT_DXT1_RGB;
-
-   case PIPE_FORMAT_RGBA_DXT1:
-       return BRW_SURFACEFORMAT_BC1_UNORM;
-
-   case PIPE_FORMAT_RGBA_DXT3:
-       return BRW_SURFACEFORMAT_BC2_UNORM;
-
-   case PIPE_FORMAT_RGBA_DXT5:
-       return BRW_SURFACEFORMAT_BC3_UNORM;
-
-   case PIPE_FORMAT_SRGBA8:
-      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM_SRGB;
-   case PIPE_FORMAT_SRGB_DXT1:
-      return BRW_SURFACEFORMAT_BC1_UNORM_SRGB;
-#endif
-
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-static unsigned brw_buffer_offset(struct brw_context *brw,
-                                  struct pipe_buffer *buffer)
-{
-   return brw->winsys->get_buffer_offset(brw->winsys,
-                                         buffer,
-                                         0);
-}
-
-static
-void brw_update_texture_surface( struct brw_context *brw,
-				 unsigned unit )
-{
-   const struct brw_texture *tObj = brw->attribs.Texture[unit];
-   struct brw_surface_state surf;
-
-   memset(&surf, 0, sizeof(surf));
-
-   surf.ss0.mipmap_layout_mode = BRW_SURFACE_MIPMAPLAYOUT_BELOW;
-   surf.ss0.surface_type = translate_tex_target(tObj->base.target);
-   surf.ss0.surface_format = translate_tex_format(tObj->base.format);
-
-   /* This is ok for all textures with channel width 8bit or less:
-    */
-/*    surf.ss0.data_return_format = BRW_SURFACERETURNFORMAT_S1; */
-
-   /* Updated in emit_reloc */
-   surf.ss1.base_addr = brw_buffer_offset( brw, tObj->buffer );
-
-   surf.ss2.mip_count = tObj->base.last_level;
-   surf.ss2.width = tObj->base.width[0] - 1;
-   surf.ss2.height = tObj->base.height[0] - 1;
-
-   surf.ss3.tile_walk = BRW_TILEWALK_XMAJOR;
-   surf.ss3.tiled_surface = 0; /* always zero */
-   surf.ss3.pitch = tObj->stride - 1;
-   surf.ss3.depth = tObj->base.depth[0] - 1;
-
-   surf.ss4.min_lod = 0;
-
-   if (tObj->base.target == PIPE_TEXTURE_CUBE) {
-      surf.ss0.cube_pos_x = 1;
-      surf.ss0.cube_pos_y = 1;
-      surf.ss0.cube_pos_z = 1;
-      surf.ss0.cube_neg_x = 1;
-      surf.ss0.cube_neg_y = 1;
-      surf.ss0.cube_neg_z = 1;
-   }
-
-   brw->wm.bind.surf_ss_offset[unit + 1] =
-      brw_cache_data( &brw->cache[BRW_SS_SURFACE], &surf );
-}
-
-
-
-#define OFFSET(TYPE, FIELD) ( (unsigned)&(((TYPE *)0)->FIELD) )
-
-
-static void upload_wm_surfaces(struct brw_context *brw )
-{
-   unsigned i;
-
-   {
-      struct brw_surface_state surf;
-
-      /* BRW_NEW_FRAMEBUFFER
-       */
-      struct pipe_surface *pipe_surface = brw->attribs.FrameBuffer.cbufs[0];/*fixme*/
-      struct brw_texture *tex = (struct brw_texture *)pipe_surface->texture;
-
-      memset(&surf, 0, sizeof(surf));
-
-      if (pipe_surface != NULL) {
-	 if (pipe_surface->block.size == 4)
-	    surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-	 else
-	    surf.ss0.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM;
-
-	 surf.ss0.surface_type = BRW_SURFACE_2D;
-
-	 surf.ss1.base_addr = brw_buffer_offset( brw, tex->buffer );
-
-	 surf.ss2.width = pipe_surface->width - 1;
-	 surf.ss2.height = pipe_surface->height - 1;
-	 surf.ss3.tile_walk = BRW_TILEWALK_XMAJOR;
-	 surf.ss3.tiled_surface = 0;
-	 surf.ss3.pitch = pipe_surface->stride - 1;
-      } else {
-	 surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-	 surf.ss0.surface_type = BRW_SURFACE_NULL;
-      }
-
-      /* BRW_NEW_BLEND */
-      surf.ss0.color_blend = (!brw->attribs.Blend->logicop_enable &&
-			      brw->attribs.Blend->blend_enable);
-
-
-      surf.ss0.writedisable_red =   !(brw->attribs.Blend->colormask & PIPE_MASK_R);
-      surf.ss0.writedisable_green = !(brw->attribs.Blend->colormask & PIPE_MASK_G);
-      surf.ss0.writedisable_blue =  !(brw->attribs.Blend->colormask & PIPE_MASK_B);
-      surf.ss0.writedisable_alpha = !(brw->attribs.Blend->colormask & PIPE_MASK_A);
-
-
-
-
-      brw->wm.bind.surf_ss_offset[0] = brw_cache_data( &brw->cache[BRW_SS_SURFACE], &surf );
-
-      brw->wm.nr_surfaces = 1;
-   }
-
-
-   /* BRW_NEW_TEXTURE
-    */
-   for (i = 0; i < brw->num_textures && i < brw->num_samplers; i++) {
-      const struct brw_texture *texUnit = brw->attribs.Texture[i];
-
-      if (texUnit &&
-	  texUnit->base.reference.count/*(texUnit->reference.count > 0) == really used */) {
-
-	 brw_update_texture_surface(brw, i);
-
-	 brw->wm.nr_surfaces = i+2;
-      }
-      else {
-	 brw->wm.bind.surf_ss_offset[i+1] = 0;
-      }
-   }
-
-   brw->wm.bind_ss_offset = brw_cache_data( &brw->cache[BRW_SS_SURF_BIND],
-					    &brw->wm.bind );
-}
-
-
-/* KW: Will find a different way to acheive this, see for example the
- * state caches with relocs in the i915 swz driver.
- */
-#if 0
-static void emit_reloc_wm_surfaces(struct brw_context *brw)
-{
-   int unit;
-
-   if (brw->state.draw_region != NULL) {
-      /* Emit framebuffer relocation */
-      dri_emit_reloc(brw_cache_buffer(brw, BRW_SS_SURFACE),
-		     DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE,
-		     0,
-		     brw->wm.bind.surf_ss_offset[0] +
-		     offsetof(struct brw_surface_state, ss1),
-		     brw->state.draw_region->buffer);
-   }
-
-   /* Emit relocations for texture buffers */
-   for (unit = 0; unit < BRW_MAX_TEX_UNIT; unit++) {
-      struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[unit];
-      struct gl_texture_object *tObj = texUnit->_Current;
-      struct intel_texture_object *intelObj = intel_texture_object(tObj);
-
-      if (texUnit->_ReallyEnabled && intelObj->mt != NULL) {
-	 dri_emit_reloc(brw_cache_buffer(brw, BRW_SS_SURFACE),
-			DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
-			0,
-			brw->wm.bind.surf_ss_offset[unit + 1] +
-			offsetof(struct brw_surface_state, ss1),
-			intelObj->mt->region->buffer);
-      }
-   }
-}
-#endif
-
-const struct brw_tracked_state brw_wm_surfaces = {
-   .dirty = {
-      .brw = (BRW_NEW_FRAMEBUFFER |
-	      BRW_NEW_BLEND |
-	      BRW_NEW_TEXTURE),
-      .cache = 0
-   },
-   .update = upload_wm_surfaces,
-};
diff --git a/src/gallium/drivers/llvmpipe/.gitignore b/src/gallium/drivers/llvmpipe/.gitignore
new file mode 100644
index 0000000000..257b72d7b2
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/.gitignore
@@ -0,0 +1 @@
+lp_tile_soa.c
diff --git a/src/gallium/drivers/llvmpipe/Makefile b/src/gallium/drivers/llvmpipe/Makefile
index 5ac09de79e..b96ee23a99 100644
--- a/src/gallium/drivers/llvmpipe/Makefile
+++ b/src/gallium/drivers/llvmpipe/Makefile
@@ -17,9 +17,12 @@ C_SOURCES = \
 	lp_bld_depth.c \
 	lp_bld_flow.c \
 	lp_bld_format_aos.c \
+	lp_bld_format_soa.c \
 	lp_bld_interp.c \
 	lp_bld_intr.c \
 	lp_bld_logic.c \
+	lp_bld_pack.c \
+	lp_bld_sample_soa.c \
 	lp_bld_swizzle.c \
 	lp_bld_struct.c \
 	lp_bld_tgsi_soa.c \
@@ -46,9 +49,13 @@ C_SOURCES = \
 	lp_state_vs.c \
 	lp_surface.c \
 	lp_tex_cache.c \
-	lp_tex_sample.c \
+	lp_tex_sample_c.c \
+	lp_tex_sample_llvm.c \
 	lp_texture.c \
 	lp_tile_cache.c \
 	lp_tile_soa.c
 
 include ../../Makefile.template
+
+lp_tile_soa.c: lp_tile_soa.py ../../auxiliary/util/u_format_parse.py ../../auxiliary/util/u_format_access.py ../../auxiliary/util/u_format.csv
+	python lp_tile_soa.py ../../auxiliary/util/u_format.csv > $@
diff --git a/src/gallium/drivers/llvmpipe/README b/src/gallium/drivers/llvmpipe/README
index 498d21dea6..89d08834a3 100644
--- a/src/gallium/drivers/llvmpipe/README
+++ b/src/gallium/drivers/llvmpipe/README
@@ -8,13 +8,16 @@ Done so far is:
 
  - the whole fragment pipeline is code generated in a single function
  
+   - input interpolation
+   
    - depth testing
  
+   - texture sampling (not all state/formats are supported) 
+   
    - fragment shader TGSI translation
      - same level of support as the TGSI SSE2 exec machine, with the exception
        we don't fallback to TGSI interpretation when an unsupported opcode is
        found, but just ignore it
-     - texture sampling via an intrinsic call
      - done in SoA layout
      - input interpolation also code generated
  
@@ -28,16 +31,17 @@ Done so far is:
      any width and length
    - not all operations are implemented for these types yet though
 
-Most mesa/progs/demos/* work. Speed is on par with Keith's softpipe-opt branch,
-which includes hand written fast implementations for common cases.
+Most mesa/progs/demos/* work. 
 
 To do (probably by this order):
 
  - code generate stipple and stencil testing
 
- - code generate texture sampling
+ - translate the remaining bits of texture sampling state
 
  - translate TGSI control flow instructions, and all other remaining opcodes
+ 
+ - integrate with the draw module for VS code generation
 
  - code generate the triangle setup and rasterization
 
@@ -93,7 +97,7 @@ Alternatively, you can build it with GNU make, if you prefer, by invoking it as
 
   make linux-llvm
 
-but the rest of these instructions assume scons is used.
+but the rest of these instructions assume that scons is used.
 
 
 Using
@@ -108,6 +112,9 @@ or
 
   export LD_LIBRARY_PATH=$PWD/build/linux-x86-debug/lib:$LD_LIBRARY_PATH
 
+For performance evaluation pass debug=no to scons, and use the corresponding
+lib directory without the "-debug" suffix.
+
 
 Unit testing
 ============
@@ -119,7 +126,7 @@ build/linux-???-debug/gallium/drivers/llvmpipe:
  - lp_test_conv: SIMD vector conversion
  - lp_test_format: pixel unpacking/packing
 
-Some of this tests can output results and benchmarks to a tab-seperated-file
+Some of this tests can output results and benchmarks to a tab-separated-file
 for posterior analysis, e.g.:
 
   build/linux-x86_64-debug/gallium/drivers/llvmpipe/lp_test_blend -o blend.tsv
@@ -133,10 +140,10 @@ Development Notes
   at the top of the lp_bld_*.c functions.  
 
 - All lp_bld_*.[ch] are isolated from the rest of the driver, and could/may be 
-  put in a standalone Gallium state -> LLVM IR translation module.
+  put in a stand-alone Gallium state -> LLVM IR translation module.
 
 - We use LLVM-C bindings for now. They are not documented, but follow the C++
   interfaces very closely, and appear to be complete enough for code
   generation. See 
   http://npcontemplation.blogspot.com/2008/06/secret-of-llvm-c-bindings.html
-  for a standalone example.
+  for a stand-alone example.
diff --git a/src/gallium/drivers/llvmpipe/SConscript b/src/gallium/drivers/llvmpipe/SConscript
index 5c29bdac56..403e4daa43 100644
--- a/src/gallium/drivers/llvmpipe/SConscript
+++ b/src/gallium/drivers/llvmpipe/SConscript
@@ -3,12 +3,19 @@ Import('*')
 env = env.Clone()
 
 env.Tool('llvm')
-if 'LLVM_VERSION' not in env:
+if not env.has_key('LLVM_VERSION'):
     print 'warning: LLVM not found: not building llvmpipe'
     Return()
 
 env.Tool('udis86')
 
+env.CodeGenerate(
+	target = 'lp_tile_soa.c',
+	script = 'lp_tile_soa.py',
+	source = ['#src/gallium/auxiliary/util/u_format.csv'],
+	command = 'python $SCRIPT $SOURCE > $TARGET'
+)
+
 llvmpipe = env.ConvenienceLibrary(
 	target = 'llvmpipe',
 	source = [
@@ -23,8 +30,11 @@ llvmpipe = env.ConvenienceLibrary(
 		'lp_bld_depth.c',
 		'lp_bld_flow.c',
 		'lp_bld_format_aos.c',
+		'lp_bld_format_soa.c',
 		'lp_bld_interp.c',
 		'lp_bld_intr.c',
+        'lp_bld_pack.c',
+		'lp_bld_sample_soa.c',
 		'lp_bld_struct.c',
 		'lp_bld_logic.c',
 		'lp_bld_swizzle.c',
@@ -52,7 +62,8 @@ llvmpipe = env.ConvenienceLibrary(
 		'lp_state_vs.c',
 		'lp_surface.c',
 		'lp_tex_cache.c',
-		'lp_tex_sample.c',
+		'lp_tex_sample_c.c',
+		'lp_tex_sample_llvm.c',
 		'lp_texture.c',
 		'lp_tile_cache.c',
 		'lp_tile_soa.c',
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_alpha.c b/src/gallium/drivers/llvmpipe/lp_bld_alpha.c
index 49c2f911af..2b4bc5c819 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_alpha.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_alpha.c
@@ -45,7 +45,7 @@
 void
 lp_build_alpha_test(LLVMBuilderRef builder,
                     const struct pipe_alpha_state *state,
-                    union lp_type type,
+                    struct lp_type type,
                     struct lp_build_mask_context *mask,
                     LLVMValueRef alpha,
                     LLVMValueRef ref)
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_alpha.h b/src/gallium/drivers/llvmpipe/lp_bld_alpha.h
index 9dbcdb4daa..634575670d 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_alpha.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_alpha.h
@@ -38,14 +38,14 @@
 #include <llvm-c/Core.h>  
 
 struct pipe_alpha_state;
-union lp_type;
+struct lp_type;
 struct lp_build_mask_context;
 
 
 void
 lp_build_alpha_test(LLVMBuilderRef builder,
                     const struct pipe_alpha_state *state,
-                    union lp_type type,
+                    struct lp_type type,
                     struct lp_build_mask_context *mask,
                     LLVMValueRef alpha,
                     LLVMValueRef ref);
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.c b/src/gallium/drivers/llvmpipe/lp_bld_arit.c
index 09a57ff33d..83ca06acf8 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_arit.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.c
@@ -48,11 +48,14 @@
 #include "util/u_memory.h"
 #include "util/u_debug.h"
 #include "util/u_string.h"
+#include "util/u_cpu_detect.h"
 
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
 #include "lp_bld_intr.h"
 #include "lp_bld_logic.h"
+#include "lp_bld_pack.h"
+#include "lp_bld_debug.h"
 #include "lp_bld_arit.h"
 
 
@@ -65,36 +68,34 @@ lp_build_min_simple(struct lp_build_context *bld,
                     LLVMValueRef a,
                     LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    const char *intrinsic = NULL;
    LLVMValueRef cond;
 
    /* TODO: optimize the constant case */
 
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
    if(type.width * type.length == 128) {
       if(type.floating) {
-         if(type.width == 32)
+         if(type.width == 32 && util_cpu_caps.has_sse)
             intrinsic = "llvm.x86.sse.min.ps";
-         if(type.width == 64)
+         if(type.width == 64 && util_cpu_caps.has_sse2)
             intrinsic = "llvm.x86.sse2.min.pd";
       }
       else {
-         if(type.width == 8 && !type.sign)
+         if(type.width == 8 && !type.sign && util_cpu_caps.has_sse2)
             intrinsic = "llvm.x86.sse2.pminu.b";
-         if(type.width == 8 && type.sign)
+         if(type.width == 8 && type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pminsb";
-         if(type.width == 16 && !type.sign)
+         if(type.width == 16 && !type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pminuw";
-         if(type.width == 16 && type.sign)
+         if(type.width == 16 && type.sign && util_cpu_caps.has_sse2)
             intrinsic = "llvm.x86.sse2.pmins.w";
-         if(type.width == 32 && !type.sign)
+         if(type.width == 32 && !type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pminud";
-         if(type.width == 32 && type.sign)
+         if(type.width == 32 && type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pminsd";
       }
    }
-#endif
 
    if(intrinsic)
       return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b);
@@ -113,36 +114,34 @@ lp_build_max_simple(struct lp_build_context *bld,
                     LLVMValueRef a,
                     LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    const char *intrinsic = NULL;
    LLVMValueRef cond;
 
    /* TODO: optimize the constant case */
 
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
    if(type.width * type.length == 128) {
       if(type.floating) {
-         if(type.width == 32)
+         if(type.width == 32 && util_cpu_caps.has_sse)
             intrinsic = "llvm.x86.sse.max.ps";
-         if(type.width == 64)
+         if(type.width == 64 && util_cpu_caps.has_sse2)
             intrinsic = "llvm.x86.sse2.max.pd";
       }
       else {
-         if(type.width == 8 && !type.sign)
+         if(type.width == 8 && !type.sign && util_cpu_caps.has_sse2)
             intrinsic = "llvm.x86.sse2.pmaxu.b";
-         if(type.width == 8 && type.sign)
+         if(type.width == 8 && type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pmaxsb";
-         if(type.width == 16 && !type.sign)
+         if(type.width == 16 && !type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pmaxuw";
-         if(type.width == 16 && type.sign)
+         if(type.width == 16 && type.sign && util_cpu_caps.has_sse2)
             intrinsic = "llvm.x86.sse2.pmaxs.w";
-         if(type.width == 32 && !type.sign)
+         if(type.width == 32 && !type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pmaxud";
-         if(type.width == 32 && type.sign)
+         if(type.width == 32 && type.sign && util_cpu_caps.has_sse4_1)
             intrinsic = "llvm.x86.sse41.pmaxsd";
       }
    }
-#endif
 
    if(intrinsic)
       return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b);
@@ -159,7 +158,7 @@ LLVMValueRef
 lp_build_comp(struct lp_build_context *bld,
               LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
 
    if(a == bld->one)
       return bld->zero;
@@ -188,7 +187,7 @@ lp_build_add(struct lp_build_context *bld,
              LLVMValueRef a,
              LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMValueRef res;
 
    if(a == bld->zero)
@@ -204,15 +203,14 @@ lp_build_add(struct lp_build_context *bld,
       if(a == bld->one || b == bld->one)
         return bld->one;
 
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-      if(type.width * type.length == 128 &&
+      if(util_cpu_caps.has_sse2 &&
+         type.width * type.length == 128 &&
          !type.floating && !type.fixed) {
          if(type.width == 8)
             intrinsic = type.sign ? "llvm.x86.sse2.padds.b" : "llvm.x86.sse2.paddus.b";
          if(type.width == 16)
             intrinsic = type.sign ? "llvm.x86.sse2.padds.w" : "llvm.x86.sse2.paddus.w";
       }
-#endif
    
       if(intrinsic)
          return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b);
@@ -241,7 +239,7 @@ lp_build_sub(struct lp_build_context *bld,
              LLVMValueRef a,
              LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMValueRef res;
 
    if(b == bld->zero)
@@ -257,15 +255,14 @@ lp_build_sub(struct lp_build_context *bld,
       if(b == bld->one)
         return bld->zero;
 
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-      if(type.width * type.length == 128 &&
+      if(util_cpu_caps.has_sse2 &&
+         type.width * type.length == 128 &&
          !type.floating && !type.fixed) {
          if(type.width == 8)
             intrinsic = type.sign ? "llvm.x86.sse2.psubs.b" : "llvm.x86.sse2.psubus.b";
          if(type.width == 16)
             intrinsic = type.sign ? "llvm.x86.sse2.psubs.w" : "llvm.x86.sse2.psubus.w";
       }
-#endif
    
       if(intrinsic)
          return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b);
@@ -284,45 +281,6 @@ lp_build_sub(struct lp_build_context *bld,
 
 
 /**
- * Build shuffle vectors that match PUNPCKLxx and PUNPCKHxx instructions.
- */
-static LLVMValueRef 
-lp_build_unpack_shuffle(unsigned n, unsigned lo_hi)
-{
-   LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
-   unsigned i, j;
-
-   assert(n <= LP_MAX_VECTOR_LENGTH);
-   assert(lo_hi < 2);
-
-   for(i = 0, j = lo_hi*n/2; i < n; i += 2, ++j) {
-      elems[i + 0] = LLVMConstInt(LLVMInt32Type(), 0 + j, 0);
-      elems[i + 1] = LLVMConstInt(LLVMInt32Type(), n + j, 0);
-   }
-
-   return LLVMConstVector(elems, n);
-}
-
-
-/**
- * Build constant int vector of width 'n' and value 'c'.
- */
-static LLVMValueRef 
-lp_build_const_vec(LLVMTypeRef type, unsigned n, long long c)
-{
-   LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
-   unsigned i;
-
-   assert(n <= LP_MAX_VECTOR_LENGTH);
-
-   for(i = 0; i < n; ++i)
-      elems[i] = LLVMConstInt(type, c, 0);
-
-   return LLVMConstVector(elems, n);
-}
-
-
-/**
  * Normalized 8bit multiplication.
  *
  * - alpha plus one
@@ -365,33 +323,30 @@ lp_build_const_vec(LLVMTypeRef type, unsigned n, long long c)
  */
 static LLVMValueRef
 lp_build_mul_u8n(LLVMBuilderRef builder,
+                 struct lp_type i16_type,
                  LLVMValueRef a, LLVMValueRef b)
 {
-   static LLVMValueRef c01 = NULL;
-   static LLVMValueRef c08 = NULL;
-   static LLVMValueRef c80 = NULL;
+   LLVMValueRef c8;
    LLVMValueRef ab;
 
-   if(!c01) c01 = lp_build_const_vec(LLVMInt16Type(), 8, 0x01);
-   if(!c08) c08 = lp_build_const_vec(LLVMInt16Type(), 8, 0x08);
-   if(!c80) c80 = lp_build_const_vec(LLVMInt16Type(), 8, 0x80);
+   c8 = lp_build_int_const_scalar(i16_type, 8);
    
 #if 0
    
    /* a*b/255 ~= (a*(b + 1)) >> 256 */
-   b = LLVMBuildAdd(builder, b, c01, "");
+   b = LLVMBuildAdd(builder, b, lp_build_int_const_scalar(i16_type, 1), "");
    ab = LLVMBuildMul(builder, a, b, "");
 
 #else
    
-   /* t/255 ~= (t + (t >> 8) + 0x80) >> 8 */
+   /* ab/255 ~= (ab + (ab >> 8) + 0x80) >> 8 */
    ab = LLVMBuildMul(builder, a, b, "");
-   ab = LLVMBuildAdd(builder, ab, LLVMBuildLShr(builder, ab, c08, ""), "");
-   ab = LLVMBuildAdd(builder, ab, c80, "");
+   ab = LLVMBuildAdd(builder, ab, LLVMBuildLShr(builder, ab, c8, ""), "");
+   ab = LLVMBuildAdd(builder, ab, lp_build_int_const_scalar(i16_type, 0x80), "");
 
 #endif
    
-   ab = LLVMBuildLShr(builder, ab, c08, "");
+   ab = LLVMBuildLShr(builder, ab, c8, "");
 
    return ab;
 }
@@ -405,7 +360,7 @@ lp_build_mul(struct lp_build_context *bld,
              LLVMValueRef a,
              LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
 
    if(a == bld->zero)
       return bld->zero;
@@ -419,44 +374,21 @@ lp_build_mul(struct lp_build_context *bld,
       return bld->undef;
 
    if(!type.floating && !type.fixed && type.norm) {
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-      if(type.width == 8 && type.length == 16) {
-         LLVMTypeRef i16x8 = LLVMVectorType(LLVMInt16Type(), 8);
-         LLVMTypeRef i8x16 = LLVMVectorType(LLVMInt8Type(), 16);
-         static LLVMValueRef ml = NULL;
-         static LLVMValueRef mh = NULL;
-         LLVMValueRef al, ah, bl, bh;
-         LLVMValueRef abl, abh;
-         LLVMValueRef ab;
-         
-         if(!ml) ml = lp_build_unpack_shuffle(16, 0);
-         if(!mh) mh = lp_build_unpack_shuffle(16, 1);
+      if(type.width == 8) {
+         struct lp_type i16_type = lp_wider_type(type);
+         LLVMValueRef al, ah, bl, bh, abl, abh, ab;
 
-         /*  PUNPCKLBW, PUNPCKHBW */
-         al = LLVMBuildShuffleVector(bld->builder, a, bld->zero, ml, "");
-         bl = LLVMBuildShuffleVector(bld->builder, b, bld->zero, ml, "");
-         ah = LLVMBuildShuffleVector(bld->builder, a, bld->zero, mh, "");
-         bh = LLVMBuildShuffleVector(bld->builder, b, bld->zero, mh, "");
-
-         /* NOP */
-         al = LLVMBuildBitCast(bld->builder, al, i16x8, "");
-         bl = LLVMBuildBitCast(bld->builder, bl, i16x8, "");
-         ah = LLVMBuildBitCast(bld->builder, ah, i16x8, "");
-         bh = LLVMBuildBitCast(bld->builder, bh, i16x8, "");
+         lp_build_unpack2(bld->builder, type, i16_type, a, &al, &ah);
+         lp_build_unpack2(bld->builder, type, i16_type, b, &bl, &bh);
 
          /* PMULLW, PSRLW, PADDW */
-         abl = lp_build_mul_u8n(bld->builder, al, bl);
-         abh = lp_build_mul_u8n(bld->builder, ah, bh);
-
-         /* PACKUSWB */
-         ab = lp_build_intrinsic_binary(bld->builder, "llvm.x86.sse2.packuswb.128" , i16x8, abl, abh);
+         abl = lp_build_mul_u8n(bld->builder, i16_type, al, bl);
+         abh = lp_build_mul_u8n(bld->builder, i16_type, ah, bh);
 
-         /* NOP */
-         ab = LLVMBuildBitCast(bld->builder, ab, i8x16, "");
+         ab = lp_build_pack2(bld->builder, i16_type, type, abl, abh);
          
          return ab;
       }
-#endif
 
       /* FIXME */
       assert(0);
@@ -477,7 +409,7 @@ lp_build_div(struct lp_build_context *bld,
              LLVMValueRef a,
              LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
 
    if(a == bld->zero)
       return bld->zero;
@@ -493,15 +425,38 @@ lp_build_div(struct lp_build_context *bld,
    if(LLVMIsConstant(a) && LLVMIsConstant(b))
       return LLVMConstFDiv(a, b);
 
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-   if(type.width == 32 && type.length == 4)
+   if(util_cpu_caps.has_sse && type.width == 32 && type.length == 4)
       return lp_build_mul(bld, a, lp_build_rcp(bld, b));
-#endif
 
    return LLVMBuildFDiv(bld->builder, a, b, "");
 }
 
 
+LLVMValueRef
+lp_build_lerp(struct lp_build_context *bld,
+              LLVMValueRef x,
+              LLVMValueRef v0,
+              LLVMValueRef v1)
+{
+   return lp_build_add(bld, v0, lp_build_mul(bld, x, lp_build_sub(bld, v1, v0)));
+}
+
+
+LLVMValueRef
+lp_build_lerp_2d(struct lp_build_context *bld,
+                 LLVMValueRef x,
+                 LLVMValueRef y,
+                 LLVMValueRef v00,
+                 LLVMValueRef v01,
+                 LLVMValueRef v10,
+                 LLVMValueRef v11)
+{
+   LLVMValueRef v0 = lp_build_lerp(bld, x, v00, v01);
+   LLVMValueRef v1 = lp_build_lerp(bld, x, v10, v11);
+   return lp_build_lerp(bld, y, v0, v1);
+}
+
+
 /**
  * Generate min(a, b)
  * Do checks for special cases.
@@ -565,33 +520,333 @@ LLVMValueRef
 lp_build_abs(struct lp_build_context *bld,
              LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
+   LLVMTypeRef vec_type = lp_build_vec_type(type);
 
    if(!type.sign)
       return a;
 
-   /* XXX: is this really necessary? */
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-   if(!type.floating && type.width*type.length == 128) {
-      LLVMTypeRef vec_type = lp_build_vec_type(type);
-      if(type.width == 8)
+   if(type.floating) {
+      /* Mask out the sign bit */
+      LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+      LLVMValueRef mask = lp_build_int_const_scalar(type, ((unsigned long long)1 << type.width) - 1);
+      a = LLVMBuildBitCast(bld->builder, a, int_vec_type, "");
+      a = LLVMBuildAnd(bld->builder, a, mask, "");
+      a = LLVMBuildBitCast(bld->builder, a, vec_type, "");
+      return a;
+   }
+
+   if(type.width*type.length == 128 && util_cpu_caps.has_ssse3) {
+      switch(type.width) {
+      case 8:
          return lp_build_intrinsic_unary(bld->builder, "llvm.x86.ssse3.pabs.b.128", vec_type, a);
-      if(type.width == 16)
+      case 16:
          return lp_build_intrinsic_unary(bld->builder, "llvm.x86.ssse3.pabs.w.128", vec_type, a);
-      if(type.width == 32)
+      case 32:
          return lp_build_intrinsic_unary(bld->builder, "llvm.x86.ssse3.pabs.d.128", vec_type, a);
+      }
    }
-#endif
 
    return lp_build_max(bld, a, LLVMBuildNeg(bld->builder, a, ""));
 }
 
 
 LLVMValueRef
+lp_build_sgn(struct lp_build_context *bld,
+             LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+   LLVMTypeRef vec_type = lp_build_vec_type(type);
+   LLVMValueRef cond;
+   LLVMValueRef res;
+
+   /* Handle non-zero case */
+   if(!type.sign) {
+      /* if not zero then sign must be positive */
+      res = bld->one;
+   }
+   else if(type.floating) {
+      /* Take the sign bit and add it to 1 constant */
+      LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+      LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1));
+      LLVMValueRef sign;
+      LLVMValueRef one;
+      sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, "");
+      sign = LLVMBuildAnd(bld->builder, sign, mask, "");
+      one = LLVMConstBitCast(bld->one, int_vec_type);
+      res = LLVMBuildOr(bld->builder, sign, one, "");
+      res = LLVMBuildBitCast(bld->builder, res, vec_type, "");
+   }
+   else
+   {
+      LLVMValueRef minus_one = lp_build_const_scalar(type, -1.0);
+      cond = lp_build_cmp(bld, PIPE_FUNC_GREATER, a, bld->zero);
+      res = lp_build_select(bld, cond, bld->one, minus_one);
+   }
+
+   /* Handle zero */
+   cond = lp_build_cmp(bld, PIPE_FUNC_EQUAL, a, bld->zero);
+   res = lp_build_select(bld, cond, bld->zero, bld->one);
+
+   return res;
+}
+
+
+enum lp_build_round_sse41_mode
+{
+   LP_BUILD_ROUND_SSE41_NEAREST = 0,
+   LP_BUILD_ROUND_SSE41_FLOOR = 1,
+   LP_BUILD_ROUND_SSE41_CEIL = 2,
+   LP_BUILD_ROUND_SSE41_TRUNCATE = 3
+};
+
+
+static INLINE LLVMValueRef
+lp_build_round_sse41(struct lp_build_context *bld,
+                     LLVMValueRef a,
+                     enum lp_build_round_sse41_mode mode)
+{
+   const struct lp_type type = bld->type;
+   LLVMTypeRef vec_type = lp_build_vec_type(type);
+   const char *intrinsic;
+
+   assert(type.floating);
+   assert(type.width*type.length == 128);
+   assert(lp_check_value(type, a));
+   assert(util_cpu_caps.has_sse4_1);
+
+   switch(type.width) {
+   case 32:
+      intrinsic = "llvm.x86.sse41.round.ps";
+      break;
+   case 64:
+      intrinsic = "llvm.x86.sse41.round.pd";
+      break;
+   default:
+      assert(0);
+      return bld->undef;
+   }
+
+   return lp_build_intrinsic_binary(bld->builder, intrinsic, vec_type, a,
+                                    LLVMConstInt(LLVMInt32Type(), mode, 0));
+}
+
+
+LLVMValueRef
+lp_build_trunc(struct lp_build_context *bld,
+               LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   if(util_cpu_caps.has_sse4_1)
+      return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_TRUNCATE);
+   else {
+      LLVMTypeRef vec_type = lp_build_vec_type(type);
+      LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+      LLVMValueRef res;
+      res = LLVMBuildFPToSI(bld->builder, a, int_vec_type, "");
+      res = LLVMBuildSIToFP(bld->builder, res, vec_type, "");
+      return res;
+   }
+}
+
+
+LLVMValueRef
+lp_build_round(struct lp_build_context *bld,
+               LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   if(util_cpu_caps.has_sse4_1)
+      return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_NEAREST);
+   else {
+      LLVMTypeRef vec_type = lp_build_vec_type(type);
+      LLVMValueRef res;
+      res = lp_build_iround(bld, a);
+      res = LLVMBuildSIToFP(bld->builder, res, vec_type, "");
+      return res;
+   }
+}
+
+
+LLVMValueRef
+lp_build_floor(struct lp_build_context *bld,
+               LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+
+   assert(type.floating);
+
+   if(util_cpu_caps.has_sse4_1)
+      return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_FLOOR);
+   else {
+      LLVMTypeRef vec_type = lp_build_vec_type(type);
+      LLVMValueRef res;
+      res = lp_build_ifloor(bld, a);
+      res = LLVMBuildSIToFP(bld->builder, res, vec_type, "");
+      return res;
+   }
+}
+
+
+LLVMValueRef
+lp_build_ceil(struct lp_build_context *bld,
+              LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   if(util_cpu_caps.has_sse4_1)
+      return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_CEIL);
+   else {
+      LLVMTypeRef vec_type = lp_build_vec_type(type);
+      LLVMValueRef res;
+      res = lp_build_iceil(bld, a);
+      res = LLVMBuildSIToFP(bld->builder, res, vec_type, "");
+      return res;
+   }
+}
+
+
+/**
+ * Convert to integer, through whichever rounding method that's fastest,
+ * typically truncating to zero.
+ */
+LLVMValueRef
+lp_build_itrunc(struct lp_build_context *bld,
+                LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+   LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   return LLVMBuildFPToSI(bld->builder, a, int_vec_type, "");
+}
+
+
+LLVMValueRef
+lp_build_iround(struct lp_build_context *bld,
+                LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+   LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+   LLVMValueRef res;
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   if(util_cpu_caps.has_sse4_1) {
+      res = lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_NEAREST);
+   }
+   else {
+      LLVMTypeRef vec_type = lp_build_vec_type(type);
+      LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1));
+      LLVMValueRef sign;
+      LLVMValueRef half;
+
+      /* get sign bit */
+      sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, "");
+      sign = LLVMBuildAnd(bld->builder, sign, mask, "");
+
+      /* sign * 0.5 */
+      half = lp_build_const_scalar(type, 0.5);
+      half = LLVMBuildBitCast(bld->builder, half, int_vec_type, "");
+      half = LLVMBuildOr(bld->builder, sign, half, "");
+      half = LLVMBuildBitCast(bld->builder, half, vec_type, "");
+
+      res = LLVMBuildAdd(bld->builder, a, half, "");
+   }
+
+   res = LLVMBuildFPToSI(bld->builder, res, int_vec_type, "");
+
+   return res;
+}
+
+
+LLVMValueRef
+lp_build_ifloor(struct lp_build_context *bld,
+                LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+   LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+   LLVMValueRef res;
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   if(util_cpu_caps.has_sse4_1) {
+      res = lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_FLOOR);
+   }
+   else {
+      /* Take the sign bit and add it to 1 constant */
+      LLVMTypeRef vec_type = lp_build_vec_type(type);
+      unsigned mantissa = lp_mantissa(type);
+      LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1));
+      LLVMValueRef sign;
+      LLVMValueRef offset;
+
+      /* sign = a < 0 ? ~0 : 0 */
+      sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, "");
+      sign = LLVMBuildAnd(bld->builder, sign, mask, "");
+      sign = LLVMBuildAShr(bld->builder, sign, lp_build_int_const_scalar(type, type.width - 1), "");
+
+      /* offset = -0.99999(9)f */
+      offset = lp_build_const_scalar(type, -(double)(((unsigned long long)1 << mantissa) - 1)/((unsigned long long)1 << mantissa));
+      offset = LLVMConstBitCast(offset, int_vec_type);
+
+      /* offset = a < 0 ? -0.99999(9)f : 0.0f */
+      offset = LLVMBuildAnd(bld->builder, offset, sign, "");
+      offset = LLVMBuildBitCast(bld->builder, offset, vec_type, "");
+
+      res = LLVMBuildAdd(bld->builder, a, offset, "");
+   }
+
+   res = LLVMBuildFPToSI(bld->builder, res, int_vec_type, "");
+
+   return res;
+}
+
+
+LLVMValueRef
+lp_build_iceil(struct lp_build_context *bld,
+               LLVMValueRef a)
+{
+   const struct lp_type type = bld->type;
+   LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
+   LLVMValueRef res;
+
+   assert(type.floating);
+   assert(lp_check_value(type, a));
+
+   if(util_cpu_caps.has_sse4_1) {
+      res = lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_CEIL);
+   }
+   else {
+      assert(0);
+      res = bld->undef;
+   }
+
+   res = LLVMBuildFPToSI(bld->builder, res, int_vec_type, "");
+
+   return res;
+}
+
+
+LLVMValueRef
 lp_build_sqrt(struct lp_build_context *bld,
               LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    char intrinsic[32];
 
@@ -609,7 +864,7 @@ LLVMValueRef
 lp_build_rcp(struct lp_build_context *bld,
              LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
 
    if(a == bld->zero)
       return bld->undef;
@@ -623,11 +878,9 @@ lp_build_rcp(struct lp_build_context *bld,
    if(LLVMIsConstant(a))
       return LLVMConstFDiv(bld->one, a);
 
-   /* XXX: is this really necessary? */
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-   if(type.width == 32 && type.length == 4)
+   if(util_cpu_caps.has_sse && type.width == 32 && type.length == 4)
+      /* FIXME: improve precision */
       return lp_build_intrinsic_unary(bld->builder, "llvm.x86.sse.rcp.ps", lp_build_vec_type(type), a);
-#endif
 
    return LLVMBuildFDiv(bld->builder, bld->one, a, "");
 }
@@ -640,15 +893,12 @@ LLVMValueRef
 lp_build_rsqrt(struct lp_build_context *bld,
                LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
 
    assert(type.floating);
 
-   /* XXX: is this really necessary? */
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-   if(type.width == 32 && type.length == 4)
+   if(util_cpu_caps.has_sse && type.width == 32 && type.length == 4)
       return lp_build_intrinsic_unary(bld->builder, "llvm.x86.sse.rsqrt.ps", lp_build_vec_type(type), a);
-#endif
 
    return lp_build_rcp(bld, lp_build_sqrt(bld, a));
 }
@@ -661,7 +911,7 @@ LLVMValueRef
 lp_build_cos(struct lp_build_context *bld,
               LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    char intrinsic[32];
 
@@ -681,7 +931,7 @@ LLVMValueRef
 lp_build_sin(struct lp_build_context *bld,
               LLVMValueRef a)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    char intrinsic[32];
 
@@ -704,7 +954,8 @@ lp_build_pow(struct lp_build_context *bld,
 {
    /* TODO: optimize the constant case */
    if(LLVMIsConstant(x) && LLVMIsConstant(y))
-      debug_printf("%s: inefficient/imprecise constant arithmetic\n");
+      debug_printf("%s: inefficient/imprecise constant arithmetic\n",
+                   __FUNCTION__);
 
    return lp_build_exp2(bld, lp_build_mul(bld, lp_build_log2(bld, x), y));
 }
@@ -752,13 +1003,14 @@ lp_build_polynomial(struct lp_build_context *bld,
                     const double *coeffs,
                     unsigned num_coeffs)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMValueRef res = NULL;
    unsigned i;
 
    /* TODO: optimize the constant case */
    if(LLVMIsConstant(x))
-      debug_printf("%s: inefficient/imprecise constant arithmetic\n");
+      debug_printf("%s: inefficient/imprecise constant arithmetic\n",
+                   __FUNCTION__);
 
    for (i = num_coeffs; i--; ) {
       LLVMValueRef coeff = lp_build_const_scalar(type, coeffs[i]);
@@ -800,7 +1052,7 @@ lp_build_exp2_approx(struct lp_build_context *bld,
                      LLVMValueRef *p_frac_part,
                      LLVMValueRef *p_exp2)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
    LLVMValueRef ipart = NULL;
@@ -812,7 +1064,8 @@ lp_build_exp2_approx(struct lp_build_context *bld,
    if(p_exp2_int_part || p_frac_part || p_exp2) {
       /* TODO: optimize the constant case */
       if(LLVMIsConstant(x))
-         debug_printf("%s: inefficient/imprecise constant arithmetic\n");
+         debug_printf("%s: inefficient/imprecise constant arithmetic\n",
+                      __FUNCTION__);
 
       assert(type.floating && type.width == 32);
 
@@ -893,7 +1146,7 @@ lp_build_log2_approx(struct lp_build_context *bld,
                      LLVMValueRef *p_floor_log2,
                      LLVMValueRef *p_log2)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
 
@@ -911,7 +1164,8 @@ lp_build_log2_approx(struct lp_build_context *bld,
    if(p_exp || p_floor_log2 || p_log2) {
       /* TODO: optimize the constant case */
       if(LLVMIsConstant(x))
-         debug_printf("%s: inefficient/imprecise constant arithmetic\n");
+         debug_printf("%s: inefficient/imprecise constant arithmetic\n",
+                      __FUNCTION__);
 
       assert(type.floating && type.width == 32);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.h b/src/gallium/drivers/llvmpipe/lp_bld_arit.h
index fc8cb25966..4e568c055e 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_arit.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.h
@@ -40,7 +40,7 @@
 #include <llvm-c/Core.h>  
 
 
-union lp_type type;
+struct lp_type;
 struct lp_build_context;
 
 
@@ -72,6 +72,26 @@ lp_build_div(struct lp_build_context *bld,
              LLVMValueRef b);
 
 LLVMValueRef
+lp_build_lerp(struct lp_build_context *bld,
+              LLVMValueRef x,
+              LLVMValueRef v0,
+              LLVMValueRef v1);
+
+/**
+ * Bilinear interpolation.
+ *
+ * Values indices are in v_{yx}.
+ */
+LLVMValueRef
+lp_build_lerp_2d(struct lp_build_context *bld,
+                 LLVMValueRef x,
+                 LLVMValueRef y,
+                 LLVMValueRef v00,
+                 LLVMValueRef v01,
+                 LLVMValueRef v10,
+                 LLVMValueRef v11);
+
+LLVMValueRef
 lp_build_min(struct lp_build_context *bld,
              LLVMValueRef a,
              LLVMValueRef b);
@@ -86,6 +106,41 @@ lp_build_abs(struct lp_build_context *bld,
              LLVMValueRef a);
 
 LLVMValueRef
+lp_build_sgn(struct lp_build_context *bld,
+             LLVMValueRef a);
+
+LLVMValueRef
+lp_build_round(struct lp_build_context *bld,
+               LLVMValueRef a);
+
+LLVMValueRef
+lp_build_floor(struct lp_build_context *bld,
+               LLVMValueRef a);
+
+LLVMValueRef
+lp_build_ceil(struct lp_build_context *bld,
+              LLVMValueRef a);
+
+LLVMValueRef
+lp_build_trunc(struct lp_build_context *bld,
+               LLVMValueRef a);
+
+LLVMValueRef
+lp_build_ifloor(struct lp_build_context *bld,
+                LLVMValueRef a);
+LLVMValueRef
+lp_build_iceil(struct lp_build_context *bld,
+               LLVMValueRef a);
+
+LLVMValueRef
+lp_build_iround(struct lp_build_context *bld,
+                LLVMValueRef a);
+
+LLVMValueRef
+lp_build_itrunc(struct lp_build_context *bld,
+                LLVMValueRef a);
+
+LLVMValueRef
 lp_build_sqrt(struct lp_build_context *bld,
               LLVMValueRef a);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend.h b/src/gallium/drivers/llvmpipe/lp_bld_blend.h
index d19e18846c..da272e549f 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_blend.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_blend.h
@@ -46,7 +46,7 @@
 
 
 struct pipe_blend_state;
-union lp_type;
+struct lp_type;
 struct lp_build_context;
 
 
@@ -74,7 +74,7 @@ lp_build_blend_func(struct lp_build_context *bld,
 LLVMValueRef
 lp_build_blend_aos(LLVMBuilderRef builder,
                    const struct pipe_blend_state *blend,
-                   union lp_type type,
+                   struct lp_type type,
                    LLVMValueRef src,
                    LLVMValueRef dst,
                    LLVMValueRef const_,
@@ -84,7 +84,7 @@ lp_build_blend_aos(LLVMBuilderRef builder,
 void
 lp_build_blend_soa(LLVMBuilderRef builder,
                    const struct pipe_blend_state *blend,
-                   union lp_type type,
+                   struct lp_type type,
                    LLVMValueRef src[4],
                    LLVMValueRef dst[4],
                    LLVMValueRef const_[4],
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c
index c11a9398f8..d14f468ba9 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c
@@ -303,7 +303,7 @@ lp_build_blend_func(struct lp_build_context *bld,
 LLVMValueRef
 lp_build_blend_aos(LLVMBuilderRef builder,
                    const struct pipe_blend_state *blend,
-                   union lp_type type,
+                   struct lp_type type,
                    LLVMValueRef src,
                    LLVMValueRef dst,
                    LLVMValueRef const_,
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
index b92254a7d6..9511299d55 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
@@ -199,7 +199,7 @@ lp_build_blend_soa_factor(struct lp_build_blend_soa_context *bld,
 void
 lp_build_blend_soa(LLVMBuilderRef builder,
                    const struct pipe_blend_state *blend,
-                   union lp_type type,
+                   struct lp_type type,
                    LLVMValueRef src[4],
                    LLVMValueRef dst[4],
                    LLVMValueRef con[4],
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_const.c b/src/gallium/drivers/llvmpipe/lp_bld_const.c
index 21487365ea..c8eaa8c394 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_const.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_const.c
@@ -42,7 +42,7 @@
 
 
 unsigned
-lp_mantissa(union lp_type type)
+lp_mantissa(struct lp_type type)
 {
    assert(type.floating);
 
@@ -72,7 +72,7 @@ lp_mantissa(union lp_type type)
  * Same as lp_const_scale(), but in terms of shifts.
  */
 unsigned
-lp_const_shift(union lp_type type)
+lp_const_shift(struct lp_type type)
 {
    if(type.floating)
       return 0;
@@ -86,7 +86,7 @@ lp_const_shift(union lp_type type)
 
 
 unsigned
-lp_const_offset(union lp_type type)
+lp_const_offset(struct lp_type type)
 {
    if(type.floating || type.fixed)
       return 0;
@@ -104,7 +104,7 @@ lp_const_offset(union lp_type type)
  * else for the fixed points types and normalized integers.
  */
 double
-lp_const_scale(union lp_type type)
+lp_const_scale(struct lp_type type)
 {
    unsigned long long llscale;
    double dscale;
@@ -122,7 +122,7 @@ lp_const_scale(union lp_type type)
  * Minimum value representable by the type.
  */
 double
-lp_const_min(union lp_type type)
+lp_const_min(struct lp_type type)
 {
    unsigned bits;
 
@@ -158,7 +158,7 @@ lp_const_min(union lp_type type)
  * Maximum value representable by the type.
  */
 double
-lp_const_max(union lp_type type)
+lp_const_max(struct lp_type type)
 {
    unsigned bits;
 
@@ -190,7 +190,7 @@ lp_const_max(union lp_type type)
 
 
 double
-lp_const_eps(union lp_type type)
+lp_const_eps(struct lp_type type)
 {
    if (type.floating) {
       switch(type.width) {
@@ -211,7 +211,7 @@ lp_const_eps(union lp_type type)
 
 
 LLVMValueRef
-lp_build_undef(union lp_type type)
+lp_build_undef(struct lp_type type)
 {
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    return LLVMGetUndef(vec_type);
@@ -219,7 +219,7 @@ lp_build_undef(union lp_type type)
                
 
 LLVMValueRef
-lp_build_zero(union lp_type type)
+lp_build_zero(struct lp_type type)
 {
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    return LLVMConstNull(vec_type);
@@ -227,7 +227,7 @@ lp_build_zero(union lp_type type)
                
 
 LLVMValueRef
-lp_build_one(union lp_type type)
+lp_build_one(struct lp_type type)
 {
    LLVMTypeRef elem_type;
    LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
@@ -269,7 +269,7 @@ lp_build_one(union lp_type type)
                
 
 LLVMValueRef
-lp_build_const_scalar(union lp_type type,
+lp_build_const_scalar(struct lp_type type,
                       double val)
 {
    LLVMTypeRef elem_type = lp_build_elem_type(type);
@@ -295,7 +295,7 @@ lp_build_const_scalar(union lp_type type,
 
 
 LLVMValueRef
-lp_build_int_const_scalar(union lp_type type,
+lp_build_int_const_scalar(struct lp_type type,
                           long long val)
 {
    LLVMTypeRef elem_type = lp_build_int_elem_type(type);
@@ -312,7 +312,7 @@ lp_build_int_const_scalar(union lp_type type,
 
 
 LLVMValueRef
-lp_build_const_aos(union lp_type type, 
+lp_build_const_aos(struct lp_type type, 
                    double r, double g, double b, double a, 
                    const unsigned char *swizzle)
 {
@@ -352,8 +352,8 @@ lp_build_const_aos(union lp_type type,
 
 
 LLVMValueRef
-lp_build_const_mask_aos(union lp_type type,
-                        boolean cond[4])
+lp_build_const_mask_aos(struct lp_type type,
+                        const boolean cond[4])
 {
    LLVMTypeRef elem_type = LLVMIntType(type.width);
    LLVMValueRef masks[LP_MAX_VECTOR_LENGTH];
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_const.h b/src/gallium/drivers/llvmpipe/lp_bld_const.h
index 1934530ea3..cb8e1c7b00 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_const.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_const.h
@@ -42,67 +42,67 @@
 #include <pipe/p_compiler.h>
 
 
-union lp_type type;
+struct lp_type;
 
 
 unsigned
-lp_mantissa(union lp_type type);
+lp_mantissa(struct lp_type type);
 
 
 unsigned
-lp_const_shift(union lp_type type);
+lp_const_shift(struct lp_type type);
 
 
 unsigned
-lp_const_offset(union lp_type type);
+lp_const_offset(struct lp_type type);
 
 
 double
-lp_const_scale(union lp_type type);
+lp_const_scale(struct lp_type type);
 
 double
-lp_const_min(union lp_type type);
+lp_const_min(struct lp_type type);
 
 
 double
-lp_const_max(union lp_type type);
+lp_const_max(struct lp_type type);
 
 
 double
-lp_const_eps(union lp_type type);
+lp_const_eps(struct lp_type type);
 
 
 LLVMValueRef
-lp_build_undef(union lp_type type);
+lp_build_undef(struct lp_type type);
 
 
 LLVMValueRef
-lp_build_zero(union lp_type type);
+lp_build_zero(struct lp_type type);
 
 
 LLVMValueRef
-lp_build_one(union lp_type type);
+lp_build_one(struct lp_type type);
 
 
 LLVMValueRef
-lp_build_const_scalar(union lp_type type,
+lp_build_const_scalar(struct lp_type type,
                       double val);
 
 
 LLVMValueRef
-lp_build_int_const_scalar(union lp_type type,
+lp_build_int_const_scalar(struct lp_type type,
                           long long val);
 
 
 LLVMValueRef
-lp_build_const_aos(union lp_type type, 
+lp_build_const_aos(struct lp_type type, 
                    double r, double g, double b, double a, 
                    const unsigned char *swizzle);
 
 
 LLVMValueRef
-lp_build_const_mask_aos(union lp_type type,
-                        boolean cond[4]);
+lp_build_const_mask_aos(struct lp_type type,
+                        const boolean cond[4]);
 
 
 #endif /* !LP_BLD_CONST_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_conv.c b/src/gallium/drivers/llvmpipe/lp_bld_conv.c
index c8954c8a34..9935209437 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_conv.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_conv.c
@@ -63,11 +63,13 @@
 
 #include "util/u_debug.h"
 #include "util/u_math.h"
+#include "util/u_cpu_detect.h"
 
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
 #include "lp_bld_intr.h"
 #include "lp_bld_arit.h"
+#include "lp_bld_pack.h"
 #include "lp_bld_conv.h"
 
 
@@ -86,7 +88,7 @@
  */
 LLVMValueRef
 lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder,
-                                        union lp_type src_type,
+                                        struct lp_type src_type,
                                         unsigned dst_width,
                                         LLVMValueRef src)
 {
@@ -122,7 +124,7 @@ lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder,
       int shift = dst_width - n;
       res = LLVMBuildShl(builder, res, lp_build_int_const_scalar(src_type, shift), "");
 
-      /* Fill in the empty lower bits for added precision? */
+      /* TODO: Fill in the empty lower bits for additional precision? */
 #if 0
       {
          LLVMValueRef msb;
@@ -152,7 +154,7 @@ lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder,
 LLVMValueRef
 lp_build_unsigned_norm_to_float(LLVMBuilderRef builder,
                                 unsigned src_width,
-                                union lp_type dst_type,
+                                struct lp_type dst_type,
                                 LLVMValueRef src)
 {
    LLVMTypeRef vec_type = lp_build_vec_type(dst_type);
@@ -198,243 +200,6 @@ lp_build_unsigned_norm_to_float(LLVMBuilderRef builder,
 
 
 /**
- * Build shuffle vectors that match PUNPCKLxx and PUNPCKHxx instructions.
- */
-static LLVMValueRef
-lp_build_const_unpack_shuffle(unsigned n, unsigned lo_hi)
-{
-   LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
-   unsigned i, j;
-
-   assert(n <= LP_MAX_VECTOR_LENGTH);
-   assert(lo_hi < 2);
-
-   /* TODO: cache results in a static table */
-
-   for(i = 0, j = lo_hi*n/2; i < n; i += 2, ++j) {
-      elems[i + 0] = LLVMConstInt(LLVMInt32Type(), 0 + j, 0);
-      elems[i + 1] = LLVMConstInt(LLVMInt32Type(), n + j, 0);
-   }
-
-   return LLVMConstVector(elems, n);
-}
-
-
-/**
- * Build shuffle vectors that match PACKxx instructions.
- */
-static LLVMValueRef
-lp_build_const_pack_shuffle(unsigned n)
-{
-   LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
-   unsigned i;
-
-   assert(n <= LP_MAX_VECTOR_LENGTH);
-
-   /* TODO: cache results in a static table */
-
-   for(i = 0; i < n; ++i)
-      elems[i] = LLVMConstInt(LLVMInt32Type(), 2*i, 0);
-
-   return LLVMConstVector(elems, n);
-}
-
-
-/**
- * Expand the bit width.
- *
- * This will only change the number of bits the values are represented, not the
- * values themselved.
- */
-static void
-lp_build_expand(LLVMBuilderRef builder,
-               union lp_type src_type,
-               union lp_type dst_type,
-               LLVMValueRef src,
-               LLVMValueRef *dst, unsigned num_dsts)
-{
-   unsigned num_tmps;
-   unsigned i;
-
-   /* Register width must remain constant */
-   assert(src_type.width * src_type.length == dst_type.width * dst_type.length);
-
-   /* We must not loose or gain channels. Only precision */
-   assert(src_type.length == dst_type.length * num_dsts);
-
-   num_tmps = 1;
-   dst[0] = src;
-
-   while(src_type.width < dst_type.width) {
-      union lp_type new_type = src_type;
-      LLVMTypeRef new_vec_type;
-
-      new_type.width *= 2;
-      new_type.length /= 2;
-      new_vec_type = lp_build_vec_type(new_type);
-
-      for(i = num_tmps; i--; ) {
-         LLVMValueRef zero;
-         LLVMValueRef shuffle_lo;
-         LLVMValueRef shuffle_hi;
-         LLVMValueRef lo;
-         LLVMValueRef hi;
-
-         zero = lp_build_zero(src_type);
-         shuffle_lo = lp_build_const_unpack_shuffle(src_type.length, 0);
-         shuffle_hi = lp_build_const_unpack_shuffle(src_type.length, 1);
-
-         /*  PUNPCKLBW, PUNPCKHBW */
-         lo = LLVMBuildShuffleVector(builder, dst[i], zero, shuffle_lo, "");
-         hi = LLVMBuildShuffleVector(builder, dst[i], zero, shuffle_hi, "");
-
-         dst[2*i + 0] = LLVMBuildBitCast(builder, lo, new_vec_type, "");
-         dst[2*i + 1] = LLVMBuildBitCast(builder, hi, new_vec_type, "");
-      }
-
-      src_type = new_type;
-
-      num_tmps *= 2;
-   }
-
-   assert(num_tmps == num_dsts);
-}
-
-
-/**
- * Non-interleaved pack.
- *
- * This will move values as
- *
- *   lo =   __ l0 __ l1 __ l2 __..  __ ln
- *   hi =   __ h0 __ h1 __ h2 __..  __ hn
- *   res =  l0 l1 l2 .. ln h0 h1 h2 .. hn
- *
- * TODO: handle saturation consistently.
- */
-static LLVMValueRef
-lp_build_pack2(LLVMBuilderRef builder,
-               union lp_type src_type,
-               union lp_type dst_type,
-               boolean clamped,
-               LLVMValueRef lo,
-               LLVMValueRef hi)
-{
-   LLVMTypeRef src_vec_type = lp_build_vec_type(src_type);
-   LLVMTypeRef dst_vec_type = lp_build_vec_type(dst_type);
-   LLVMValueRef shuffle;
-   LLVMValueRef res;
-
-   /* Register width must remain constant */
-   assert(src_type.width * src_type.length == dst_type.width * dst_type.length);
-
-   /* We must not loose or gain channels. Only precision */
-   assert(src_type.length * 2 == dst_type.length);
-
-   assert(!src_type.floating);
-   assert(!dst_type.floating);
-
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-   if(src_type.width * src_type.length == 128) {
-      /* All X86 non-interleaved pack instructions all take signed inputs and
-       * saturate them, so saturate beforehand. */
-      if(!src_type.sign && !clamped) {
-         struct lp_build_context bld;
-         unsigned dst_bits = dst_type.sign ? dst_type.width - 1 : dst_type.width;
-         LLVMValueRef dst_max = lp_build_int_const_scalar(src_type, ((unsigned long long)1 << dst_bits) - 1);
-         lp_build_context_init(&bld, builder, src_type);
-         lo = lp_build_min(&bld, lo, dst_max);
-         hi = lp_build_min(&bld, hi, dst_max);
-      }
-
-      switch(src_type.width) {
-      case 32:
-         if(dst_type.sign)
-            res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packssdw.128", src_vec_type, lo, hi);
-         else
-            /* PACKUSDW is the only instrinsic with a consistent signature */
-            return lp_build_intrinsic_binary(builder, "llvm.x86.sse41.packusdw", dst_vec_type, lo, hi);
-         break;
-
-      case 16:
-         if(dst_type.sign)
-            res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packsswb.128", src_vec_type, lo, hi);
-         else
-            res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packuswb.128", src_vec_type, lo, hi);
-         break;
-
-      default:
-         assert(0);
-         return LLVMGetUndef(dst_vec_type);
-         break;
-      }
-
-      res = LLVMBuildBitCast(builder, res, dst_vec_type, "");
-      return res;
-   }
-#endif
-
-   lo = LLVMBuildBitCast(builder, lo, dst_vec_type, "");
-   hi = LLVMBuildBitCast(builder, hi, dst_vec_type, "");
-
-   shuffle = lp_build_const_pack_shuffle(dst_type.length);
-
-   res = LLVMBuildShuffleVector(builder, lo, hi, shuffle, "");
-
-   return res;
-}
-
-
-/**
- * Truncate the bit width.
- *
- * TODO: Handle saturation consistently.
- */
-static LLVMValueRef
-lp_build_trunc(LLVMBuilderRef builder,
-               union lp_type src_type,
-               union lp_type dst_type,
-               boolean clamped,
-               const LLVMValueRef *src, unsigned num_srcs)
-{
-   LLVMValueRef tmp[LP_MAX_VECTOR_LENGTH];
-   unsigned i;
-
-   /* Register width must remain constant */
-   assert(src_type.width * src_type.length == dst_type.width * dst_type.length);
-
-   /* We must not loose or gain channels. Only precision */
-   assert(src_type.length * num_srcs == dst_type.length);
-
-   for(i = 0; i < num_srcs; ++i)
-      tmp[i] = src[i];
-
-   while(src_type.width > dst_type.width) {
-      union lp_type new_type = src_type;
-
-      new_type.width /= 2;
-      new_type.length *= 2;
-
-      /* Take in consideration the sign changes only in the last step */
-      if(new_type.width == dst_type.width)
-         new_type.sign = dst_type.sign;
-
-      num_srcs /= 2;
-
-      for(i = 0; i < num_srcs; ++i)
-         tmp[i] = lp_build_pack2(builder, src_type, new_type, clamped,
-                                 tmp[2*i + 0], tmp[2*i + 1]);
-
-      src_type = new_type;
-   }
-
-   assert(num_srcs == 1);
-
-   return tmp[0];
-}
-
-
-/**
  * Generic type conversion.
  *
  * TODO: Take a precision argument, or even better, add a new precision member
@@ -442,12 +207,12 @@ lp_build_trunc(LLVMBuilderRef builder,
  */
 void
 lp_build_conv(LLVMBuilderRef builder,
-              union lp_type src_type,
-              union lp_type dst_type,
+              struct lp_type src_type,
+              struct lp_type dst_type,
               const LLVMValueRef *src, unsigned num_srcs,
               LLVMValueRef *dst, unsigned num_dsts)
 {
-   union lp_type tmp_type;
+   struct lp_type tmp_type;
    LLVMValueRef tmp[LP_MAX_VECTOR_LENGTH];
    unsigned num_tmps;
    unsigned i;
@@ -470,7 +235,7 @@ lp_build_conv(LLVMBuilderRef builder,
     * Clamp if necessary
     */
 
-   if(src_type.value != dst_type.value) {
+   if(memcmp(&src_type, &dst_type, sizeof src_type) != 0) {
       struct lp_build_context bld;
       double src_min = lp_const_min(src_type);
       double dst_min = lp_const_min(dst_type);
@@ -565,7 +330,7 @@ lp_build_conv(LLVMBuilderRef builder,
 
    if(tmp_type.width > dst_type.width) {
       assert(num_dsts == 1);
-      tmp[0] = lp_build_trunc(builder, tmp_type, dst_type, TRUE, tmp, num_tmps);
+      tmp[0] = lp_build_pack(builder, tmp_type, dst_type, TRUE, tmp, num_tmps);
       tmp_type.width = dst_type.width;
       tmp_type.length = dst_type.length;
       num_tmps = 1;
@@ -573,7 +338,7 @@ lp_build_conv(LLVMBuilderRef builder,
 
    if(tmp_type.width < dst_type.width) {
       assert(num_tmps == 1);
-      lp_build_expand(builder, tmp_type, dst_type, tmp[0], tmp, num_dsts);
+      lp_build_unpack(builder, tmp_type, dst_type, tmp[0], tmp, num_dsts);
       tmp_type.width = dst_type.width;
       tmp_type.length = dst_type.length;
       num_tmps = num_dsts;
@@ -656,8 +421,8 @@ lp_build_conv(LLVMBuilderRef builder,
  */
 void
 lp_build_conv_mask(LLVMBuilderRef builder,
-                   union lp_type src_type,
-                   union lp_type dst_type,
+                   struct lp_type src_type,
+                   struct lp_type dst_type,
                    const LLVMValueRef *src, unsigned num_srcs,
                    LLVMValueRef *dst, unsigned num_dsts)
 {
@@ -689,11 +454,11 @@ lp_build_conv_mask(LLVMBuilderRef builder,
 
    if(src_type.width > dst_type.width) {
       assert(num_dsts == 1);
-      dst[0] = lp_build_trunc(builder, src_type, dst_type, TRUE, src, num_srcs);
+      dst[0] = lp_build_pack(builder, src_type, dst_type, TRUE, src, num_srcs);
    }
    else if(src_type.width < dst_type.width) {
       assert(num_srcs == 1);
-      lp_build_expand(builder, src_type, dst_type, src[0], dst, num_dsts);
+      lp_build_unpack(builder, src_type, dst_type, src[0], dst, num_dsts);
    }
    else {
       assert(num_srcs == num_dsts);
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_conv.h b/src/gallium/drivers/llvmpipe/lp_bld_conv.h
index 05c1ef2a10..948e68fae4 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_conv.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_conv.h
@@ -40,33 +40,33 @@
 #include <llvm-c/Core.h>  
 
 
-union lp_type type;
+struct lp_type;
 
 
 LLVMValueRef
 lp_build_clamped_float_to_unsigned_norm(LLVMBuilderRef builder,
-                                        union lp_type src_type,
+                                        struct lp_type src_type,
                                         unsigned dst_width,
                                         LLVMValueRef src);
 
 LLVMValueRef
 lp_build_unsigned_norm_to_float(LLVMBuilderRef builder,
                                 unsigned src_width,
-                                union lp_type dst_type,
+                                struct lp_type dst_type,
                                 LLVMValueRef src);
 
 
 void
 lp_build_conv(LLVMBuilderRef builder,
-              union lp_type src_type,
-              union lp_type dst_type,
+              struct lp_type src_type,
+              struct lp_type dst_type,
               const LLVMValueRef *srcs, unsigned num_srcs,
               LLVMValueRef *dsts, unsigned num_dsts);
 
 void
 lp_build_conv_mask(LLVMBuilderRef builder,
-                   union lp_type src_type,
-                   union lp_type dst_type,
+                   struct lp_type src_type,
+                   struct lp_type dst_type,
                    const LLVMValueRef *src, unsigned num_srcs,
                    LLVMValueRef *dst, unsigned num_dsts);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_debug.c b/src/gallium/drivers/llvmpipe/lp_bld_debug.c
index 30925b5f41..59d8f492e6 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_debug.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_debug.c
@@ -30,10 +30,27 @@
 #include <udis86.h>
 #endif
 
+#include "util/u_math.h"
 #include "util/u_debug.h"
 #include "lp_bld_debug.h"
 
 
+/**
+ * Check alignment.
+ *
+ * It is important that this check is not implemented as a macro or inlined
+ * function, as the compiler assumptions in respect to alignment of global
+ * and stack variables would often make the check a no op, defeating the
+ * whole purpose of the exercise.
+ */
+boolean
+lp_check_alignment(const void *ptr, unsigned alignment)
+{
+   assert(util_is_pot(alignment));
+   return ((uintptr_t)ptr & (alignment - 1)) == 0;
+}
+
+
 void
 lp_disassemble(const void* func)
 {
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_debug.h b/src/gallium/drivers/llvmpipe/lp_bld_debug.h
index ecdafef76d..583e6132b4 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_debug.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_debug.h
@@ -53,6 +53,10 @@ lp_build_name(LLVMValueRef val, const char *format, ...)
 }
 
 
+boolean
+lp_check_alignment(const void *ptr, unsigned alignment);
+
+
 void
 lp_disassemble(const void* func);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.c b/src/gallium/drivers/llvmpipe/lp_bld_depth.c
index 2cd6e6b921..98ec1cb1b9 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_depth.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.c
@@ -71,11 +71,11 @@
 /**
  * Return a type appropriate for depth/stencil testing.
  */
-union lp_type
+struct lp_type
 lp_depth_type(const struct util_format_description *format_desc,
               unsigned length)
 {
-   union lp_type type;
+   struct lp_type type;
    unsigned swizzle;
 
    assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS);
@@ -85,7 +85,7 @@ lp_depth_type(const struct util_format_description *format_desc,
    swizzle = format_desc->swizzle[0];
    assert(swizzle < 4);
 
-   type.value = 0;
+   memset(&type, 0, sizeof type);
    type.width = format_desc->block.bits;
 
    if(format_desc->channel[swizzle].type == UTIL_FORMAT_TYPE_FLOAT) {
@@ -114,7 +114,7 @@ lp_depth_type(const struct util_format_description *format_desc,
 void
 lp_build_depth_test(LLVMBuilderRef builder,
                     const struct pipe_depth_state *state,
-                    union lp_type type,
+                    struct lp_type type,
                     const struct util_format_description *format_desc,
                     struct lp_build_mask_context *mask,
                     LLVMValueRef src,
@@ -179,12 +179,13 @@ lp_build_depth_test(LLVMBuilderRef builder,
       padding_right = 0;
       for(chan = 0; chan < z_swizzle; ++chan)
          padding_right += format_desc->channel[chan].size;
-      padding_left = format_desc->block.bits - format_desc->channel[z_swizzle].size;
+      padding_left = format_desc->block.bits -
+                     (padding_right + format_desc->channel[z_swizzle].size);
 
       if(padding_left || padding_right) {
-         const long long mask_left = ((long long)1 << (format_desc->block.bits - padding_left)) - 1;
-         const long long mask_right = ((long long)1 << (padding_right)) - 1;
-         z_bitmask = lp_build_int_const_scalar(type, mask_left & mask_right);
+         const unsigned long long mask_left = ((unsigned long long)1 << (format_desc->block.bits - padding_left)) - 1;
+         const unsigned long long mask_right = ((unsigned long long)1 << (padding_right)) - 1;
+         z_bitmask = lp_build_int_const_scalar(type, mask_left ^ mask_right);
       }
 
       if(padding_left)
@@ -209,6 +210,4 @@ lp_build_depth_test(LLVMBuilderRef builder,
       dst = lp_build_select(&bld, z_bitmask, src, dst);
       LLVMBuildStore(builder, dst, dst_ptr);
    }
-
-   assert(!state->occlusion_count);
 }
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.h b/src/gallium/drivers/llvmpipe/lp_bld_depth.h
index 5d2e042fcc..79d6981bb5 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_depth.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.h
@@ -41,11 +41,11 @@
  
 struct pipe_depth_state;
 struct util_format_description;
-union lp_type;
+struct lp_type;
 struct lp_build_mask_context;
 
 
-union lp_type
+struct lp_type
 lp_depth_type(const struct util_format_description *format_desc,
               unsigned length);
 
@@ -53,7 +53,7 @@ lp_depth_type(const struct util_format_description *format_desc,
 void
 lp_build_depth_test(LLVMBuilderRef builder,
                     const struct pipe_depth_state *state,
-                    union lp_type type,
+                    struct lp_type type,
                     const struct util_format_description *format_desc,
                     struct lp_build_mask_context *mask,
                     LLVMValueRef src,
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_flow.c b/src/gallium/drivers/llvmpipe/lp_bld_flow.c
index 9d99e1a9d9..dcc25fbff8 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_flow.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_flow.c
@@ -32,59 +32,261 @@
  */
 
 #include "util/u_debug.h"
+#include "util/u_memory.h"
 
 #include "lp_bld_type.h"
 #include "lp_bld_flow.h"
 
 
+#define LP_BUILD_FLOW_MAX_VARIABLES 32
+#define LP_BUILD_FLOW_MAX_DEPTH 32
+
+
+/**
+ * Enumeration of all possible flow constructs.
+ */
+enum lp_build_flow_construct_kind {
+   lP_BUILD_FLOW_SCOPE,
+   LP_BUILD_FLOW_SKIP,
+};
+
+
+/**
+ * Variable declaration scope.
+ */
+struct lp_build_flow_scope
+{
+   /** Number of variables declared in this scope */
+   unsigned num_variables;
+};
+
+
+/**
+ * Early exit. Useful to skip to the end of a function or block when
+ * the execution mask becomes zero or when there is an error condition.
+ */
+struct lp_build_flow_skip
+{
+   /** Block to skip to */
+   LLVMBasicBlockRef block;
+
+   /** Number of variables declared at the beginning */
+   unsigned num_variables;
+
+   LLVMValueRef *phi;
+};
+
+
+/**
+ * Union of all possible flow constructs' data
+ */
+union lp_build_flow_construct_data
+{
+   struct lp_build_flow_scope scope;
+   struct lp_build_flow_skip skip;
+};
+
+
+/**
+ * Element of the flow construct stack.
+ */
+struct lp_build_flow_construct
+{
+   enum lp_build_flow_construct_kind kind;
+   union lp_build_flow_construct_data data;
+};
+
+
+/**
+ * All necessary data to generate LLVM control flow constructs.
+ *
+ * Besides keeping track of the control flow construct themselves we also
+ * need to keep track of variables in order to generate SSA Phi values.
+ */
+struct lp_build_flow_context
+{
+   LLVMBuilderRef builder;
+
+   /**
+    * Control flow stack.
+    */
+   struct lp_build_flow_construct constructs[LP_BUILD_FLOW_MAX_DEPTH];
+   unsigned num_constructs;
+
+   /**
+    * Variable stack
+    */
+   LLVMValueRef *variables[LP_BUILD_FLOW_MAX_VARIABLES];
+   unsigned num_variables;
+};
+
+
+struct lp_build_flow_context *
+lp_build_flow_create(LLVMBuilderRef builder)
+{
+   struct lp_build_flow_context *flow;
+
+   flow = CALLOC_STRUCT(lp_build_flow_context);
+   if(!flow)
+      return NULL;
+
+   flow->builder = builder;
+
+   return flow;
+}
+
+
 void
-lp_build_mask_begin(struct lp_build_mask_context *mask,
-                    LLVMBuilderRef builder,
-                    union lp_type type,
-                    LLVMValueRef value)
+lp_build_flow_destroy(struct lp_build_flow_context *flow)
 {
-   memset(mask, 0, sizeof *mask);
+   assert(flow->num_constructs == 0);
+   assert(flow->num_variables == 0);
+   FREE(flow);
+}
 
-   mask->builder = builder;
-   mask->reg_type = LLVMIntType(type.width * type.length);
-   mask->value = value;
+
+static union lp_build_flow_construct_data *
+lp_build_flow_push(struct lp_build_flow_context *flow,
+                   enum lp_build_flow_construct_kind kind)
+{
+   assert(flow->num_constructs < LP_BUILD_FLOW_MAX_DEPTH);
+   if(flow->num_constructs >= LP_BUILD_FLOW_MAX_DEPTH)
+      return NULL;
+
+   flow->constructs[flow->num_constructs].kind = kind;
+   return &flow->constructs[flow->num_constructs++].data;
+}
+
+
+static union lp_build_flow_construct_data *
+lp_build_flow_peek(struct lp_build_flow_context *flow,
+                   enum lp_build_flow_construct_kind kind)
+{
+   assert(flow->num_constructs);
+   if(!flow->num_constructs)
+      return NULL;
+
+   assert(flow->constructs[flow->num_constructs - 1].kind == kind);
+   if(flow->constructs[flow->num_constructs - 1].kind != kind)
+      return NULL;
+
+   return &flow->constructs[flow->num_constructs - 1].data;
 }
 
 
+static union lp_build_flow_construct_data *
+lp_build_flow_pop(struct lp_build_flow_context *flow,
+                  enum lp_build_flow_construct_kind kind)
+{
+   assert(flow->num_constructs);
+   if(!flow->num_constructs)
+      return NULL;
+
+   assert(flow->constructs[flow->num_constructs - 1].kind == kind);
+   if(flow->constructs[flow->num_constructs - 1].kind != kind)
+      return NULL;
+
+   return &flow->constructs[--flow->num_constructs].data;
+}
+
+
+/**
+ * Begin a variable scope.
+ *
+ *
+ */
 void
-lp_build_mask_update(struct lp_build_mask_context *mask,
-                     LLVMValueRef value)
+lp_build_flow_scope_begin(struct lp_build_flow_context *flow)
 {
+   struct lp_build_flow_scope *scope;
 
-   LLVMValueRef cond;
-   LLVMBasicBlockRef current_block;
-   LLVMBasicBlockRef next_block;
-   LLVMBasicBlockRef new_block;
+   scope = &lp_build_flow_push(flow, lP_BUILD_FLOW_SCOPE)->scope;
+   if(!scope)
+      return;
 
-   if(mask->value)
-      mask->value = LLVMBuildAnd(mask->builder, mask->value, value, "");
-   else
-      mask->value = value;
+   scope->num_variables = 0;
+}
 
-   /* FIXME: disabled until we have proper control flow helpers */
-#if 0
-   cond = LLVMBuildICmp(mask->builder,
-                        LLVMIntEQ,
-                        LLVMBuildBitCast(mask->builder, mask->value, mask->reg_type, ""),
-                        LLVMConstNull(mask->reg_type),
-                        "");
 
-   current_block = LLVMGetInsertBlock(mask->builder);
+/**
+ * Declare a variable.
+ *
+ * A variable is a named entity which can have different LLVMValueRef's at
+ * different points of the program. This is relevant for control flow because
+ * when there are mutiple branches to a same location we need to replace
+ * the variable's value with a Phi function as explained in
+ * http://en.wikipedia.org/wiki/Static_single_assignment_form .
+ *
+ * We keep track of variables by keeping around a pointer to where their
+ * current.
+ *
+ * There are a few cautions to observe:
+ *
+ * - Variable's value must not be NULL. If there is no initial value then
+ *   LLVMGetUndef() should be used.
+ *
+ * - Variable's value must be kept up-to-date. If the variable is going to be
+ *   modified by a function then a pointer should be passed so that its value
+ *   is accurate. Failure to do this will cause some of the variables'
+ *   transient values to be lost, leading to wrong results.
+ *
+ * - A program should be written from top to bottom, by always appending
+ *   instructions to the bottom with a single LLVMBuilderRef. Inserting and/or
+ *   modifying existing statements will most likely lead to wrong results.
+ *
+ */
+void
+lp_build_flow_scope_declare(struct lp_build_flow_context *flow,
+                            LLVMValueRef *variable)
+{
+   struct lp_build_flow_scope *scope;
+
+   scope = &lp_build_flow_peek(flow, lP_BUILD_FLOW_SCOPE)->scope;
+   if(!scope)
+      return;
 
-   if(!mask->skip_block) {
-      LLVMValueRef function = LLVMGetBasicBlockParent(current_block);
-      mask->skip_block = LLVMAppendBasicBlock(function, "skip");
+   assert(*variable);
+   if(!*variable)
+      return;
+
+   assert(flow->num_variables < LP_BUILD_FLOW_MAX_VARIABLES);
+   if(flow->num_variables >= LP_BUILD_FLOW_MAX_VARIABLES)
+      return;
+
+   flow->variables[flow->num_variables++] = variable;
+   ++scope->num_variables;
+}
+
+
+void
+lp_build_flow_scope_end(struct lp_build_flow_context *flow)
+{
+   struct lp_build_flow_scope *scope;
+
+   scope = &lp_build_flow_pop(flow, lP_BUILD_FLOW_SCOPE)->scope;
+   if(!scope)
+      return;
 
-      mask->phi = LLVMBuildPhi(mask->builder, LLVMTypeOf(mask->value), "");
+   assert(flow->num_variables >= scope->num_variables);
+   if(flow->num_variables < scope->num_variables) {
+      flow->num_variables = 0;
+      return;
    }
 
+   flow->num_variables -= scope->num_variables;
+}
+
+
+static LLVMBasicBlockRef
+lp_build_flow_insert_block(struct lp_build_flow_context *flow)
+{
+   LLVMBasicBlockRef current_block;
+   LLVMBasicBlockRef next_block;
+   LLVMBasicBlockRef new_block;
+
+   current_block = LLVMGetInsertBlock(flow->builder);
+
    next_block = LLVMGetNextBasicBlock(current_block);
-   assert(next_block);
    if(next_block) {
       new_block = LLVMInsertBasicBlock(next_block, "");
    }
@@ -93,30 +295,148 @@ lp_build_mask_update(struct lp_build_mask_context *mask,
       new_block = LLVMAppendBasicBlock(function, "");
    }
 
-   LLVMAddIncoming(mask->phi, &mask->value, &current_block, 1);
-   LLVMBuildCondBr(mask->builder, cond, mask->skip_block, new_block);
+   return new_block;
+}
+
+void
+lp_build_flow_skip_begin(struct lp_build_flow_context *flow)
+{
+   struct lp_build_flow_skip *skip;
+   LLVMBuilderRef builder;
+   unsigned i;
+
+   skip = &lp_build_flow_push(flow, LP_BUILD_FLOW_SKIP)->skip;
+   if(!skip)
+      return;
+
+   skip->block = lp_build_flow_insert_block(flow);
+   skip->num_variables = flow->num_variables;
+   if(!skip->num_variables) {
+      skip->phi = NULL;
+      return;
+   }
 
-   LLVMPositionBuilderAtEnd(mask->builder, new_block);
-#endif
+   skip->phi = MALLOC(skip->num_variables * sizeof *skip->phi);
+   if(!skip->phi) {
+      skip->num_variables = 0;
+      return;
+   }
+
+   builder = LLVMCreateBuilder();
+   LLVMPositionBuilderAtEnd(builder, skip->block);
+
+   for(i = 0; i < skip->num_variables; ++i)
+      skip->phi[i] = LLVMBuildPhi(builder, LLVMTypeOf(*flow->variables[i]), "");
+
+   LLVMDisposeBuilder(builder);
 }
 
 
-LLVMValueRef
-lp_build_mask_end(struct lp_build_mask_context *mask)
+void
+lp_build_flow_skip_cond_break(struct lp_build_flow_context *flow,
+                              LLVMValueRef cond)
+{
+   struct lp_build_flow_skip *skip;
+   LLVMBasicBlockRef current_block;
+   LLVMBasicBlockRef new_block;
+   unsigned i;
+
+   skip = &lp_build_flow_peek(flow, LP_BUILD_FLOW_SKIP)->skip;
+   if(!skip)
+      return;
+
+   current_block = LLVMGetInsertBlock(flow->builder);
+
+   new_block = lp_build_flow_insert_block(flow);
+
+   for(i = 0; i < skip->num_variables; ++i) {
+      assert(*flow->variables[i]);
+      LLVMAddIncoming(skip->phi[i], flow->variables[i], &current_block, 1);
+   }
+
+   LLVMBuildCondBr(flow->builder, cond, skip->block, new_block);
+
+   LLVMPositionBuilderAtEnd(flow->builder, new_block);
+ }
+
+
+void
+lp_build_flow_skip_end(struct lp_build_flow_context *flow)
 {
-   if(mask->skip_block) {
-      LLVMBasicBlockRef current_block = LLVMGetInsertBlock(mask->builder);
+   struct lp_build_flow_skip *skip;
+   LLVMBasicBlockRef current_block;
+   unsigned i;
 
-      LLVMAddIncoming(mask->phi, &mask->value, &current_block, 1);
-      LLVMBuildBr(mask->builder, mask->skip_block);
+   skip = &lp_build_flow_pop(flow, LP_BUILD_FLOW_SKIP)->skip;
+   if(!skip)
+      return;
 
-      LLVMPositionBuilderAtEnd(mask->builder, mask->skip_block);
+   current_block = LLVMGetInsertBlock(flow->builder);
 
-      mask->value = mask->phi;
-      mask->phi = NULL;
-      mask->skip_block = NULL;
+   for(i = 0; i < skip->num_variables; ++i) {
+      assert(*flow->variables[i]);
+      LLVMAddIncoming(skip->phi[i], flow->variables[i], &current_block, 1);
+      *flow->variables[i] = skip->phi[i];
    }
 
+   LLVMBuildBr(flow->builder, skip->block);
+   LLVMPositionBuilderAtEnd(flow->builder, skip->block);
+
+   FREE(skip->phi);
+}
+
+
+static void
+lp_build_mask_check(struct lp_build_mask_context *mask)
+{
+   LLVMBuilderRef builder = mask->flow->builder;
+   LLVMValueRef cond;
+
+   cond = LLVMBuildICmp(builder,
+                        LLVMIntEQ,
+                        LLVMBuildBitCast(builder, mask->value, mask->reg_type, ""),
+                        LLVMConstNull(mask->reg_type),
+                        "");
+
+   lp_build_flow_skip_cond_break(mask->flow, cond);
+}
+
+
+void
+lp_build_mask_begin(struct lp_build_mask_context *mask,
+                    struct lp_build_flow_context *flow,
+                    struct lp_type type,
+                    LLVMValueRef value)
+{
+   memset(mask, 0, sizeof *mask);
+
+   mask->flow = flow;
+   mask->reg_type = LLVMIntType(type.width * type.length);
+   mask->value = value;
+
+   lp_build_flow_scope_begin(flow);
+   lp_build_flow_scope_declare(flow, &mask->value);
+   lp_build_flow_skip_begin(flow);
+
+   lp_build_mask_check(mask);
+}
+
+
+void
+lp_build_mask_update(struct lp_build_mask_context *mask,
+                     LLVMValueRef value)
+{
+   mask->value = LLVMBuildAnd( mask->flow->builder, mask->value, value, "");
+
+   lp_build_mask_check(mask);
+}
+
+
+LLVMValueRef
+lp_build_mask_end(struct lp_build_mask_context *mask)
+{
+   lp_build_flow_skip_end(mask->flow);
+   lp_build_flow_scope_end(mask->flow);
    return mask->value;
 }
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_flow.h b/src/gallium/drivers/llvmpipe/lp_bld_flow.h
index 1b634ff038..e61999ff06 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_flow.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_flow.h
@@ -38,27 +38,53 @@
 #include <llvm-c/Core.h>  
 
 
-union lp_type;
+struct lp_type;
+
+
+struct lp_build_flow_context;
+
+
+struct lp_build_flow_context *
+lp_build_flow_create(LLVMBuilderRef builder);
+
+void
+lp_build_flow_destroy(struct lp_build_flow_context *flow);
+
+void
+lp_build_flow_scope_begin(struct lp_build_flow_context *flow);
+
+void
+lp_build_flow_scope_declare(struct lp_build_flow_context *flow,
+                            LLVMValueRef *variable);
+
+void
+lp_build_flow_scope_end(struct lp_build_flow_context *flow);
+
+void
+lp_build_flow_skip_begin(struct lp_build_flow_context *flow);
+
+void
+lp_build_flow_skip_cond_break(struct lp_build_flow_context *flow,
+                              LLVMValueRef cond);
+
+void
+lp_build_flow_skip_end(struct lp_build_flow_context *flow);
 
 
 struct lp_build_mask_context
 {
-   LLVMBuilderRef builder;
+   struct lp_build_flow_context *flow;
 
    LLVMTypeRef reg_type;
 
    LLVMValueRef value;
-
-   LLVMValueRef phi;
-
-   LLVMBasicBlockRef skip_block;
 };
 
 
 void
 lp_build_mask_begin(struct lp_build_mask_context *mask,
-                    LLVMBuilderRef builder,
-                    union lp_type type,
+                    struct lp_build_flow_context *flow,
+                    struct lp_type type,
                     LLVMValueRef value);
 
 /**
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_format.h b/src/gallium/drivers/llvmpipe/lp_bld_format.h
index 01c8a752d1..c087fc986e 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_format.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_format.h
@@ -25,27 +25,21 @@
  *
  **************************************************************************/
 
-#ifndef LP_BLD_H
-#define LP_BLD_H
+#ifndef LP_BLD_FORMAT_H
+#define LP_BLD_FORMAT_H
 
 
 /**
  * @file
- * LLVM IR building helpers interfaces.
- *
- * We use LLVM-C bindings for now. They are not documented, but follow the C++
- * interfaces very closely, and appear to be complete enough for code
- * genration. See
- * http://npcontemplation.blogspot.com/2008/06/secret-of-llvm-c-bindings.html
- * for a standalone example.
+ * Pixel format helpers.
  */
 
 #include <llvm-c/Core.h>  
- 
-#include "pipe/p_format.h"
 
+#include "pipe/p_format.h"
 
-union lp_type;
+struct util_format_description;
+struct lp_type;
 
 
 /**
@@ -56,9 +50,9 @@ union lp_type;
  * @return RGBA in a 4 floats vector.
  */
 LLVMValueRef
-lp_build_unpack_rgba(LLVMBuilderRef builder,
-                     enum pipe_format format, 
-                     LLVMValueRef packed);
+lp_build_unpack_rgba_aos(LLVMBuilderRef builder,
+                         enum pipe_format format,
+                         LLVMValueRef packed);
 
 
 /**
@@ -67,9 +61,9 @@ lp_build_unpack_rgba(LLVMBuilderRef builder,
  * @param rgba 4 float vector with the unpacked components.
  */
 LLVMValueRef
-lp_build_pack_rgba(LLVMBuilderRef builder,
-                   enum pipe_format format,
-                   LLVMValueRef rgba);
+lp_build_pack_rgba_aos(LLVMBuilderRef builder,
+                       enum pipe_format format,
+                       LLVMValueRef rgba);
 
 
 /**
@@ -81,9 +75,9 @@ lp_build_pack_rgba(LLVMBuilderRef builder,
  * @return RGBA in a 4 floats vector.
  */
 LLVMValueRef
-lp_build_load_rgba(LLVMBuilderRef builder,
-                   enum pipe_format format, 
-                   LLVMValueRef ptr);
+lp_build_load_rgba_aos(LLVMBuilderRef builder,
+                       enum pipe_format format,
+                       LLVMValueRef ptr);
 
 
 /**
@@ -92,10 +86,34 @@ lp_build_load_rgba(LLVMBuilderRef builder,
  * @param rgba 4 float vector with the unpacked components.
  */
 void 
-lp_build_store_rgba(LLVMBuilderRef builder,
-                    enum pipe_format format,
-                    LLVMValueRef ptr,
-                    LLVMValueRef rgba);
+lp_build_store_rgba_aos(LLVMBuilderRef builder,
+                        enum pipe_format format,
+                        LLVMValueRef ptr,
+                        LLVMValueRef rgba);
 
-
-#endif /* !LP_BLD_H */
+LLVMValueRef
+lp_build_gather(LLVMBuilderRef builder,
+                unsigned length,
+                unsigned src_width,
+                unsigned dst_width,
+                LLVMValueRef base_ptr,
+                LLVMValueRef offsets);
+
+
+void
+lp_build_unpack_rgba_soa(LLVMBuilderRef builder,
+                         const struct util_format_description *format_desc,
+                         struct lp_type type,
+                         LLVMValueRef packed,
+                         LLVMValueRef *rgba);
+
+
+void
+lp_build_load_rgba_soa(LLVMBuilderRef builder,
+                       const struct util_format_description *format_desc,
+                       struct lp_type type,
+                       LLVMValueRef base_ptr,
+                       LLVMValueRef offsets,
+                       LLVMValueRef *rgba);
+
+#endif /* !LP_BLD_FORMAT_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c b/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c
index dcbc0076c7..b9b5d84bed 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c
@@ -32,9 +32,9 @@
 
 
 LLVMValueRef
-lp_build_unpack_rgba(LLVMBuilderRef builder,
-                     enum pipe_format format,
-                     LLVMValueRef packed)
+lp_build_unpack_rgba_aos(LLVMBuilderRef builder,
+                         enum pipe_format format,
+                         LLVMValueRef packed)
 {
    const struct util_format_description *desc;
    LLVMTypeRef type;
@@ -152,9 +152,9 @@ lp_build_unpack_rgba(LLVMBuilderRef builder,
 
 
 LLVMValueRef
-lp_build_pack_rgba(LLVMBuilderRef builder,
-                   enum pipe_format format,
-                   LLVMValueRef rgba)
+lp_build_pack_rgba_aos(LLVMBuilderRef builder,
+                       enum pipe_format format,
+                       LLVMValueRef rgba)
 {
    const struct util_format_description *desc;
    LLVMTypeRef type;
@@ -250,9 +250,9 @@ lp_build_pack_rgba(LLVMBuilderRef builder,
 
 
 LLVMValueRef
-lp_build_load_rgba(LLVMBuilderRef builder,
-                   enum pipe_format format,
-                   LLVMValueRef ptr)
+lp_build_load_rgba_aos(LLVMBuilderRef builder,
+                       enum pipe_format format,
+                       LLVMValueRef ptr)
 {
    const struct util_format_description *desc;
    LLVMTypeRef type;
@@ -272,15 +272,15 @@ lp_build_load_rgba(LLVMBuilderRef builder,
 
    packed = LLVMBuildLoad(builder, ptr, "");
 
-   return lp_build_unpack_rgba(builder, format, packed);
+   return lp_build_unpack_rgba_aos(builder, format, packed);
 }
 
 
 void
-lp_build_store_rgba(LLVMBuilderRef builder,
-                    enum pipe_format format,
-                    LLVMValueRef ptr,
-                    LLVMValueRef rgba)
+lp_build_store_rgba_aos(LLVMBuilderRef builder,
+                        enum pipe_format format,
+                        LLVMValueRef ptr,
+                        LLVMValueRef rgba)
 {
    const struct util_format_description *desc;
    LLVMTypeRef type;
@@ -294,7 +294,7 @@ lp_build_store_rgba(LLVMBuilderRef builder,
 
    type = LLVMIntType(desc->block.bits);
 
-   packed = lp_build_pack_rgba(builder, format, rgba);
+   packed = lp_build_pack_rgba_aos(builder, format, rgba);
 
    ptr = LLVMBuildBitCast(builder, ptr, LLVMPointerType(type, 0), "");
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c
new file mode 100644
index 0000000000..66bebdcdec
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_bld_format_soa.c
@@ -0,0 +1,209 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+#include "util/u_format.h"
+
+#include "lp_bld_type.h"
+#include "lp_bld_const.h"
+#include "lp_bld_conv.h"
+#include "lp_bld_format.h"
+
+
+/**
+ * Gather elements from scatter positions in memory into a single vector.
+ *
+ * @param src_width src element width
+ * @param dst_width result element width (source will be expanded to fit)
+ * @param length length of the offsets,
+ * @param base_ptr base pointer, should be a i8 pointer type.
+ * @param offsets vector with offsets
+ */
+LLVMValueRef
+lp_build_gather(LLVMBuilderRef builder,
+                unsigned length,
+                unsigned src_width,
+                unsigned dst_width,
+                LLVMValueRef base_ptr,
+                LLVMValueRef offsets)
+{
+   LLVMTypeRef src_type = LLVMIntType(src_width);
+   LLVMTypeRef src_ptr_type = LLVMPointerType(src_type, 0);
+   LLVMTypeRef dst_elem_type = LLVMIntType(dst_width);
+   LLVMTypeRef dst_vec_type = LLVMVectorType(dst_elem_type, length);
+   LLVMValueRef res;
+   unsigned i;
+
+   res = LLVMGetUndef(dst_vec_type);
+   for(i = 0; i < length; ++i) {
+      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
+      LLVMValueRef elem_offset;
+      LLVMValueRef elem_ptr;
+      LLVMValueRef elem;
+
+      elem_offset = LLVMBuildExtractElement(builder, offsets, index, "");
+      elem_ptr = LLVMBuildGEP(builder, base_ptr, &elem_offset, 1, "");
+      elem_ptr = LLVMBuildBitCast(builder, elem_ptr, src_ptr_type, "");
+      elem = LLVMBuildLoad(builder, elem_ptr, "");
+
+      assert(src_width <= dst_width);
+      if(src_width > dst_width)
+         elem = LLVMBuildTrunc(builder, elem, dst_elem_type, "");
+      if(src_width < dst_width)
+         elem = LLVMBuildZExt(builder, elem, dst_elem_type, "");
+
+      res = LLVMBuildInsertElement(builder, res, elem, index, "");
+   }
+
+   return res;
+}
+
+
+static LLVMValueRef
+lp_build_format_swizzle(struct lp_type type,
+                        const LLVMValueRef *inputs,
+                        enum util_format_swizzle swizzle)
+{
+   switch (swizzle) {
+   case UTIL_FORMAT_SWIZZLE_X:
+   case UTIL_FORMAT_SWIZZLE_Y:
+   case UTIL_FORMAT_SWIZZLE_Z:
+   case UTIL_FORMAT_SWIZZLE_W:
+      return inputs[swizzle];
+   case UTIL_FORMAT_SWIZZLE_0:
+      return lp_build_zero(type);
+   case UTIL_FORMAT_SWIZZLE_1:
+      return lp_build_one(type);
+   case UTIL_FORMAT_SWIZZLE_NONE:
+      return lp_build_undef(type);
+   default:
+      assert(0);
+      return lp_build_undef(type);
+   }
+}
+
+
+void
+lp_build_unpack_rgba_soa(LLVMBuilderRef builder,
+                         const struct util_format_description *format_desc,
+                         struct lp_type type,
+                         LLVMValueRef packed,
+                         LLVMValueRef *rgba)
+{
+   LLVMValueRef inputs[4];
+   unsigned start;
+   unsigned chan;
+
+   /* FIXME: Support more formats */
+   assert(format_desc->layout == UTIL_FORMAT_LAYOUT_ARITH ||
+          (format_desc->layout == UTIL_FORMAT_LAYOUT_ARRAY &&
+           format_desc->block.bits == format_desc->channel[0].size));
+   assert(format_desc->block.width == 1);
+   assert(format_desc->block.height == 1);
+   assert(format_desc->block.bits <= 32);
+
+   /* Decode the input vector components */
+   start = 0;
+   for (chan = 0; chan < 4; ++chan) {
+      unsigned width = format_desc->channel[chan].size;
+      unsigned stop = start + width;
+      LLVMValueRef input;
+
+      input = packed;
+
+      switch(format_desc->channel[chan].type) {
+      case UTIL_FORMAT_TYPE_VOID:
+         input = NULL;
+         break;
+
+      case UTIL_FORMAT_TYPE_UNSIGNED:
+         if(type.floating) {
+            if(start)
+               input = LLVMBuildLShr(builder, input, lp_build_int_const_scalar(type, start), "");
+            if(stop < format_desc->block.bits) {
+               unsigned mask = ((unsigned long long)1 << width) - 1;
+               input = LLVMBuildAnd(builder, input, lp_build_int_const_scalar(type, mask), "");
+            }
+
+            if(format_desc->channel[chan].normalized)
+               input = lp_build_unsigned_norm_to_float(builder, width, type, input);
+            else
+               input = LLVMBuildFPToSI(builder, input, lp_build_vec_type(type), "");
+         }
+         else {
+            /* FIXME */
+            assert(0);
+            input = lp_build_undef(type);
+         }
+         break;
+
+      default:
+         /* fall through */
+         input = lp_build_undef(type);
+         break;
+      }
+
+      inputs[chan] = input;
+
+      start = stop;
+   }
+
+   if(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS) {
+      enum util_format_swizzle swizzle = format_desc->swizzle[0];
+      LLVMValueRef depth = lp_build_format_swizzle(type, inputs, swizzle);
+      rgba[2] = rgba[1] = rgba[0] = depth;
+      rgba[3] = lp_build_one(type);
+   }
+   else {
+      for (chan = 0; chan < 4; ++chan) {
+         enum util_format_swizzle swizzle = format_desc->swizzle[chan];
+         rgba[chan] = lp_build_format_swizzle(type, inputs, swizzle);
+      }
+   }
+}
+
+
+void
+lp_build_load_rgba_soa(LLVMBuilderRef builder,
+                       const struct util_format_description *format_desc,
+                       struct lp_type type,
+                       LLVMValueRef base_ptr,
+                       LLVMValueRef offsets,
+                       LLVMValueRef *rgba)
+{
+   LLVMValueRef packed;
+
+   assert(format_desc->block.width == 1);
+   assert(format_desc->block.height == 1);
+   assert(format_desc->block.bits <= type.width);
+
+   packed = lp_build_gather(builder,
+                            type.length, format_desc->block.bits, type.width,
+                            base_ptr, offsets);
+
+   lp_build_unpack_rgba_soa(builder, format_desc, type, packed, rgba);
+}
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_interp.c b/src/gallium/drivers/llvmpipe/lp_bld_interp.c
index cfe20a0d75..338dbca6d1 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_interp.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_interp.c
@@ -292,7 +292,7 @@ void
 lp_build_interp_soa_init(struct lp_build_interp_soa_context *bld,
                          const struct tgsi_token *tokens,
                          LLVMBuilderRef builder,
-                         union lp_type type,
+                         struct lp_type type,
                          LLVMValueRef a0_ptr,
                          LLVMValueRef dadx_ptr,
                          LLVMValueRef dady_ptr,
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_interp.h b/src/gallium/drivers/llvmpipe/lp_bld_interp.h
index 9194f6233a..9c57a10879 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_interp.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_interp.h
@@ -83,7 +83,7 @@ void
 lp_build_interp_soa_init(struct lp_build_interp_soa_context *bld,
                          const struct tgsi_token *tokens,
                          LLVMBuilderRef builder,
-                         union lp_type type,
+                         struct lp_type type,
                          LLVMValueRef a0_ptr,
                          LLVMValueRef dadx_ptr,
                          LLVMValueRef dady_ptr,
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_logic.c b/src/gallium/drivers/llvmpipe/lp_bld_logic.c
index 8631efd6c3..db22a8028a 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_logic.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_logic.c
@@ -33,6 +33,8 @@
  */
 
 
+#include "util/u_cpu_detect.h"
+
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
 #include "lp_bld_intr.h"
@@ -45,7 +47,7 @@ lp_build_cmp(struct lp_build_context *bld,
              LLVMValueRef a,
              LLVMValueRef b)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMTypeRef vec_type = lp_build_vec_type(type);
    LLVMTypeRef int_vec_type = lp_build_int_vec_type(type);
    LLVMValueRef zeros = LLVMConstNull(int_vec_type);
@@ -65,7 +67,7 @@ lp_build_cmp(struct lp_build_context *bld,
 
 #if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
    if(type.width * type.length == 128) {
-      if(type.floating) {
+      if(type.floating && util_cpu_caps.has_sse) {
          LLVMValueRef args[3];
          unsigned cc;
          boolean swap;
@@ -114,7 +116,7 @@ lp_build_cmp(struct lp_build_context *bld,
          res = LLVMBuildBitCast(bld->builder, res, int_vec_type, "");
          return res;
       }
-      else {
+      else if(util_cpu_caps.has_sse2) {
          static const struct {
             unsigned swap:1;
             unsigned eq:1;
@@ -301,7 +303,7 @@ lp_build_select(struct lp_build_context *bld,
                 LLVMValueRef a,
                 LLVMValueRef b)
 {
-   union lp_type type = bld->type;
+   struct lp_type type = bld->type;
    LLVMValueRef res;
 
    if(a == b)
@@ -313,8 +315,6 @@ lp_build_select(struct lp_build_context *bld,
       b = LLVMBuildBitCast(bld->builder, b, int_vec_type, "");
    }
 
-   /* TODO: On SSE4 we could do this with a single instruction -- PBLENDVB */
-
    a = LLVMBuildAnd(bld->builder, a, mask, "");
 
    /* This often gets translated to PANDN, but sometimes the NOT is
@@ -339,9 +339,9 @@ LLVMValueRef
 lp_build_select_aos(struct lp_build_context *bld,
                     LLVMValueRef a,
                     LLVMValueRef b,
-                    boolean cond[4])
+                    const boolean cond[4])
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    const unsigned n = type.length;
    unsigned i, j;
 
@@ -376,9 +376,9 @@ lp_build_select_aos(struct lp_build_context *bld,
 
       return LLVMBuildShuffleVector(bld->builder, a, b, LLVMConstVector(shuffles, n), "");
    }
+   else {
 #if 0
-   else if(0) {
-      /* FIXME: Unfortunately select of vectors do not work */
+      /* XXX: Unfortunately select of vectors do not work */
       /* Use a select */
       LLVMTypeRef elem_type = LLVMInt1Type();
       LLVMValueRef cond[LP_MAX_VECTOR_LENGTH];
@@ -388,10 +388,9 @@ lp_build_select_aos(struct lp_build_context *bld,
             cond[j + i] = LLVMConstInt(elem_type, cond[i] ? 1 : 0, 0);
 
       return LLVMBuildSelect(bld->builder, LLVMConstVector(cond, n), a, b, "");
-   }
-#endif
-   else {
+#else
       LLVMValueRef mask = lp_build_const_mask_aos(type, cond);
       return lp_build_select(bld, mask, a, b);
+#endif
    }
 }
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_logic.h b/src/gallium/drivers/llvmpipe/lp_bld_logic.h
index 29b9e1c45b..d67500ef70 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_logic.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_logic.h
@@ -42,7 +42,7 @@
 #include "pipe/p_defines.h" /* For PIPE_FUNC_xxx */
 
 
-union lp_type type;
+struct lp_type;
 struct lp_build_context;
 
 
@@ -66,7 +66,7 @@ LLVMValueRef
 lp_build_select_aos(struct lp_build_context *bld,
                     LLVMValueRef a,
                     LLVMValueRef b,
-                    boolean cond[4]);
+                    const boolean cond[4]);
 
 
 #endif /* !LP_BLD_LOGIC_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_pack.c b/src/gallium/drivers/llvmpipe/lp_bld_pack.c
new file mode 100644
index 0000000000..fe82fda039
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_bld_pack.c
@@ -0,0 +1,419 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+/**
+ * @file
+ * Helper functions for packing/unpacking.
+ *
+ * Pack/unpacking is necessary for conversion between types of different
+ * bit width.
+ *
+ * They are also commonly used when an computation needs higher
+ * precision for the intermediate values. For example, if one needs the
+ * function:
+ *
+ *   c = compute(a, b);
+ *
+ * to use more precision for intermediate results then one should implement it
+ * as:
+ *
+ *   LLVMValueRef
+ *   compute(LLVMBuilderRef builder struct lp_type type, LLVMValueRef a, LLVMValueRef b)
+ *   {
+ *      struct lp_type wide_type = lp_wider_type(type);
+ *      LLVMValueRef al, ah, bl, bh, cl, ch, c;
+ *
+ *      lp_build_unpack2(builder, type, wide_type, a, &al, &ah);
+ *      lp_build_unpack2(builder, type, wide_type, b, &bl, &bh);
+ *
+ *      cl = compute_half(al, bl);
+ *      ch = compute_half(ah, bh);
+ *
+ *      c = lp_build_pack2(bld->builder, wide_type, type, cl, ch);
+ *
+ *      return c;
+ *   }
+ *
+ * where compute_half() would do the computation for half the elements with
+ * twice the precision.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#include "util/u_debug.h"
+#include "util/u_math.h"
+#include "util/u_cpu_detect.h"
+
+#include "lp_bld_type.h"
+#include "lp_bld_const.h"
+#include "lp_bld_intr.h"
+#include "lp_bld_arit.h"
+#include "lp_bld_pack.h"
+
+
+/**
+ * Build shuffle vectors that match PUNPCKLxx and PUNPCKHxx instructions.
+ */
+static LLVMValueRef
+lp_build_const_unpack_shuffle(unsigned n, unsigned lo_hi)
+{
+   LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
+   unsigned i, j;
+
+   assert(n <= LP_MAX_VECTOR_LENGTH);
+   assert(lo_hi < 2);
+
+   /* TODO: cache results in a static table */
+
+   for(i = 0, j = lo_hi*n/2; i < n; i += 2, ++j) {
+      elems[i + 0] = LLVMConstInt(LLVMInt32Type(), 0 + j, 0);
+      elems[i + 1] = LLVMConstInt(LLVMInt32Type(), n + j, 0);
+   }
+
+   return LLVMConstVector(elems, n);
+}
+
+
+/**
+ * Build shuffle vectors that match PACKxx instructions.
+ */
+static LLVMValueRef
+lp_build_const_pack_shuffle(unsigned n)
+{
+   LLVMValueRef elems[LP_MAX_VECTOR_LENGTH];
+   unsigned i;
+
+   assert(n <= LP_MAX_VECTOR_LENGTH);
+
+   /* TODO: cache results in a static table */
+
+   for(i = 0; i < n; ++i)
+      elems[i] = LLVMConstInt(LLVMInt32Type(), 2*i, 0);
+
+   return LLVMConstVector(elems, n);
+}
+
+
+/**
+ * Interleave vector elements.
+ *
+ * Matches the PUNPCKLxx and PUNPCKHxx SSE instructions.
+ */
+LLVMValueRef
+lp_build_interleave2(LLVMBuilderRef builder,
+                     struct lp_type type,
+                     LLVMValueRef a,
+                     LLVMValueRef b,
+                     unsigned lo_hi)
+{
+   LLVMValueRef shuffle;
+
+   shuffle = lp_build_const_unpack_shuffle(type.length, lo_hi);
+
+   return LLVMBuildShuffleVector(builder, a, b, shuffle, "");
+}
+
+
+/**
+ * Double the bit width.
+ *
+ * This will only change the number of bits the values are represented, not the
+ * values themselves.
+ */
+void
+lp_build_unpack2(LLVMBuilderRef builder,
+                 struct lp_type src_type,
+                 struct lp_type dst_type,
+                 LLVMValueRef src,
+                 LLVMValueRef *dst_lo,
+                 LLVMValueRef *dst_hi)
+{
+   LLVMValueRef msb;
+   LLVMTypeRef dst_vec_type;
+
+   assert(!src_type.floating);
+   assert(!dst_type.floating);
+   assert(dst_type.sign == src_type.sign);
+   assert(dst_type.width == src_type.width * 2);
+   assert(dst_type.length * 2 == src_type.length);
+
+   if(src_type.sign) {
+      /* Replicate the sign bit in the most significant bits */
+      msb = LLVMBuildAShr(builder, src, lp_build_int_const_scalar(src_type, src_type.width - 1), "");
+   }
+   else
+      /* Most significant bits always zero */
+      msb = lp_build_zero(src_type);
+
+   /* Interleave bits */
+   if(util_cpu_caps.little_endian) {
+      *dst_lo = lp_build_interleave2(builder, src_type, src, msb, 0);
+      *dst_hi = lp_build_interleave2(builder, src_type, src, msb, 1);
+   }
+   else {
+      *dst_lo = lp_build_interleave2(builder, src_type, msb, src, 0);
+      *dst_hi = lp_build_interleave2(builder, src_type, msb, src, 1);
+   }
+
+   /* Cast the result into the new type (twice as wide) */
+
+   dst_vec_type = lp_build_vec_type(dst_type);
+
+   *dst_lo = LLVMBuildBitCast(builder, *dst_lo, dst_vec_type, "");
+   *dst_hi = LLVMBuildBitCast(builder, *dst_hi, dst_vec_type, "");
+}
+
+
+/**
+ * Expand the bit width.
+ *
+ * This will only change the number of bits the values are represented, not the
+ * values themselves.
+ */
+void
+lp_build_unpack(LLVMBuilderRef builder,
+                struct lp_type src_type,
+                struct lp_type dst_type,
+                LLVMValueRef src,
+                LLVMValueRef *dst, unsigned num_dsts)
+{
+   unsigned num_tmps;
+   unsigned i;
+
+   /* Register width must remain constant */
+   assert(src_type.width * src_type.length == dst_type.width * dst_type.length);
+
+   /* We must not loose or gain channels. Only precision */
+   assert(src_type.length == dst_type.length * num_dsts);
+
+   num_tmps = 1;
+   dst[0] = src;
+
+   while(src_type.width < dst_type.width) {
+      struct lp_type tmp_type = src_type;
+
+      tmp_type.width *= 2;
+      tmp_type.length /= 2;
+
+      for(i = num_tmps; i--; ) {
+         lp_build_unpack2(builder, src_type, tmp_type, dst[i], &dst[2*i + 0], &dst[2*i + 1]);
+      }
+
+      src_type = tmp_type;
+
+      num_tmps *= 2;
+   }
+
+   assert(num_tmps == num_dsts);
+}
+
+
+/**
+ * Non-interleaved pack.
+ *
+ * This will move values as
+ *
+ *   lo =   __ l0 __ l1 __ l2 __..  __ ln
+ *   hi =   __ h0 __ h1 __ h2 __..  __ hn
+ *   res =  l0 l1 l2 .. ln h0 h1 h2 .. hn
+ *
+ * This will only change the number of bits the values are represented, not the
+ * values themselves.
+ *
+ * It is assumed the values are already clamped into the destination type range.
+ * Values outside that range will produce undefined results. Use
+ * lp_build_packs2 instead.
+ */
+LLVMValueRef
+lp_build_pack2(LLVMBuilderRef builder,
+               struct lp_type src_type,
+               struct lp_type dst_type,
+               LLVMValueRef lo,
+               LLVMValueRef hi)
+{
+   LLVMTypeRef src_vec_type = lp_build_vec_type(src_type);
+   LLVMTypeRef dst_vec_type = lp_build_vec_type(dst_type);
+   LLVMValueRef shuffle;
+   LLVMValueRef res;
+
+   dst_vec_type = lp_build_vec_type(dst_type);
+
+   assert(!src_type.floating);
+   assert(!dst_type.floating);
+   assert(src_type.width == dst_type.width * 2);
+   assert(src_type.length * 2 == dst_type.length);
+
+   if(util_cpu_caps.has_sse2 && src_type.width * src_type.length == 128) {
+      switch(src_type.width) {
+      case 32:
+         if(dst_type.sign) {
+            res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packssdw.128", src_vec_type, lo, hi);
+         }
+         else {
+            if (util_cpu_caps.has_sse4_1) {
+               /* PACKUSDW is the only instrinsic with a consistent signature */
+               return lp_build_intrinsic_binary(builder, "llvm.x86.sse41.packusdw", dst_vec_type, lo, hi);
+            }
+            else {
+               assert(0);
+               return LLVMGetUndef(dst_vec_type);
+            }
+         }
+         break;
+
+      case 16:
+         if(dst_type.sign)
+            res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packsswb.128", src_vec_type, lo, hi);
+         else
+            res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packuswb.128", src_vec_type, lo, hi);
+         break;
+
+      default:
+         assert(0);
+         return LLVMGetUndef(dst_vec_type);
+         break;
+      }
+
+      res = LLVMBuildBitCast(builder, res, dst_vec_type, "");
+      return res;
+   }
+
+   lo = LLVMBuildBitCast(builder, lo, dst_vec_type, "");
+   hi = LLVMBuildBitCast(builder, hi, dst_vec_type, "");
+
+   shuffle = lp_build_const_pack_shuffle(dst_type.length);
+
+   res = LLVMBuildShuffleVector(builder, lo, hi, shuffle, "");
+
+   return res;
+}
+
+
+
+/**
+ * Non-interleaved pack and saturate.
+ *
+ * Same as lp_build_pack2 but will saturate values so that they fit into the
+ * destination type.
+ */
+LLVMValueRef
+lp_build_packs2(LLVMBuilderRef builder,
+                struct lp_type src_type,
+                struct lp_type dst_type,
+                LLVMValueRef lo,
+                LLVMValueRef hi)
+{
+   boolean clamp;
+
+   assert(!src_type.floating);
+   assert(!dst_type.floating);
+   assert(src_type.sign == dst_type.sign);
+   assert(src_type.width == dst_type.width * 2);
+   assert(src_type.length * 2 == dst_type.length);
+
+   clamp = TRUE;
+
+   /* All X86 SSE non-interleaved pack instructions take signed inputs and
+    * saturate them, so no need to clamp for those cases. */
+   if(util_cpu_caps.has_sse2 &&
+      src_type.width * src_type.length == 128 &&
+      src_type.sign)
+      clamp = FALSE;
+
+   if(clamp) {
+      struct lp_build_context bld;
+      unsigned dst_bits = dst_type.sign ? dst_type.width - 1 : dst_type.width;
+      LLVMValueRef dst_max = lp_build_int_const_scalar(src_type, ((unsigned long long)1 << dst_bits) - 1);
+      lp_build_context_init(&bld, builder, src_type);
+      lo = lp_build_min(&bld, lo, dst_max);
+      hi = lp_build_min(&bld, hi, dst_max);
+      /* FIXME: What about lower bound? */
+   }
+
+   return lp_build_pack2(builder, src_type, dst_type, lo, hi);
+}
+
+
+/**
+ * Truncate the bit width.
+ *
+ * TODO: Handle saturation consistently.
+ */
+LLVMValueRef
+lp_build_pack(LLVMBuilderRef builder,
+              struct lp_type src_type,
+              struct lp_type dst_type,
+              boolean clamped,
+              const LLVMValueRef *src, unsigned num_srcs)
+{
+   LLVMValueRef (*pack2)(LLVMBuilderRef builder,
+                         struct lp_type src_type,
+                         struct lp_type dst_type,
+                         LLVMValueRef lo,
+                         LLVMValueRef hi);
+   LLVMValueRef tmp[LP_MAX_VECTOR_LENGTH];
+   unsigned i;
+
+
+   /* Register width must remain constant */
+   assert(src_type.width * src_type.length == dst_type.width * dst_type.length);
+
+   /* We must not loose or gain channels. Only precision */
+   assert(src_type.length * num_srcs == dst_type.length);
+
+   if(clamped)
+      pack2 = &lp_build_pack2;
+   else
+      pack2 = &lp_build_packs2;
+
+   for(i = 0; i < num_srcs; ++i)
+      tmp[i] = src[i];
+
+   while(src_type.width > dst_type.width) {
+      struct lp_type tmp_type = src_type;
+
+      tmp_type.width /= 2;
+      tmp_type.length *= 2;
+
+      /* Take in consideration the sign changes only in the last step */
+      if(tmp_type.width == dst_type.width)
+         tmp_type.sign = dst_type.sign;
+
+      num_srcs /= 2;
+
+      for(i = 0; i < num_srcs; ++i)
+         tmp[i] = pack2(builder, src_type, tmp_type, tmp[2*i + 0], tmp[2*i + 1]);
+
+      src_type = tmp_type;
+   }
+
+   assert(num_srcs == 1);
+
+   return tmp[0];
+}
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_pack.h b/src/gallium/drivers/llvmpipe/lp_bld_pack.h
new file mode 100644
index 0000000000..fb2a34984a
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_bld_pack.h
@@ -0,0 +1,95 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Helper functions for packing/unpacking conversions.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#ifndef LP_BLD_PACK_H
+#define LP_BLD_PACK_H
+
+
+#include <llvm-c/Core.h>  
+
+
+struct lp_type;
+
+
+LLVMValueRef
+lp_build_interleave2(LLVMBuilderRef builder,
+                     struct lp_type type,
+                     LLVMValueRef a,
+                     LLVMValueRef b,
+                     unsigned lo_hi);
+
+
+void
+lp_build_unpack2(LLVMBuilderRef builder,
+                 struct lp_type src_type,
+                 struct lp_type dst_type,
+                 LLVMValueRef src,
+                 LLVMValueRef *dst_lo,
+                 LLVMValueRef *dst_hi);
+
+
+void
+lp_build_unpack(LLVMBuilderRef builder,
+                struct lp_type src_type,
+                struct lp_type dst_type,
+                LLVMValueRef src,
+                LLVMValueRef *dst, unsigned num_dsts);
+
+
+LLVMValueRef
+lp_build_packs2(LLVMBuilderRef builder,
+                struct lp_type src_type,
+                struct lp_type dst_type,
+                LLVMValueRef lo,
+                LLVMValueRef hi);
+
+
+LLVMValueRef
+lp_build_pack2(LLVMBuilderRef builder,
+               struct lp_type src_type,
+               struct lp_type dst_type,
+               LLVMValueRef lo,
+               LLVMValueRef hi);
+
+
+LLVMValueRef
+lp_build_pack(LLVMBuilderRef builder,
+              struct lp_type src_type,
+              struct lp_type dst_type,
+              boolean clamped,
+              const LLVMValueRef *src, unsigned num_srcs);
+
+
+#endif /* !LP_BLD_PACK_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_sample.h b/src/gallium/drivers/llvmpipe/lp_bld_sample.h
new file mode 100644
index 0000000000..403d0e4836
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_bld_sample.h
@@ -0,0 +1,135 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Texture sampling.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#ifndef LP_BLD_SAMPLE_H
+#define LP_BLD_SAMPLE_H
+
+
+#include <llvm-c/Core.h>
+
+struct pipe_texture;
+struct pipe_sampler_state;
+struct lp_type;
+
+
+/**
+ * Sampler static state.
+ *
+ * These are the bits of state from pipe_texture and pipe_sampler_state that
+ * are embedded in the generated code.
+ */
+struct lp_sampler_static_state
+{
+   /* pipe_texture's state */
+   enum pipe_format format;
+   unsigned target:2;
+   unsigned pot_width:1;
+   unsigned pot_height:1;
+   unsigned pot_depth:1;
+
+   /* pipe_sampler_state's state */
+   unsigned wrap_s:3;
+   unsigned wrap_t:3;
+   unsigned wrap_r:3;
+   unsigned min_img_filter:2;
+   unsigned min_mip_filter:2;
+   unsigned mag_img_filter:2;
+   unsigned compare_mode:1;
+   unsigned compare_func:3;
+   unsigned normalized_coords:1;
+   unsigned prefilter:4;
+};
+
+
+/**
+ * Sampler dynamic state.
+ *
+ * These are the bits of state from pipe_texture and pipe_sampler_state that
+ * are computed in runtime.
+ *
+ * There are obtained through callbacks, as we don't want to tie the texture
+ * sampling code generation logic to any particular texture layout or pipe
+ * driver.
+ */
+struct lp_sampler_dynamic_state
+{
+
+   /** Obtain the base texture width. */
+   LLVMValueRef
+   (*width)( struct lp_sampler_dynamic_state *state,
+             LLVMBuilderRef builder,
+             unsigned unit);
+
+   /** Obtain the base texture height. */
+   LLVMValueRef
+   (*height)( struct lp_sampler_dynamic_state *state,
+              LLVMBuilderRef builder,
+              unsigned unit);
+
+   LLVMValueRef
+   (*stride)( struct lp_sampler_dynamic_state *state,
+              LLVMBuilderRef builder,
+              unsigned unit);
+
+   LLVMValueRef
+   (*data_ptr)( struct lp_sampler_dynamic_state *state,
+                LLVMBuilderRef builder,
+                unsigned unit);
+
+};
+
+
+/**
+ * Derive the sampler static state.
+ */
+void
+lp_sampler_static_state(struct lp_sampler_static_state *state,
+                        const struct pipe_texture *texture,
+                        const struct pipe_sampler_state *sampler);
+
+
+void
+lp_build_sample_soa(LLVMBuilderRef builder,
+                    const struct lp_sampler_static_state *static_state,
+                    struct lp_sampler_dynamic_state *dynamic_state,
+                    struct lp_type fp_type,
+                    unsigned unit,
+                    unsigned num_coords,
+                    const LLVMValueRef *coords,
+                    LLVMValueRef lodbias,
+                    LLVMValueRef *texel);
+
+
+
+#endif /* LP_BLD_SAMPLE_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c
new file mode 100644
index 0000000000..1a47ca32d2
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c
@@ -0,0 +1,416 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Texture sampling.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "util/u_debug.h"
+#include "util/u_memory.h"
+#include "util/u_math.h"
+#include "util/u_format.h"
+#include "lp_bld_debug.h"
+#include "lp_bld_type.h"
+#include "lp_bld_const.h"
+#include "lp_bld_arit.h"
+#include "lp_bld_logic.h"
+#include "lp_bld_swizzle.h"
+#include "lp_bld_format.h"
+#include "lp_bld_sample.h"
+
+
+void
+lp_sampler_static_state(struct lp_sampler_static_state *state,
+                        const struct pipe_texture *texture,
+                        const struct pipe_sampler_state *sampler)
+{
+   memset(state, 0, sizeof *state);
+
+   if(!texture)
+      return;
+
+   if(!sampler)
+      return;
+
+   state->format            = texture->format;
+   state->target            = texture->target;
+   state->pot_width         = util_is_pot(texture->width[0]);
+   state->pot_height        = util_is_pot(texture->height[0]);
+   state->pot_depth         = util_is_pot(texture->depth[0]);
+
+   state->wrap_s            = sampler->wrap_s;
+   state->wrap_t            = sampler->wrap_t;
+   state->wrap_r            = sampler->wrap_r;
+   state->min_img_filter    = sampler->min_img_filter;
+   state->min_mip_filter    = sampler->min_mip_filter;
+   state->mag_img_filter    = sampler->mag_img_filter;
+   if(sampler->compare_mode) {
+      state->compare_mode      = sampler->compare_mode;
+      state->compare_func      = sampler->compare_func;
+   }
+   state->normalized_coords = sampler->normalized_coords;
+   state->prefilter         = sampler->prefilter;
+}
+
+
+
+/**
+ * Keep all information for sampling code generation in a single place.
+ */
+struct lp_build_sample_context
+{
+   LLVMBuilderRef builder;
+
+   const struct lp_sampler_static_state *static_state;
+
+   struct lp_sampler_dynamic_state *dynamic_state;
+
+   const struct util_format_description *format_desc;
+
+   /** Incoming coordinates type and build context */
+   struct lp_type coord_type;
+   struct lp_build_context coord_bld;
+
+   /** Integer coordinates */
+   struct lp_type int_coord_type;
+   struct lp_build_context int_coord_bld;
+
+   /** Output texels type and build context */
+   struct lp_type texel_type;
+   struct lp_build_context texel_bld;
+};
+
+
+static void
+lp_build_sample_texel(struct lp_build_sample_context *bld,
+                      LLVMValueRef x,
+                      LLVMValueRef y,
+                      LLVMValueRef y_stride,
+                      LLVMValueRef data_ptr,
+                      LLVMValueRef *texel)
+{
+   struct lp_build_context *int_coord_bld = &bld->int_coord_bld;
+   LLVMValueRef x_stride;
+   LLVMValueRef offset;
+
+   x_stride = lp_build_const_scalar(bld->int_coord_type, bld->format_desc->block.bits/8);
+
+   if(bld->format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS) {
+      LLVMValueRef x_lo, x_hi;
+      LLVMValueRef y_lo, y_hi;
+      LLVMValueRef x_stride_lo, x_stride_hi;
+      LLVMValueRef y_stride_lo, y_stride_hi;
+      LLVMValueRef x_offset_lo, x_offset_hi;
+      LLVMValueRef y_offset_lo, y_offset_hi;
+      LLVMValueRef offset_lo, offset_hi;
+
+      x_lo = LLVMBuildAnd(bld->builder, x, int_coord_bld->one, "");
+      y_lo = LLVMBuildAnd(bld->builder, y, int_coord_bld->one, "");
+
+      x_hi = LLVMBuildLShr(bld->builder, x, int_coord_bld->one, "");
+      y_hi = LLVMBuildLShr(bld->builder, y, int_coord_bld->one, "");
+
+      x_stride_lo = x_stride;
+      y_stride_lo = lp_build_const_scalar(bld->int_coord_type, 2*bld->format_desc->block.bits/8);
+
+      x_stride_hi = lp_build_const_scalar(bld->int_coord_type, 4*bld->format_desc->block.bits/8);
+      y_stride_hi = LLVMBuildShl(bld->builder, y_stride, int_coord_bld->one, "");
+
+      x_offset_lo = lp_build_mul(int_coord_bld, x_lo, x_stride_lo);
+      y_offset_lo = lp_build_mul(int_coord_bld, y_lo, y_stride_lo);
+      offset_lo = lp_build_add(int_coord_bld, x_offset_lo, y_offset_lo);
+
+      x_offset_hi = lp_build_mul(int_coord_bld, x_hi, x_stride_hi);
+      y_offset_hi = lp_build_mul(int_coord_bld, y_hi, y_stride_hi);
+      offset_hi = lp_build_add(int_coord_bld, x_offset_hi, y_offset_hi);
+
+      offset = lp_build_add(int_coord_bld, offset_hi, offset_lo);
+   }
+   else {
+      LLVMValueRef x_offset;
+      LLVMValueRef y_offset;
+
+      x_offset = lp_build_mul(int_coord_bld, x, x_stride);
+      y_offset = lp_build_mul(int_coord_bld, y, y_stride);
+
+      offset = lp_build_add(int_coord_bld, x_offset, y_offset);
+   }
+
+   lp_build_load_rgba_soa(bld->builder,
+                          bld->format_desc,
+                          bld->texel_type,
+                          data_ptr,
+                          offset,
+                          texel);
+}
+
+
+static LLVMValueRef
+lp_build_sample_wrap(struct lp_build_sample_context *bld,
+                     LLVMValueRef coord,
+                     LLVMValueRef length,
+                     boolean is_pot,
+                     unsigned wrap_mode)
+{
+   struct lp_build_context *int_coord_bld = &bld->int_coord_bld;
+   LLVMValueRef length_minus_one;
+
+   length_minus_one = lp_build_sub(int_coord_bld, length, int_coord_bld->one);
+
+   switch(wrap_mode) {
+   case PIPE_TEX_WRAP_REPEAT:
+      if(is_pot)
+         coord = LLVMBuildAnd(bld->builder, coord, length_minus_one, "");
+      else
+         /* Signed remainder won't give the right results for negative
+          * dividends but unsigned remainder does.*/
+         coord = LLVMBuildURem(bld->builder, coord, length, "");
+      break;
+
+   case PIPE_TEX_WRAP_CLAMP:
+      coord = lp_build_max(int_coord_bld, coord, int_coord_bld->zero);
+      coord = lp_build_min(int_coord_bld, coord, length_minus_one);
+      break;
+
+   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+   case PIPE_TEX_WRAP_MIRROR_REPEAT:
+   case PIPE_TEX_WRAP_MIRROR_CLAMP:
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+      /* FIXME */
+      _debug_printf("warning: failed to translate texture wrap mode %u\n", wrap_mode);
+      coord = lp_build_max(int_coord_bld, coord, int_coord_bld->zero);
+      coord = lp_build_min(int_coord_bld, coord, length_minus_one);
+      break;
+
+   default:
+      assert(0);
+   }
+
+   return coord;
+}
+
+
+static void
+lp_build_sample_2d_nearest_soa(struct lp_build_sample_context *bld,
+                               LLVMValueRef s,
+                               LLVMValueRef t,
+                               LLVMValueRef width,
+                               LLVMValueRef height,
+                               LLVMValueRef stride,
+                               LLVMValueRef data_ptr,
+                               LLVMValueRef *texel)
+{
+   LLVMValueRef x;
+   LLVMValueRef y;
+
+   x = lp_build_ifloor(&bld->coord_bld, s);
+   y = lp_build_ifloor(&bld->coord_bld, t);
+
+   x = lp_build_sample_wrap(bld, x, width,  bld->static_state->pot_width,  bld->static_state->wrap_s);
+   y = lp_build_sample_wrap(bld, y, height, bld->static_state->pot_height, bld->static_state->wrap_t);
+
+   lp_build_sample_texel(bld, x, y, stride, data_ptr, texel);
+}
+
+
+static void
+lp_build_sample_2d_linear_soa(struct lp_build_sample_context *bld,
+                              LLVMValueRef s,
+                              LLVMValueRef t,
+                              LLVMValueRef width,
+                              LLVMValueRef height,
+                              LLVMValueRef stride,
+                              LLVMValueRef data_ptr,
+                              LLVMValueRef *texel)
+{
+   LLVMValueRef half;
+   LLVMValueRef s_ipart;
+   LLVMValueRef t_ipart;
+   LLVMValueRef s_fpart;
+   LLVMValueRef t_fpart;
+   LLVMValueRef x0, x1;
+   LLVMValueRef y0, y1;
+   LLVMValueRef neighbors[2][2][4];
+   unsigned chan;
+
+   half = lp_build_const_scalar(bld->coord_type, 0.5);
+   s = lp_build_sub(&bld->coord_bld, s, half);
+   t = lp_build_sub(&bld->coord_bld, t, half);
+
+   s_ipart = lp_build_floor(&bld->coord_bld, s);
+   t_ipart = lp_build_floor(&bld->coord_bld, t);
+
+   s_fpart = lp_build_sub(&bld->coord_bld, s, s_ipart);
+   t_fpart = lp_build_sub(&bld->coord_bld, t, t_ipart);
+
+   x0 = lp_build_itrunc(&bld->coord_bld, s_ipart);
+   y0 = lp_build_itrunc(&bld->coord_bld, t_ipart);
+
+   x0 = lp_build_sample_wrap(bld, x0, width,  bld->static_state->pot_width,  bld->static_state->wrap_s);
+   y0 = lp_build_sample_wrap(bld, y0, height, bld->static_state->pot_height, bld->static_state->wrap_t);
+
+   x1 = lp_build_add(&bld->int_coord_bld, x0, bld->int_coord_bld.one);
+   y1 = lp_build_add(&bld->int_coord_bld, y0, bld->int_coord_bld.one);
+
+   x1 = lp_build_sample_wrap(bld, x1, width,  bld->static_state->pot_width,  bld->static_state->wrap_s);
+   y1 = lp_build_sample_wrap(bld, y1, height, bld->static_state->pot_height, bld->static_state->wrap_t);
+
+   lp_build_sample_texel(bld, x0, y0, stride, data_ptr, neighbors[0][0]);
+   lp_build_sample_texel(bld, x1, y0, stride, data_ptr, neighbors[0][1]);
+   lp_build_sample_texel(bld, x0, y1, stride, data_ptr, neighbors[1][0]);
+   lp_build_sample_texel(bld, x1, y1, stride, data_ptr, neighbors[1][1]);
+
+   /* TODO: Don't interpolate missing channels */
+   for(chan = 0; chan < 4; ++chan) {
+      texel[chan] = lp_build_lerp_2d(&bld->texel_bld,
+                                     s_fpart, t_fpart,
+                                     neighbors[0][0][chan],
+                                     neighbors[0][1][chan],
+                                     neighbors[1][0][chan],
+                                     neighbors[1][1][chan]);
+   }
+}
+
+
+static void
+lp_build_sample_compare(struct lp_build_sample_context *bld,
+                        LLVMValueRef p,
+                        LLVMValueRef *texel)
+{
+   struct lp_build_context *texel_bld = &bld->texel_bld;
+   LLVMValueRef res;
+   unsigned chan;
+
+   if(!bld->static_state->compare_mode)
+      return;
+
+   /* TODO: Compare before swizzling, to avoid redundant computations */
+   res = NULL;
+   for(chan = 0; chan < 4; ++chan) {
+      LLVMValueRef cmp;
+      cmp = lp_build_cmp(texel_bld, bld->static_state->compare_func, p, texel[chan]);
+      cmp = lp_build_select(texel_bld, cmp, texel_bld->one, texel_bld->zero);
+
+      if(res)
+         res = lp_build_add(texel_bld, res, cmp);
+      else
+         res = cmp;
+   }
+
+   assert(res);
+   res = lp_build_mul(texel_bld, res, lp_build_const_scalar(texel_bld->type, 0.25));
+
+   /* XXX returning result for default GL_DEPTH_TEXTURE_MODE = GL_LUMINANCE */
+   for(chan = 0; chan < 3; ++chan)
+      texel[chan] = res;
+   texel[3] = texel_bld->one;
+}
+
+
+void
+lp_build_sample_soa(LLVMBuilderRef builder,
+                    const struct lp_sampler_static_state *static_state,
+                    struct lp_sampler_dynamic_state *dynamic_state,
+                    struct lp_type type,
+                    unsigned unit,
+                    unsigned num_coords,
+                    const LLVMValueRef *coords,
+                    LLVMValueRef lodbias,
+                    LLVMValueRef *texel)
+{
+   struct lp_build_sample_context bld;
+   LLVMValueRef width;
+   LLVMValueRef height;
+   LLVMValueRef stride;
+   LLVMValueRef data_ptr;
+   LLVMValueRef s;
+   LLVMValueRef t;
+   LLVMValueRef p;
+
+   /* Setup our build context */
+   memset(&bld, 0, sizeof bld);
+   bld.builder = builder;
+   bld.static_state = static_state;
+   bld.dynamic_state = dynamic_state;
+   bld.format_desc = util_format_description(static_state->format);
+   bld.coord_type = type;
+   bld.int_coord_type = lp_int_type(type);
+   bld.texel_type = type;
+   lp_build_context_init(&bld.coord_bld, builder, bld.coord_type);
+   lp_build_context_init(&bld.int_coord_bld, builder, bld.int_coord_type);
+   lp_build_context_init(&bld.texel_bld, builder, bld.texel_type);
+
+   /* Get the dynamic state */
+   width = dynamic_state->width(dynamic_state, builder, unit);
+   height = dynamic_state->height(dynamic_state, builder, unit);
+   stride = dynamic_state->stride(dynamic_state, builder, unit);
+   data_ptr = dynamic_state->data_ptr(dynamic_state, builder, unit);
+
+   s = coords[0];
+   t = coords[1];
+   p = coords[2];
+
+   width = lp_build_broadcast_scalar(&bld.int_coord_bld, width);
+   height = lp_build_broadcast_scalar(&bld.int_coord_bld, height);
+   stride = lp_build_broadcast_scalar(&bld.int_coord_bld, stride);
+
+   if(static_state->target == PIPE_TEXTURE_1D)
+      t = bld.coord_bld.zero;
+
+   if(static_state->normalized_coords) {
+      LLVMTypeRef coord_vec_type = lp_build_vec_type(bld.coord_type);
+      LLVMValueRef fp_width = LLVMBuildSIToFP(builder, width, coord_vec_type, "");
+      LLVMValueRef fp_height = LLVMBuildSIToFP(builder, height, coord_vec_type, "");
+      s = lp_build_mul(&bld.coord_bld, s, fp_width);
+      t = lp_build_mul(&bld.coord_bld, t, fp_height);
+   }
+
+   switch (static_state->min_img_filter) {
+   case PIPE_TEX_FILTER_NEAREST:
+      lp_build_sample_2d_nearest_soa(&bld, s, t, width, height, stride, data_ptr, texel);
+      break;
+   case PIPE_TEX_FILTER_LINEAR:
+   case PIPE_TEX_FILTER_ANISO:
+      lp_build_sample_2d_linear_soa(&bld, s, t, width, height, stride, data_ptr, texel);
+      break;
+   default:
+      assert(0);
+   }
+
+   /* FIXME: respect static_state->min_mip_filter */;
+   /* FIXME: respect static_state->mag_img_filter */;
+   /* FIXME: respect static_state->prefilter */;
+
+   lp_build_sample_compare(&bld, p, texel);
+}
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_struct.c b/src/gallium/drivers/llvmpipe/lp_bld_struct.c
index 14d2b10df9..3998ac374f 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_struct.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_struct.c
@@ -42,17 +42,30 @@
 
 
 LLVMValueRef
+lp_build_struct_get_ptr(LLVMBuilderRef builder,
+                        LLVMValueRef ptr,
+                        unsigned member,
+                        const char *name)
+{
+   LLVMValueRef indices[2];
+   LLVMValueRef member_ptr;
+   indices[0] = LLVMConstInt(LLVMInt32Type(), 0, 0);
+   indices[1] = LLVMConstInt(LLVMInt32Type(), member, 0);
+   member_ptr = LLVMBuildGEP(builder, ptr, indices, Elements(indices), "");
+   lp_build_name(member_ptr, "%s.%s_ptr", LLVMGetValueName(ptr), name);
+   return member_ptr;
+}
+
+
+LLVMValueRef
 lp_build_struct_get(LLVMBuilderRef builder,
                     LLVMValueRef ptr,
                     unsigned member,
                     const char *name)
 {
-   LLVMValueRef indices[2];
    LLVMValueRef member_ptr;
    LLVMValueRef res;
-   indices[0] = LLVMConstInt(LLVMInt32Type(), 0, 0);
-   indices[1] = LLVMConstInt(LLVMInt32Type(), member, 0);
-   member_ptr = LLVMBuildGEP(builder, ptr, indices, Elements(indices), "");
+   member_ptr = lp_build_struct_get_ptr(builder, ptr, member, name);
    res = LLVMBuildLoad(builder, member_ptr, "");
    lp_build_name(res, "%s.%s", LLVMGetValueName(ptr), name);
    return res;
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_struct.h b/src/gallium/drivers/llvmpipe/lp_bld_struct.h
index cbefdc9f81..740392f561 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_struct.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_struct.h
@@ -53,6 +53,18 @@
              offsetof(_ctype, _cmember))
 
 
+/**
+ * Get value pointer to a structure member.
+ */
+LLVMValueRef
+lp_build_struct_get_ptr(LLVMBuilderRef builder,
+                        LLVMValueRef ptr,
+                        unsigned member,
+                        const char *name);
+
+/**
+ * Get the value of a structure member.
+ */
 LLVMValueRef
 lp_build_struct_get(LLVMBuilderRef builder,
                     LLVMValueRef ptr,
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c
index ac7eed9379..64e81f7b1f 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.c
@@ -64,7 +64,7 @@ LLVMValueRef
 lp_build_broadcast_scalar(struct lp_build_context *bld,
                           LLVMValueRef scalar)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    LLVMValueRef res;
    unsigned i;
 
@@ -83,7 +83,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld,
                        LLVMValueRef a,
                        unsigned channel)
 {
-   const union lp_type type = bld->type;
+   const struct lp_type type = bld->type;
    const unsigned n = type.length;
    unsigned i, j;
 
@@ -115,7 +115,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld,
        *   YY00 YY00 .... YY00
        *   YYYY YYYY .... YYYY  <= output
        */
-      union lp_type type4 = type;
+      struct lp_type type4 = type;
       const char shifts[4][2] = {
          { 1,  2},
          {-1,  2},
@@ -161,7 +161,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld,
 LLVMValueRef
 lp_build_swizzle1_aos(struct lp_build_context *bld,
                       LLVMValueRef a,
-                      unsigned char swizzle[4])
+                      const unsigned char swizzle[4])
 {
    const unsigned n = bld->type.length;
    unsigned i, j;
@@ -192,7 +192,7 @@ LLVMValueRef
 lp_build_swizzle2_aos(struct lp_build_context *bld,
                       LLVMValueRef a,
                       LLVMValueRef b,
-                      unsigned char swizzle[4])
+                      const unsigned char swizzle[4])
 {
    const unsigned n = bld->type.length;
    unsigned i, j;
@@ -201,11 +201,12 @@ lp_build_swizzle2_aos(struct lp_build_context *bld,
       return lp_build_swizzle1_aos(bld, a, swizzle);
 
    if(a == b) {
-      swizzle[0] %= 4;
-      swizzle[1] %= 4;
-      swizzle[2] %= 4;
-      swizzle[3] %= 4;
-      return lp_build_swizzle1_aos(bld, a, swizzle);
+      unsigned char swizzle1[4];
+      swizzle1[0] = swizzle[0] % 4;
+      swizzle1[1] = swizzle[1] % 4;
+      swizzle1[2] = swizzle[2] % 4;
+      swizzle1[3] = swizzle[3] % 4;
+      return lp_build_swizzle1_aos(bld, a, swizzle1);
    }
 
    if(swizzle[0] % 4 == 0 &&
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h
index d7dd6a8a60..b9472127a6 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_swizzle.h
@@ -40,7 +40,7 @@
 #include <llvm-c/Core.h>  
 
 
-union lp_type type;
+struct lp_type;
 struct lp_build_context;
 
 
@@ -73,7 +73,7 @@ lp_build_broadcast_aos(struct lp_build_context *bld,
 LLVMValueRef
 lp_build_swizzle1_aos(struct lp_build_context *bld,
                       LLVMValueRef a,
-                      unsigned char swizzle[4]);
+                      const unsigned char swizzle[4]);
 
 
 /**
@@ -85,7 +85,7 @@ LLVMValueRef
 lp_build_swizzle2_aos(struct lp_build_context *bld,
                       LLVMValueRef a,
                       LLVMValueRef b,
-                      unsigned char swizzle[4]);
+                      const unsigned char swizzle[4]);
 
 
 #endif /* !LP_BLD_SWIZZLE_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h b/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h
index 912db24aec..eddb7a83fa 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi.h
@@ -39,31 +39,46 @@
 
 
 struct tgsi_token;
-union lp_type;
+struct lp_type;
 struct lp_build_context;
 struct lp_build_mask_context;
 
 
-typedef void
-(*lp_emit_fetch_texel_soa_callback)( LLVMBuilderRef builder,
-                                void *context,
-                                unsigned unit,
-                                unsigned num_coords,
-                                const LLVMValueRef *coords,
-                                LLVMValueRef lodbias,
-                                LLVMValueRef *texel);
+/**
+ * Sampler code generation interface.
+ *
+ * Although texture sampling is a requirement for TGSI translation, it is
+ * a very different problem with several different approaches to it. This
+ * structure establishes an interface for texture sampling code generation, so
+ * that we can easily use different texture sampling strategies.
+ */
+struct lp_build_sampler_soa
+{
+   void
+   (*destroy)( struct lp_build_sampler_soa *sampler );
+
+   void
+   (*emit_fetch_texel)( struct lp_build_sampler_soa *sampler,
+                        LLVMBuilderRef builder,
+                        struct lp_type type,
+                        unsigned unit,
+                        unsigned num_coords,
+                        const LLVMValueRef *coords,
+                        LLVMValueRef lodbias,
+                        LLVMValueRef *texel);
+};
+
 
 void
 lp_build_tgsi_soa(LLVMBuilderRef builder,
                   const struct tgsi_token *tokens,
-                  union lp_type type,
+                  struct lp_type type,
                   struct lp_build_mask_context *mask,
                   LLVMValueRef consts_ptr,
                   const LLVMValueRef *pos,
                   const LLVMValueRef (*inputs)[4],
                   LLVMValueRef (*outputs)[4],
-                  lp_emit_fetch_texel_soa_callback emit_fetch_texel,
-                  void *emit_fetch_texel_context);
+                  struct lp_build_sampler_soa *sampler);
 
 
 #endif /* LP_BLD_TGSI_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c
index d4d18febec..adc81569ed 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c
@@ -78,6 +78,11 @@
 #define CHAN_Z 2
 #define CHAN_W 3
 
+#define QUAD_TOP_LEFT     0
+#define QUAD_TOP_RIGHT    1
+#define QUAD_BOTTOM_LEFT  2
+#define QUAD_BOTTOM_RIGHT 3
+
 
 struct lp_build_tgsi_soa_context
 {
@@ -88,8 +93,7 @@ struct lp_build_tgsi_soa_context
    const LLVMValueRef (*inputs)[NUM_CHANNELS];
    LLVMValueRef (*outputs)[NUM_CHANNELS];
 
-   lp_emit_fetch_texel_soa_callback emit_fetch_texel;
-   void *emit_fetch_texel_context;
+   struct lp_build_sampler_soa *sampler;
 
    LLVMValueRef immediates[LP_MAX_IMMEDIATES][NUM_CHANNELS];
    LLVMValueRef temps[LP_MAX_TEMPS][NUM_CHANNELS];
@@ -98,6 +102,51 @@ struct lp_build_tgsi_soa_context
 };
 
 
+static const unsigned char
+swizzle_left[4] = {
+   QUAD_TOP_LEFT,     QUAD_TOP_LEFT,
+   QUAD_BOTTOM_LEFT,  QUAD_BOTTOM_LEFT
+};
+
+static const unsigned char
+swizzle_right[4] = {
+   QUAD_TOP_RIGHT,    QUAD_TOP_RIGHT,
+   QUAD_BOTTOM_RIGHT, QUAD_BOTTOM_RIGHT
+};
+
+static const unsigned char
+swizzle_top[4] = {
+   QUAD_TOP_LEFT,     QUAD_TOP_RIGHT,
+   QUAD_TOP_LEFT,     QUAD_TOP_RIGHT
+};
+
+static const unsigned char
+swizzle_bottom[4] = {
+   QUAD_BOTTOM_LEFT,  QUAD_BOTTOM_RIGHT,
+   QUAD_BOTTOM_LEFT,  QUAD_BOTTOM_RIGHT
+};
+
+
+static LLVMValueRef
+emit_ddx(struct lp_build_tgsi_soa_context *bld,
+         LLVMValueRef src)
+{
+   LLVMValueRef src_left  = lp_build_swizzle1_aos(&bld->base, src, swizzle_left);
+   LLVMValueRef src_right = lp_build_swizzle1_aos(&bld->base, src, swizzle_right);
+   return lp_build_sub(&bld->base, src_right, src_left);
+}
+
+
+static LLVMValueRef
+emit_ddy(struct lp_build_tgsi_soa_context *bld,
+         LLVMValueRef src)
+{
+   LLVMValueRef src_top    = lp_build_swizzle1_aos(&bld->base, src, swizzle_top);
+   LLVMValueRef src_bottom = lp_build_swizzle1_aos(&bld->base, src, swizzle_bottom);
+   return lp_build_sub(&bld->base, src_top, src_bottom);
+}
+
+
 /**
  * Register fetch.
  */
@@ -168,6 +217,7 @@ emit_fetch(
       break;
 
    case TGSI_UTIL_SIGN_SET:
+      /* TODO: Use bitwese OR for floating point */
       res = lp_build_abs( &bld->base, res );
       res = LLVMBuildNeg( bld->base.builder, res, "" );
       break;
@@ -185,6 +235,36 @@ emit_fetch(
 
 
 /**
+ * Register fetch with derivatives.
+ */
+static void
+emit_fetch_deriv(
+   struct lp_build_tgsi_soa_context *bld,
+   const struct tgsi_full_instruction *inst,
+   unsigned index,
+   const unsigned chan_index,
+   LLVMValueRef *res,
+   LLVMValueRef *ddx,
+   LLVMValueRef *ddy)
+{
+   LLVMValueRef src;
+
+   src = emit_fetch(bld, inst, index, chan_index);
+
+   if(res)
+      *res = src;
+
+   /* TODO: use interpolation coeffs for inputs */
+
+   if(ddx)
+      *ddx = emit_ddx(bld, src);
+
+   if(ddy)
+      *ddy = emit_ddy(bld, src);
+}
+
+
+/**
  * Register store.
  */
 static void
@@ -239,17 +319,18 @@ emit_store(
  * High-level instruction translators.
  */
 
+
 static void
 emit_tex( struct lp_build_tgsi_soa_context *bld,
           const struct tgsi_full_instruction *inst,
           boolean apply_lodbias,
-          boolean projected)
+          boolean projected,
+          LLVMValueRef *texel)
 {
    const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
    LLVMValueRef lodbias;
    LLVMValueRef oow;
    LLVMValueRef coords[3];
-   LLVMValueRef texel[4];
    unsigned num_coords;
    unsigned i;
 
@@ -289,12 +370,11 @@ emit_tex( struct lp_build_tgsi_soa_context *bld,
          coords[i] = lp_build_mul(&bld->base, coords[i], oow);
    }
 
-   bld->emit_fetch_texel(bld->base.builder, bld->emit_fetch_texel_context,
-                         unit, num_coords, coords, lodbias, texel);
-
-   FOR_EACH_DST0_ENABLED_CHANNEL( inst, i ) {
-      emit_store( bld, inst, 0, i, texel[i] );
-   }
+   bld->sampler->emit_fetch_texel(bld->sampler,
+                                  bld->base.builder,
+                                  bld->base.type,
+                                  unit, num_coords, coords, lodbias,
+                                  texel);
 }
 
 
@@ -347,14 +427,6 @@ emit_kil(
 }
 
 
-static void
-emit_kilp(
-   struct lp_build_tgsi_soa_context *bld )
-{
-   /* XXX todo / fix me */
-}
-
-
 /**
  * Check if inst src/dest regs use indirect addressing into temporary
  * register file.
@@ -382,25 +454,35 @@ indirect_temp_reference(const struct tgsi_full_instruction *inst)
 static int
 emit_instruction(
    struct lp_build_tgsi_soa_context *bld,
-   struct tgsi_full_instruction *inst )
+   const struct tgsi_full_instruction *inst,
+   const struct tgsi_opcode_info *info)
 {
    unsigned chan_index;
    LLVMValueRef src0, src1, src2;
    LLVMValueRef tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-   LLVMValueRef dst0;
+   LLVMValueRef res;
+   LLVMValueRef dst0[NUM_CHANNELS];
 
    /* we can't handle indirect addressing into temp register file yet */
    if (indirect_temp_reference(inst))
       return FALSE;
 
+   assert(info->num_dst <= 1);
+   if(info->num_dst) {
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         dst0[chan_index] = bld->base.undef;
+      }
+   }
+
    switch (inst->Instruction.Opcode) {
 #if 0
    case TGSI_OPCODE_ARL:
+      /* FIXME */
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
          emit_flr(bld, 0, 0);
          emit_f2it( bld, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 #endif
@@ -408,19 +490,17 @@ emit_instruction(
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_SWZ:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = emit_fetch( bld, inst, 0, chan_index );
       }
       break;
 
    case TGSI_OPCODE_LIT:
       if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) ) {
-         emit_store( bld, inst, 0, CHAN_X, bld->base.one);
+         dst0[CHAN_X] = bld->base.one;
       }
       if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) ) {
          src0 = emit_fetch( bld, inst, 0, CHAN_X );
-         dst0 = lp_build_max( &bld->base, src0, bld->base.zero);
-         emit_store( bld, inst, 0, CHAN_Y, dst0);
+         dst0[CHAN_Y] = lp_build_max( &bld->base, src0, bld->base.zero);
       }
       if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) ) {
          /* XMM[1] = SrcReg[0].yyyy */
@@ -432,20 +512,19 @@ emit_instruction(
          tmp1 = lp_build_pow( &bld->base, tmp1, tmp2);
          tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
          tmp2 = lp_build_cmp(&bld->base, PIPE_FUNC_GREATER, tmp0, bld->base.zero);
-         dst0 = lp_build_select(&bld->base, tmp2, tmp1, bld->base.zero);
-         emit_store( bld, inst, 0, CHAN_Z, dst0);
+         dst0[CHAN_Z] = lp_build_select(&bld->base, tmp2, tmp1, bld->base.zero);
       }
       if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) ) {
-         emit_store( bld, inst, 0, CHAN_W, bld->base.one);
+         dst0[CHAN_W] = bld->base.one;
       }
       break;
 
    case TGSI_OPCODE_RCP:
    /* TGSI_OPCODE_RECIP */
       src0 = emit_fetch( bld, inst, 0, CHAN_X );
-      dst0 = lp_build_rcp(&bld->base, src0);
+      res = lp_build_rcp(&bld->base, src0);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, dst0 );
+         dst0[chan_index] = res;
       }
       break;
 
@@ -453,9 +532,9 @@ emit_instruction(
    /* TGSI_OPCODE_RECIPSQRT */
       src0 = emit_fetch( bld, inst, 0, CHAN_X );
       src0 = lp_build_abs(&bld->base, src0);
-      dst0 = lp_build_rsqrt(&bld->base, src0);
+      res = lp_build_rsqrt(&bld->base, src0);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, dst0 );
+         dst0[chan_index] = res;
       }
       break;
 
@@ -479,16 +558,15 @@ emit_instruction(
          lp_build_exp2_approx(&bld->base, src0, p_exp2_int_part, p_frac_part, p_exp2);
 
          if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ))
-            emit_store( bld, inst, 0, CHAN_X, tmp0);
+            dst0[CHAN_X] = tmp0;
          if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ))
-            emit_store( bld, inst, 0, CHAN_Y, tmp1);
+            dst0[CHAN_Y] = tmp1;
          if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ))
-            emit_store( bld, inst, 0, CHAN_Z, tmp2);
+            dst0[CHAN_Z] = tmp2;
       }
       /* dst.w = 1.0 */
       if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_W )) {
-         tmp0 = bld->base.one;
-         emit_store( bld, inst, 0, CHAN_W, tmp0);
+         dst0[CHAN_W] = bld->base.one;
       }
       break;
 
@@ -514,20 +592,18 @@ emit_instruction(
 
          /* dst.x = floor(lg2(abs(src.x))) */
          if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ))
-            emit_store( bld, inst, 0, CHAN_X, tmp0);
+            dst0[CHAN_X] = tmp0;
          /* dst.y = abs(src)/ex2(floor(lg2(abs(src.x)))) */
          if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y )) {
-            tmp1 = lp_build_div( &bld->base, src0, tmp1);
-            emit_store( bld, inst, 0, CHAN_Y, tmp1);
+            dst0[CHAN_Y] = lp_build_div( &bld->base, src0, tmp1);
          }
          /* dst.z = lg2(abs(src.x)) */
          if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ))
-            emit_store( bld, inst, 0, CHAN_Z, tmp2);
+            dst0[CHAN_Z] = tmp2;
       }
       /* dst.w = 1.0 */
       if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_W )) {
-         tmp0 = bld->base.one;
-         emit_store( bld, inst, 0, CHAN_W, tmp0);
+         dst0[CHAN_W] = bld->base.one;
       }
       break;
 
@@ -535,8 +611,7 @@ emit_instruction(
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
-         dst0 = lp_build_mul(&bld->base, src0, src1);
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_mul(&bld->base, src0, src1);
       }
       break;
 
@@ -544,8 +619,7 @@ emit_instruction(
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
-         dst0 = lp_build_add(&bld->base, src0, src1);
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_add(&bld->base, src0, src1);
       }
       break;
 
@@ -563,7 +637,7 @@ emit_instruction(
       tmp1 = lp_build_mul( &bld->base, tmp1, tmp2);
       tmp0 = lp_build_add( &bld->base, tmp0, tmp1);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
@@ -585,28 +659,24 @@ emit_instruction(
       tmp1 = lp_build_mul( &bld->base, tmp1, tmp2);
       tmp0 = lp_build_add( &bld->base, tmp0, tmp1);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
    case TGSI_OPCODE_DST:
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) {
-         tmp0 = bld->base.one;
-         emit_store( bld, inst, 0, CHAN_X, tmp0);
+         dst0[CHAN_X] = bld->base.one;
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) {
          tmp0 = emit_fetch( bld, inst, 0, CHAN_Y );
          tmp1 = emit_fetch( bld, inst, 1, CHAN_Y );
-         tmp0 = lp_build_mul( &bld->base, tmp0, tmp1);
-         emit_store( bld, inst, 0, CHAN_Y, tmp0);
+         dst0[CHAN_Y] = lp_build_mul( &bld->base, tmp0, tmp1);
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) {
-         tmp0 = emit_fetch( bld, inst, 0, CHAN_Z );
-         emit_store( bld, inst, 0, CHAN_Z, tmp0);
+         dst0[CHAN_Z] = emit_fetch( bld, inst, 0, CHAN_Z );
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) {
-         tmp0 = emit_fetch( bld, inst, 1, CHAN_W );
-         emit_store( bld, inst, 0, CHAN_W, tmp0);
+         dst0[CHAN_W] = emit_fetch( bld, inst, 1, CHAN_W );
       }
       break;
 
@@ -614,8 +684,7 @@ emit_instruction(
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
-         dst0 = lp_build_min( &bld->base, src0, src1 );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_min( &bld->base, src0, src1 );
       }
       break;
 
@@ -623,8 +692,7 @@ emit_instruction(
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
-         dst0 = lp_build_max( &bld->base, src0, src1 );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_max( &bld->base, src0, src1 );
       }
       break;
 
@@ -634,8 +702,7 @@ emit_instruction(
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_LESS, src0, src1 );
-         dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
       }
       break;
 
@@ -645,8 +712,7 @@ emit_instruction(
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_GEQUAL, src0, src1 );
-         dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
       }
       break;
 
@@ -658,7 +724,7 @@ emit_instruction(
          tmp2 = emit_fetch( bld, inst, 2, chan_index );
          tmp0 = lp_build_mul( &bld->base, tmp0, tmp1);
          tmp0 = lp_build_add( &bld->base, tmp0, tmp2);
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
@@ -666,8 +732,7 @@ emit_instruction(
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
          tmp1 = emit_fetch( bld, inst, 1, chan_index );
-         tmp0 = lp_build_sub( &bld->base, tmp0, tmp1);
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = lp_build_sub( &bld->base, tmp0, tmp1);
       }
       break;
 
@@ -678,13 +743,19 @@ emit_instruction(
          src2 = emit_fetch( bld, inst, 2, chan_index );
          tmp0 = lp_build_sub( &bld->base, src1, src2 );
          tmp0 = lp_build_mul( &bld->base, src0, tmp0 );
-         dst0 = lp_build_add( &bld->base, tmp0, src2 );
-         emit_store( bld, inst, 0, chan_index, dst0 );
+         dst0[chan_index] = lp_build_add( &bld->base, tmp0, src2 );
       }
       break;
 
    case TGSI_OPCODE_CND:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         src0 = emit_fetch( bld, inst, 0, chan_index );
+         src1 = emit_fetch( bld, inst, 1, chan_index );
+         src2 = emit_fetch( bld, inst, 2, chan_index );
+         tmp1 = lp_build_const_scalar(bld->base.type, 0.5);
+         tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_GREATER, src2, tmp1);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, src0, src1 );
+      }
       break;
 
    case TGSI_OPCODE_DP2A:
@@ -698,45 +769,49 @@ emit_instruction(
       tmp1 = emit_fetch( bld, inst, 2, CHAN_X );  /* xmm1 = src[2].x */
       tmp0 = lp_build_add( &bld->base, tmp0, tmp1);              /* xmm0 = xmm0 + xmm1 */
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);  /* dest[ch] = xmm0 */
+         dst0[chan_index] = tmp0;  /* dest[ch] = xmm0 */
       }
       break;
 
-#if 0
    case TGSI_OPCODE_FRC:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         emit_frc( bld, 0, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         src0 = emit_fetch( bld, inst, 0, chan_index );
+         tmp0 = lp_build_floor(&bld->base, src0);
+         tmp0 = lp_build_sub(&bld->base, tmp0, src0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
    case TGSI_OPCODE_CLAMP:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         tmp0 = emit_fetch( bld, inst, 0, chan_index );
+         src1 = emit_fetch( bld, inst, 1, chan_index );
+         src2 = emit_fetch( bld, inst, 2, chan_index );
+         tmp0 = lp_build_max(&bld->base, tmp0, src1);
+         tmp0 = lp_build_min(&bld->base, tmp0, src2);
+         dst0[chan_index] = tmp0;
+      }
       break;
 
    case TGSI_OPCODE_FLR:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         emit_flr( bld, 0, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = lp_build_floor(&bld->base, tmp0);
       }
       break;
 
    case TGSI_OPCODE_ROUND:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         emit_rnd( bld, 0, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = lp_build_round(&bld->base, tmp0);
       }
       break;
-#endif
 
    case TGSI_OPCODE_EX2: {
       tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
       tmp0 = lp_build_exp2( &bld->base, tmp0);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
    }
@@ -745,16 +820,16 @@ emit_instruction(
       tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
       tmp0 = lp_build_log2( &bld->base, tmp0);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
    case TGSI_OPCODE_POW:
       src0 = emit_fetch( bld, inst, 0, CHAN_X );
       src1 = emit_fetch( bld, inst, 1, CHAN_X );
-      dst0 = lp_build_pow( &bld->base, src0, src1 );
+      res = lp_build_pow( &bld->base, src0, src1 );
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, dst0 );
+         dst0[chan_index] = res;
       }
       break;
 
@@ -775,7 +850,7 @@ emit_instruction(
          tmp5 = tmp3;
          tmp5 = lp_build_mul( &bld->base, tmp5, tmp4);
          tmp2 = lp_build_sub( &bld->base, tmp2, tmp5);
-         emit_store( bld, inst, 0, CHAN_X, tmp2);
+         dst0[CHAN_X] = tmp2;
       }
       if( IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) ||
           IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) ) {
@@ -786,31 +861,30 @@ emit_instruction(
          tmp3 = lp_build_mul( &bld->base, tmp3, tmp2);
          tmp1 = lp_build_mul( &bld->base, tmp1, tmp5);
          tmp3 = lp_build_sub( &bld->base, tmp3, tmp1);
-         emit_store( bld, inst, 0, CHAN_Y, tmp3);
+         dst0[CHAN_Y] = tmp3;
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) {
          tmp5 = lp_build_mul( &bld->base, tmp5, tmp4);
          tmp0 = lp_build_mul( &bld->base, tmp0, tmp2);
          tmp5 = lp_build_sub( &bld->base, tmp5, tmp0);
-         emit_store( bld, inst, 0, CHAN_Z, tmp5);
+         dst0[CHAN_Z] = tmp5;
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) {
-         tmp0 = bld->base.one;
-         emit_store( bld, inst, 0, CHAN_W, tmp0);
+         dst0[CHAN_W] = bld->base.one;
       }
       break;
 
    case TGSI_OPCODE_ABS:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         tmp0 = lp_build_abs( &bld->base, tmp0 ) ;
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = lp_build_abs( &bld->base, tmp0 );
       }
       break;
 
    case TGSI_OPCODE_RCC:
+      /* deprecated? */
+      assert(0);
       return 0;
-      break;
 
    case TGSI_OPCODE_DPH:
       tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
@@ -827,7 +901,7 @@ emit_instruction(
       tmp1 = emit_fetch( bld, inst, 1, CHAN_W );
       tmp0 = lp_build_add( &bld->base, tmp0, tmp1);
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
@@ -835,25 +909,27 @@ emit_instruction(
       tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
       tmp0 = lp_build_cos( &bld->base, tmp0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
    case TGSI_OPCODE_DDX:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         emit_fetch_deriv( bld, inst, 0, chan_index, NULL, &dst0[chan_index], NULL);
+      }
       break;
 
    case TGSI_OPCODE_DDY:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         emit_fetch_deriv( bld, inst, 0, chan_index, NULL, NULL, &dst0[chan_index]);
+      }
       break;
 
-#if 0
    case TGSI_OPCODE_KILP:
       /* predicated kill */
-      emit_kilp( bld );
-      return 0; /* XXX fix me */
+      /* FIXME */
+      return 0;
       break;
-#endif
 
    case TGSI_OPCODE_KIL:
       /* conditional kill */
@@ -885,13 +961,14 @@ emit_instruction(
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_EQUAL, src0, src1 );
-         dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
       }
       break;
 
    case TGSI_OPCODE_SFL:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         dst0[chan_index] = bld->base.zero;
+      }
       break;
 
    case TGSI_OPCODE_SGT:
@@ -899,8 +976,7 @@ emit_instruction(
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_GREATER, src0, src1 );
-         dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
       }
       break;
 
@@ -908,7 +984,7 @@ emit_instruction(
       tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
       tmp0 = lp_build_sin( &bld->base, tmp0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 
@@ -917,8 +993,7 @@ emit_instruction(
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_LEQUAL, src0, src1 );
-         dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
       }
       break;
 
@@ -927,85 +1002,99 @@ emit_instruction(
          src0 = emit_fetch( bld, inst, 0, chan_index );
          src1 = emit_fetch( bld, inst, 1, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_NOTEQUAL, src0, src1 );
-         dst0 = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, bld->base.one, bld->base.zero );
       }
       break;
 
    case TGSI_OPCODE_STR:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         dst0[chan_index] = bld->base.one;
+      }
       break;
 
    case TGSI_OPCODE_TEX:
-      emit_tex( bld, inst, FALSE, FALSE );
+      emit_tex( bld, inst, FALSE, FALSE, dst0 );
       break;
 
    case TGSI_OPCODE_TXD:
+      /* FIXME */
       return 0;
       break;
 
    case TGSI_OPCODE_UP2H:
+      /* deprecated */
+      assert (0);
       return 0;
       break;
 
    case TGSI_OPCODE_UP2US:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_UP4B:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_UP4UB:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_X2D:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_ARA:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
 #if 0
    case TGSI_OPCODE_ARR:
+      /* FIXME */
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
          emit_rnd( bld, 0, 0 );
          emit_f2it( bld, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = tmp0;
       }
       break;
 #endif
 
    case TGSI_OPCODE_BRA:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_CAL:
+      /* FIXME */
       return 0;
       break;
 
-#if 0
    case TGSI_OPCODE_RET:
-      emit_ret( bld );
+      /* FIXME */
+      return 0;
       break;
-#endif
 
    case TGSI_OPCODE_END:
       break;
 
-#if 0
    case TGSI_OPCODE_SSG:
    /* TGSI_OPCODE_SGN */
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         emit_sgn( bld, 0, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = lp_build_sgn( &bld->base, tmp0 );
       }
       break;
-#endif
 
    case TGSI_OPCODE_CMP:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
@@ -1013,34 +1102,29 @@ emit_instruction(
          src1 = emit_fetch( bld, inst, 1, chan_index );
          src2 = emit_fetch( bld, inst, 2, chan_index );
          tmp0 = lp_build_cmp( &bld->base, PIPE_FUNC_LESS, src0, bld->base.zero );
-         dst0 = lp_build_select( &bld->base, tmp0, src1, src2);
-         emit_store( bld, inst, 0, chan_index, dst0);
+         dst0[chan_index] = lp_build_select( &bld->base, tmp0, src1, src2);
       }
       break;
 
    case TGSI_OPCODE_SCS:
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) {
          tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
-         tmp0 = lp_build_cos( &bld->base, tmp0 );
-         emit_store( bld, inst, 0, CHAN_X, tmp0);
+         dst0[CHAN_X] = lp_build_cos( &bld->base, tmp0 );
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) {
          tmp0 = emit_fetch( bld, inst, 0, CHAN_X );
-         tmp0 = lp_build_sin( &bld->base, tmp0 );
-         emit_store( bld, inst, 0, CHAN_Y, tmp0);
+         dst0[CHAN_Y] = lp_build_sin( &bld->base, tmp0 );
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z ) {
-         tmp0 = bld->base.zero;
-         emit_store( bld, inst, 0, CHAN_Z, tmp0);
+         dst0[CHAN_Z] = bld->base.zero;
       }
       IF_IS_DST0_CHANNEL_ENABLED( inst, CHAN_W ) {
-         tmp0 = bld->base.one;
-         emit_store( bld, inst, 0, CHAN_W, tmp0);
+         dst0[CHAN_W] = bld->base.one;
       }
       break;
 
    case TGSI_OPCODE_TXB:
-      emit_tex( bld, inst, TRUE, FALSE );
+      emit_tex( bld, inst, TRUE, FALSE, dst0 );
       break;
 
    case TGSI_OPCODE_NRM:
@@ -1099,38 +1183,35 @@ emit_instruction(
 
             /* dst.x = xmm1 * src.x */
             if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_X)) {
-               tmp4 = lp_build_mul( &bld->base, tmp4, tmp1);
-               emit_store(bld, inst, 0, CHAN_X, tmp4);
+               dst0[CHAN_X] = lp_build_mul( &bld->base, tmp4, tmp1);
             }
 
             /* dst.y = xmm1 * src.y */
             if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_Y)) {
-               tmp5 = lp_build_mul( &bld->base, tmp5, tmp1);
-               emit_store(bld, inst, 0, CHAN_Y, tmp5);
+               dst0[CHAN_Y] = lp_build_mul( &bld->base, tmp5, tmp1);
             }
 
             /* dst.z = xmm1 * src.z */
             if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_Z)) {
-               tmp6 = lp_build_mul( &bld->base, tmp6, tmp1);
-               emit_store(bld, inst, 0, CHAN_Z, tmp6);
+               dst0[CHAN_Z] = lp_build_mul( &bld->base, tmp6, tmp1);
             }
 
             /* dst.w = xmm1 * src.w */
             if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_X) && dims == 4) {
-               tmp7 = lp_build_mul( &bld->base, tmp7, tmp1);
-               emit_store(bld, inst, 0, CHAN_W, tmp7);
+               dst0[CHAN_W] = lp_build_mul( &bld->base, tmp7, tmp1);
             }
          }
 
-         /* dst0.w = 1.0 */
+         /* dst.w = 1.0 */
          if (IS_DST0_CHANNEL_ENABLED(inst, CHAN_W) && dims == 3) {
-            tmp0 = bld->base.one;
-            emit_store(bld, inst, 0, CHAN_W, tmp0);
+            dst0[CHAN_W] = bld->base.one;
          }
       }
       break;
 
    case TGSI_OPCODE_DIV:
+      /* deprecated */
+      assert( 0 );
       return 0;
       break;
 
@@ -1143,118 +1224,157 @@ emit_instruction(
       tmp1 = lp_build_mul( &bld->base, tmp1, tmp2);              /* xmm1 = xmm1 * xmm2 */
       tmp0 = lp_build_add( &bld->base, tmp0, tmp1);              /* xmm0 = xmm0 + xmm1 */
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
-         emit_store( bld, inst, 0, chan_index, tmp0);  /* dest[ch] = xmm0 */
+         dst0[chan_index] = tmp0;  /* dest[ch] = xmm0 */
       }
       break;
 
    case TGSI_OPCODE_TXL:
-      emit_tex( bld, inst, TRUE, FALSE );
+      emit_tex( bld, inst, TRUE, FALSE, dst0 );
       break;
 
    case TGSI_OPCODE_TXP:
-      emit_tex( bld, inst, FALSE, TRUE );
+      emit_tex( bld, inst, FALSE, TRUE, dst0 );
       break;
       
    case TGSI_OPCODE_BRK:
+      /* FIXME */
       return 0;
       break;
 
    case TGSI_OPCODE_IF:
+      /* FIXME */
       return 0;
       break;
 
    case TGSI_OPCODE_BGNFOR:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_REP:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_ELSE:
+      /* FIXME */
       return 0;
       break;
 
    case TGSI_OPCODE_ENDIF:
+      /* FIXME */
       return 0;
       break;
 
    case TGSI_OPCODE_ENDFOR:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_ENDREP:
+      /* deprecated */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_PUSHA:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_POPA:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_CEIL:
-      return 0;
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         tmp0 = emit_fetch( bld, inst, 0, chan_index );
+         dst0[chan_index] = lp_build_ceil(&bld->base, tmp0);
+      }
       break;
 
    case TGSI_OPCODE_I2F:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_NOT:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
-#if 0
    case TGSI_OPCODE_TRUNC:
       FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
          tmp0 = emit_fetch( bld, inst, 0, chan_index );
-         emit_f2it( bld, 0 );
-         emit_i2f( bld, 0 );
-         emit_store( bld, inst, 0, chan_index, tmp0);
+         dst0[chan_index] = lp_build_trunc(&bld->base, tmp0);
       }
       break;
-#endif
 
    case TGSI_OPCODE_SHL:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_SHR:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_AND:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_OR:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_MOD:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_XOR:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_SAD:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_TXF:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_TXQ:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
    case TGSI_OPCODE_CONT:
+      /* deprecated? */
+      assert(0);
       return 0;
       break;
 
@@ -1266,10 +1386,28 @@ emit_instruction(
       return 0;
       break;
 
+   case TGSI_OPCODE_NOISE1:
+   case TGSI_OPCODE_NOISE2:
+   case TGSI_OPCODE_NOISE3:
+   case TGSI_OPCODE_NOISE4:
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         dst0[chan_index] = bld->base.zero;
+      }
+      break;
+
+   case TGSI_OPCODE_NOP:
+      break;
+
    default:
       return 0;
    }
    
+   if(info->num_dst) {
+      FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) {
+         emit_store( bld, inst, 0, chan_index, dst0[chan_index]);
+      }
+   }
+
    return 1;
 }
 
@@ -1277,14 +1415,13 @@ emit_instruction(
 void
 lp_build_tgsi_soa(LLVMBuilderRef builder,
                   const struct tgsi_token *tokens,
-                  union lp_type type,
+                  struct lp_type type,
                   struct lp_build_mask_context *mask,
                   LLVMValueRef consts_ptr,
                   const LLVMValueRef *pos,
                   const LLVMValueRef (*inputs)[NUM_CHANNELS],
                   LLVMValueRef (*outputs)[NUM_CHANNELS],
-                  lp_emit_fetch_texel_soa_callback emit_fetch_texel,
-                  void *emit_fetch_texel_context)
+                  struct lp_build_sampler_soa *sampler)
 {
    struct lp_build_tgsi_soa_context bld;
    struct tgsi_parse_context parse;
@@ -1299,8 +1436,7 @@ lp_build_tgsi_soa(LLVMBuilderRef builder,
    bld.inputs = inputs;
    bld.outputs = outputs;
    bld.consts_ptr = consts_ptr;
-   bld.emit_fetch_texel = emit_fetch_texel;
-   bld.emit_fetch_texel_context = emit_fetch_texel_context;
+   bld.sampler = sampler;
 
    tgsi_parse_init( &parse, tokens );
 
@@ -1309,16 +1445,18 @@ lp_build_tgsi_soa(LLVMBuilderRef builder,
 
       switch( parse.FullToken.Token.Type ) {
       case TGSI_TOKEN_TYPE_DECLARATION:
-         /* Input already interpolated */
+         /* Inputs already interpolated */
          break;
 
       case TGSI_TOKEN_TYPE_INSTRUCTION:
-         if (!emit_instruction( &bld, &parse.FullToken.FullInstruction )) {
+         {
             unsigned opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
             const struct tgsi_opcode_info *info = tgsi_get_opcode_info(opcode);
-	    _debug_printf("warning: failed to translate tgsi opcode %s to LLVM\n",
-	                  info ? info->mnemonic : "<invalid>");
-	 }
+            if (!emit_instruction( &bld, &parse.FullToken.FullInstruction, info ))
+               _debug_printf("warning: failed to translate tgsi opcode %s to LLVM\n",
+                             info ? info->mnemonic : "<invalid>");
+         }
+
          break;
 
       case TGSI_TOKEN_TYPE_IMMEDIATE:
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_type.c b/src/gallium/drivers/llvmpipe/lp_bld_type.c
index 8e0026fd97..1320a26721 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_type.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_type.c
@@ -33,7 +33,7 @@
 
 
 LLVMTypeRef
-lp_build_elem_type(union lp_type type)
+lp_build_elem_type(struct lp_type type)
 {
    if (type.floating) {
       switch(type.width) {
@@ -55,7 +55,7 @@ lp_build_elem_type(union lp_type type)
 
 
 LLVMTypeRef
-lp_build_vec_type(union lp_type type)
+lp_build_vec_type(struct lp_type type)
 {
    LLVMTypeRef elem_type = lp_build_elem_type(type);
    return LLVMVectorType(elem_type, type.length);
@@ -69,7 +69,7 @@ lp_build_vec_type(union lp_type type)
  * type and check for identity.
  */
 boolean
-lp_check_elem_type(union lp_type type, LLVMTypeRef elem_type) 
+lp_check_elem_type(struct lp_type type, LLVMTypeRef elem_type) 
 {
    LLVMTypeKind elem_kind;
 
@@ -107,7 +107,7 @@ lp_check_elem_type(union lp_type type, LLVMTypeRef elem_type)
 
 
 boolean
-lp_check_vec_type(union lp_type type, LLVMTypeRef vec_type) 
+lp_check_vec_type(struct lp_type type, LLVMTypeRef vec_type) 
 {
    LLVMTypeRef elem_type;
 
@@ -128,7 +128,7 @@ lp_check_vec_type(union lp_type type, LLVMTypeRef vec_type)
 
 
 boolean
-lp_check_value(union lp_type type, LLVMValueRef val) 
+lp_check_value(struct lp_type type, LLVMValueRef val) 
 {
    LLVMTypeRef vec_type;
 
@@ -143,24 +143,55 @@ lp_check_value(union lp_type type, LLVMValueRef val)
 
 
 LLVMTypeRef
-lp_build_int_elem_type(union lp_type type)
+lp_build_int_elem_type(struct lp_type type)
 {
    return LLVMIntType(type.width);
 }
 
 
 LLVMTypeRef
-lp_build_int_vec_type(union lp_type type)
+lp_build_int_vec_type(struct lp_type type)
 {
    LLVMTypeRef elem_type = lp_build_int_elem_type(type);
    return LLVMVectorType(elem_type, type.length);
 }
 
 
+struct lp_type
+lp_int_type(struct lp_type type)
+{
+   struct lp_type res_type;
+
+   memset(&res_type, 0, sizeof res_type);
+   res_type.width = type.width;
+   res_type.length = type.length;
+
+   return res_type;
+}
+
+
+/**
+ * Return the type with twice the bit width (hence half the number of elements).
+ */
+struct lp_type
+lp_wider_type(struct lp_type type)
+{
+   struct lp_type res_type;
+
+   memcpy(&res_type, &type, sizeof res_type);
+   res_type.width *= 2;
+   res_type.length /= 2;
+
+   assert(res_type.length);
+
+   return res_type;
+}
+
+
 void
 lp_build_context_init(struct lp_build_context *bld,
                       LLVMBuilderRef builder,
-                      union lp_type type)
+                      struct lp_type type)
 {
    bld->builder = builder;
    bld->type = type;
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_type.h b/src/gallium/drivers/llvmpipe/lp_bld_type.h
index 3ce566be64..46c298fa20 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_type.h
+++ b/src/gallium/drivers/llvmpipe/lp_bld_type.h
@@ -56,58 +56,55 @@
  * on the types used for intermediate computations, such as signed vs unsigned,
  * normalized values, or fixed point.
  */
-union lp_type {
-   struct {
-      /** 
-       * Floating-point. Cannot be used with fixed. Integer numbers are
-       * represented by this zero.
-       */
-      unsigned floating:1;
-
-      /** 
-       * Fixed-point. Cannot be used with floating. Integer numbers are
-       * represented by this zero.
-       */
-      unsigned fixed:1;
-      
-      /** 
-       * Whether it can represent negative values or not.
-       *
-       * If this is not set for floating point, it means that all values are
-       * assumed to be positive.
-       */
-      unsigned sign:1;
-
-      /**
-       * Whether values are normalized to fit [0, 1] interval, or [-1, 1]
-       * interval for signed types.
-       *
-       * For integer types it means the representable integer range should be
-       * interpreted as the interval above.
-       *
-       * For floating and fixed point formats it means the values should be
-       * clamped to the interval above.
-       */
-      unsigned norm:1;
-
-      /**
-       * Element width.
-       *
-       * For fixed point values, the fixed point is assumed to be at half the
-       * width.
-       */
-      unsigned width:14;
-
-      /** 
-       * Vector length.
-       *
-       * width*length should be a power of two greater or equal to eight.
-       *
-       * @sa LP_MAX_VECTOR_LENGTH
-       */
-      unsigned length:14;
-   };
-   uint32_t value;
+struct lp_type {
+   /**
+    * Floating-point. Cannot be used with fixed. Integer numbers are
+    * represented by this zero.
+    */
+   unsigned floating:1;
+
+   /**
+    * Fixed-point. Cannot be used with floating. Integer numbers are
+    * represented by this zero.
+    */
+   unsigned fixed:1;
+
+   /**
+    * Whether it can represent negative values or not.
+    *
+    * If this is not set for floating point, it means that all values are
+    * assumed to be positive.
+    */
+   unsigned sign:1;
+
+   /**
+    * Whether values are normalized to fit [0, 1] interval, or [-1, 1]
+    * interval for signed types.
+    *
+    * For integer types it means the representable integer range should be
+    * interpreted as the interval above.
+    *
+    * For floating and fixed point formats it means the values should be
+    * clamped to the interval above.
+    */
+   unsigned norm:1;
+
+   /**
+    * Element width.
+    *
+    * For fixed point values, the fixed point is assumed to be at half the
+    * width.
+    */
+   unsigned width:14;
+
+   /**
+    * Vector length.
+    *
+    * width*length should be a power of two greater or equal to eight.
+    *
+    * @sa LP_MAX_VECTOR_LENGTH
+    */
+   unsigned length:14;
 };
 
 
@@ -124,7 +121,7 @@ struct lp_build_context
     * This not only describes the input/output LLVM types, but also whether
     * to normalize/clamp the results.
     */
-   union lp_type type;
+   struct lp_type type;
 
    /** Same as lp_build_undef(type) */
    LLVMValueRef undef;
@@ -138,37 +135,45 @@ struct lp_build_context
 
 
 LLVMTypeRef
-lp_build_elem_type(union lp_type type);
+lp_build_elem_type(struct lp_type type);
 
 
 LLVMTypeRef
-lp_build_vec_type(union lp_type type);
+lp_build_vec_type(struct lp_type type);
 
 
 boolean
-lp_check_elem_type(union lp_type type, LLVMTypeRef elem_type);
+lp_check_elem_type(struct lp_type type, LLVMTypeRef elem_type);
 
 
 boolean
-lp_check_vec_type(union lp_type type, LLVMTypeRef vec_type);
+lp_check_vec_type(struct lp_type type, LLVMTypeRef vec_type);
 
 
 boolean
-lp_check_value(union lp_type type, LLVMValueRef val);
+lp_check_value(struct lp_type type, LLVMValueRef val);
 
 
 LLVMTypeRef
-lp_build_int_elem_type(union lp_type type);
+lp_build_int_elem_type(struct lp_type type);
 
 
 LLVMTypeRef
-lp_build_int_vec_type(union lp_type type);
+lp_build_int_vec_type(struct lp_type type);
+
+
+struct lp_type
+lp_int_type(struct lp_type type);
+
+
+struct lp_type
+lp_wider_type(struct lp_type type);
 
 
 void
 lp_build_context_init(struct lp_build_context *bld,
                       LLVMBuilderRef builder,
-                      union lp_type type);
+                      struct lp_type type);
 
 
 #endif /* !LP_BLD_TYPE_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_clear.c b/src/gallium/drivers/llvmpipe/lp_clear.c
index 580cca5b46..bdcff94b9b 100644
--- a/src/gallium/drivers/llvmpipe/lp_clear.c
+++ b/src/gallium/drivers/llvmpipe/lp_clear.c
@@ -67,6 +67,7 @@ llvmpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba,
          util_pack_color(rgba, ps->format, &cv);
          lp_tile_cache_clear(llvmpipe->cbuf_cache[i], rgba, cv);
       }
+      llvmpipe->dirty_render_cache = TRUE;
    }
 
    if (buffers & PIPE_CLEAR_DEPTHSTENCIL) {
diff --git a/src/gallium/drivers/llvmpipe/lp_context.c b/src/gallium/drivers/llvmpipe/lp_context.c
index 233d1df0e1..202cb8ef43 100644
--- a/src/gallium/drivers/llvmpipe/lp_context.c
+++ b/src/gallium/drivers/llvmpipe/lp_context.c
@@ -107,11 +107,16 @@ static void llvmpipe_destroy( struct pipe_context *pipe )
    if (llvmpipe->draw)
       draw_destroy( llvmpipe->draw );
 
-   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++)
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
       lp_destroy_tile_cache(llvmpipe->cbuf_cache[i]);
+      pipe_surface_reference(&llvmpipe->framebuffer.cbufs[i], NULL);
+   }
+   pipe_surface_reference(&llvmpipe->framebuffer.zsbuf, NULL);
 
-   for (i = 0; i < PIPE_MAX_SAMPLERS; i++)
+   for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
       lp_destroy_tex_tile_cache(llvmpipe->tex_cache[i]);
+      pipe_texture_reference(&llvmpipe->texture[i], NULL);
+   }
 
    for (i = 0; i < Elements(llvmpipe->constants); i++) {
       if (llvmpipe->constants[i].buffer) {
@@ -141,8 +146,6 @@ llvmpipe_is_texture_referenced( struct pipe_context *pipe,
          return PIPE_REFERENCED_FOR_WRITE;
    }
    
-   /* FIXME: we also need to do the same for the texture cache */
-   
    return PIPE_UNREFERENCED;
 }
 
diff --git a/src/gallium/drivers/llvmpipe/lp_context.h b/src/gallium/drivers/llvmpipe/lp_context.h
index 8d5a0d4f1f..7df340554e 100644
--- a/src/gallium/drivers/llvmpipe/lp_context.h
+++ b/src/gallium/drivers/llvmpipe/lp_context.h
@@ -88,9 +88,6 @@ struct llvmpipe_context {
    /** Mapped vertex buffers */
    ubyte *mapped_vbuffer[PIPE_MAX_ATTRIBS];
    
-   /** Mapped constant buffers */
-   void *mapped_constants[PIPE_SHADER_TYPES];
-
    /** Vertex format */
    struct vertex_info vertex_info;
    struct vertex_info vertex_info_vbuf;
diff --git a/src/gallium/drivers/llvmpipe/lp_draw_arrays.c b/src/gallium/drivers/llvmpipe/lp_draw_arrays.c
index 89772e62d3..0aa13a1fc6 100644
--- a/src/gallium/drivers/llvmpipe/lp_draw_arrays.c
+++ b/src/gallium/drivers/llvmpipe/lp_draw_arrays.c
@@ -45,54 +45,6 @@
 
 
 
-static void
-llvmpipe_map_constant_buffers(struct llvmpipe_context *lp)
-{
-   struct pipe_screen *screen = lp->pipe.screen;
-   uint i, size;
-
-   for (i = 0; i < PIPE_SHADER_TYPES; i++) {
-      if (lp->constants[i].buffer && lp->constants[i].buffer->size)
-         lp->mapped_constants[i] = screen->buffer_map(screen, lp->constants[i].buffer,
-                                                      PIPE_BUFFER_USAGE_CPU_READ);
-   }
-
-   if (lp->constants[PIPE_SHADER_VERTEX].buffer)
-      size = lp->constants[PIPE_SHADER_VERTEX].buffer->size;
-   else
-      size = 0;
-
-   lp->jit_context.constants = lp->mapped_constants[PIPE_SHADER_FRAGMENT];
-
-   draw_set_mapped_constant_buffer(lp->draw,
-                                   lp->mapped_constants[PIPE_SHADER_VERTEX],
-                                   size);
-}
-
-
-static void
-llvmpipe_unmap_constant_buffers(struct llvmpipe_context *lp)
-{
-   struct pipe_screen *screen = lp->pipe.screen;
-   uint i;
-
-   /* really need to flush all prims since the vert/frag shaders const buffers
-    * are going away now.
-    */
-   draw_flush(lp->draw);
-
-   draw_set_mapped_constant_buffer(lp->draw, NULL, 0);
-
-   lp->jit_context.constants = NULL;
-
-   for (i = 0; i < 2; i++) {
-      if (lp->constants[i].buffer && lp->constants[i].buffer->size)
-         screen->buffer_unmap(screen, lp->constants[i].buffer);
-      lp->mapped_constants[i] = NULL;
-   }
-}
-
-
 boolean
 llvmpipe_draw_arrays(struct pipe_context *pipe, unsigned mode,
                      unsigned start, unsigned count)
@@ -124,7 +76,6 @@ llvmpipe_draw_range_elements(struct pipe_context *pipe,
       llvmpipe_update_derived( lp );
 
    llvmpipe_map_transfers(lp);
-   llvmpipe_map_constant_buffers(lp);
 
    /*
     * Map vertex buffers
@@ -163,7 +114,6 @@ llvmpipe_draw_range_elements(struct pipe_context *pipe,
 
 
    /* Note: leave drawing surfaces mapped */
-   llvmpipe_unmap_constant_buffers(lp);
 
    lp->dirty_render_cache = TRUE;
    
diff --git a/src/gallium/drivers/llvmpipe/lp_flush.c b/src/gallium/drivers/llvmpipe/lp_flush.c
index b5c1c95bb7..cd8381fe30 100644
--- a/src/gallium/drivers/llvmpipe/lp_flush.c
+++ b/src/gallium/drivers/llvmpipe/lp_flush.c
@@ -58,8 +58,10 @@ llvmpipe_flush( struct pipe_context *pipe,
        * in the hope that a later clear will wipe them out.
        */
       for (i = 0; i < llvmpipe->framebuffer.nr_cbufs; i++)
-         if (llvmpipe->cbuf_cache[i])
+         if (llvmpipe->cbuf_cache[i]) {
+            lp_tile_cache_map_transfers(llvmpipe->cbuf_cache[i]);
             lp_flush_tile_cache(llvmpipe->cbuf_cache[i]);
+         }
 
       /* Need this call for hardware buffers before swapbuffers.
        *
@@ -71,8 +73,10 @@ llvmpipe_flush( struct pipe_context *pipe,
    }
    else if (flags & PIPE_FLUSH_RENDER_CACHE) {
       for (i = 0; i < llvmpipe->framebuffer.nr_cbufs; i++)
-         if (llvmpipe->cbuf_cache[i])
+         if (llvmpipe->cbuf_cache[i]) {
+            lp_tile_cache_map_transfers(llvmpipe->cbuf_cache[i]);
             lp_flush_tile_cache(llvmpipe->cbuf_cache[i]);
+         }
 
       /* FIXME: untile zsbuf! */
      
diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c
index d288460a1b..1126bf90b9 100644
--- a/src/gallium/drivers/llvmpipe/lp_jit.c
+++ b/src/gallium/drivers/llvmpipe/lp_jit.c
@@ -36,6 +36,7 @@
 #include <llvm-c/Transforms/Scalar.h>
 
 #include "util/u_memory.h"
+#include "util/u_cpu_detect.h"
 #include "lp_screen.h"
 #include "lp_bld_intr.h"
 #include "lp_jit.h"
@@ -44,15 +45,47 @@
 static void
 lp_jit_init_globals(struct llvmpipe_screen *screen)
 {
-   /* struct lp_jit_context */
+   LLVMTypeRef texture_type;
+
+   /* struct lp_jit_texture */
    {
       LLVMTypeRef elem_types[4];
+
+      elem_types[LP_JIT_TEXTURE_WIDTH]  = LLVMInt32Type();
+      elem_types[LP_JIT_TEXTURE_HEIGHT] = LLVMInt32Type();
+      elem_types[LP_JIT_TEXTURE_STRIDE] = LLVMInt32Type();
+      elem_types[LP_JIT_TEXTURE_DATA]   = LLVMPointerType(LLVMInt8Type(), 0);
+
+      texture_type = LLVMStructType(elem_types, Elements(elem_types), 0);
+
+      LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, width,
+                             screen->target, texture_type,
+                             LP_JIT_TEXTURE_WIDTH);
+      LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, height,
+                             screen->target, texture_type,
+                             LP_JIT_TEXTURE_HEIGHT);
+      LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, stride,
+                             screen->target, texture_type,
+                             LP_JIT_TEXTURE_STRIDE);
+      LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, data,
+                             screen->target, texture_type,
+                             LP_JIT_TEXTURE_DATA);
+      LP_CHECK_STRUCT_SIZE(struct lp_jit_texture,
+                           screen->target, texture_type);
+
+      LLVMAddTypeName(screen->module, "texture", texture_type);
+   }
+
+   /* struct lp_jit_context */
+   {
+      LLVMTypeRef elem_types[5];
       LLVMTypeRef context_type;
 
       elem_types[0] = LLVMPointerType(LLVMFloatType(), 0); /* constants */
       elem_types[1] = LLVMPointerType(LLVMInt8Type(), 0);  /* samplers */
       elem_types[2] = LLVMFloatType();                     /* alpha_ref_value */
       elem_types[3] = LLVMPointerType(LLVMInt8Type(), 0);  /* blend_color */
+      elem_types[4] = LLVMArrayType(texture_type, PIPE_MAX_SAMPLERS); /* textures */
 
       context_type = LLVMStructType(elem_types, Elements(elem_types), 0);
 
@@ -64,6 +97,9 @@ lp_jit_init_globals(struct llvmpipe_screen *screen)
                              screen->target, context_type, 2);
       LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, blend_color,
                              screen->target, context_type, 3);
+      LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, textures,
+                             screen->target, context_type,
+                             LP_JIT_CONTEXT_TEXTURES_INDEX);
       LP_CHECK_STRUCT_SIZE(struct lp_jit_context,
                            screen->target, context_type);
 
@@ -112,12 +148,25 @@ lp_jit_screen_init(struct llvmpipe_screen *screen)
 {
    char *error = NULL;
 
+   util_cpu_detect();
+
+#if 0
+   /* For simulating less capable machines */
+   util_cpu_caps.has_sse3 = 0;
+   util_cpu_caps.has_sse4_1 = 0;
+#endif
+
+#ifdef LLVM_NATIVE_ARCH
+   LLVMLinkInJIT();
+   LLVMInitializeNativeTarget();
+#endif
+
    screen->module = LLVMModuleCreateWithName("llvmpipe");
 
    screen->provider = LLVMCreateModuleProviderForExistingModule(screen->module);
 
    if (LLVMCreateJITCompiler(&screen->engine, screen->provider, 1, &error)) {
-      fprintf(stderr, "%s\n", error);
+      _debug_printf("%s\n", error);
       LLVMDisposeMessage(error);
       abort();
    }
@@ -128,8 +177,15 @@ lp_jit_screen_init(struct llvmpipe_screen *screen)
    LLVMAddTargetData(screen->target, screen->pass);
    /* These are the passes currently listed in llvm-c/Transforms/Scalar.h,
     * but there are more on SVN. */
+   /* TODO: Add more passes */
    LLVMAddConstantPropagationPass(screen->pass);
-   LLVMAddInstructionCombiningPass(screen->pass);
+   if(util_cpu_caps.has_sse4_1) {
+      /* FIXME: There is a bug in this pass, whereby the combination of fptosi
+       * and sitofp (necessary for trunc/floor/ceil/round implementation)
+       * somehow becomes invalid code.
+       */
+      LLVMAddInstructionCombiningPass(screen->pass);
+   }
    LLVMAddPromoteMemoryToRegisterPass(screen->pass);
    LLVMAddGVNPass(screen->pass);
    LLVMAddCFGSimplificationPass(screen->pass);
diff --git a/src/gallium/drivers/llvmpipe/lp_jit.h b/src/gallium/drivers/llvmpipe/lp_jit.h
index a7fb60f9f5..58f716ede2 100644
--- a/src/gallium/drivers/llvmpipe/lp_jit.h
+++ b/src/gallium/drivers/llvmpipe/lp_jit.h
@@ -38,11 +38,31 @@
 
 #include "lp_bld_struct.h"
 
+#include "pipe/p_state.h"
+
 
 struct tgsi_sampler;
 struct llvmpipe_screen;
 
 
+struct lp_jit_texture
+{
+   uint32_t width;
+   uint32_t height;
+   uint32_t stride;
+   const void *data;
+};
+
+
+enum {
+   LP_JIT_TEXTURE_WIDTH = 0,
+   LP_JIT_TEXTURE_HEIGHT,
+   LP_JIT_TEXTURE_STRIDE,
+   LP_JIT_TEXTURE_DATA
+};
+
+
+
 /**
  * This structure is passed directly to the generated fragment shader.
  *
@@ -60,11 +80,12 @@ struct lp_jit_context
 
    struct tgsi_sampler **samplers;
 
-   /* TODO: alpha reference value */
    float alpha_ref_value;
 
-   /* TODO: blend constant color */
+   /* FIXME: store (also?) in floats */
    uint8_t *blend_color;
+
+   struct lp_jit_texture textures[PIPE_MAX_SAMPLERS];
 };
 
 
@@ -80,6 +101,11 @@ struct lp_jit_context
 #define lp_jit_context_blend_color(_builder, _ptr) \
    lp_build_struct_get(_builder, _ptr, 3, "blend_color")
 
+#define LP_JIT_CONTEXT_TEXTURES_INDEX 4
+
+#define lp_jit_context_textures(_builder, _ptr) \
+   lp_build_struct_get_ptr(_builder, _ptr, LP_JIT_CONTEXT_TEXTURES_INDEX, "textures")
+
 
 typedef void
 (*lp_jit_frag_func)(struct lp_jit_context *context,
diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c
index 125035771e..0518927458 100644
--- a/src/gallium/drivers/llvmpipe/lp_screen.c
+++ b/src/gallium/drivers/llvmpipe/lp_screen.c
@@ -27,8 +27,6 @@
 
 
 #include "util/u_memory.h"
-#include "util/u_simple_screen.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_screen.h"
 
@@ -67,8 +65,6 @@ llvmpipe_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_GLSL:
       return 1;
-   case PIPE_CAP_S3TC:
-      return 0;
    case PIPE_CAP_ANISOTROPIC_FILTER:
       return 0;
    case PIPE_CAP_POINT_SPRITE:
@@ -86,7 +82,7 @@ llvmpipe_get_param(struct pipe_screen *screen, int param)
    case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
       return 13; /* max 4Kx4K */
    case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
-      return 8;  /* max 128x128x128 */
+      return 9;  /* max 256x256x256 */
    case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
       return 13; /* max 4Kx4K */
    case PIPE_CAP_TGSI_CONT_SUPPORTED:
@@ -196,8 +192,7 @@ static void
 llvmpipe_destroy_screen( struct pipe_screen *_screen )
 {
    struct llvmpipe_screen *screen = llvmpipe_screen(_screen);
-
-   struct pipe_winsys *winsys = _screen->winsys;
+   struct llvmpipe_winsys *winsys = screen->winsys;
 
    lp_jit_screen_cleanup(screen);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c
index d145f6d6bb..c43b3da450 100644
--- a/src/gallium/drivers/llvmpipe/lp_setup.c
+++ b/src/gallium/drivers/llvmpipe/lp_setup.c
@@ -44,6 +44,7 @@
 #include "pipe/p_thread.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
+#include "lp_bld_debug.h"
 #include "lp_tile_cache.h"
 #include "lp_tile_soa.h"
 
@@ -114,6 +115,7 @@ struct setup_context {
 /**
  * Execute fragment shader for the four fragments in the quad.
  */
+ALIGN_STACK
 static void
 shade_quads(struct llvmpipe_context *llvmpipe,
             struct quad_header *quads[],
@@ -123,7 +125,7 @@ shade_quads(struct llvmpipe_context *llvmpipe,
    struct quad_header *quad = quads[0];
    const unsigned x = quad->input.x0;
    const unsigned y = quad->input.y0;
-   uint8_t *tile = lp_get_cached_tile(llvmpipe->cbuf_cache[0], x, y);
+   uint8_t *tile;
    uint8_t *color;
    void *depth;
    uint32_t ALIGN16_ATTRIB mask[4][NUM_CHANNELS];
@@ -149,7 +151,13 @@ shade_quads(struct llvmpipe_context *llvmpipe,
          mask[q][chan_index] = quads[q]->inout.mask & (1 << chan_index) ? ~0 : 0;
 
    /* color buffer */
-   color = &TILE_PIXEL(tile, x & (TILE_SIZE-1), y & (TILE_SIZE-1), 0);
+   if(llvmpipe->framebuffer.nr_cbufs >= 1 &&
+      llvmpipe->framebuffer.cbufs[0]) {
+      tile = lp_get_cached_tile(llvmpipe->cbuf_cache[0], x, y);
+      color = &TILE_PIXEL(tile, x & (TILE_SIZE-1), y & (TILE_SIZE-1), 0);
+   }
+   else
+      color = NULL;
 
    /* depth buffer */
    if(llvmpipe->zsbuf_map) {
@@ -162,12 +170,12 @@ shade_quads(struct llvmpipe_context *llvmpipe,
    else
       depth = NULL;
 
-   /* TODO: blend color */
+   /* XXX: This will most likely fail on 32bit x86 without -mstackrealign */
+   assert(lp_check_alignment(mask, 16));
 
-   assert((((uintptr_t)mask) & 0xf) == 0);
-   assert((((uintptr_t)depth) & 0xf) == 0);
-   assert((((uintptr_t)color) & 0xf) == 0);
-   assert((((uintptr_t)llvmpipe->jit_context.blend_color) & 0xf) == 0);
+   assert(lp_check_alignment(depth, 16));
+   assert(lp_check_alignment(color, 16));
+   assert(lp_check_alignment(llvmpipe->jit_context.blend_color, 16));
 
    /* run shader */
    fs->current->jit_function( &llvmpipe->jit_context,
@@ -270,11 +278,13 @@ clip_emit_quad( struct setup_context *setup, struct quad_header *quad )
        * until we codegenerate single-quad variants of the fragment pipeline
        * we need this hack. */
       const unsigned nr_quads = TILE_VECTOR_HEIGHT*TILE_VECTOR_WIDTH/QUAD_SIZE;
-      struct quad_header quads[nr_quads];
-      struct quad_header *quad_ptrs[nr_quads];
+      struct quad_header quads[4];
+      struct quad_header *quad_ptrs[4];
       int x0 = block_x(quad->input.x0);
       unsigned i;
 
+      assert(nr_quads == 4);
+
       for(i = 0; i < nr_quads; ++i) {
          int x = x0 + 2*i;
          if(x == quad->input.x0)
diff --git a/src/gallium/drivers/llvmpipe/lp_state.h b/src/gallium/drivers/llvmpipe/lp_state.h
index fb10329887..7b26ce61a3 100644
--- a/src/gallium/drivers/llvmpipe/lp_state.h
+++ b/src/gallium/drivers/llvmpipe/lp_state.h
@@ -36,6 +36,7 @@
 #include "pipe/p_state.h"
 #include "tgsi/tgsi_scan.h"
 #include "lp_jit.h"
+#include "lp_bld_sample.h" /* for struct lp_sampler_static_state */
 
 
 #define LP_NEW_VIEWPORT      0x1
@@ -57,16 +58,20 @@
 
 struct tgsi_sampler;
 struct vertex_info;
-
+struct pipe_context;
+struct llvmpipe_context;
 
 struct lp_fragment_shader;
 
 
 struct lp_fragment_shader_variant_key
 {
+   enum pipe_format zsbuf_format;
    struct pipe_depth_state depth;
    struct pipe_alpha_state alpha;
    struct pipe_blend_state blend;
+
+   struct lp_sampler_static_state sampler[PIPE_MAX_SAMPLERS];
 };
 
 
diff --git a/src/gallium/drivers/llvmpipe/lp_state_derived.c b/src/gallium/drivers/llvmpipe/lp_state_derived.c
index 6fbb057937..30fb41ea65 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_derived.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_derived.c
@@ -93,6 +93,23 @@ llvmpipe_get_vertex_info(struct llvmpipe_context *llvmpipe)
       vinfo->num_attribs = 0;
       for (i = 0; i < lpfs->info.num_inputs; i++) {
          int src;
+         enum interp_mode interp;
+
+         switch (lpfs->info.input_interpolate[i]) {
+         case TGSI_INTERPOLATE_CONSTANT:
+            interp = INTERP_CONSTANT;
+            break;
+         case TGSI_INTERPOLATE_LINEAR:
+            interp = INTERP_LINEAR;
+            break;
+         case TGSI_INTERPOLATE_PERSPECTIVE:
+            interp = INTERP_PERSPECTIVE;
+            break;
+         default:
+            assert(0);
+            interp = INTERP_LINEAR;
+         }
+
          switch (lpfs->info.input_semantic_name[i]) {
          case TGSI_SEMANTIC_POSITION:
             src = draw_find_vs_output(llvmpipe->draw,
@@ -108,7 +125,7 @@ llvmpipe_get_vertex_info(struct llvmpipe_context *llvmpipe)
 
          case TGSI_SEMANTIC_FOG:
             src = draw_find_vs_output(llvmpipe->draw, TGSI_SEMANTIC_FOG, 0);
-            draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
+            draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src);
             break;
 
          case TGSI_SEMANTIC_GENERIC:
@@ -116,7 +133,7 @@ llvmpipe_get_vertex_info(struct llvmpipe_context *llvmpipe)
             /* this includes texcoords and varying vars */
             src = draw_find_vs_output(llvmpipe->draw, TGSI_SEMANTIC_GENERIC,
                                       lpfs->info.input_semantic_index[i]);
-            draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
+            draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src);
             break;
 
          default:
@@ -250,7 +267,9 @@ void llvmpipe_update_derived( struct llvmpipe_context *llvmpipe )
 
    if (llvmpipe->dirty & (LP_NEW_FS |
                           LP_NEW_BLEND |
-                          LP_NEW_DEPTH_STENCIL_ALPHA))
+                          LP_NEW_DEPTH_STENCIL_ALPHA |
+                          LP_NEW_SAMPLER |
+                          LP_NEW_TEXTURE))
       llvmpipe_update_fs( llvmpipe );
 
 
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index 94170bd716..7728ba6076 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -83,8 +83,10 @@
 #include "lp_bld_debug.h"
 #include "lp_screen.h"
 #include "lp_context.h"
+#include "lp_buffer.h"
 #include "lp_state.h"
 #include "lp_quad.h"
+#include "lp_tex_sample.h"
 
 
 static const unsigned char quad_offset_x[4] = {0, 1, 0, 1};
@@ -130,21 +132,20 @@ generate_pos0(LLVMBuilderRef builder,
  * Generate the depth test.
  */
 static void
-generate_depth(struct llvmpipe_context *lp,
-               LLVMBuilderRef builder,
-               const struct pipe_depth_state *state,
-               union lp_type src_type,
+generate_depth(LLVMBuilderRef builder,
+               const struct lp_fragment_shader_variant_key *key,
+               struct lp_type src_type,
                struct lp_build_mask_context *mask,
                LLVMValueRef src,
                LLVMValueRef dst_ptr)
 {
    const struct util_format_description *format_desc;
-   union lp_type dst_type;
+   struct lp_type dst_type;
 
-   if(!lp->framebuffer.zsbuf)
+   if(!key->depth.enabled)
       return;
 
-   format_desc = util_format_description(lp->framebuffer.zsbuf->format);
+   format_desc = util_format_description(key->zsbuf_format);
    assert(format_desc);
 
    /* Pick the depth type. */
@@ -164,7 +165,7 @@ generate_depth(struct llvmpipe_context *lp,
 #endif
 
    lp_build_depth_test(builder,
-                       state,
+                       &key->depth,
                        dst_type,
                        format_desc,
                        mask,
@@ -173,107 +174,6 @@ generate_depth(struct llvmpipe_context *lp,
 }
 
 
-struct build_fetch_texel_context
-{
-   LLVMValueRef context_ptr;
-
-   LLVMValueRef samplers_ptr;
-
-   /** Coords/texels store */
-   LLVMValueRef store_ptr;
-};
-
-
-void PIPE_CDECL
-lp_fetch_texel_soa( struct tgsi_sampler **samplers,
-                    uint32_t unit,
-                    float *store )
-{
-   struct tgsi_sampler *sampler = samplers[unit];
-
-#if 0
-   uint j;
-
-   debug_printf("%s sampler: %p (%p) store: %p\n",
-                __FUNCTION__,
-                sampler, *sampler,
-                store );
-
-   debug_printf("lodbias %f\n", store[12]);
-
-   for (j = 0; j < 4; j++)
-      debug_printf("sample %d texcoord %f %f\n",
-                   j,
-                   store[0+j],
-                   store[4+j]);
-#endif
-
-   {
-      float rgba[NUM_CHANNELS][QUAD_SIZE];
-      sampler->get_samples(sampler,
-                           &store[0],
-                           &store[4],
-                           &store[8],
-                           0.0f, /*store[12],  lodbias */
-                           rgba);
-      memcpy(store, rgba, sizeof rgba);
-   }
-
-#if 0
-   for (j = 0; j < 4; j++)
-      debug_printf("sample %d result %f %f %f %f\n",
-                   j,
-                   store[0+j],
-                   store[4+j],
-                   store[8+j],
-                   store[12+j]);
-#endif
-}
-
-
-static void
-emit_fetch_texel( LLVMBuilderRef builder,
-                  void *context,
-                  unsigned unit,
-                  unsigned num_coords,
-                  const LLVMValueRef *coords,
-                  LLVMValueRef lodbias,
-                  LLVMValueRef *texel)
-{
-   struct build_fetch_texel_context *bld = context;
-   LLVMTypeRef vec_type = LLVMTypeOf(coords[0]);
-   LLVMValueRef args[3];
-   unsigned i;
-
-   if(!bld->samplers_ptr)
-      bld->samplers_ptr = lp_jit_context_samplers(builder, bld->context_ptr);
-
-   if(!bld->store_ptr)
-      bld->store_ptr = LLVMBuildArrayAlloca(builder,
-                                            vec_type,
-                                            LLVMConstInt(LLVMInt32Type(), 4, 0),
-                                            "texel_store");
-
-   for (i = 0; i < num_coords; i++) {
-      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
-      LLVMValueRef coord_ptr = LLVMBuildGEP(builder, bld->store_ptr, &index, 1, "");
-      LLVMBuildStore(builder, coords[i], coord_ptr);
-   }
-
-   args[0] = bld->samplers_ptr;
-   args[1] = LLVMConstInt(LLVMInt32Type(), unit, 0);
-   args[2] = bld->store_ptr;
-
-   lp_build_intrinsic(builder, "fetch_texel", LLVMVoidType(), args, 3);
-
-   for (i = 0; i < NUM_CHANNELS; ++i) {
-      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
-      LLVMValueRef texel_ptr = LLVMBuildGEP(builder, bld->store_ptr, &index, 1, "");
-      texel[i] = LLVMBuildLoad(builder, texel_ptr, "");
-   }
-}
-
-
 /**
  * Generate the fragment shader, depth/stencil test, and alpha tests.
  */
@@ -282,11 +182,11 @@ generate_fs(struct llvmpipe_context *lp,
             struct lp_fragment_shader *shader,
             const struct lp_fragment_shader_variant_key *key,
             LLVMBuilderRef builder,
-            union lp_type type,
+            struct lp_type type,
             LLVMValueRef context_ptr,
             unsigned i,
             const struct lp_build_interp_soa_context *interp,
-            struct build_fetch_texel_context *sampler,
+            struct lp_build_sampler_soa *sampler,
             LLVMValueRef *pmask,
             LLVMValueRef *color,
             LLVMValueRef depth_ptr)
@@ -298,6 +198,7 @@ generate_fs(struct llvmpipe_context *lp,
    LLVMValueRef consts_ptr;
    LLVMValueRef outputs[PIPE_MAX_SHADER_OUTPUTS][NUM_CHANNELS];
    LLVMValueRef z = interp->pos[2];
+   struct lp_build_flow_context *flow;
    struct lp_build_mask_context mask;
    boolean early_depth_test;
    unsigned attrib;
@@ -309,25 +210,35 @@ generate_fs(struct llvmpipe_context *lp,
 
    consts_ptr = lp_jit_context_constants(builder, context_ptr);
 
-   lp_build_mask_begin(&mask, builder, type, *pmask);
+   flow = lp_build_flow_create(builder);
+
+   memset(outputs, 0, sizeof outputs);
+
+   lp_build_flow_scope_begin(flow);
+
+   /* Declare the color and z variables */
+   for(chan = 0; chan < NUM_CHANNELS; ++chan) {
+      color[chan] = LLVMGetUndef(vec_type);
+      lp_build_flow_scope_declare(flow, &color[chan]);
+   }
+   lp_build_flow_scope_declare(flow, &z);
+
+   lp_build_mask_begin(&mask, flow, type, *pmask);
 
    early_depth_test =
-      lp->depth_stencil->depth.enabled &&
-      lp->framebuffer.zsbuf &&
-      !lp->depth_stencil->alpha.enabled &&
-      !lp->fs->info.uses_kill &&
-      !lp->fs->info.writes_z;
+      key->depth.enabled &&
+      !key->alpha.enabled &&
+      !shader->info.uses_kill &&
+      !shader->info.writes_z;
 
    if(early_depth_test)
-      generate_depth(lp, builder, &key->depth,
+      generate_depth(builder, key,
                      type, &mask,
                      z, depth_ptr);
 
-   memset(outputs, 0, sizeof outputs);
-
    lp_build_tgsi_soa(builder, tokens, type, &mask,
                      consts_ptr, interp->pos, interp->inputs,
-                     outputs, emit_fetch_texel, sampler);
+                     outputs, sampler);
 
    for (attrib = 0; attrib < shader->info.num_outputs; ++attrib) {
       for(chan = 0; chan < NUM_CHANNELS; ++chan) {
@@ -368,12 +279,16 @@ generate_fs(struct llvmpipe_context *lp,
    }
 
    if(!early_depth_test)
-      generate_depth(lp, builder, &key->depth,
+      generate_depth(builder, key,
                      type, &mask,
                      z, depth_ptr);
 
    lp_build_mask_end(&mask);
 
+   lp_build_flow_scope_end(flow);
+
+   lp_build_flow_destroy(flow);
+
    *pmask = mask.value;
 
 }
@@ -385,13 +300,15 @@ generate_fs(struct llvmpipe_context *lp,
 static void
 generate_blend(const struct pipe_blend_state *blend,
                LLVMBuilderRef builder,
-               union lp_type type,
+               struct lp_type type,
                LLVMValueRef context_ptr,
                LLVMValueRef mask,
                LLVMValueRef *src,
                LLVMValueRef dst_ptr)
 {
    struct lp_build_context bld;
+   struct lp_build_flow_context *flow;
+   struct lp_build_mask_context mask_ctx;
    LLVMTypeRef vec_type;
    LLVMTypeRef int_vec_type;
    LLVMValueRef const_ptr;
@@ -400,11 +317,14 @@ generate_blend(const struct pipe_blend_state *blend,
    LLVMValueRef res[4];
    unsigned chan;
 
+   lp_build_context_init(&bld, builder, type);
+
+   flow = lp_build_flow_create(builder);
+   lp_build_mask_begin(&mask_ctx, flow, type, mask);
+
    vec_type = lp_build_vec_type(type);
    int_vec_type = lp_build_int_vec_type(type);
 
-   lp_build_context_init(&bld, builder, type);
-
    const_ptr = lp_jit_context_blend_color(builder, context_ptr);
    const_ptr = LLVMBuildBitCast(builder, const_ptr,
                                 LLVMPointerType(vec_type, 0), "");
@@ -422,11 +342,16 @@ generate_blend(const struct pipe_blend_state *blend,
    lp_build_blend_soa(builder, blend, type, src, dst, con, res);
 
    for(chan = 0; chan < 4; ++chan) {
-      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), chan, 0);
-      lp_build_name(res[chan], "res.%c", "rgba"[chan]);
-      res[chan] = lp_build_select(&bld, mask, res[chan], dst[chan]);
-      LLVMBuildStore(builder, res[chan], LLVMBuildGEP(builder, dst_ptr, &index, 1, ""));
+      if(blend->colormask & (1 << chan)) {
+         LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), chan, 0);
+         lp_build_name(res[chan], "res.%c", "rgba"[chan]);
+         res[chan] = lp_build_select(&bld, mask, res[chan], dst[chan]);
+         LLVMBuildStore(builder, res[chan], LLVMBuildGEP(builder, dst_ptr, &index, 1, ""));
+      }
    }
+
+   lp_build_mask_end(&mask_ctx);
+   lp_build_flow_destroy(flow);
 }
 
 
@@ -440,8 +365,8 @@ generate_fragment(struct llvmpipe_context *lp,
 {
    struct llvmpipe_screen *screen = llvmpipe_screen(lp->pipe.screen);
    struct lp_fragment_shader_variant *variant;
-   union lp_type fs_type;
-   union lp_type blend_type;
+   struct lp_type fs_type;
+   struct lp_type blend_type;
    LLVMTypeRef fs_elem_type;
    LLVMTypeRef fs_vec_type;
    LLVMTypeRef fs_int_vec_type;
@@ -462,7 +387,7 @@ generate_fragment(struct llvmpipe_context *lp,
    LLVMBuilderRef builder;
    LLVMValueRef x0;
    LLVMValueRef y0;
-   struct build_fetch_texel_context sampler;
+   struct lp_build_sampler_soa *sampler;
    struct lp_build_interp_soa_context interp;
    LLVMValueRef fs_mask[LP_MAX_VECTOR_LENGTH];
    LLVMValueRef fs_out_color[NUM_CHANNELS][LP_MAX_VECTOR_LENGTH];
@@ -477,7 +402,6 @@ generate_fragment(struct llvmpipe_context *lp,
    if(key->depth.enabled) {
       debug_printf("depth.func = %s\n", debug_dump_func(key->depth.func, TRUE));
       debug_printf("depth.writemask = %u\n", key->depth.writemask);
-      debug_printf("depth.occlusion_count = %u\n", key->depth.occlusion_count);
    }
    if(key->alpha.enabled) {
       debug_printf("alpha.func = %s\n", debug_dump_func(key->alpha.func, TRUE));
@@ -507,7 +431,7 @@ generate_fragment(struct llvmpipe_context *lp,
    /* TODO: actually pick these based on the fs and color buffer
     * characteristics. */
 
-   fs_type.value = 0;
+   memset(&fs_type, 0, sizeof fs_type);
    fs_type.floating = TRUE; /* floating point values */
    fs_type.sign = TRUE;     /* values are signed */
    fs_type.norm = FALSE;    /* values are not limited to [0,1] or [-1,1] */
@@ -515,7 +439,7 @@ generate_fragment(struct llvmpipe_context *lp,
    fs_type.length = 4;      /* 4 element per vector */
    num_fs = 4;
 
-   blend_type.value = 0;
+   memset(&blend_type, 0, sizeof blend_type);
    blend_type.floating = FALSE; /* values are integers */
    blend_type.sign = FALSE;     /* values are unsigned */
    blend_type.norm = TRUE;      /* values are in [0,1] or [-1,1] */
@@ -586,8 +510,13 @@ generate_fragment(struct llvmpipe_context *lp,
                             a0_ptr, dadx_ptr, dady_ptr,
                             x0, y0, 2, 0);
 
-   memset(&sampler, 0, sizeof sampler);
-   sampler.context_ptr = context_ptr;
+#if 0
+   /* C texture sampling */
+   sampler = lp_c_sampler_soa_create(context_ptr);
+#else
+   /* code generated texture sampling */
+   sampler = lp_llvm_sampler_soa_create(key->sampler, context_ptr);
+#endif
 
    for(i = 0; i < num_fs; ++i) {
       LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
@@ -606,7 +535,7 @@ generate_fragment(struct llvmpipe_context *lp,
                   context_ptr,
                   i,
                   &interp,
-                  &sampler,
+                  sampler,
                   &fs_mask[i],
                   out_color,
                   depth_ptr_i);
@@ -615,6 +544,8 @@ generate_fragment(struct llvmpipe_context *lp,
          fs_out_color[chan][i] = out_color[chan];
    }
 
+   sampler->destroy(sampler);
+
    /* 
     * Convert the fs's output color and mask to fit to the blending type. 
     */
@@ -651,6 +582,11 @@ generate_fragment(struct llvmpipe_context *lp,
     * Translate the LLVM IR into machine code.
     */
 
+   if(LLVMVerifyFunction(variant->function, LLVMPrintMessageAction)) {
+      LLVMDumpValue(variant->function);
+      abort();
+   }
+
    LLVMRunFunctionPassManager(screen->pass, variant->function);
 
 #ifdef DEBUG
@@ -658,11 +594,6 @@ generate_fragment(struct llvmpipe_context *lp,
    debug_printf("\n");
 #endif
 
-   if(LLVMVerifyFunction(variant->function, LLVMPrintMessageAction)) {
-      LLVMDumpValue(variant->function);
-      abort();
-   }
-
    variant->jit_function = (lp_jit_frag_func)LLVMGetPointerToGlobal(screen->engine, variant->function);
 
 #ifdef DEBUG
@@ -741,16 +672,29 @@ llvmpipe_delete_fs_state(struct pipe_context *pipe, void *fs)
 void
 llvmpipe_set_constant_buffer(struct pipe_context *pipe,
                              uint shader, uint index,
-                             const struct pipe_constant_buffer *buf)
+                             const struct pipe_constant_buffer *constants)
 {
    struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);
+   struct pipe_buffer *buffer = constants ? constants->buffer : NULL;
+   unsigned size = buffer ? buffer->size : 0;
+   const void *data = buffer ? llvmpipe_buffer(buffer)->data : NULL;
 
    assert(shader < PIPE_SHADER_TYPES);
    assert(index == 0);
 
+   if(shader == PIPE_SHADER_VERTEX)
+      draw_flush(llvmpipe->draw);
+
    /* note: reference counting */
-   pipe_buffer_reference(&llvmpipe->constants[shader].buffer,
-			 buf ? buf->buffer : NULL);
+   pipe_buffer_reference(&llvmpipe->constants[shader].buffer, buffer);
+
+   if(shader == PIPE_SHADER_FRAGMENT) {
+      llvmpipe->jit_context.constants = data;
+   }
+
+   if(shader == PIPE_SHADER_VERTEX) {
+      draw_set_mapped_constant_buffer(llvmpipe->draw, data, size);
+   }
 
    llvmpipe->dirty |= LP_NEW_CONSTANTS;
 }
@@ -765,18 +709,45 @@ llvmpipe_set_constant_buffer(struct pipe_context *pipe,
  */
 static void
 make_variant_key(struct llvmpipe_context *lp,
+                 struct lp_fragment_shader *shader,
                  struct lp_fragment_shader_variant_key *key)
 {
+   unsigned i;
+
    memset(key, 0, sizeof *key);
 
-   memcpy(&key->depth, &lp->depth_stencil->depth, sizeof key->depth);
+   if(lp->framebuffer.zsbuf &&
+      lp->depth_stencil->depth.enabled) {
+      key->zsbuf_format = lp->framebuffer.zsbuf->format;
+      memcpy(&key->depth, &lp->depth_stencil->depth, sizeof key->depth);
+   }
 
    key->alpha.enabled = lp->depth_stencil->alpha.enabled;
    if(key->alpha.enabled)
       key->alpha.func = lp->depth_stencil->alpha.func;
    /* alpha.ref_value is passed in jit_context */
 
-   memcpy(&key->blend, lp->blend, sizeof key->blend);
+   if(lp->framebuffer.cbufs[0]) {
+      const struct util_format_description *format_desc;
+      unsigned chan;
+
+      memcpy(&key->blend, lp->blend, sizeof key->blend);
+
+      format_desc = util_format_description(lp->framebuffer.cbufs[0]->format);
+      assert(format_desc->layout == UTIL_FORMAT_COLORSPACE_RGB ||
+             format_desc->layout == UTIL_FORMAT_COLORSPACE_SRGB);
+
+      /* mask out color channels not present in the color buffer */
+      for(chan = 0; chan < 4; ++chan) {
+         enum util_format_swizzle swizzle = format_desc->swizzle[chan];
+         if(swizzle > 4)
+            key->blend.colormask &= ~(1 << chan);
+      }
+   }
+
+   for(i = 0; i < PIPE_MAX_SAMPLERS; ++i)
+      if(shader->info.file_mask[TGSI_FILE_SAMPLER] & (1 << i))
+         lp_sampler_static_state(&key->sampler[i], lp->texture[i], lp->sampler[i]);
 }
 
 
@@ -787,7 +758,7 @@ llvmpipe_update_fs(struct llvmpipe_context *lp)
    struct lp_fragment_shader_variant_key key;
    struct lp_fragment_shader_variant *variant;
 
-   make_variant_key(lp, &key);
+   make_variant_key(lp, shader, &key);
 
    variant = shader->variants;
    while(variant) {
diff --git a/src/gallium/drivers/llvmpipe/lp_state_sampler.c b/src/gallium/drivers/llvmpipe/lp_state_sampler.c
index 4fef541b1e..c69d90c723 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_sampler.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_sampler.c
@@ -98,6 +98,16 @@ llvmpipe_set_sampler_textures(struct pipe_context *pipe,
 
       pipe_texture_reference(&llvmpipe->texture[i], tex);
       lp_tex_tile_cache_set_texture(llvmpipe->tex_cache[i], tex);
+
+      if(tex) {
+         struct llvmpipe_texture *lp_tex = llvmpipe_texture(tex);
+         struct lp_jit_texture *jit_tex = &llvmpipe->jit_context.textures[i];
+         jit_tex->width = tex->width[0];
+         jit_tex->height = tex->height[0];
+         jit_tex->stride = lp_tex->stride[0];
+         if(!lp_tex->dt)
+            jit_tex->data = lp_tex->data;
+      }
    }
 
    llvmpipe->num_textures = num;
diff --git a/src/gallium/drivers/llvmpipe/lp_state_surface.c b/src/gallium/drivers/llvmpipe/lp_state_surface.c
index 177a26b7b1..c06ce8b75c 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_surface.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_surface.c
@@ -53,10 +53,11 @@ llvmpipe_set_framebuffer_state(struct pipe_context *pipe,
       /* check if changing cbuf */
       if (lp->framebuffer.cbufs[i] != fb->cbufs[i]) {
          /* flush old */
+         lp_tile_cache_map_transfers(lp->cbuf_cache[i]);
          lp_flush_tile_cache(lp->cbuf_cache[i]);
 
          /* assign new */
-         lp->framebuffer.cbufs[i] = fb->cbufs[i];
+         pipe_surface_reference(&lp->framebuffer.cbufs[i], fb->cbufs[i]);
 
          /* update cache */
          lp_tile_cache_set_surface(lp->cbuf_cache[i], fb->cbufs[i]);
@@ -81,7 +82,7 @@ llvmpipe_set_framebuffer_state(struct pipe_context *pipe,
       }
 
       /* assign new */
-      lp->framebuffer.zsbuf = fb->zsbuf;
+      pipe_surface_reference(&lp->framebuffer.zsbuf, fb->zsbuf);
 
       /* Tell draw module how deep the Z/depth buffer is */
       if (lp->framebuffer.zsbuf) {
diff --git a/src/gallium/drivers/llvmpipe/lp_test.h b/src/gallium/drivers/llvmpipe/lp_test.h
index 69aaae26e0..39d80726e6 100644
--- a/src/gallium/drivers/llvmpipe/lp_test.h
+++ b/src/gallium/drivers/llvmpipe/lp_test.h
@@ -56,6 +56,9 @@
 #include "lp_bld_type.h"
 
 
+#define LP_TEST_NUM_SAMPLES 32
+
+
 void
 write_tsv_header(FILE *fp);
 
@@ -68,17 +71,28 @@ boolean
 test_all(unsigned verbose, FILE *fp);
 
 
+#if defined(PIPE_CC_MSVC)
+
+unsigned __int64 __rdtsc();
+#pragma intrinsic(__rdtsc)
+#define rdtsc() __rdtsc()
+
+#elif defined(PIPE_CC_GCC) && (defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64))
+
 static INLINE uint64_t
 rdtsc(void)
 {
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
    uint32_t hi, lo;
    __asm__ __volatile__ ("rdtsc" : "=a"(lo), "=d"(hi));
    return ((uint64_t)lo) | (((uint64_t)hi) << 32);
+}
+
 #else
-   return 0;
+
+#define rdtsc() 0
+
 #endif
-}
+
 
 
 float
@@ -86,43 +100,43 @@ random_float(void);
 
 
 void
-dump_type(FILE *fp, union lp_type type);
+dump_type(FILE *fp, struct lp_type type);
 
 
 double
-read_elem(union lp_type type, const void *src, unsigned index);
+read_elem(struct lp_type type, const void *src, unsigned index);
 
 
 void
-write_elem(union lp_type type, void *dst, unsigned index, double src);
+write_elem(struct lp_type type, void *dst, unsigned index, double src);
 
 
 void
-random_elem(union lp_type type, void *dst, unsigned index);
+random_elem(struct lp_type type, void *dst, unsigned index);
 
 
 void
-read_vec(union lp_type type, const void *src, double *dst);
+read_vec(struct lp_type type, const void *src, double *dst);
 
 
 void
-write_vec(union lp_type type, void *dst, const double *src);
+write_vec(struct lp_type type, void *dst, const double *src);
 
 
 void
-random_vec(union lp_type type, void *dst);
+random_vec(struct lp_type type, void *dst);
 
 
 boolean
-compare_vec_with_eps(union lp_type type, const void *res, const void *ref, double eps);
+compare_vec_with_eps(struct lp_type type, const void *res, const void *ref, double eps);
 
 
 boolean
-compare_vec(union lp_type type, const void *res, const void *ref);
+compare_vec(struct lp_type type, const void *res, const void *ref);
 
 
 void
-dump_vec(FILE *fp, union lp_type type, const void *src);
+dump_vec(FILE *fp, struct lp_type type, const void *src);
 
 
 #endif /* !LP_TEST_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_test_blend.c b/src/gallium/drivers/llvmpipe/lp_test_blend.c
index 8dfad468e3..e3af81cffb 100644
--- a/src/gallium/drivers/llvmpipe/lp_test_blend.c
+++ b/src/gallium/drivers/llvmpipe/lp_test_blend.c
@@ -80,7 +80,7 @@ static void
 write_tsv_row(FILE *fp,
               const struct pipe_blend_state *blend,
               enum vector_mode mode,
-              union lp_type type,
+              struct lp_type type,
               double cycles,
               boolean success)
 {
@@ -125,7 +125,7 @@ static void
 dump_blend_type(FILE *fp,
                 const struct pipe_blend_state *blend,
                 enum vector_mode mode,
-                union lp_type type)
+                struct lp_type type)
 {
    fprintf(fp, "%s", mode ? "soa" : "aos");
 
@@ -153,7 +153,7 @@ static LLVMValueRef
 add_blend_test(LLVMModuleRef module,
                const struct pipe_blend_state *blend,
                enum vector_mode mode,
-               union lp_type type)
+               struct lp_type type)
 {
    LLVMTypeRef ret_type;
    LLVMTypeRef vec_type;
@@ -467,7 +467,7 @@ test_one(unsigned verbose,
          FILE *fp,
          const struct pipe_blend_state *blend,
          enum vector_mode mode,
-         union lp_type type)
+         struct lp_type type)
 {
    LLVMModuleRef module = NULL;
    LLVMValueRef func = NULL;
@@ -477,8 +477,8 @@ test_one(unsigned verbose,
    char *error = NULL;
    blend_test_ptr_t blend_test_ptr;
    boolean success;
-   const unsigned n = 32;
-   int64_t cycles[n];
+   const unsigned n = LP_TEST_NUM_SAMPLES;
+   int64_t cycles[LP_TEST_NUM_SAMPLES];
    double cycles_avg = 0.0;
    unsigned i, j;
 
@@ -765,10 +765,10 @@ blend_funcs[] = {
 };
 
 
-const union lp_type blend_types[] = {
+const struct lp_type blend_types[] = {
    /* float, fixed,  sign,  norm, width, len */
-   {{  TRUE, FALSE, FALSE,  TRUE,    32,   4 }}, /* f32 x 4 */
-   {{ FALSE, FALSE, FALSE,  TRUE,     8,  16 }}, /* u8n x 16 */
+   {   TRUE, FALSE, FALSE,  TRUE,    32,   4 }, /* f32 x 4 */
+   {  FALSE, FALSE, FALSE,  TRUE,     8,  16 }, /* u8n x 16 */
 };
 
 
@@ -788,7 +788,7 @@ test_all(unsigned verbose, FILE *fp)
    const unsigned *alpha_dst_factor;
    struct pipe_blend_state blend;
    enum vector_mode mode;
-   const union lp_type *type;
+   const struct lp_type *type;
    bool success = TRUE;
 
    for(rgb_func = blend_funcs; rgb_func < &blend_funcs[num_funcs]; ++rgb_func) {
@@ -841,27 +841,27 @@ test_some(unsigned verbose, FILE *fp, unsigned long n)
    const unsigned *alpha_dst_factor;
    struct pipe_blend_state blend;
    enum vector_mode mode;
-   const union lp_type *type;
+   const struct lp_type *type;
    unsigned long i;
    bool success = TRUE;
 
    for(i = 0; i < n; ++i) {
-      rgb_func = &blend_funcs[random() % num_funcs];
-      alpha_func = &blend_funcs[random() % num_funcs];
-      rgb_src_factor = &blend_factors[random() % num_factors];
-      alpha_src_factor = &blend_factors[random() % num_factors];
+      rgb_func = &blend_funcs[rand() % num_funcs];
+      alpha_func = &blend_funcs[rand() % num_funcs];
+      rgb_src_factor = &blend_factors[rand() % num_factors];
+      alpha_src_factor = &blend_factors[rand() % num_factors];
       
       do {
-         rgb_dst_factor = &blend_factors[random() % num_factors];
+         rgb_dst_factor = &blend_factors[rand() % num_factors];
       } while(*rgb_dst_factor == PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE);
 
       do {
-         alpha_dst_factor = &blend_factors[random() % num_factors];
+         alpha_dst_factor = &blend_factors[rand() % num_factors];
       } while(*alpha_dst_factor == PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE);
 
-      mode = random() & 1;
+      mode = rand() & 1;
 
-      type = &blend_types[random() % num_types];
+      type = &blend_types[rand() % num_types];
 
       memset(&blend, 0, sizeof blend);
       blend.blend_enable      = 1;
diff --git a/src/gallium/drivers/llvmpipe/lp_test_conv.c b/src/gallium/drivers/llvmpipe/lp_test_conv.c
index e6489834af..ac2a6d05e3 100644
--- a/src/gallium/drivers/llvmpipe/lp_test_conv.c
+++ b/src/gallium/drivers/llvmpipe/lp_test_conv.c
@@ -59,8 +59,8 @@ write_tsv_header(FILE *fp)
 
 static void
 write_tsv_row(FILE *fp,
-              union lp_type src_type,
-              union lp_type dst_type,
+              struct lp_type src_type,
+              struct lp_type dst_type,
               double cycles,
               boolean success)
 {
@@ -80,8 +80,8 @@ write_tsv_row(FILE *fp,
 
 static void
 dump_conv_types(FILE *fp,
-               union lp_type src_type,
-               union lp_type dst_type)
+               struct lp_type src_type,
+               struct lp_type dst_type)
 {
    fprintf(fp, "src_type=");
    dump_type(fp, src_type);
@@ -96,8 +96,8 @@ dump_conv_types(FILE *fp,
 
 static LLVMValueRef
 add_conv_test(LLVMModuleRef module,
-              union lp_type src_type, unsigned num_srcs,
-              union lp_type dst_type, unsigned num_dsts)
+              struct lp_type src_type, unsigned num_srcs,
+              struct lp_type dst_type, unsigned num_dsts)
 {
    LLVMTypeRef args[2];
    LLVMValueRef func;
@@ -145,8 +145,8 @@ add_conv_test(LLVMModuleRef module,
 static boolean
 test_one(unsigned verbose,
          FILE *fp,
-         union lp_type src_type,
-         union lp_type dst_type)
+         struct lp_type src_type,
+         struct lp_type dst_type)
 {
    LLVMModuleRef module = NULL;
    LLVMValueRef func = NULL;
@@ -156,8 +156,8 @@ test_one(unsigned verbose,
    char *error = NULL;
    conv_test_ptr_t conv_test_ptr;
    boolean success;
-   const unsigned n = 32;
-   int64_t cycles[n];
+   const unsigned n = LP_TEST_NUM_SAMPLES;
+   int64_t cycles[LP_TEST_NUM_SAMPLES];
    double cycles_avg = 0.0;
    unsigned num_srcs;
    unsigned num_dsts;
@@ -343,35 +343,35 @@ test_one(unsigned verbose,
 }
 
 
-const union lp_type conv_types[] = {
+const struct lp_type conv_types[] = {
    /* float, fixed,  sign,  norm, width, len */
 
-   {{  TRUE, FALSE,  TRUE,  TRUE,    32,   4 }},
-   {{  TRUE, FALSE,  TRUE, FALSE,    32,   4 }},
-   {{  TRUE, FALSE, FALSE,  TRUE,    32,   4 }},
-   {{  TRUE, FALSE, FALSE, FALSE,    32,   4 }},
+   {   TRUE, FALSE,  TRUE,  TRUE,    32,   4 },
+   {   TRUE, FALSE,  TRUE, FALSE,    32,   4 },
+   {   TRUE, FALSE, FALSE,  TRUE,    32,   4 },
+   {   TRUE, FALSE, FALSE, FALSE,    32,   4 },
 
    /* TODO: test fixed formats too */
 
-   {{ FALSE, FALSE,  TRUE,  TRUE,    16,   8 }},
-   {{ FALSE, FALSE,  TRUE, FALSE,    16,   8 }},
-   {{ FALSE, FALSE, FALSE,  TRUE,    16,   8 }},
-   {{ FALSE, FALSE, FALSE, FALSE,    16,   8 }},
-
-   {{ FALSE, FALSE,  TRUE,  TRUE,    32,   4 }},
-   {{ FALSE, FALSE,  TRUE, FALSE,    32,   4 }},
-   {{ FALSE, FALSE, FALSE,  TRUE,    32,   4 }},
-   {{ FALSE, FALSE, FALSE, FALSE,    32,   4 }},
-
-   {{ FALSE, FALSE,  TRUE,  TRUE,    16,   8 }},
-   {{ FALSE, FALSE,  TRUE, FALSE,    16,   8 }},
-   {{ FALSE, FALSE, FALSE,  TRUE,    16,   8 }},
-   {{ FALSE, FALSE, FALSE, FALSE,    16,   8 }},
-
-   {{ FALSE, FALSE,  TRUE,  TRUE,     8,  16 }},
-   {{ FALSE, FALSE,  TRUE, FALSE,     8,  16 }},
-   {{ FALSE, FALSE, FALSE,  TRUE,     8,  16 }},
-   {{ FALSE, FALSE, FALSE, FALSE,     8,  16 }},
+   {  FALSE, FALSE,  TRUE,  TRUE,    16,   8 },
+   {  FALSE, FALSE,  TRUE, FALSE,    16,   8 },
+   {  FALSE, FALSE, FALSE,  TRUE,    16,   8 },
+   {  FALSE, FALSE, FALSE, FALSE,    16,   8 },
+
+   {  FALSE, FALSE,  TRUE,  TRUE,    32,   4 },
+   {  FALSE, FALSE,  TRUE, FALSE,    32,   4 },
+   {  FALSE, FALSE, FALSE,  TRUE,    32,   4 },
+   {  FALSE, FALSE, FALSE, FALSE,    32,   4 },
+
+   {  FALSE, FALSE,  TRUE,  TRUE,    16,   8 },
+   {  FALSE, FALSE,  TRUE, FALSE,    16,   8 },
+   {  FALSE, FALSE, FALSE,  TRUE,    16,   8 },
+   {  FALSE, FALSE, FALSE, FALSE,    16,   8 },
+
+   {  FALSE, FALSE,  TRUE,  TRUE,     8,  16 },
+   {  FALSE, FALSE,  TRUE, FALSE,     8,  16 },
+   {  FALSE, FALSE, FALSE,  TRUE,     8,  16 },
+   {  FALSE, FALSE, FALSE, FALSE,     8,  16 },
 };
 
 
@@ -381,8 +381,8 @@ const unsigned num_types = sizeof(conv_types)/sizeof(conv_types[0]);
 boolean
 test_all(unsigned verbose, FILE *fp)
 {
-   const union lp_type *src_type;
-   const union lp_type *dst_type;
+   const struct lp_type *src_type;
+   const struct lp_type *dst_type;
    bool success = TRUE;
 
    for(src_type = conv_types; src_type < &conv_types[num_types]; ++src_type) {
@@ -407,16 +407,16 @@ test_all(unsigned verbose, FILE *fp)
 boolean
 test_some(unsigned verbose, FILE *fp, unsigned long n)
 {
-   const union lp_type *src_type;
-   const union lp_type *dst_type;
+   const struct lp_type *src_type;
+   const struct lp_type *dst_type;
    unsigned long i;
    bool success = TRUE;
 
    for(i = 0; i < n; ++i) {
-      src_type = &conv_types[random() % num_types];
+      src_type = &conv_types[rand() % num_types];
       
       do {
-         dst_type = &conv_types[random() % num_types];
+         dst_type = &conv_types[rand() % num_types];
       } while (src_type == dst_type || src_type->norm != dst_type->norm);
 
       if(!test_one(verbose, fp, *src_type, *dst_type))
diff --git a/src/gallium/drivers/llvmpipe/lp_test_format.c b/src/gallium/drivers/llvmpipe/lp_test_format.c
index 1d192355ee..5dc8297fe9 100644
--- a/src/gallium/drivers/llvmpipe/lp_test_format.c
+++ b/src/gallium/drivers/llvmpipe/lp_test_format.c
@@ -35,9 +35,9 @@
 #include <llvm-c/Target.h>
 #include <llvm-c/Transforms/Scalar.h>
 
+#include "util/u_cpu_detect.h"
 #include "util/u_format.h"
 
-#include "lp_bld_flow.h"
 #include "lp_bld_format.h"
 
 
@@ -103,7 +103,6 @@ add_load_rgba_test(LLVMModuleRef module,
    LLVMBasicBlockRef block;
    LLVMBuilderRef builder;
    LLVMValueRef rgba;
-   struct lp_build_loop_state loop;
 
    args[0] = LLVMPointerType(LLVMInt8Type(), 0);
    args[1] = LLVMPointerType(LLVMVectorType(LLVMFloatType(), 4), 0);
@@ -117,13 +116,9 @@ add_load_rgba_test(LLVMModuleRef module,
    builder = LLVMCreateBuilder();
    LLVMPositionBuilderAtEnd(builder, block);
 
-   lp_build_loop_begin(builder, LLVMConstInt(LLVMInt32Type(), 1, 0), &loop);
-
-   rgba = lp_build_load_rgba(builder, format, ptr);
+   rgba = lp_build_load_rgba_aos(builder, format, ptr);
    LLVMBuildStore(builder, rgba, rgba_ptr);
 
-   lp_build_loop_end(builder, LLVMConstInt(LLVMInt32Type(), 4, 0), NULL, &loop);
-
    LLVMBuildRetVoid(builder);
 
    LLVMDisposeBuilder(builder);
@@ -160,7 +155,7 @@ add_store_rgba_test(LLVMModuleRef module,
 
    rgba = LLVMBuildLoad(builder, rgba_ptr, "");
 
-   lp_build_store_rgba(builder, format, ptr, rgba);
+   lp_build_store_rgba_aos(builder, format, ptr, rgba);
 
    LLVMBuildRetVoid(builder);
 
@@ -264,6 +259,13 @@ int main(int argc, char **argv)
    unsigned i;
    int ret;
 
+#ifdef LLVM_NATIVE_ARCH
+   LLVMLinkInJIT();
+   LLVMInitializeNativeTarget();
+#endif
+
+   util_cpu_detect();
+
    for (i = 0; i < sizeof(test_cases)/sizeof(test_cases[0]); ++i)
       if(!test_format(&test_cases[i]))
         ret = 1;
diff --git a/src/gallium/drivers/llvmpipe/lp_test_main.c b/src/gallium/drivers/llvmpipe/lp_test_main.c
index 49213fb4f0..d4767ff52b 100644
--- a/src/gallium/drivers/llvmpipe/lp_test_main.c
+++ b/src/gallium/drivers/llvmpipe/lp_test_main.c
@@ -34,13 +34,15 @@
  */
 
 
+#include "util/u_cpu_detect.h"
+
 #include "lp_bld_const.h"
 #include "lp_test.h"
 
 
 void
 dump_type(FILE *fp,
-          union lp_type type)
+          struct lp_type type)
 {
    fprintf(fp, "%s%s%u%sx%u",
            type.sign ? (type.floating || type.fixed ? "" : "s") : "u",
@@ -52,7 +54,7 @@ dump_type(FILE *fp,
 
 
 double
-read_elem(union lp_type type, const void *src, unsigned index)
+read_elem(struct lp_type type, const void *src, unsigned index)
 {
    double scale = lp_const_scale(type);
    double value;
@@ -115,7 +117,7 @@ read_elem(union lp_type type, const void *src, unsigned index)
 
 
 void
-write_elem(union lp_type type, void *dst, unsigned index, double value)
+write_elem(struct lp_type type, void *dst, unsigned index, double value)
 {
    assert(index < type.length);
    if(!type.sign && value < 0.0)
@@ -184,11 +186,11 @@ write_elem(union lp_type type, void *dst, unsigned index, double value)
 
 
 void
-random_elem(union lp_type type, void *dst, unsigned index)
+random_elem(struct lp_type type, void *dst, unsigned index)
 {
    double value;
    assert(index < type.length);
-   value = (double)random()/(double)RAND_MAX;
+   value = (double)rand()/(double)RAND_MAX;
    if(!type.norm) {
       unsigned long long mask;
       if (type.floating)
@@ -199,17 +201,17 @@ random_elem(union lp_type type, void *dst, unsigned index)
          mask = ((unsigned long long)1 << (type.width - 1)) - 1;
       else
          mask = ((unsigned long long)1 << type.width) - 1;
-      value += (double)(mask & random());
+      value += (double)(mask & rand());
    }
    if(!type.sign)
-      if(random() & 1)
+      if(rand() & 1)
          value = -value;
    write_elem(type, dst, index, value);
 }
 
 
 void
-read_vec(union lp_type type, const void *src, double *dst)
+read_vec(struct lp_type type, const void *src, double *dst)
 {
    unsigned i;
    for (i = 0; i < type.length; ++i)
@@ -218,7 +220,7 @@ read_vec(union lp_type type, const void *src, double *dst)
 
 
 void
-write_vec(union lp_type type, void *dst, const double *src)
+write_vec(struct lp_type type, void *dst, const double *src)
 {
    unsigned i;
    for (i = 0; i < type.length; ++i)
@@ -229,12 +231,12 @@ write_vec(union lp_type type, void *dst, const double *src)
 float
 random_float(void)
 {
-    return (float)((double)random()/(double)RAND_MAX);
+    return (float)((double)rand()/(double)RAND_MAX);
 }
 
 
 void
-random_vec(union lp_type type, void *dst)
+random_vec(struct lp_type type, void *dst)
 {
    unsigned i;
    for (i = 0; i < type.length; ++i)
@@ -243,7 +245,7 @@ random_vec(union lp_type type, void *dst)
 
 
 boolean
-compare_vec_with_eps(union lp_type type, const void *res, const void *ref, double eps)
+compare_vec_with_eps(struct lp_type type, const void *res, const void *ref, double eps)
 {
    unsigned i;
    for (i = 0; i < type.length; ++i) {
@@ -259,7 +261,7 @@ compare_vec_with_eps(union lp_type type, const void *res, const void *ref, doubl
 
 
 boolean
-compare_vec(union lp_type type, const void *res, const void *ref)
+compare_vec(struct lp_type type, const void *res, const void *ref)
 {
    double eps = lp_const_eps(type);
    return compare_vec_with_eps(type, res, ref, eps);
@@ -267,7 +269,7 @@ compare_vec(union lp_type type, const void *res, const void *ref)
 
 
 void
-dump_vec(FILE *fp, union lp_type type, const void *src)
+dump_vec(FILE *fp, struct lp_type type, const void *src)
 {
    unsigned i;
    for (i = 0; i < type.length; ++i) {
@@ -365,6 +367,13 @@ int main(int argc, char **argv)
          n = atoi(argv[i]);
    }
 
+#ifdef LLVM_NATIVE_ARCH
+   LLVMLinkInJIT();
+   LLVMInitializeNativeTarget();
+#endif
+
+   util_cpu_detect();
+
    if(fp) {
       /* Warm up the caches */
       test_some(0, NULL, 100);
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_cache.c b/src/gallium/drivers/llvmpipe/lp_tex_cache.c
index 23a94b5b0d..773e848242 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_cache.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_cache.c
@@ -154,7 +154,7 @@ lp_tex_tile_cache_validate_texture(struct llvmpipe_tex_tile_cache *tc)
       if (lpt->timestamp != tc->timestamp) {
          /* texture was modified, invalidate all cached tiles */
          uint i;
-         _debug_printf("INV %d %d\n", tc->timestamp, lpt->timestamp);
+         debug_printf("INV %d %d\n", tc->timestamp, lpt->timestamp);
          for (i = 0; i < NUM_ENTRIES; i++) {
             tc->entries[i].addr.bits.invalid = 1;
          }
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample.h b/src/gallium/drivers/llvmpipe/lp_tex_sample.h
index 628ec3f1ef..9ad1bde956 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_sample.h
+++ b/src/gallium/drivers/llvmpipe/lp_tex_sample.h
@@ -29,10 +29,13 @@
 #define LP_TEX_SAMPLE_H
 
 
+#include <llvm-c/Core.h>
+
 #include "tgsi/tgsi_exec.h"
 
 
 struct llvmpipe_tex_tile_cache;
+struct lp_sampler_static_state;
 
 
 /**
@@ -75,4 +78,24 @@ lp_get_samples(struct tgsi_sampler *tgsi_sampler,
                float rgba[NUM_CHANNELS][QUAD_SIZE]);
 
 
+/**
+ * Texture sampling code generator that just calls lp_get_samples C function
+ * for the actual sampling computation.
+ *
+ * @param context_ptr LLVM value with the pointer to the struct lp_jit_context.
+ */
+struct lp_build_sampler_soa *
+lp_c_sampler_soa_create(LLVMValueRef context_ptr);
+
+
+/**
+ * Pure-LLVM texture sampling code generator.
+ *
+ * @param context_ptr LLVM value with the pointer to the struct lp_jit_context.
+ */
+struct lp_build_sampler_soa *
+lp_llvm_sampler_soa_create(const struct lp_sampler_static_state *key,
+                           LLVMValueRef context_ptr);
+
+
 #endif /* LP_TEX_SAMPLE_H */
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
index 94eb6dad5a..a1365a045f 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_sample.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
@@ -1578,3 +1578,136 @@ out:
    tgsi_sampler->get_samples( tgsi_sampler, s, t, p, lodbias, rgba );
 }
 
+
+void PIPE_CDECL
+lp_fetch_texel_soa( struct tgsi_sampler **samplers,
+                    uint32_t unit,
+                    float *store )
+{
+   struct tgsi_sampler *sampler = samplers[unit];
+
+#if 0
+   uint j;
+
+   debug_printf("%s sampler: %p (%p) store: %p\n",
+                __FUNCTION__,
+                sampler, *sampler,
+                store );
+
+   debug_printf("lodbias %f\n", store[12]);
+
+   for (j = 0; j < 4; j++)
+      debug_printf("sample %d texcoord %f %f\n",
+                   j,
+                   store[0+j],
+                   store[4+j]);
+#endif
+
+   {
+      float rgba[NUM_CHANNELS][QUAD_SIZE];
+      sampler->get_samples(sampler,
+                           &store[0],
+                           &store[4],
+                           &store[8],
+                           0.0f, /*store[12],  lodbias */
+                           rgba);
+      memcpy(store, rgba, sizeof rgba);
+   }
+
+#if 0
+   for (j = 0; j < 4; j++)
+      debug_printf("sample %d result %f %f %f %f\n",
+                   j,
+                   store[0+j],
+                   store[4+j],
+                   store[8+j],
+                   store[12+j]);
+#endif
+}
+
+
+#include "lp_bld_type.h"
+#include "lp_bld_intr.h"
+#include "lp_bld_tgsi.h"
+
+
+struct lp_c_sampler_soa
+{
+   struct lp_build_sampler_soa base;
+
+   LLVMValueRef context_ptr;
+
+   LLVMValueRef samplers_ptr;
+
+   /** Coords/texels store */
+   LLVMValueRef store_ptr;
+};
+
+
+static void
+lp_c_sampler_soa_destroy(struct lp_build_sampler_soa *sampler)
+{
+   FREE(sampler);
+}
+
+
+static void
+lp_c_sampler_soa_emit_fetch_texel(struct lp_build_sampler_soa *_sampler,
+                                  LLVMBuilderRef builder,
+                                  struct lp_type type,
+                                  unsigned unit,
+                                  unsigned num_coords,
+                                  const LLVMValueRef *coords,
+                                  LLVMValueRef lodbias,
+                                  LLVMValueRef *texel)
+{
+   struct lp_c_sampler_soa *sampler = (struct lp_c_sampler_soa *)_sampler;
+   LLVMTypeRef vec_type = LLVMTypeOf(coords[0]);
+   LLVMValueRef args[3];
+   unsigned i;
+
+   if(!sampler->samplers_ptr)
+      sampler->samplers_ptr = lp_jit_context_samplers(builder, sampler->context_ptr);
+
+   if(!sampler->store_ptr)
+      sampler->store_ptr = LLVMBuildArrayAlloca(builder,
+                                            vec_type,
+                                            LLVMConstInt(LLVMInt32Type(), 4, 0),
+                                            "texel_store");
+
+   for (i = 0; i < num_coords; i++) {
+      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
+      LLVMValueRef coord_ptr = LLVMBuildGEP(builder, sampler->store_ptr, &index, 1, "");
+      LLVMBuildStore(builder, coords[i], coord_ptr);
+   }
+
+   args[0] = sampler->samplers_ptr;
+   args[1] = LLVMConstInt(LLVMInt32Type(), unit, 0);
+   args[2] = sampler->store_ptr;
+
+   lp_build_intrinsic(builder, "fetch_texel", LLVMVoidType(), args, 3);
+
+   for (i = 0; i < NUM_CHANNELS; ++i) {
+      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
+      LLVMValueRef texel_ptr = LLVMBuildGEP(builder, sampler->store_ptr, &index, 1, "");
+      texel[i] = LLVMBuildLoad(builder, texel_ptr, "");
+   }
+}
+
+
+struct lp_build_sampler_soa *
+lp_c_sampler_soa_create(LLVMValueRef context_ptr)
+{
+   struct lp_c_sampler_soa *sampler;
+
+   sampler = CALLOC_STRUCT(lp_c_sampler_soa);
+   if(!sampler)
+      return NULL;
+
+   sampler->base.destroy = lp_c_sampler_soa_destroy;
+   sampler->base.emit_fetch_texel = lp_c_sampler_soa_emit_fetch_texel;
+   sampler->context_ptr = context_ptr;
+
+   return &sampler->base;
+}
+
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c
new file mode 100644
index 0000000000..d2a6ae21f5
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c
@@ -0,0 +1,196 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 VMware, Inc.
+ * All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * Texture sampling code generation
+ *
+ * This file is nothing more than ugly glue between three largely independent
+ * entities:
+ * - TGSI -> LLVM translation (i.e., lp_build_tgsi_soa)
+ * - texture sampling code generation (i.e., lp_build_sample_soa)
+ * - LLVM pipe driver
+ *
+ * All interesting code is in the functions mentioned above. There is really
+ * nothing to see here.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+#include "lp_bld_debug.h"
+#include "lp_bld_type.h"
+#include "lp_bld_intr.h"
+#include "lp_bld_sample.h"
+#include "lp_bld_tgsi.h"
+#include "lp_state.h"
+#include "lp_tex_sample.h"
+
+
+/**
+ * This provides the bridge between the sampler state store in lp_jit_context
+ * and lp_jit_texture and the sampler code generator. It provides the
+ * texture layout information required by the texture sampler code generator
+ * in terms of the state stored in lp_jit_context and lp_jit_texture in runtime.
+ */
+struct llvmpipe_sampler_dynamic_state
+{
+   struct lp_sampler_dynamic_state base;
+
+   const struct lp_sampler_static_state *static_state;
+
+   LLVMValueRef context_ptr;
+};
+
+
+/**
+ * This is the bridge between our sampler and the TGSI translator.
+ */
+struct lp_llvm_sampler_soa
+{
+   struct lp_build_sampler_soa base;
+
+   struct llvmpipe_sampler_dynamic_state dynamic_state;
+};
+
+
+/**
+ * Fetch the specified member of the lp_jit_texture structure.
+ *
+ * @sa http://llvm.org/docs/GetElementPtr.html
+ */
+static LLVMValueRef
+lp_llvm_texture_member(struct lp_sampler_dynamic_state *base,
+                       LLVMBuilderRef builder,
+                       unsigned unit,
+                       unsigned member_index,
+                       const char *member_name)
+{
+   struct llvmpipe_sampler_dynamic_state *state = (struct llvmpipe_sampler_dynamic_state *)base;
+   LLVMValueRef indices[4];
+   LLVMValueRef ptr;
+   LLVMValueRef res;
+
+   assert(unit < PIPE_MAX_SAMPLERS);
+
+   /* context[0] */
+   indices[0] = LLVMConstInt(LLVMInt32Type(), 0, 0);
+   /* context[0].textures */
+   indices[1] = LLVMConstInt(LLVMInt32Type(), LP_JIT_CONTEXT_TEXTURES_INDEX, 0);
+   /* context[0].textures[unit] */
+   indices[2] = LLVMConstInt(LLVMInt32Type(), unit, 0);
+   /* context[0].textures[unit].member */
+   indices[3] = LLVMConstInt(LLVMInt32Type(), member_index, 0);
+
+   ptr = LLVMBuildGEP(builder, state->context_ptr, indices, Elements(indices), "");
+
+   res = LLVMBuildLoad(builder, ptr, "");
+
+   lp_build_name(res, "context.texture%u.%s", unit, member_name);
+
+   return res;
+}
+
+
+/**
+ * Helper macro to instantiate the functions that generate the code to fetch
+ * the members of lp_jit_texture to fulfill the sampler code generator requests.
+ *
+ * This complexity is the price we have to pay to keep the texture sampler code
+ * generator a reusable module without dependencies to llvmpipe internals.
+ */
+#define LP_LLVM_TEXTURE_MEMBER(_name, _index) \
+   static LLVMValueRef \
+   lp_llvm_texture_##_name( struct lp_sampler_dynamic_state *base, \
+                            LLVMBuilderRef builder, \
+                            unsigned unit) \
+   { \
+      return lp_llvm_texture_member(base, builder, unit, _index, #_name ); \
+   }
+
+
+LP_LLVM_TEXTURE_MEMBER(width,    LP_JIT_TEXTURE_WIDTH)
+LP_LLVM_TEXTURE_MEMBER(height,   LP_JIT_TEXTURE_HEIGHT)
+LP_LLVM_TEXTURE_MEMBER(stride,   LP_JIT_TEXTURE_STRIDE)
+LP_LLVM_TEXTURE_MEMBER(data_ptr, LP_JIT_TEXTURE_DATA)
+
+
+static void
+lp_llvm_sampler_soa_destroy(struct lp_build_sampler_soa *sampler)
+{
+   FREE(sampler);
+}
+
+
+static void
+lp_llvm_sampler_soa_emit_fetch_texel(struct lp_build_sampler_soa *base,
+                                     LLVMBuilderRef builder,
+                                     struct lp_type type,
+                                     unsigned unit,
+                                     unsigned num_coords,
+                                     const LLVMValueRef *coords,
+                                     LLVMValueRef lodbias,
+                                     LLVMValueRef *texel)
+{
+   struct lp_llvm_sampler_soa *sampler = (struct lp_llvm_sampler_soa *)base;
+
+   assert(unit < PIPE_MAX_SAMPLERS);
+
+   lp_build_sample_soa(builder,
+                       &sampler->dynamic_state.static_state[unit],
+                       &sampler->dynamic_state.base,
+                       type,
+                       unit,
+                       num_coords,
+                       coords,
+                       lodbias,
+                       texel);
+}
+
+
+struct lp_build_sampler_soa *
+lp_llvm_sampler_soa_create(const struct lp_sampler_static_state *static_state,
+                           LLVMValueRef context_ptr)
+{
+   struct lp_llvm_sampler_soa *sampler;
+
+   sampler = CALLOC_STRUCT(lp_llvm_sampler_soa);
+   if(!sampler)
+      return NULL;
+
+   sampler->base.destroy = lp_llvm_sampler_soa_destroy;
+   sampler->base.emit_fetch_texel = lp_llvm_sampler_soa_emit_fetch_texel;
+   sampler->dynamic_state.base.width = lp_llvm_texture_width;
+   sampler->dynamic_state.base.height = lp_llvm_texture_height;
+   sampler->dynamic_state.base.stride = lp_llvm_texture_stride;
+   sampler->dynamic_state.base.data_ptr = lp_llvm_texture_data_ptr;
+   sampler->dynamic_state.static_state = static_state;
+   sampler->dynamic_state.context_ptr = context_ptr;
+
+   return &sampler->base;
+}
+
diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c
index 724d437833..a00f2495df 100644
--- a/src/gallium/drivers/llvmpipe/lp_texture.c
+++ b/src/gallium/drivers/llvmpipe/lp_texture.c
@@ -66,16 +66,24 @@ llvmpipe_texture_layout(struct llvmpipe_screen *screen,
    pf_get_block(lpt->base.format, &lpt->base.block);
 
    for (level = 0; level <= pt->last_level; level++) {
+      unsigned nblocksx, nblocksy;
+
       pt->width[level] = width;
       pt->height[level] = height;
       pt->depth[level] = depth;
       pt->nblocksx[level] = pf_get_nblocksx(&pt->block, width);  
-      pt->nblocksy[level] = pf_get_nblocksy(&pt->block, height);  
-      lpt->stride[level] = align(pt->nblocksx[level]*pt->block.size, 16);
+      pt->nblocksy[level] = pf_get_nblocksy(&pt->block, height);
+
+      /* Allocate storage for whole quads. This is particularly important
+       * for depth surfaces, which are currently stored in a swizzled format. */
+      nblocksx = pf_get_nblocksx(&pt->block, align(width, 2));
+      nblocksy = pf_get_nblocksy(&pt->block, align(height, 2));
+
+      lpt->stride[level] = align(nblocksx*pt->block.size, 16);
 
       lpt->level_offset[level] = buffer_size;
 
-      buffer_size += (pt->nblocksy[level] *
+      buffer_size += (nblocksy *
                       ((pt->target == PIPE_TEXTURE_CUBE) ? 6 : depth) *
                       lpt->stride[level]);
 
@@ -353,17 +361,9 @@ llvmpipe_transfer_map( struct pipe_screen *_screen,
 
    if(lpt->dt) {
       struct llvmpipe_winsys *winsys = screen->winsys;
-      unsigned flags = 0;
-
-      if (transfer->usage != PIPE_TRANSFER_READ) {
-         flags |= PIPE_BUFFER_USAGE_CPU_WRITE;
-      }
-
-      if (transfer->usage != PIPE_TRANSFER_WRITE) {
-         flags |= PIPE_BUFFER_USAGE_CPU_READ;
-      }
 
-      map = winsys->displaytarget_map(winsys, lpt->dt, flags);
+      map = winsys->displaytarget_map(winsys, lpt->dt,
+                                      pipe_transfer_buffer_flags(transfer));
       if (map == NULL)
          return NULL;
    }
@@ -373,7 +373,7 @@ llvmpipe_transfer_map( struct pipe_screen *_screen,
    /* May want to different things here depending on read/write nature
     * of the map:
     */
-   if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) 
+   if (transfer->texture && (transfer->usage & PIPE_TRANSFER_WRITE))
    {
       /* Do something to notify sharing contexts of a texture change.
        * In llvmpipe, that would mean flushing the texture cache.
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.c b/src/gallium/drivers/llvmpipe/lp_tile_cache.c
index 143afec3d3..ec3e002d62 100644
--- a/src/gallium/drivers/llvmpipe/lp_tile_cache.c
+++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.c
@@ -44,10 +44,53 @@
 #include "lp_tile_cache.h"
 
 
+#define MAX_WIDTH 4096
+#define MAX_HEIGHT 4096
+
+
+enum llvmpipe_tile_status
+{
+   LP_TILE_STATUS_UNDEFINED = 0,
+   LP_TILE_STATUS_CLEAR = 1,
+   LP_TILE_STATUS_DEFINED = 2
+};
+
+
+struct llvmpipe_cached_tile
+{
+   enum llvmpipe_tile_status status;
+
+   /** color in SOA format */
+   uint8_t *color;
+};
+
+
+struct llvmpipe_tile_cache
+{
+   struct pipe_screen *screen;
+   struct pipe_surface *surface;  /**< the surface we're caching */
+   struct pipe_transfer *transfer;
+   void *transfer_map;
+
+   struct llvmpipe_cached_tile entries[MAX_WIDTH/TILE_SIZE][MAX_HEIGHT/TILE_SIZE];
+
+   uint8_t clear_color[4];  /**< for color bufs */
+   uint clear_val;        /**< for z+stencil, or packed color clear value */
+
+   struct llvmpipe_cached_tile *last_tile;  /**< most recently retrieved tile */
+};
+
+
 struct llvmpipe_tile_cache *
 lp_create_tile_cache( struct pipe_screen *screen )
 {
    struct llvmpipe_tile_cache *tc;
+   int maxLevels, maxTexSize;
+
+   /* sanity checking: max sure MAX_WIDTH/HEIGHT >= largest texture image */
+   maxLevels = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS);
+   maxTexSize = 1 << (maxLevels - 1);
+   assert(MAX_WIDTH >= maxTexSize);
 
    tc = CALLOC_STRUCT( llvmpipe_tile_cache );
    if(!tc)
@@ -193,44 +236,41 @@ lp_flush_tile_cache(struct llvmpipe_tile_cache *tc)
    if(!pt)
       return;
 
+   assert(tc->transfer_map);
+
    /* push the tile to all positions marked as clear */
    for (y = 0; y < pt->height; y += TILE_SIZE) {
       for (x = 0; x < pt->width; x += TILE_SIZE) {
          struct llvmpipe_cached_tile *tile = &tc->entries[y/TILE_SIZE][x/TILE_SIZE];
 
-         switch(tile->status) {
-         case LP_TILE_STATUS_UNDEFINED:
-            break;
-
-         case LP_TILE_STATUS_CLEAR: {
-            /**
-             * Actually clear the tiles which were flagged as being in a clear state.
-             */
-
-            struct pipe_screen *screen = pt->texture->screen;
-            unsigned tw = TILE_SIZE;
-            unsigned th = TILE_SIZE;
-            void *dst;
+         if(tile->status != LP_TILE_STATUS_UNDEFINED) {
+            unsigned w = TILE_SIZE;
+            unsigned h = TILE_SIZE;
+
+            if (!pipe_clip_tile(x, y, &w, &h, pt)) {
+               switch(tile->status) {
+               case LP_TILE_STATUS_CLEAR:
+                  /* Actually clear the tiles which were flagged as being in a
+                   * clear state. */
+                  util_fill_rect(tc->transfer_map, &pt->block, pt->stride,
+                                 x, y, w, h,
+                                 tc->clear_val);
+                  break;
+
+               case LP_TILE_STATUS_DEFINED:
+                  lp_tile_write_4ub(pt->format,
+                                    tile->color,
+                                    tc->transfer_map, pt->stride,
+                                    x, y, w, h);
+                  break;
+
+               default:
+                  assert(0);
+                  break;
+               }
+            }
 
-            if (pipe_clip_tile(x, y, &tw, &th, pt))
-               continue;
-
-            dst = screen->transfer_map(screen, pt);
-            assert(dst);
-            if(!dst)
-               continue;
-
-            util_fill_rect(dst, &pt->block, pt->stride,
-                           x, y, tw,  th,
-                           tc->clear_val);
-
-            screen->transfer_unmap(screen, pt);
-            break;
-         }
-
-         case LP_TILE_STATUS_DEFINED:
-            lp_put_tile_rgba_soa(pt, x, y, tile->color);
-            break;
+            tile->status = LP_TILE_STATUS_UNDEFINED;
          }
       }
    }
@@ -248,6 +288,9 @@ lp_get_cached_tile(struct llvmpipe_tile_cache *tc,
    struct llvmpipe_cached_tile *tile = &tc->entries[y/TILE_SIZE][x/TILE_SIZE];
    struct pipe_transfer *pt = tc->transfer;
    
+   assert(tc->surface);
+   assert(tc->transfer);
+
    switch(tile->status) {
    case LP_TILE_STATUS_CLEAR:
       /* don't get tile from framebuffer, just clear it */
@@ -255,11 +298,22 @@ lp_get_cached_tile(struct llvmpipe_tile_cache *tc,
       tile->status = LP_TILE_STATUS_DEFINED;
       break;
 
-   case LP_TILE_STATUS_UNDEFINED:
-      /* get new tile data from transfer */
-      lp_get_tile_rgba_soa(pt, x, y, tile->color);
+   case LP_TILE_STATUS_UNDEFINED: {
+      unsigned w = TILE_SIZE;
+      unsigned h = TILE_SIZE;
+
+      x &= ~(TILE_SIZE - 1);
+      y &= ~(TILE_SIZE - 1);
+
+      if (!pipe_clip_tile(x, y, &w, &h, tc->transfer))
+         lp_tile_read_4ub(pt->format,
+                          tile->color,
+                          tc->transfer_map, tc->transfer->stride,
+                          x, y, w, h);
+
       tile->status = LP_TILE_STATUS_DEFINED;
       break;
+   }
 
    case LP_TILE_STATUS_DEFINED:
       /* nothing to do */
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.h b/src/gallium/drivers/llvmpipe/lp_tile_cache.h
index 6d8ba5ece7..161bab3799 100644
--- a/src/gallium/drivers/llvmpipe/lp_tile_cache.h
+++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.h
@@ -33,42 +33,7 @@
 #include "lp_tile_soa.h"
 
 
-enum llvmpipe_tile_status
-{
-   LP_TILE_STATUS_UNDEFINED = 0,
-   LP_TILE_STATUS_CLEAR = 1,
-   LP_TILE_STATUS_DEFINED = 2
-};
-
-
-struct llvmpipe_cached_tile
-{
-   enum llvmpipe_tile_status status;
-
-   /** color in SOA format */
-   uint8_t *color;
-};
-
-
-/** XXX move these */
-#define MAX_WIDTH 2048
-#define MAX_HEIGHT 2048
-
-
-struct llvmpipe_tile_cache
-{
-   struct pipe_screen *screen;
-   struct pipe_surface *surface;  /**< the surface we're caching */
-   struct pipe_transfer *transfer;
-   void *transfer_map;
-
-   struct llvmpipe_cached_tile entries[MAX_WIDTH/TILE_SIZE][MAX_HEIGHT/TILE_SIZE];
-
-   uint8_t clear_color[4];  /**< for color bufs */
-   uint clear_val;        /**< for z+stencil, or packed color clear value */
-
-   struct llvmpipe_cached_tile *last_tile;  /**< most recently retrieved tile */
-};
+struct llvmpipe_tile_cache;  /* opaque */
 
 
 extern struct llvmpipe_tile_cache *
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_soa.c b/src/gallium/drivers/llvmpipe/lp_tile_soa.c
deleted file mode 100644
index 4e4ccb31cc..0000000000
--- a/src/gallium/drivers/llvmpipe/lp_tile_soa.c
+++ /dev/null
@@ -1,931 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/**
- * RGBA/float tile get/put functions.
- * Usable both by drivers and state trackers.
- */
-
-
-#include "pipe/p_defines.h"
-#include "pipe/p_inlines.h"
-
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "util/u_rect.h"
-#include "util/u_tile.h"
-#include "lp_tile_cache.h"
-#include "lp_tile_soa.h"
-
-
-const unsigned char
-tile_offset[TILE_VECTOR_HEIGHT][TILE_VECTOR_WIDTH] = {
-   {  0,  1,  4,  5,  8,  9, 12, 13},
-   {  2,  3,  6,  7, 10, 11, 14, 15}
-};
-
-
-
-/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/
-
-static void
-a8r8g8b8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         const unsigned pixel = *src++;
-         TILE_PIXEL(p, j, i, 0) = (pixel >> 16) & 0xff;
-         TILE_PIXEL(p, j, i, 1) = (pixel >>  8) & 0xff;
-         TILE_PIXEL(p, j, i, 2) = (pixel >>  0) & 0xff;
-         TILE_PIXEL(p, j, i, 3) = (pixel >> 24) & 0xff;
-      }
-   }
-}
-
-
-static void
-a8r8g8b8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r, g, b, a;
-         r = TILE_PIXEL(p, j, i, 0);
-         g = TILE_PIXEL(p, j, i, 1);
-         b = TILE_PIXEL(p, j, i, 2);
-         a = TILE_PIXEL(p, j, i, 3);
-         *dst++ = (a << 24) | (r << 16) | (g << 8) | b;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/
-
-static void
-x8r8g8b8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         const unsigned pixel = *src++;
-         TILE_PIXEL(p, j, i, 0) = (pixel >> 16) & 0xff;
-         TILE_PIXEL(p, j, i, 1) = (pixel >>  8) & 0xff;
-         TILE_PIXEL(p, j, i, 2) = (pixel >>  0) & 0xff;
-         TILE_PIXEL(p, j, i, 3) = 0xff;
-      }
-   }
-}
-
-
-static void
-x8r8g8b8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r, g, b;
-         r = TILE_PIXEL(p, j, i, 0);
-         g = TILE_PIXEL(p, j, i, 1);
-         b = TILE_PIXEL(p, j, i, 2);
-         *dst++ = (0xff << 24) | (r << 16) | (g << 8) | b;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_B8G8R8A8_UNORM ***/
-
-static void
-b8g8r8a8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         const unsigned pixel = *src++;
-         TILE_PIXEL(p, j, i, 0) = (pixel >>  8) & 0xff;
-         TILE_PIXEL(p, j, i, 1) = (pixel >> 16) & 0xff;
-         TILE_PIXEL(p, j, i, 2) = (pixel >> 24) & 0xff;
-         TILE_PIXEL(p, j, i, 3) = (pixel >>  0) & 0xff;
-      }
-   }
-}
-
-
-static void
-b8g8r8a8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r, g, b, a;
-         r = TILE_PIXEL(p, j, i, 0);
-         g = TILE_PIXEL(p, j, i, 1);
-         b = TILE_PIXEL(p, j, i, 2);
-         a = TILE_PIXEL(p, j, i, 3);
-         *dst++ = (b << 24) | (g << 16) | (r << 8) | a;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_A1R5G5B5_UNORM ***/
-
-static void
-a1r5g5b5_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         const ushort pixel = *src++;
-         TILE_PIXEL(p, j, i, 0) = ((pixel >> 10) & 0x1f) * 255 / 31;
-         TILE_PIXEL(p, j, i, 1) = ((pixel >>  5) & 0x1f) * 255 / 31;
-         TILE_PIXEL(p, j, i, 2) = ((pixel      ) & 0x1f) * 255 / 31;
-         TILE_PIXEL(p, j, i, 3) = ((pixel >> 15)       ) * 255;
-      }
-   }
-}
-
-
-static void
-a1r5g5b5_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r, g, b, a;
-         r = TILE_PIXEL(p, j, i, 0);
-         g = TILE_PIXEL(p, j, i, 1);
-         b = TILE_PIXEL(p, j, i, 2);
-         a = TILE_PIXEL(p, j, i, 3);
-         r = r >> 3;  /* 5 bits */
-         g = g >> 3;  /* 5 bits */
-         b = b >> 3;  /* 5 bits */
-         a = a >> 7;  /* 1 bit */
-         *dst++ = (a << 15) | (r << 10) | (g << 5) | b;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_A4R4G4B4_UNORM ***/
-
-static void
-a4r4g4b4_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         const ushort pixel = *src++;
-         TILE_PIXEL(p, j, i, 0) = ((pixel >>  8) & 0xf) * 255 / 15;
-         TILE_PIXEL(p, j, i, 1) = ((pixel >>  4) & 0xf) * 255 / 15;
-         TILE_PIXEL(p, j, i, 2) = ((pixel      ) & 0xf) * 255 / 15;
-         TILE_PIXEL(p, j, i, 3) = ((pixel >> 12)      ) * 255 / 15;
-      }
-   }
-}
-
-
-static void
-a4r4g4b4_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r, g, b, a;
-         r = TILE_PIXEL(p, j, i, 0);
-         g = TILE_PIXEL(p, j, i, 1);
-         b = TILE_PIXEL(p, j, i, 2);
-         a = TILE_PIXEL(p, j, i, 3);
-         r >>= 4;
-         g >>= 4;
-         b >>= 4;
-         a >>= 4;
-         *dst++ = (a << 12) | (r << 16) | (g << 4) | b;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_R5G6B5_UNORM ***/
-
-static void
-r5g6b5_get_tile_rgba(const ushort *src,
-                     unsigned w, unsigned h,
-                     uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         const ushort pixel = *src++;
-         TILE_PIXEL(p, j, i, 0) = ((pixel >> 11) & 0x1f) * 255 / 31;
-         TILE_PIXEL(p, j, i, 1) = ((pixel >>  5) & 0x3f) * 255 / 63;
-         TILE_PIXEL(p, j, i, 2) = ((pixel      ) & 0x1f) * 255 / 31;
-         TILE_PIXEL(p, j, i, 3) = 255;
-      }
-   }
-}
-
-
-static void
-r5g6b5_put_tile_rgba(ushort *dst,
-                     unsigned w, unsigned h,
-                     const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         uint r = (uint) TILE_PIXEL(p, j, i, 0) * 31 / 255;
-         uint g = (uint) TILE_PIXEL(p, j, i, 1) * 63 / 255;
-         uint b = (uint) TILE_PIXEL(p, j, i, 2) * 31 / 255;
-         *dst++ = (r << 11) | (g << 5) | (b);
-      }
-   }
-}
-
-
-
-/*** PIPE_FORMAT_Z16_UNORM ***/
-
-/**
- * Return each Z value as four floats in [0,1].
- */
-static void
-z16_get_tile_rgba(const ushort *src,
-                  unsigned w, unsigned h,
-                  uint8_t *p)
-{
-   const float scale = 1.0f / 65535.0f;
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = *src++ * scale;
-      }
-   }
-}
-
-
-
-
-/*** PIPE_FORMAT_L8_UNORM ***/
-
-static void
-l8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, src++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) = *src;
-         TILE_PIXEL(p, j, i, 3) = 255;
-      }
-   }
-}
-
-
-static void
-l8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r;
-         r = TILE_PIXEL(p, j, i, 0);
-         *dst++ = (ubyte) r;
-      }
-   }
-}
-
-
-
-/*** PIPE_FORMAT_A8_UNORM ***/
-
-static void
-a8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, src++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) = 0;
-         TILE_PIXEL(p, j, i, 3) = *src;
-      }
-   }
-}
-
-
-static void
-a8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned a;
-         a = TILE_PIXEL(p, j, i, 3);
-         *dst++ = (ubyte) a;
-      }
-   }
-}
-
-
-
-/*** PIPE_FORMAT_R16_SNORM ***/
-
-static void
-r16_get_tile_rgba(const short *src,
-                  unsigned w, unsigned h,
-                  uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, src++) {
-         TILE_PIXEL(p, j, i, 0) = MAX2(src[0] >> 7, 0);
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) = 0;
-         TILE_PIXEL(p, j, i, 3) = 255;
-      }
-   }
-}
-
-
-static void
-r16_put_tile_rgba(short *dst,
-                  unsigned w, unsigned h,
-                  const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, dst++) {
-         dst[0] = TILE_PIXEL(p, j, i, 0) << 7;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_R16G16B16A16_SNORM ***/
-
-static void
-r16g16b16a16_get_tile_rgba(const short *src,
-                           unsigned w, unsigned h,
-                           uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, src += 4) {
-         TILE_PIXEL(p, j, i, 0) = src[0] >> 8;
-         TILE_PIXEL(p, j, i, 1) = src[1] >> 8;
-         TILE_PIXEL(p, j, i, 2) = src[2] >> 8;
-         TILE_PIXEL(p, j, i, 3) = src[3] >> 8;
-      }
-   }
-}
-
-
-static void
-r16g16b16a16_put_tile_rgba(short *dst,
-                           unsigned w, unsigned h,
-                           const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, dst += 4) {
-         dst[0] = TILE_PIXEL(p, j, i, 0) << 8;
-         dst[1] = TILE_PIXEL(p, j, i, 1) << 8;
-         dst[2] = TILE_PIXEL(p, j, i, 2) << 8;
-         dst[3] = TILE_PIXEL(p, j, i, 3) << 8;
-      }
-   }
-}
-
-
-
-/*** PIPE_FORMAT_I8_UNORM ***/
-
-static void
-i8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++, src++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = *src;
-      }
-   }
-}
-
-
-static void
-i8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r;
-         r = TILE_PIXEL(p, j, i, 0);
-         *dst++ = (ubyte) r;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_A8L8_UNORM ***/
-
-static void
-a8l8_get_tile_rgba(const ushort *src,
-                   unsigned w, unsigned h,
-                   uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         ushort ra = *src++;
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) = ra & 0xff;
-         TILE_PIXEL(p, j, i, 3) = ra >> 8;
-      }
-   }
-}
-
-
-static void
-a8l8_put_tile_rgba(ushort *dst,
-                   unsigned w, unsigned h,
-                   const uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         unsigned r, a;
-         r = TILE_PIXEL(p, j, i, 0);
-         a = TILE_PIXEL(p, j, i, 3);
-         *dst++ = (a << 8) | r;
-      }
-   }
-}
-
-
-
-
-/*** PIPE_FORMAT_Z32_UNORM ***/
-
-/**
- * Return each Z value as four floats in [0,1].
- */
-static void
-z32_get_tile_rgba(const unsigned *src,
-                  unsigned w, unsigned h,
-                  uint8_t *p)
-{
-   const double scale = 1.0 / (double) 0xffffffff;
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = (float) (*src++ * scale);
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_S8Z24_UNORM ***/
-
-/**
- * Return Z component as four float in [0,1].  Stencil part ignored.
- */
-static void
-s8z24_get_tile_rgba(const unsigned *src,
-                    unsigned w, unsigned h,
-                    uint8_t *p)
-{
-   const double scale = 1.0 / ((1 << 24) - 1);
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = (float) (scale * (*src++ & 0xffffff));
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_Z24S8_UNORM ***/
-
-/**
- * Return Z component as four float in [0,1].  Stencil part ignored.
- */
-static void
-z24s8_get_tile_rgba(const unsigned *src,
-                    unsigned w, unsigned h,
-                    uint8_t *p)
-{
-   const double scale = 1.0 / ((1 << 24) - 1);
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = (float) (scale * (*src++ >> 8));
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_Z32_FLOAT ***/
-
-/**
- * Return each Z value as four floats in [0,1].
- */
-static void
-z32f_get_tile_rgba(const float *src,
-                   unsigned w, unsigned h,
-                   uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = *src++;
-      }
-   }
-}
-
-
-/*** PIPE_FORMAT_YCBCR / PIPE_FORMAT_YCBCR_REV ***/
-
-/**
- * Convert YCbCr (or YCrCb) to RGBA.
- */
-static void
-ycbcr_get_tile_rgba(const ushort *src,
-                    unsigned w, unsigned h,
-                    uint8_t *p,
-                    boolean rev)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      /* do two texels at a time */
-      for (j = 0; j < (w & ~1); j += 2, src += 2) {
-         const ushort t0 = src[0];
-         const ushort t1 = src[1];
-         const ubyte y0 = (t0 >> 8) & 0xff;  /* luminance */
-         const ubyte y1 = (t1 >> 8) & 0xff;  /* luminance */
-         ubyte cb, cr;
-         float r, g, b;
-
-         if (rev) {
-            cb = t1 & 0xff;         /* chroma U */
-            cr = t0 & 0xff;         /* chroma V */
-         }
-         else {
-            cb = t0 & 0xff;         /* chroma U */
-            cr = t1 & 0xff;         /* chroma V */
-         }
-
-         /* even pixel: y0,cr,cb */
-         r = 1.164f * (y0-16) + 1.596f * (cr-128);
-         g = 1.164f * (y0-16) - 0.813f * (cr-128) - 0.391f * (cb-128);
-         b = 1.164f * (y0-16) + 2.018f * (cb-128);
-         TILE_PIXEL(p, j, i, 0) = r;
-         TILE_PIXEL(p, j, i, 1) = g;
-         TILE_PIXEL(p, j, i, 2) = b;
-         TILE_PIXEL(p, j, i, 3) = 255;
-
-         /* odd pixel: use y1,cr,cb */
-         r = 1.164f * (y1-16) + 1.596f * (cr-128);
-         g = 1.164f * (y1-16) - 0.813f * (cr-128) - 0.391f * (cb-128);
-         b = 1.164f * (y1-16) + 2.018f * (cb-128);
-         TILE_PIXEL(p, j + 1, i, 0) = r;
-         TILE_PIXEL(p, j + 1, i, 1) = g;
-         TILE_PIXEL(p, j + 1, i, 2) = b;
-         TILE_PIXEL(p, j + 1, i, 3) = 255;
-      }
-      /* do the last texel */
-      if (w & 1) {
-         const ushort t0 = src[0];
-         const ushort t1 = src[1];
-         const ubyte y0 = (t0 >> 8) & 0xff;  /* luminance */
-         ubyte cb, cr;
-         float r, g, b;
-
-         if (rev) {
-            cb = t1 & 0xff;         /* chroma U */
-            cr = t0 & 0xff;         /* chroma V */
-         }
-         else {
-            cb = t0 & 0xff;         /* chroma U */
-            cr = t1 & 0xff;         /* chroma V */
-         }
-
-         /* even pixel: y0,cr,cb */
-         r = 1.164f * (y0-16) + 1.596f * (cr-128);
-         g = 1.164f * (y0-16) - 0.813f * (cr-128) - 0.391f * (cb-128);
-         b = 1.164f * (y0-16) + 2.018f * (cb-128);
-         TILE_PIXEL(p, j, i, 0) = r;
-         TILE_PIXEL(p, j, i, 1) = g;
-         TILE_PIXEL(p, j, i, 2) = b;
-         TILE_PIXEL(p, j, i, 3) = 255;
-      }
-   }
-}
-
-
-static void
-fake_get_tile_rgba(const ushort *src,
-                   unsigned w, unsigned h,
-                   uint8_t *p)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      for (j = 0; j < w; j++) {
-         TILE_PIXEL(p, j, i, 0) =
-         TILE_PIXEL(p, j, i, 1) =
-         TILE_PIXEL(p, j, i, 2) =
-         TILE_PIXEL(p, j, i, 3) = (i ^ j) & 1 ? 255 : 0;
-      }
-   }
-}
-
-
-static void
-lp_tile_raw_to_rgba_soa(enum pipe_format format,
-                        void *src,
-                        uint w, uint h,
-                        uint8_t *p)
-{
-   switch (format) {
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      a8r8g8b8_get_tile_rgba((unsigned *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_X8R8G8B8_UNORM:
-      x8r8g8b8_get_tile_rgba((unsigned *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      b8g8r8a8_get_tile_rgba((unsigned *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_A1R5G5B5_UNORM:
-      a1r5g5b5_get_tile_rgba((ushort *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_A4R4G4B4_UNORM:
-      a4r4g4b4_get_tile_rgba((ushort *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_R5G6B5_UNORM:
-      r5g6b5_get_tile_rgba((ushort *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_L8_UNORM:
-      l8_get_tile_rgba((ubyte *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_A8_UNORM:
-      a8_get_tile_rgba((ubyte *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_I8_UNORM:
-      i8_get_tile_rgba((ubyte *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_A8L8_UNORM:
-      a8l8_get_tile_rgba((ushort *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_R16_SNORM:
-      r16_get_tile_rgba((short *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      r16g16b16a16_get_tile_rgba((short *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_Z16_UNORM:
-      z16_get_tile_rgba((ushort *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_Z32_UNORM:
-      z32_get_tile_rgba((unsigned *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_S8Z24_UNORM:
-   case PIPE_FORMAT_X8Z24_UNORM:
-      s8z24_get_tile_rgba((unsigned *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_Z24S8_UNORM:
-   case PIPE_FORMAT_Z24X8_UNORM:
-      z24s8_get_tile_rgba((unsigned *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_Z32_FLOAT:
-      z32f_get_tile_rgba((float *) src, w, h, p);
-      break;
-   case PIPE_FORMAT_YCBCR:
-      ycbcr_get_tile_rgba((ushort *) src, w, h, p, FALSE);
-      break;
-   case PIPE_FORMAT_YCBCR_REV:
-      ycbcr_get_tile_rgba((ushort *) src, w, h, p, TRUE);
-      break;
-   default:
-      debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(format));
-      fake_get_tile_rgba(src, w, h, p);
-   }
-}
-
-
-void
-lp_get_tile_rgba_soa(struct pipe_transfer *pt,
-                     uint x, uint y,
-                     uint8_t *p)
-{
-   uint w = TILE_SIZE, h = TILE_SIZE;
-   void *packed;
-
-   if (pipe_clip_tile(x, y, &w, &h, pt))
-      return;
-
-   packed = MALLOC(pf_get_nblocks(&pt->block, w, h) * pt->block.size);
-
-   if (!packed)
-      return;
-
-   if(pt->format == PIPE_FORMAT_YCBCR || pt->format == PIPE_FORMAT_YCBCR_REV)
-      assert((x & 1) == 0);
-
-   pipe_get_tile_raw(pt, x, y, w, h, packed, 0);
-
-   lp_tile_raw_to_rgba_soa(pt->format, packed, w, h, p);
-
-   FREE(packed);
-}
-
-
-void
-lp_put_tile_rgba_soa(struct pipe_transfer *pt,
-                     uint x, uint y,
-                     const uint8_t *p)
-{
-   uint w = TILE_SIZE, h = TILE_SIZE;
-   void *packed;
-
-   if (pipe_clip_tile(x, y, &w, &h, pt))
-      return;
-
-   packed = MALLOC(pf_get_nblocks(&pt->block, w, h) * pt->block.size);
-
-   if (!packed)
-      return;
-
-   switch (pt->format) {
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      a8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_X8R8G8B8_UNORM:
-      x8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      b8g8r8a8_put_tile_rgba((unsigned *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_A1R5G5B5_UNORM:
-      a1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_R5G6B5_UNORM:
-      r5g6b5_put_tile_rgba((ushort *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_R8G8B8A8_UNORM:
-      assert(0);
-      break;
-   case PIPE_FORMAT_A4R4G4B4_UNORM:
-      a4r4g4b4_put_tile_rgba((ushort *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_L8_UNORM:
-      l8_put_tile_rgba((ubyte *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_A8_UNORM:
-      a8_put_tile_rgba((ubyte *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_I8_UNORM:
-      i8_put_tile_rgba((ubyte *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_A8L8_UNORM:
-      a8l8_put_tile_rgba((ushort *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_R16_SNORM:
-      r16_put_tile_rgba((short *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      r16g16b16a16_put_tile_rgba((short *) packed, w, h, p);
-      break;
-   case PIPE_FORMAT_Z16_UNORM:
-      /*z16_put_tile_rgba((ushort *) packed, w, h, p);*/
-      break;
-   case PIPE_FORMAT_Z32_UNORM:
-      /*z32_put_tile_rgba((unsigned *) packed, w, h, p);*/
-      break;
-   case PIPE_FORMAT_S8Z24_UNORM:
-   case PIPE_FORMAT_X8Z24_UNORM:
-      /*s8z24_put_tile_rgba((unsigned *) packed, w, h, p);*/
-      break;
-   case PIPE_FORMAT_Z24S8_UNORM:
-   case PIPE_FORMAT_Z24X8_UNORM:
-      /*z24s8_put_tile_rgba((unsigned *) packed, w, h, p);*/
-      break;
-   default:
-      debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(pt->format));
-   }
-
-   pipe_put_tile_raw(pt, x, y, w, h, packed, 0);
-
-   FREE(packed);
-}
-
-
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_soa.h b/src/gallium/drivers/llvmpipe/lp_tile_soa.h
index 3d8c703b73..040b01865d 100644
--- a/src/gallium/drivers/llvmpipe/lp_tile_soa.h
+++ b/src/gallium/drivers/llvmpipe/lp_tile_soa.h
@@ -64,14 +64,18 @@ tile_offset[TILE_VECTOR_HEIGHT][TILE_VECTOR_WIDTH];
 
 
 void
-lp_get_tile_rgba_soa(struct pipe_transfer *pt,
-                     uint x, uint y,
-                     uint8_t *p);
+lp_tile_read_4ub(enum pipe_format format,
+                 uint8_t *dst,
+                 const void *src, unsigned src_stride,
+                 unsigned x, unsigned y, unsigned w, unsigned h);
+
 
 void
-lp_put_tile_rgba_soa(struct pipe_transfer *pt,
-                     uint x, uint y,
-                     const uint8_t *p);
+lp_tile_write_4ub(enum pipe_format format,
+                  const uint8_t *src,
+                  void *dst, unsigned dst_stride,
+                  unsigned x, unsigned y, unsigned w, unsigned h);
+
 
 
 #ifdef __cplusplus
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_soa.py b/src/gallium/drivers/llvmpipe/lp_tile_soa.py
new file mode 100644
index 0000000000..004c5c979e
--- /dev/null
+++ b/src/gallium/drivers/llvmpipe/lp_tile_soa.py
@@ -0,0 +1,278 @@
+#!/usr/bin/env python
+
+'''
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Pixel format accessor functions.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+'''
+
+
+import sys
+import os.path
+
+sys.path.insert(0, os.path.join(os.path.dirname(sys.argv[0]), '../../auxiliary/util'))
+
+from u_format_access import *
+
+
+def generate_format_read(format, dst_type, dst_native_type, dst_suffix):
+    '''Generate the function to read pixels from a particular format'''
+
+    name = short_name(format)
+
+    src_native_type = native_type(format)
+
+    print 'static void'
+    print 'lp_tile_%s_read_%s(%s *dst, const uint8_t *src, unsigned src_stride, unsigned x0, unsigned y0, unsigned w, unsigned h)' % (name, dst_suffix, dst_native_type)
+    print '{'
+    print '   unsigned x, y;'
+    print '   const uint8_t *src_row = src + y0*src_stride;'
+    print '   for (y = 0; y < h; ++y) {'
+    print '      const %s *src_pixel = (const %s *)(src_row + x0*%u);' % (src_native_type, src_native_type, format.stride())
+    print '      for (x = 0; x < w; ++x) {'
+
+    names = ['']*4
+    if format.colorspace == 'rgb':
+        for i in range(4):
+            swizzle = format.out_swizzle[i]
+            if swizzle < 4:
+                names[swizzle] += 'rgba'[i]
+    elif format.colorspace == 'zs':
+        swizzle = format.out_swizzle[0]
+        if swizzle < 4:
+            names[swizzle] = 'z'
+        else:
+            assert False
+    else:
+        assert False
+
+    if format.layout == ARITH:
+        print '         %s pixel = *src_pixel++;' % src_native_type
+        shift = 0;
+        for i in range(4):
+            src_type = format.in_types[i]
+            width = src_type.size
+            if names[i]:
+                value = 'pixel'
+                mask = (1 << width) - 1
+                if shift:
+                    value = '(%s >> %u)' % (value, shift)
+                if shift + width < format.block_size():
+                    value = '(%s & 0x%x)' % (value, mask)
+                value = conversion_expr(src_type, dst_type, dst_native_type, value)
+                print '         %s %s = %s;' % (dst_native_type, names[i], value)
+            shift += width
+    elif format.layout == ARRAY:
+        for i in range(4):
+            src_type = format.in_types[i]
+            if names[i]:
+                value = '(*src_pixel++)'
+                value = conversion_expr(src_type, dst_type, dst_native_type, value)
+                print '         %s %s = %s;' % (dst_native_type, names[i], value)
+    else:
+        assert False
+
+    for i in range(4):
+        if format.colorspace == 'rgb':
+            swizzle = format.out_swizzle[i]
+            if swizzle < 4:
+                value = names[swizzle]
+            elif swizzle == SWIZZLE_0:
+                value = '0'
+            elif swizzle == SWIZZLE_1:
+                value = '1'
+            else:
+                assert False
+        elif format.colorspace == 'zs':
+            if i < 3:
+                value = 'z'
+            else:
+                value = '1'
+        else:
+            assert False
+        print '         TILE_PIXEL(dst, x, y, %u) = %s; /* %s */' % (i, value, 'rgba'[i])
+
+    print '      }'
+    print '      src_row += src_stride;'
+    print '   }'
+    print '}'
+    print
+    
+
+def generate_format_write(format, src_type, src_native_type, src_suffix):
+    '''Generate the function to write pixels to a particular format'''
+
+    name = short_name(format)
+
+    dst_native_type = native_type(format)
+
+    print 'static void'
+    print 'lp_tile_%s_write_%s(const %s *src, uint8_t *dst, unsigned dst_stride, unsigned x0, unsigned y0, unsigned w, unsigned h)' % (name, src_suffix, src_native_type)
+    print '{'
+    print '   unsigned x, y;'
+    print '   uint8_t *dst_row = dst + y0*dst_stride;'
+    print '   for (y = 0; y < h; ++y) {'
+    print '      %s *dst_pixel = (%s *)(dst_row + x0*%u);' % (dst_native_type, dst_native_type, format.stride())
+    print '      for (x = 0; x < w; ++x) {'
+
+    inv_swizzle = [None]*4
+    if format.colorspace == 'rgb':
+        for i in range(4):
+            swizzle = format.out_swizzle[i]
+            if swizzle < 4:
+                inv_swizzle[swizzle] = i
+    elif format.colorspace == 'zs':
+        swizzle = format.out_swizzle[0]
+        if swizzle < 4:
+            inv_swizzle[swizzle] = 0
+    else:
+        assert False
+
+    if format.layout == ARITH:
+        print '         %s pixel = 0;' % dst_native_type
+        shift = 0;
+        for i in range(4):
+            dst_type = format.in_types[i]
+            width = dst_type.size
+            if inv_swizzle[i] is not None:
+                value = 'TILE_PIXEL(src, x, y, %u)' % inv_swizzle[i]
+                value = conversion_expr(src_type, dst_type, dst_native_type, value)
+                if shift:
+                    value = '(%s << %u)' % (value, shift)
+                print '         pixel |= %s;' % value
+            shift += width
+        print '         *dst_pixel++ = pixel;'
+    elif format.layout == ARRAY:
+        for i in range(4):
+            dst_type = format.in_types[i]
+            if inv_swizzle[i] is not None:
+                value = 'TILE_PIXEL(src, x, y, %u)' % inv_swizzle[i]
+                value = conversion_expr(src_type, dst_type, dst_native_type, value)
+                print '         *dst_pixel++ = %s;' % value
+    else:
+        assert False
+
+    print '      }'
+    print '      dst_row += dst_stride;'
+    print '   }'
+    print '}'
+    print
+    
+
+def generate_read(formats, dst_type, dst_native_type, dst_suffix):
+    '''Generate the dispatch function to read pixels from any format'''
+
+    for format in formats:
+        if is_format_supported(format):
+            generate_format_read(format, dst_type, dst_native_type, dst_suffix)
+
+    print 'void'
+    print 'lp_tile_read_%s(enum pipe_format format, %s *dst, const void *src, unsigned src_stride, unsigned x, unsigned y, unsigned w, unsigned h)' % (dst_suffix, dst_native_type)
+    print '{'
+    print '   void (*func)(%s *dst, const uint8_t *src, unsigned src_stride, unsigned x0, unsigned y0, unsigned w, unsigned h);' % dst_native_type
+    print '   switch(format) {'
+    for format in formats:
+        if is_format_supported(format):
+            print '   case %s:' % format.name
+            print '      func = &lp_tile_%s_read_%s;' % (short_name(format), dst_suffix)
+            print '      break;'
+    print '   default:'
+    print '      debug_printf("unsupported format\\n");'
+    print '      return;'
+    print '   }'
+    print '   func(dst, (const uint8_t *)src, src_stride, x, y, w, h);'
+    print '}'
+    print
+
+
+def generate_write(formats, src_type, src_native_type, src_suffix):
+    '''Generate the dispatch function to write pixels to any format'''
+
+    for format in formats:
+        if is_format_supported(format):
+            generate_format_write(format, src_type, src_native_type, src_suffix)
+
+    print 'void'
+    print 'lp_tile_write_%s(enum pipe_format format, const %s *src, void *dst, unsigned dst_stride, unsigned x, unsigned y, unsigned w, unsigned h)' % (src_suffix, src_native_type)
+    
+    print '{'
+    print '   void (*func)(const %s *src, uint8_t *dst, unsigned dst_stride, unsigned x0, unsigned y0, unsigned w, unsigned h);' % src_native_type
+    print '   switch(format) {'
+    for format in formats:
+        if is_format_supported(format):
+            print '   case %s:' % format.name
+            print '      func = &lp_tile_%s_write_%s;' % (short_name(format), src_suffix)
+            print '      break;'
+    print '   default:'
+    print '      debug_printf("unsupported format\\n");'
+    print '      return;'
+    print '   }'
+    print '   func(src, (uint8_t *)dst, dst_stride, x, y, w, h);'
+    print '}'
+    print
+
+
+def main():
+    formats = []
+    for arg in sys.argv[1:]:
+        formats.extend(parse(arg))
+
+    print '/* This file is autogenerated by lp_tile_soa.py from u_format.csv. Do not edit directly. */'
+    print
+    # This will print the copyright message on the top of this file
+    print __doc__.strip()
+    print
+    print '#include "pipe/p_compiler.h"'
+    print '#include "util/u_format.h"'
+    print '#include "util/u_math.h"'
+    print '#include "lp_tile_soa.h"'
+    print
+    print 'const unsigned char'
+    print 'tile_offset[TILE_VECTOR_HEIGHT][TILE_VECTOR_WIDTH] = {'
+    print '   {  0,  1,  4,  5,  8,  9, 12, 13},'
+    print '   {  2,  3,  6,  7, 10, 11, 14, 15}'
+    print '};'
+    print
+
+    generate_clamp()
+
+    type = Type(UNSIGNED, True, 8)
+    native_type = 'uint8_t'
+    suffix = '4ub'
+
+    generate_read(formats, type, native_type, suffix)
+    generate_write(formats, type, native_type, suffix)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/src/gallium/drivers/nouveau/Makefile b/src/gallium/drivers/nouveau/Makefile
index dbe8a6e7bf..0cb66041d5 100644
--- a/src/gallium/drivers/nouveau/Makefile
+++ b/src/gallium/drivers/nouveau/Makefile
@@ -3,6 +3,7 @@ include $(TOP)/configs/current
 
 LIBNAME = nouveau
 
-C_SOURCES = nouveau_screen.c
+C_SOURCES = nouveau_screen.c \
+	    nouveau_context.c
 
 include ../../Makefile.template
diff --git a/src/gallium/drivers/nouveau/nouveau_context.c b/src/gallium/drivers/nouveau/nouveau_context.c
new file mode 100644
index 0000000000..23443869e6
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_context.c
@@ -0,0 +1,41 @@
+#include <pipe/p_defines.h>
+#include <pipe/p_context.h>
+
+#include "nouveau/nouveau_screen.h"
+#include "nouveau/nouveau_context.h"
+
+#include "nouveau/nouveau_bo.h"
+
+static unsigned int
+nouveau_reference_flags(struct nouveau_bo *bo)
+{
+	uint32_t bo_flags;
+	int flags = 0;
+
+	bo_flags = nouveau_bo_pending(bo);
+	if (bo_flags & NOUVEAU_BO_RD)
+		flags |= PIPE_REFERENCED_FOR_READ;
+	if (bo_flags & NOUVEAU_BO_WR)
+		flags |= PIPE_REFERENCED_FOR_WRITE;
+
+	return flags;
+}
+
+unsigned int
+nouveau_is_texture_referenced(struct pipe_context *pipe,
+			      struct pipe_texture *pt,
+			      unsigned face, unsigned level)
+{
+	struct nouveau_miptree *mt = nouveau_miptree(pt);
+
+	return nouveau_reference_flags(mt->bo);
+}
+
+unsigned int
+nouveau_is_buffer_referenced(struct pipe_context *pipe, struct pipe_buffer *pb)
+{
+	struct nouveau_bo *bo = nouveau_bo(pb);
+
+	return nouveau_reference_flags(bo);
+}
+
diff --git a/src/gallium/drivers/nouveau/nouveau_context.h b/src/gallium/drivers/nouveau/nouveau_context.h
new file mode 100644
index 0000000000..6a28d40da7
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_context.h
@@ -0,0 +1,11 @@
+#ifndef __NOUVEAU_CONTEXT_H__
+#define __NOUVEAU_CONTEXT_H__
+
+unsigned int
+nouveau_is_texture_referenced(struct pipe_context *, struct pipe_texture *,
+			      unsigned face, unsigned level);
+
+unsigned int
+nouveau_is_buffer_referenced(struct pipe_context *, struct pipe_buffer *);
+
+#endif
diff --git a/src/gallium/drivers/nv04/nv04_context.c b/src/gallium/drivers/nv04/nv04_context.c
index 17166c9f51..10d984ace9 100644
--- a/src/gallium/drivers/nv04/nv04_context.c
+++ b/src/gallium/drivers/nv04/nv04_context.c
@@ -64,30 +64,6 @@ nv04_init_hwctx(struct nv04_context *nv04)
 	return TRUE;
 }
 
-static unsigned int
-nv04_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv04_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-
 struct pipe_context *
 nv04_create(struct pipe_screen *pscreen, unsigned pctx_id)
 {
@@ -113,8 +89,8 @@ nv04_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv04->pipe.clear = nv04_clear;
 	nv04->pipe.flush = nv04_flush;
 
-	nv04->pipe.is_texture_referenced = nv04_is_texture_referenced;
-	nv04->pipe.is_buffer_referenced = nv04_is_buffer_referenced;
+	nv04->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv04->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
 
 	nv04_init_surface_functions(nv04);
 	nv04_init_state_functions(nv04);
diff --git a/src/gallium/drivers/nv04/nv04_context.h b/src/gallium/drivers/nv04/nv04_context.h
index 2842b2c90d..55326c787a 100644
--- a/src/gallium/drivers/nv04/nv04_context.h
+++ b/src/gallium/drivers/nv04/nv04_context.h
@@ -13,6 +13,7 @@
 
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
+#include "nouveau/nouveau_context.h"
 
 #define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
 	struct nv04_screen *ctx = nv04->screen
diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c
index ff2febb668..170ce3eb7e 100644
--- a/src/gallium/drivers/nv04/nv04_screen.c
+++ b/src/gallium/drivers/nv04/nv04_screen.c
@@ -16,8 +16,6 @@ nv04_screen_get_param(struct pipe_screen *screen, int param)
 		return 0;
 	case PIPE_CAP_GLSL:
 		return 0;
-	case PIPE_CAP_S3TC:
-		return 0;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 0;
 	case PIPE_CAP_POINT_SPRITE:
diff --git a/src/gallium/drivers/nv04/nv04_surface_2d.c b/src/gallium/drivers/nv04/nv04_surface_2d.c
index f88e138c79..8c7eb367e2 100644
--- a/src/gallium/drivers/nv04/nv04_surface_2d.c
+++ b/src/gallium/drivers/nv04/nv04_surface_2d.c
@@ -1,5 +1,6 @@
 #include "pipe/p_context.h"
 #include "pipe/p_format.h"
+#include "util/u_math.h"
 #include "util/u_memory.h"
 
 #include "nouveau/nouveau_winsys.h"
@@ -12,10 +13,13 @@ nv04_surface_format(enum pipe_format format)
 {
 	switch (format) {
 	case PIPE_FORMAT_A8_UNORM:
+	case PIPE_FORMAT_L8_UNORM:
+	case PIPE_FORMAT_I8_UNORM:
 		return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8;
 	case PIPE_FORMAT_R16_SNORM:
 	case PIPE_FORMAT_R5G6B5_UNORM:
 	case PIPE_FORMAT_Z16_UNORM:
+	case PIPE_FORMAT_A8L8_UNORM:
 		return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5;
 	case PIPE_FORMAT_X8R8G8B8_UNORM:
 	case PIPE_FORMAT_A8R8G8B8_UNORM:
@@ -35,6 +39,7 @@ nv04_rect_format(enum pipe_format format)
 	case PIPE_FORMAT_A8_UNORM:
 		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8;
 	case PIPE_FORMAT_R5G6B5_UNORM:
+	case PIPE_FORMAT_A8L8_UNORM:
 	case PIPE_FORMAT_Z16_UNORM:
 		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5;
 	case PIPE_FORMAT_A8R8G8B8_UNORM:
@@ -50,6 +55,10 @@ static INLINE int
 nv04_scaled_image_format(enum pipe_format format)
 {
 	switch (format) {
+	case PIPE_FORMAT_A8_UNORM:
+	case PIPE_FORMAT_L8_UNORM:
+	case PIPE_FORMAT_I8_UNORM:
+		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8;
 	case PIPE_FORMAT_A1R5G5B5_UNORM:
 		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5;
 	case PIPE_FORMAT_A8R8G8B8_UNORM:
@@ -58,6 +67,7 @@ nv04_scaled_image_format(enum pipe_format format)
 		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8;
 	case PIPE_FORMAT_R5G6B5_UNORM:
 	case PIPE_FORMAT_R16_SNORM:
+	case PIPE_FORMAT_A8L8_UNORM:
 		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5;
 	default:
 		return -1;
@@ -107,17 +117,20 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx,
 	struct nouveau_bo *src_bo = nouveau_bo(ctx->buf(src));
 	struct nouveau_bo *dst_bo = nouveau_bo(ctx->buf(dst));
 	const unsigned src_pitch = ((struct nv04_surface *)src)->pitch;
+        /* Max width & height may not be the same on all HW, but must be POT */
 	const unsigned max_w = 1024;
 	const unsigned max_h = 1024;
-	const unsigned sub_w = w > max_w ? max_w : w;
-	const unsigned sub_h = h > max_h ? max_h : h;
-	unsigned cx;
-	unsigned cy;
+	unsigned sub_w = w > max_w ? max_w : w;
+	unsigned sub_h = h > max_h ? max_h : h;
+	unsigned x;
+	unsigned y;
+
+        /* Swizzled surfaces must be POT  */
+	assert(util_is_pot(dst->width) && util_is_pot(dst->height));
 
-#if 0
-	/* That's the way she likes it */
-	assert(src_pitch == ((struct nv04_surface *)dst)->pitch);
-#endif
+        /* If area is too large to copy in one shot we must copy it in POT chunks to meet alignment requirements */
+	assert(sub_w == w || util_is_pot(sub_w));
+	assert(sub_h == h || util_is_pot(sub_h));
 
 	BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1);
 	OUT_RELOCo(chan, dst_bo,
@@ -125,41 +138,46 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx,
 
 	BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_FORMAT, 1);
 	OUT_RING  (chan, nv04_surface_format(dst->format) |
-	                 log2i(w) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT |
-	                 log2i(h) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT);
- 
+	                 log2i(dst->width) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT |
+	                 log2i(dst->height) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT);
+
 	BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE, 1);
 	OUT_RELOCo(chan, src_bo,
 	                 NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
 	BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE, 1);
 	OUT_RING  (chan, swzsurf->handle);
 
-	for (cy = 0; cy < h; cy += sub_h) {
-	  for (cx = 0; cx < w; cx += sub_w) {
+	for (y = 0; y < h; y += sub_h) {
+	  sub_h = MIN2(sub_h, h - y);
+
+	  for (x = 0; x < w; x += sub_w) {
+	    sub_w = MIN2(sub_w, w - x);
+
+	    /* Must be 64-byte aligned */
+	    assert(!((dst->offset + nv04_swizzle_bits(dx+x, dy+y) * dst->texture->block.size) & 63));
+
 	    BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_OFFSET, 1);
-	    OUT_RELOCl(chan, dst_bo, dst->offset + nv04_swizzle_bits(cx+dx, cy+dy) *
-			     dst->texture->block.size, NOUVEAU_BO_GART |
-			     NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	    OUT_RELOCl(chan, dst_bo, dst->offset + nv04_swizzle_bits(dx+x, dy+y) * dst->texture->block.size,
+                             NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 
 	    BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION, 9);
 	    OUT_RING  (chan, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE);
 	    OUT_RING  (chan, nv04_scaled_image_format(src->format));
 	    OUT_RING  (chan, NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY);
 	    OUT_RING  (chan, 0);
-	    OUT_RING  (chan, sub_h << 16 | sub_w);
+	    OUT_RING  (chan, sub_h << NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT | sub_w);
 	    OUT_RING  (chan, 0);
-	    OUT_RING  (chan, sub_h << 16 | sub_w);
+	    OUT_RING  (chan, sub_h << NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_SHIFT | sub_w);
 	    OUT_RING  (chan, 1 << 20);
 	    OUT_RING  (chan, 1 << 20);
 
 	    BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_SIZE, 4);
-	    OUT_RING  (chan, sub_h << 16 | sub_w);
+	    OUT_RING  (chan, sub_h << NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_SHIFT | sub_w);
 	    OUT_RING  (chan, src_pitch |
 			     NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER |
 			     NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE);
-	    OUT_RELOCl(chan, src_bo, src->offset + (cy+sy) * src_pitch +
-			     (cx+sx) * src->texture->block.size, NOUVEAU_BO_GART |
-			     NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	    OUT_RELOCl(chan, src_bo, src->offset + (sy+y) * src_pitch + (sx+x) * src->texture->block.size,
+                             NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
 	    OUT_RING  (chan, 0);
 	  }
 	}
@@ -214,43 +232,6 @@ nv04_surface_copy_m2mf(struct nv04_surface_2d *ctx,
 }
 
 static int
-nv04_surface_copy_m2mf_swizzle(struct nv04_surface_2d *ctx,
-			       struct pipe_surface *dst, int dx, int dy,
-			       struct pipe_surface *src, int sx, int sy)
-{
-	struct nouveau_channel *chan = ctx->m2mf->channel;
-	struct nouveau_grobj *m2mf = ctx->m2mf;
-	struct nouveau_bo *src_bo = nouveau_bo(ctx->buf(src));
-	struct nouveau_bo *dst_bo = nouveau_bo(ctx->buf(dst));
-	unsigned src_pitch = ((struct nv04_surface *)src)->pitch;
-	unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch;
-	unsigned dst_offset = dst->offset + nv04_swizzle_bits(dx, dy) *
-	                      dst->texture->block.size;
-	unsigned src_offset = src->offset + sy * src_pitch +
-	                      sx * src->texture->block.size;
-
-	BEGIN_RING(chan, m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2);
-	OUT_RELOCo(chan, src_bo,
-		   NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
-	OUT_RELOCo(chan, dst_bo,
-		   NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
-
-	BEGIN_RING(chan, m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
-	OUT_RELOCl(chan, src_bo, src_offset,
-		   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
-	OUT_RELOCl(chan, dst_bo, dst_offset,
-		   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR);
-	OUT_RING  (chan, src_pitch);
-	OUT_RING  (chan, dst_pitch);
-	OUT_RING  (chan, 1 * src->texture->block.size);
-	OUT_RING  (chan, 1);
-	OUT_RING  (chan, 0x0101);
-	OUT_RING  (chan, 0);
-
-	return 0;
-}
-
-static int
 nv04_surface_copy_blit(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 		       int dx, int dy, struct pipe_surface *src, int sx, int sy,
 		       int w, int h)
@@ -299,61 +280,10 @@ nv04_surface_copy(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 	assert(src->format == dst->format);
 
 	/* Setup transfer to swizzle the texture to vram if needed */
-	if (src_linear && !dst_linear) {
-		int x,y;
-
-		if ((w>1) && (h>1)) {
-			int potWidth = 1<<log2i(w);
-			int potHeight = 1<<log2i(h);
-			int remainWidth = w-potWidth;
-			int remainHeight = h-potHeight;
-			int squareDim = (potWidth>potHeight ? potHeight : potWidth);
-
-			/* top left is always POT, but we can only swizzle squares */
-			for (y=0; y<potHeight; y+=squareDim) {
-				for (x=0; x<potWidth; x+= squareDim) {
-					nv04_surface_copy_swizzle(ctx, dst, dx+x, dy+y,
-					                          src, sx+x, sy+y,
-					                          squareDim, squareDim);
-				}
-			}
-
-			/* top right */
-			if (remainWidth>0) {
-			nv04_surface_copy(ctx, dst, dx+potWidth, dy,
-				                  src, sx+potWidth, sy,
-				                  remainWidth, potHeight);
-			}
-
-			/* bottom left */
-			if (remainHeight>0) {
-				nv04_surface_copy(ctx, dst, dx, dy+potHeight,
-			                  src, sx, sy+potHeight,
-				                  potWidth, remainHeight);
-			}
-
-			/* bottom right */
-			if ((remainWidth>0) && (remainHeight>0)) {
-				nv04_surface_copy(ctx, dst, dx+potWidth, dy+potHeight,
-				                  src, sx+potWidth, sy+potHeight,
-				                  remainWidth, remainHeight);
-			}
-		} else if (w==1) {
-			/* We have a column to copy to a swizzled texture */
-			for (y=0; y<h; y++) {
-				nv04_surface_copy_m2mf_swizzle(ctx, dst, dx, dy+y,
-				                               src, sx, sy+y);
-			}
-		} else if (h==1) {
-			/* We have a row to copy to a swizzled texture */
-			for (x=0; x<w; x++) {
-				nv04_surface_copy_m2mf_swizzle(ctx, dst, dx+x, dy,
-				                               src, sx+x, sy);
-			}
-		}
-
-		return;
-	}
+        if (src_linear && !dst_linear && w > 1 && h > 1) {
+           nv04_surface_copy_swizzle(ctx, dst, dx, dy, src, sx, sy, w, h);
+           return;
+        }
 
 	/* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback
 	 * to NV_MEMORY_TO_MEMORY_FORMAT in this case.
diff --git a/src/gallium/drivers/nv04/nv04_transfer.c b/src/gallium/drivers/nv04/nv04_transfer.c
index 854b855d64..6618660743 100644
--- a/src/gallium/drivers/nv04/nv04_transfer.c
+++ b/src/gallium/drivers/nv04/nv04_transfer.c
@@ -13,22 +13,6 @@ struct nv04_transfer {
 	bool direct;
 };
 
-static unsigned nv04_usage_tx_to_buf(unsigned tx_usage)
-{
-	switch (tx_usage) {
-		case PIPE_TRANSFER_READ:
-			return PIPE_BUFFER_USAGE_CPU_READ;
-		case PIPE_TRANSFER_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_WRITE;
-		case PIPE_TRANSFER_READ_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
 static void
 nv04_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
                              struct pipe_texture *template)
@@ -86,7 +70,7 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		tx->direct = true;
 		tx->surface = pscreen->get_tex_surface(pscreen, pt,
 	                                               0, 0, 0,
-	                                               nv04_usage_tx_to_buf(usage));
+	                                               pipe_transfer_buffer_flags(&tx->base));
 		return &tx->base;
 	}
 
@@ -103,7 +87,7 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 
 	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
 	                                       face, level, zslice,
-	                                       nv04_usage_tx_to_buf(usage));
+	                                       pipe_transfer_buffer_flags(&tx->base));
 
 	pipe_texture_reference(&tx_tex, NULL);
 
@@ -114,7 +98,7 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
-	if (usage != PIPE_TRANSFER_WRITE) {
+	if (usage & PIPE_TRANSFER_READ) {
 		struct nv04_screen *nvscreen = nv04_screen(pscreen);
 		struct pipe_surface *src;
 
@@ -140,7 +124,7 @@ nv04_transfer_del(struct pipe_transfer *ptx)
 {
 	struct nv04_transfer *tx = (struct nv04_transfer *)ptx;
 
-	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+	if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) {
 		struct pipe_screen *pscreen = ptx->texture->screen;
 		struct nv04_screen *nvscreen = nv04_screen(pscreen);
 		struct pipe_surface *dst;
@@ -170,7 +154,7 @@ nv04_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
 	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
 	struct nv04_miptree *mt = (struct nv04_miptree *)tx->surface->texture;
 	void *map = pipe_buffer_map(pscreen, mt->buffer,
-	                            nv04_usage_tx_to_buf(ptx->usage));
+	                            pipe_transfer_buffer_flags(ptx));
 
 	return map + ns->base.offset +
 	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c
index a127b134ec..933176fc32 100644
--- a/src/gallium/drivers/nv10/nv10_context.c
+++ b/src/gallium/drivers/nv10/nv10_context.c
@@ -257,29 +257,6 @@ nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
 {
 }
 
-static unsigned int
-nv10_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv10_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
 struct pipe_context *
 nv10_create(struct pipe_screen *pscreen, unsigned pctx_id)
 {
@@ -305,8 +282,8 @@ nv10_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv10->pipe.clear = nv10_clear;
 	nv10->pipe.flush = nv10_flush;
 
-	nv10->pipe.is_texture_referenced = nv10_is_texture_referenced;
-	nv10->pipe.is_buffer_referenced = nv10_is_buffer_referenced;
+	nv10->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv10->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
 
 	nv10_init_surface_functions(nv10);
 	nv10_init_state_functions(nv10);
diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h
index f1e003c953..36a6aa7a74 100644
--- a/src/gallium/drivers/nv10/nv10_context.h
+++ b/src/gallium/drivers/nv10/nv10_context.h
@@ -13,6 +13,7 @@
 
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
+#include "nouveau/nouveau_context.h"
 
 #define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
 	struct nv10_screen *ctx = nv10->screen
diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c
index 4469b22d91..ee5901e743 100644
--- a/src/gallium/drivers/nv10/nv10_screen.c
+++ b/src/gallium/drivers/nv10/nv10_screen.c
@@ -15,8 +15,6 @@ nv10_screen_get_param(struct pipe_screen *screen, int param)
 		return 0;
 	case PIPE_CAP_GLSL:
 		return 0;
-	case PIPE_CAP_S3TC:
-		return 0;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 1;
 	case PIPE_CAP_POINT_SPRITE:
diff --git a/src/gallium/drivers/nv10/nv10_transfer.c b/src/gallium/drivers/nv10/nv10_transfer.c
index c06b8d34c7..8feb85e4bd 100644
--- a/src/gallium/drivers/nv10/nv10_transfer.c
+++ b/src/gallium/drivers/nv10/nv10_transfer.c
@@ -13,22 +13,6 @@ struct nv10_transfer {
 	bool direct;
 };
 
-static unsigned nv10_usage_tx_to_buf(unsigned tx_usage)
-{
-	switch (tx_usage) {
-		case PIPE_TRANSFER_READ:
-			return PIPE_BUFFER_USAGE_CPU_READ;
-		case PIPE_TRANSFER_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_WRITE;
-		case PIPE_TRANSFER_READ_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
 static void
 nv10_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
                              struct pipe_texture *template)
@@ -86,7 +70,7 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		tx->direct = true;
 		tx->surface = pscreen->get_tex_surface(pscreen, pt,
 	                                               0, 0, 0,
-	                                               nv10_usage_tx_to_buf(usage));
+	                                               pipe_transfer_buffer_flags(&tx->base));
 		return &tx->base;
 	}
 
@@ -103,7 +87,7 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 
 	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
 	                                       face, level, zslice,
-	                                       nv10_usage_tx_to_buf(usage));
+	                                       pipe_transfer_buffer_flags(&tx->base));
 
 	pipe_texture_reference(&tx_tex, NULL);
 
@@ -114,7 +98,7 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
-	if (usage != PIPE_TRANSFER_WRITE) {
+	if (usage & PIPE_TRANSFER_READ) {
 		struct nv10_screen *nvscreen = nv10_screen(pscreen);
 		struct pipe_surface *src;
 
@@ -140,7 +124,7 @@ nv10_transfer_del(struct pipe_transfer *ptx)
 {
 	struct nv10_transfer *tx = (struct nv10_transfer *)ptx;
 
-	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+	if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) {
 		struct pipe_screen *pscreen = ptx->texture->screen;
 		struct nv10_screen *nvscreen = nv10_screen(pscreen);
 		struct pipe_surface *dst;
@@ -170,7 +154,7 @@ nv10_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
 	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
 	struct nv10_miptree *mt = (struct nv10_miptree *)tx->surface->texture;
 	void *map = pipe_buffer_map(pscreen, mt->buffer,
-	                            nv10_usage_tx_to_buf(ptx->usage));
+	                            pipe_transfer_buffer_flags(ptx));
 
 	return map + ns->base.offset +
 	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
diff --git a/src/gallium/drivers/nv20/nv20_context.c b/src/gallium/drivers/nv20/nv20_context.c
index b32d0d83ba..9a48739661 100644
--- a/src/gallium/drivers/nv20/nv20_context.c
+++ b/src/gallium/drivers/nv20/nv20_context.c
@@ -380,30 +380,6 @@ nv20_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
 {
 }
 
-
-static unsigned int
-nv20_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv20_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
 struct pipe_context *
 nv20_create(struct pipe_screen *pscreen, unsigned pctx_id)
 {
@@ -429,8 +405,8 @@ nv20_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv20->pipe.clear = nv20_clear;
 	nv20->pipe.flush = nv20_flush;
 
-	nv20->pipe.is_texture_referenced = nv20_is_texture_referenced;
-	nv20->pipe.is_buffer_referenced = nv20_is_buffer_referenced;
+	nv20->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv20->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
 
 	nv20_init_surface_functions(nv20);
 	nv20_init_state_functions(nv20);
diff --git a/src/gallium/drivers/nv20/nv20_context.h b/src/gallium/drivers/nv20/nv20_context.h
index fc932f1f90..a4eaa95660 100644
--- a/src/gallium/drivers/nv20/nv20_context.h
+++ b/src/gallium/drivers/nv20/nv20_context.h
@@ -13,6 +13,7 @@
 
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
+#include "nouveau/nouveau_context.h"
 
 #define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
 	struct nv20_screen *ctx = nv20->screen
diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c
index e6924ad71e..4eeacd1afd 100644
--- a/src/gallium/drivers/nv20/nv20_screen.c
+++ b/src/gallium/drivers/nv20/nv20_screen.c
@@ -15,8 +15,6 @@ nv20_screen_get_param(struct pipe_screen *screen, int param)
 		return 0;
 	case PIPE_CAP_GLSL:
 		return 0;
-	case PIPE_CAP_S3TC:
-		return 0;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 1;
 	case PIPE_CAP_POINT_SPRITE:
diff --git a/src/gallium/drivers/nv20/nv20_transfer.c b/src/gallium/drivers/nv20/nv20_transfer.c
index 5018995596..81b4f1a917 100644
--- a/src/gallium/drivers/nv20/nv20_transfer.c
+++ b/src/gallium/drivers/nv20/nv20_transfer.c
@@ -13,22 +13,6 @@ struct nv20_transfer {
 	bool direct;
 };
 
-static unsigned nv20_usage_tx_to_buf(unsigned tx_usage)
-{
-	switch (tx_usage) {
-		case PIPE_TRANSFER_READ:
-			return PIPE_BUFFER_USAGE_CPU_READ;
-		case PIPE_TRANSFER_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_WRITE;
-		case PIPE_TRANSFER_READ_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
 static void
 nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
                              struct pipe_texture *template)
@@ -86,7 +70,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		tx->direct = true;
 		tx->surface = pscreen->get_tex_surface(pscreen, pt,
 	                                               0, 0, 0,
-	                                               nv20_usage_tx_to_buf(usage));
+	                                               pipe_transfer_buffer_flags(&tx->base));
 		return &tx->base;
 	}
 
@@ -103,7 +87,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 
 	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
 	                                       face, level, zslice,
-	                                       nv20_usage_tx_to_buf(usage));
+	                                       pipe_transfer_buffer_flags(&tx->base));
 
 	pipe_texture_reference(&tx_tex, NULL);
 
@@ -114,7 +98,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
-	if (usage != PIPE_TRANSFER_WRITE) {
+	if (usage & PIPE_TRANSFER_READ) {
 		struct nv20_screen *nvscreen = nv20_screen(pscreen);
 		struct pipe_surface *src;
 
@@ -140,7 +124,7 @@ nv20_transfer_del(struct pipe_transfer *ptx)
 {
 	struct nv20_transfer *tx = (struct nv20_transfer *)ptx;
 
-	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+	if (!tx->direct && (ptx->usage = PIPE_TRANSFER_WRITE)) {
 		struct pipe_screen *pscreen = ptx->texture->screen;
 		struct nv20_screen *nvscreen = nv20_screen(pscreen);
 		struct pipe_surface *dst;
@@ -170,7 +154,7 @@ nv20_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
 	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
 	struct nv20_miptree *mt = (struct nv20_miptree *)tx->surface->texture;
 	void *map = pipe_buffer_map(pscreen, mt->buffer,
-	                            nv20_usage_tx_to_buf(ptx->usage));
+	                            pipe_transfer_buffer_flags(ptx));
 
 	return map + ns->base.offset +
 	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c
index f827bdc78b..d8300fd69f 100644
--- a/src/gallium/drivers/nv30/nv30_context.c
+++ b/src/gallium/drivers/nv30/nv30_context.c
@@ -10,7 +10,7 @@ nv30_flush(struct pipe_context *pipe, unsigned flags,
 	   struct pipe_fence_handle **fence)
 {
 	struct nv30_context *nv30 = nv30_context(pipe);
-	
+
 	if (flags & PIPE_FLUSH_TEXTURE_CACHE) {
 		BEGIN_RING(rankine, 0x1fd8, 1);
 		OUT_RING  (2);
@@ -31,29 +31,6 @@ nv30_destroy(struct pipe_context *pipe)
 	FREE(nv30);
 }
 
-static unsigned int
-nv30_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv30_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
 struct pipe_context *
 nv30_create(struct pipe_screen *pscreen, unsigned pctx_id)
 {
@@ -78,8 +55,8 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv30->pipe.clear = nv30_clear;
 	nv30->pipe.flush = nv30_flush;
 
-	nv30->pipe.is_texture_referenced = nv30_is_texture_referenced;
-	nv30->pipe.is_buffer_referenced = nv30_is_buffer_referenced;
+	nv30->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv30->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
 
 	nv30_init_query_functions(nv30);
 	nv30_init_surface_functions(nv30);
@@ -95,4 +72,3 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id)
 
 	return &nv30->pipe;
 }
-	
diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h
index 4229c0a0e1..8d49366dfc 100644
--- a/src/gallium/drivers/nv30/nv30_context.h
+++ b/src/gallium/drivers/nv30/nv30_context.h
@@ -13,6 +13,7 @@
 
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
+#include "nouveau/nouveau_context.h"
 
 #define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
 	struct nv30_screen *ctx = nv30->screen
diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c
index 822e1d8def..f5f17d4071 100644
--- a/src/gallium/drivers/nv30/nv30_fragtex.c
+++ b/src/gallium/drivers/nv30/nv30_fragtex.c
@@ -30,7 +30,7 @@ nv30_texture_formats[] = {
 	_(I8_UNORM      , L8      ,   S1,   S1,   S1,   S1, X, X, X, X),
 	_(A8L8_UNORM    , A8L8    ,   S1,   S1,   S1,   S1, X, X, X, Y),
 //	_(Z16_UNORM     , Z16     ,   S1,   S1,   S1,  ONE, X, X, X, X),
-//	_(Z24S8_UNORM   , Z24     ,   S1,   S1,   S1,  ONE, X, X, X, X),
+	_(Z24S8_UNORM   , A8R8G8B8,   S1,   S1,   S1,  ONE, X, X, X, X),
 	_(DXT1_RGB      , DXT1    ,   S1,   S1,   S1,  ONE, X, Y, Z, W),
 	_(DXT1_RGBA     , DXT1    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
 	_(DXT3_RGBA     , DXT3    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
@@ -69,13 +69,13 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit)
 
 	tf = nv30_fragtex_format(pt->format);
 	if (!tf)
-		assert(0);
+		return NULL;
 
 	txf  = tf->format;
 	txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0);
-	txf |= log2i(pt->width[0]) << 20;
-	txf |= log2i(pt->height[0]) << 24;
-	txf |= log2i(pt->depth[0]) << 28;
+	txf |= log2i(pt->width[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT;
+	txf |= log2i(pt->height[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT;
+	txf |= log2i(pt->depth[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT;
 	txf |= NV34TCL_TX_FORMAT_NO_BORDER | 0x10000;
 
 	switch (pt->target) {
diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c
index 7f8054de73..17acca61ab 100644
--- a/src/gallium/drivers/nv30/nv30_miptree.c
+++ b/src/gallium/drivers/nv30/nv30_miptree.c
@@ -96,6 +96,11 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 		case PIPE_FORMAT_A8R8G8B8_UNORM:
 		case PIPE_FORMAT_X8R8G8B8_UNORM:
 		case PIPE_FORMAT_R16_SNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM:
+		case PIPE_FORMAT_A8L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_I8_UNORM:
 		{
 			if (debug_get_bool_option("NOUVEAU_NO_SWIZZLE", FALSE))
 				mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c
index f8285e4455..bb40e1803d 100644
--- a/src/gallium/drivers/nv30/nv30_screen.c
+++ b/src/gallium/drivers/nv30/nv30_screen.c
@@ -10,6 +10,22 @@
 #define NV34TCL_CHIPSET_3X_MASK 0x00000010
 #define NV35TCL_CHIPSET_3X_MASK 0x000001e0
 
+/* FIXME: It seems I should not include directly ../../winsys/drm/nouveau/drm/nouveau_drm_api.h
+ * to get the pointer to the context front buffer, so I copied nouveau_winsys here.
+ * nv30_screen_surface_format_supported() can then use it to enforce creating fbo
+ * with same number of bits everywhere.
+ */
+struct nouveau_winsys {
+	struct pipe_winsys base;
+
+	struct pipe_screen *pscreen;
+
+	unsigned nr_pctx;
+	struct pipe_context **pctx;
+
+	struct pipe_surface *front;
+};
+
 static int
 nv30_screen_get_param(struct pipe_screen *pscreen, int param)
 {
@@ -22,8 +38,6 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 1;
 	case PIPE_CAP_GLSL:
 		return 0;
-	case PIPE_CAP_S3TC:
-		return 0;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 1;
 	case PIPE_CAP_POINT_SPRITE:
@@ -85,21 +99,18 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen,
 				     enum pipe_texture_target target,
 				     unsigned tex_usage, unsigned geom_flags)
 {
+	struct pipe_surface *front = ((struct nouveau_winsys *) pscreen->winsys)->front;
+
 	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
-		switch (format) {
-		case PIPE_FORMAT_A8R8G8B8_UNORM:
-		case PIPE_FORMAT_R5G6B5_UNORM:
-			return TRUE;
-		default:
-			break;
-		}
+		return (format == front->format);
 	} else
 	if (tex_usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL) {
 		switch (format) {
 		case PIPE_FORMAT_Z24S8_UNORM:
 		case PIPE_FORMAT_Z24X8_UNORM:
+			return (front->format == PIPE_FORMAT_A8R8G8B8_UNORM);
 		case PIPE_FORMAT_Z16_UNORM:
-			return TRUE;
+			return (front->format == PIPE_FORMAT_R5G6B5_UNORM);
 		default:
 			break;
 		}
diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c
index 44b6a74715..f90681b0f9 100644
--- a/src/gallium/drivers/nv30/nv30_state_fb.c
+++ b/src/gallium/drivers/nv30/nv30_state_fb.c
@@ -8,15 +8,15 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 	struct nouveau_channel *chan = nv30->screen->base.channel;
 	struct nouveau_grobj *rankine = nv30->screen->rankine;
 	struct nv04_surface *rt[2], *zeta = NULL;
-	uint32_t rt_enable, rt_format;
-	int i, colour_format = 0, zeta_format = 0;
+	uint32_t rt_enable = 0, rt_format = 0;
+	int i, colour_format = 0, zeta_format = 0, depth_only = 0;
 	struct nouveau_stateobj *so = so_new(64, 10);
 	unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM;
 	unsigned w = fb->width;
 	unsigned h = fb->height;
 	struct nv30_miptree *nv30mt;
+	int colour_bits = 32, zeta_bits = 32;
 
-	rt_enable = 0;
 	for (i = 0; i < fb->nr_cbufs; i++) {
 		if (colour_format) {
 			assert(colour_format == fb->cbufs[i]->format);
@@ -35,18 +35,35 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 		zeta = (struct nv04_surface *)fb->zsbuf;
 	}
 
-	if (!(rt[0]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
-		assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
-		for (i = 1; i < fb->nr_cbufs; i++)
-			assert(!(rt[i]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
+	if (rt_enable & (NV34TCL_RT_ENABLE_COLOR0|NV34TCL_RT_ENABLE_COLOR1)) {
+		/* Render to at least a colour buffer */
+		if (!(rt[0]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+			assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
+			for (i = 1; i < fb->nr_cbufs; i++)
+				assert(!(rt[i]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
 
-		/* FIXME: NV34TCL_RT_FORMAT_LOG2_[WIDTH/HEIGHT] */
-		rt_format = NV34TCL_RT_FORMAT_TYPE_SWIZZLED |
-		log2i(fb->width) << 16 /*NV34TCL_RT_FORMAT_LOG2_WIDTH_SHIFT*/ |
-		log2i(fb->height) << 24 /*NV34TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT*/;
+			rt_format = NV34TCL_RT_FORMAT_TYPE_SWIZZLED |
+				(log2i(rt[0]->base.width) << NV34TCL_RT_FORMAT_LOG2_WIDTH_SHIFT) |
+				(log2i(rt[0]->base.height) << NV34TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT);
+		}
+		else
+			rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR;
+	} else if (fb->zsbuf) {
+		depth_only = 1;
+
+		/* Render to depth buffer only */
+		if (!(zeta->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+			assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
+
+			rt_format = NV34TCL_RT_FORMAT_TYPE_SWIZZLED |
+				(log2i(zeta->base.width) << NV34TCL_RT_FORMAT_LOG2_WIDTH_SHIFT) |
+				(log2i(zeta->base.height) << NV34TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT);
+		}
+		else
+			rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR;
+	} else {
+		return FALSE;
 	}
-	else
-		rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR;
 
 	switch (colour_format) {
 	case PIPE_FORMAT_A8R8G8B8_UNORM:
@@ -55,6 +72,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 		break;
 	case PIPE_FORMAT_R5G6B5_UNORM:
 		rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5;
+		colour_bits = 16;
 		break;
 	default:
 		assert(0);
@@ -63,6 +81,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 	switch (zeta_format) {
 	case PIPE_FORMAT_Z16_UNORM:
 		rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16;
+		zeta_bits = 16;
 		break;
 	case PIPE_FORMAT_Z24S8_UNORM:
 	case PIPE_FORMAT_Z24X8_UNORM:
@@ -73,21 +92,27 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 		assert(0);
 	}
 
-	if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) {
-		uint32_t pitch = rt[0]->pitch;
+	if (colour_bits != zeta_bits) {
+		return FALSE;
+	}
+
+	if (depth_only || (rt_enable & NV34TCL_RT_ENABLE_COLOR0)) {
+		struct nv04_surface *rt0 = (depth_only ? zeta : rt[0]);
+		uint32_t pitch = rt0->pitch;
+
 		if (zeta) {
 			pitch |= (zeta->pitch << 16);
 		} else {
 			pitch |= (pitch << 16);
 		}
 
-		nv30mt = (struct nv30_miptree *)rt[0]->base.texture;
+		nv30mt = (struct nv30_miptree *) rt0->base.texture;
 		so_method(so, rankine, NV34TCL_DMA_COLOR0, 1);
 		so_reloc (so, nouveau_bo(nv30mt->buffer), 0, rt_flags | NOUVEAU_BO_OR,
 			      chan->vram->handle, chan->gart->handle);
 		so_method(so, rankine, NV34TCL_COLOR0_PITCH, 2);
 		so_data  (so, pitch);
-		so_reloc (so, nouveau_bo(nv30mt->buffer), rt[0]->base.offset,
+		so_reloc (so, nouveau_bo(nv30mt->buffer), rt0->base.offset,
 			      rt_flags | NOUVEAU_BO_LOW, 0, 0);
 	}
 
diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c
index 2367571878..98011decf7 100644
--- a/src/gallium/drivers/nv30/nv30_transfer.c
+++ b/src/gallium/drivers/nv30/nv30_transfer.c
@@ -13,22 +13,6 @@ struct nv30_transfer {
 	bool direct;
 };
 
-static unsigned nv30_usage_tx_to_buf(unsigned tx_usage)
-{
-	switch (tx_usage) {
-		case PIPE_TRANSFER_READ:
-			return PIPE_BUFFER_USAGE_CPU_READ;
-		case PIPE_TRANSFER_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_WRITE;
-		case PIPE_TRANSFER_READ_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
 static void
 nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
                              struct pipe_texture *template)
@@ -86,7 +70,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		tx->direct = true;
 		tx->surface = pscreen->get_tex_surface(pscreen, pt,
 	                                               face, level, zslice,
-	                                               nv30_usage_tx_to_buf(usage));
+	                                               pipe_transfer_buffer_flags(&tx->base));
 		return &tx->base;
 	}
 
@@ -103,7 +87,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 
 	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
 	                                       0, 0, 0,
-	                                       nv30_usage_tx_to_buf(usage));
+	                                       pipe_transfer_buffer_flags(&tx->base));
 
 	pipe_texture_reference(&tx_tex, NULL);
 
@@ -114,7 +98,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
-	if (usage != PIPE_TRANSFER_WRITE) {
+	if (usage & PIPE_TRANSFER_READ) {
 		struct nv30_screen *nvscreen = nv30_screen(pscreen);
 		struct pipe_surface *src;
 
@@ -140,7 +124,7 @@ nv30_transfer_del(struct pipe_transfer *ptx)
 {
 	struct nv30_transfer *tx = (struct nv30_transfer *)ptx;
 
-	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+	if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) {
 		struct pipe_screen *pscreen = ptx->texture->screen;
 		struct nv30_screen *nvscreen = nv30_screen(pscreen);
 		struct pipe_surface *dst;
@@ -170,7 +154,7 @@ nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
 	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
 	struct nv30_miptree *mt = (struct nv30_miptree *)tx->surface->texture;
 	void *map = pipe_buffer_map(pscreen, mt->buffer,
-	                            nv30_usage_tx_to_buf(ptx->usage));
+	                            pipe_transfer_buffer_flags(ptx));
 
 	return map + ns->base.offset +
 	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c
index 8eba6a43ef..7f008274a4 100644
--- a/src/gallium/drivers/nv40/nv40_context.c
+++ b/src/gallium/drivers/nv40/nv40_context.c
@@ -10,7 +10,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags,
 	   struct pipe_fence_handle **fence)
 {
 	struct nv40_context *nv40 = nv40_context(pipe);
-	
+
 	if (flags & PIPE_FLUSH_TEXTURE_CACHE) {
 		BEGIN_RING(curie, 0x1fd8, 1);
 		OUT_RING  (2);
@@ -31,29 +31,6 @@ nv40_destroy(struct pipe_context *pipe)
 	FREE(nv40);
 }
 
-static unsigned int
-nv40_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv40_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
 struct pipe_context *
 nv40_create(struct pipe_screen *pscreen, unsigned pctx_id)
 {
@@ -78,8 +55,8 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv40->pipe.clear = nv40_clear;
 	nv40->pipe.flush = nv40_flush;
 
-	nv40->pipe.is_texture_referenced = nv40_is_texture_referenced;
-	nv40->pipe.is_buffer_referenced = nv40_is_buffer_referenced;
+	nv40->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv40->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
 
 	nv40_init_query_functions(nv40);
 	nv40_init_surface_functions(nv40);
@@ -95,4 +72,3 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id)
 
 	return &nv40->pipe;
 }
-	
diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h
index 97bc83292d..a3d594167a 100644
--- a/src/gallium/drivers/nv40/nv40_context.h
+++ b/src/gallium/drivers/nv40/nv40_context.h
@@ -13,6 +13,7 @@
 
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
+#include "nouveau/nouveau_context.h"
 
 #define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
 	struct nv40_screen *ctx = nv40->screen
diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c
index 5d2a4216c5..bd13dfddd1 100644
--- a/src/gallium/drivers/nv40/nv40_screen.c
+++ b/src/gallium/drivers/nv40/nv40_screen.c
@@ -21,8 +21,6 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 1;
 	case PIPE_CAP_GLSL:
 		return 0;
-	case PIPE_CAP_S3TC:
-		return 1;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 1;
 	case PIPE_CAP_POINT_SPRITE:
diff --git a/src/gallium/drivers/nv40/nv40_transfer.c b/src/gallium/drivers/nv40/nv40_transfer.c
index 6d92ac3db9..92caee6f38 100644
--- a/src/gallium/drivers/nv40/nv40_transfer.c
+++ b/src/gallium/drivers/nv40/nv40_transfer.c
@@ -13,22 +13,6 @@ struct nv40_transfer {
 	bool direct;
 };
 
-static unsigned nv40_usage_tx_to_buf(unsigned tx_usage)
-{
-	switch (tx_usage) {
-		case PIPE_TRANSFER_READ:
-			return PIPE_BUFFER_USAGE_CPU_READ;
-		case PIPE_TRANSFER_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_WRITE;
-		case PIPE_TRANSFER_READ_WRITE:
-			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
 static void
 nv40_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
                              struct pipe_texture *template)
@@ -86,7 +70,7 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		tx->direct = true;
 		tx->surface = pscreen->get_tex_surface(pscreen, pt,
 	                                               face, level, zslice,
-	                                               nv40_usage_tx_to_buf(usage));
+	                                               pipe_transfer_buffer_flags(&tx->base));
 		return &tx->base;
 	}
 
@@ -103,7 +87,7 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 
 	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
 	                                       0, 0, 0,
-	                                       nv40_usage_tx_to_buf(usage));
+	                                       pipe_transfer_buffer_flags(&tx->base));
 
 	pipe_texture_reference(&tx_tex, NULL);
 
@@ -114,7 +98,7 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
-	if (usage != PIPE_TRANSFER_WRITE) {
+	if (usage & PIPE_TRANSFER_READ) {
 		struct nv40_screen *nvscreen = nv40_screen(pscreen);
 		struct pipe_surface *src;
 
@@ -140,7 +124,7 @@ nv40_transfer_del(struct pipe_transfer *ptx)
 {
 	struct nv40_transfer *tx = (struct nv40_transfer *)ptx;
 
-	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+	if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) {
 		struct pipe_screen *pscreen = ptx->texture->screen;
 		struct nv40_screen *nvscreen = nv40_screen(pscreen);
 		struct pipe_surface *dst;
@@ -170,7 +154,7 @@ nv40_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
 	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
 	struct nv40_miptree *mt = (struct nv40_miptree *)tx->surface->texture;
 	void *map = pipe_buffer_map(pscreen, mt->buffer,
-	                            nv40_usage_tx_to_buf(ptx->usage));
+	                            pipe_transfer_buffer_flags(ptx));
 
 	return map + ns->base.offset +
 	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c
index 6e8f4f9750..219e7a7862 100644
--- a/src/gallium/drivers/nv50/nv50_context.c
+++ b/src/gallium/drivers/nv50/nv50_context.c
@@ -33,15 +33,9 @@ nv50_flush(struct pipe_context *pipe, unsigned flags,
 {
 	struct nv50_context *nv50 = nv50_context(pipe);
 	struct nouveau_channel *chan = nv50->screen->base.channel;
-	struct nouveau_grobj *eng2d = nv50->screen->eng2d;
 
-	/* We need this in the ddx for reliable composite, not sure what we're
-	 * actually flushing. We generate all our own flushes with flags = 0. */
-	WAIT_RING(chan, 3);
-	BEGIN_RING(chan, eng2d, 0x0110, 1);
-	OUT_RING  (chan, 0);
-
-	FIRE_RING(chan);
+	if (flags & PIPE_FLUSH_FRAME)
+		FIRE_RING(chan);
 }
 
 static void
@@ -59,29 +53,6 @@ nv50_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
 {
 }
 
-static unsigned int
-nv50_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv50_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
 struct pipe_context *
 nv50_create(struct pipe_screen *pscreen, unsigned pctx_id)
 {
@@ -107,8 +78,11 @@ nv50_create(struct pipe_screen *pscreen, unsigned pctx_id)
 
 	nv50->pipe.flush = nv50_flush;
 
-	nv50->pipe.is_texture_referenced = nv50_is_texture_referenced;
-	nv50->pipe.is_buffer_referenced = nv50_is_buffer_referenced;
+	nv50->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv50->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
+
+	screen->base.channel->user_private = nv50;
+	screen->base.channel->flush_notify = nv50_state_flush_notify;
 
 	nv50_init_surface_functions(nv50);
 	nv50_init_state_functions(nv50);
diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h
index 1e9e8e49bf..33667e8765 100644
--- a/src/gallium/drivers/nv50/nv50_context.h
+++ b/src/gallium/drivers/nv50/nv50_context.h
@@ -14,6 +14,7 @@
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
 #include "nouveau/nouveau_stateobj.h"
+#include "nouveau/nouveau_context.h"
 
 #include "nv50_screen.h"
 #include "nv50_program.h"
@@ -116,9 +117,11 @@ struct nv50_state {
 	unsigned miptree_nr;
 	struct nouveau_stateobj *vertprog;
 	struct nouveau_stateobj *fragprog;
+	struct nouveau_stateobj *programs;
 	struct nouveau_stateobj *vtxfmt;
 	struct nouveau_stateobj *vtxbuf;
 	struct nouveau_stateobj *vtxattr;
+	unsigned vtxelt_nr;
 };
 
 struct nv50_context {
@@ -151,6 +154,8 @@ struct nv50_context {
 	unsigned sampler_nr;
 	struct nv50_miptree *miptree[PIPE_MAX_SAMPLERS];
 	unsigned miptree_nr;
+
+	uint16_t vbo_fifo;
 };
 
 static INLINE struct nv50_context *
@@ -190,10 +195,17 @@ extern void nv50_clear(struct pipe_context *pipe, unsigned buffers,
 /* nv50_program.c */
 extern void nv50_vertprog_validate(struct nv50_context *nv50);
 extern void nv50_fragprog_validate(struct nv50_context *nv50);
+extern void nv50_linkage_validate(struct nv50_context *nv50);
 extern void nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p);
 
 /* nv50_state_validate.c */
 extern boolean nv50_state_validate(struct nv50_context *nv50);
+extern void nv50_state_flush_notify(struct nouveau_channel *chan);
+
+extern void nv50_so_init_sifc(struct nv50_context *nv50,
+			      struct nouveau_stateobj *so,
+			      struct nouveau_bo *bo, unsigned reloc,
+			      unsigned size);
 
 /* nv50_tex.c */
 extern void nv50_tex_validate(struct nv50_context *);
diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c
index 03b9243b82..93479a0314 100644
--- a/src/gallium/drivers/nv50/nv50_miptree.c
+++ b/src/gallium/drivers/nv50/nv50_miptree.c
@@ -148,6 +148,7 @@ nv50_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 	mt->image_nr = 1;
 	mt->level[0].pitch = *stride;
 	mt->level[0].image_offset = CALLOC(1, sizeof(unsigned));
+	mt->level[0].tile_mode = bo->tile_mode;
 
 	nouveau_bo_ref(bo, &mt->base.bo);
 	return &mt->base.base;
diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index 4a838529de..c7145bb9be 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -32,6 +32,7 @@
 #include "nv50_context.h"
 
 #define NV50_SU_MAX_TEMP 64
+#define NV50_SU_MAX_ADDR 7
 //#define NV50_PROGRAM_DUMP
 
 /* ARL - gallium craps itself on progs/vp/arl.txt
@@ -79,7 +80,8 @@ struct nv50_reg {
 		P_ATTR,
 		P_RESULT,
 		P_CONST,
-		P_IMMD
+		P_IMMD,
+		P_ADDR
 	} type;
 	int index;
 
@@ -90,11 +92,16 @@ struct nv50_reg {
 	int acc; /* instruction where this reg is last read (first insn == 1) */
 };
 
+/* arbitrary limits */
+#define MAX_IF_DEPTH 4
+#define MAX_LOOP_DEPTH 4
+
 struct nv50_pc {
 	struct nv50_program *p;
 
 	/* hw resources */
 	struct nv50_reg *r_temp[NV50_SU_MAX_TEMP];
+	struct nv50_reg r_addr[NV50_SU_MAX_ADDR];
 
 	/* tgsi resources */
 	struct nv50_reg *temp;
@@ -108,15 +115,28 @@ struct nv50_pc {
 	struct nv50_reg *immd;
 	float *immd_buf;
 	int immd_nr;
+	struct nv50_reg **addr;
+	int addr_nr;
 
 	struct nv50_reg *temp_temp[16];
 	unsigned temp_temp_nr;
 
+	/* broadcast and destination replacement regs */
+	struct nv50_reg *r_brdc;
+	struct nv50_reg *r_dst[4];
+
 	unsigned interp_mode[32];
 	/* perspective interpolation registers */
 	struct nv50_reg *iv_p;
 	struct nv50_reg *iv_c;
 
+	struct nv50_program_exec *if_cond;
+	struct nv50_program_exec *if_insn[MAX_IF_DEPTH];
+	struct nv50_program_exec *br_join[MAX_IF_DEPTH];
+	struct nv50_program_exec *br_loop[MAX_LOOP_DEPTH]; /* for BRK branch */
+	int if_lvl, loop_lvl;
+	unsigned loop_pos[MAX_LOOP_DEPTH];
+
 	/* current instruction and total number of insns */
 	unsigned insn_cur;
 	unsigned insn_nr;
@@ -124,6 +144,36 @@ struct nv50_pc {
 	boolean allow32;
 };
 
+static INLINE void
+ctor_reg(struct nv50_reg *reg, unsigned type, int index, int hw)
+{
+	reg->type = type;
+	reg->index = index;
+	reg->hw = hw;
+	reg->neg = 0;
+	reg->rhw = -1;
+	reg->acc = 0;
+}
+
+static INLINE unsigned
+popcnt4(uint32_t val)
+{
+	static const unsigned cnt[16]
+	= { 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4 };
+	return cnt[val & 0xf];
+}
+
+static void
+terminate_mbb(struct nv50_pc *pc)
+{
+	int i;
+
+	/* remove records of temporary address register values */
+	for (i = 0; i < NV50_SU_MAX_ADDR; ++i)
+		if (pc->r_addr[i].index < 0)
+			pc->r_addr[i].rhw = -1;
+}
+
 static void
 alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg)
 {
@@ -173,6 +223,10 @@ alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg)
 	assert(0);
 }
 
+/* XXX: For shaders that aren't executed linearly (e.g. shaders that
+ * contain loops), we need to assign all hw regs to TGSI TEMPs early,
+ * lest we risk temp_temps overwriting regs alloc'd "later".
+ */
 static struct nv50_reg *
 alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst)
 {
@@ -184,11 +238,8 @@ alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst)
 
 	for (i = 0; i < NV50_SU_MAX_TEMP; i++) {
 		if (!pc->r_temp[i]) {
-			r = CALLOC_STRUCT(nv50_reg);
-			r->type = P_TEMP;
-			r->index = -1;
-			r->hw = i;
-			r->rhw = -1;
+			r = MALLOC_STRUCT(nv50_reg);
+			ctor_reg(r, P_TEMP, -1, i);
 			pc->r_temp[i] = r;
 			return r;
 		}
@@ -254,10 +305,8 @@ alloc_temp4(struct nv50_pc *pc, struct nv50_reg *dst[4], int idx)
 		return alloc_temp4(pc, dst, idx + 4);
 
 	for (i = 0; i < 4; i++) {
-		dst[i] = CALLOC_STRUCT(nv50_reg);
-		dst[i]->type = P_TEMP;
-		dst[i]->index = -1;
-		dst[i]->hw = idx + i;
+		dst[i] = MALLOC_STRUCT(nv50_reg);
+		ctor_reg(dst[i], P_TEMP, -1, idx + i);
 		pc->r_temp[idx + i] = dst[i];
 	}
 
@@ -309,7 +358,7 @@ ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w)
 static struct nv50_reg *
 alloc_immd(struct nv50_pc *pc, float f)
 {
-	struct nv50_reg *r = CALLOC_STRUCT(nv50_reg);
+	struct nv50_reg *r = MALLOC_STRUCT(nv50_reg);
 	unsigned hw;
 
 	for (hw = 0; hw < pc->immd_nr * 4; hw++)
@@ -319,9 +368,7 @@ alloc_immd(struct nv50_pc *pc, float f)
 	if (hw == pc->immd_nr * 4)
 		hw = ctor_immd(pc, f, -f, 0.5 * f, 0) * 4;
 
-	r->type = P_IMMD;
-	r->hw = hw;
-	r->index = -1;
+	ctor_reg(r, P_IMMD, -1, hw);
 	return r;
 }
 
@@ -423,9 +470,68 @@ set_immd(struct nv50_pc *pc, struct nv50_reg *imm, struct nv50_program_exec *e)
 	e->inst[1] |= (val >> 6) << 2;
 }
 
+static void
+emit_set_addr(struct nv50_pc *pc, struct nv50_reg *dst, unsigned val)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	assert(val <= 0xffff);
+	e->inst[0] = 0xd0000000 | ((val & 0xffff) << 9);
+	e->inst[1] = 0x20000000;
+	e->inst[0] |= dst->hw << 2;
+	set_long(pc, e);
+
+	emit(pc, e);
+}
+
+static struct nv50_reg *
+alloc_addr(struct nv50_pc *pc, struct nv50_reg *ref)
+{
+	int i;
+	struct nv50_reg *a = NULL;
+
+	if (!ref) {
+		for (i = 0; i < NV50_SU_MAX_ADDR; ++i) {
+			if (pc->r_addr[i].index >= 0)
+				continue;
+			if (pc->r_addr[i].rhw >= 0 &&
+			    pc->r_addr[i].acc == pc->insn_cur)
+				continue;
+
+			pc->r_addr[i].rhw = -1;
+			pc->r_addr[i].index = i;
+			return &pc->r_addr[i];
+		}
+		assert(0);
+		return NULL;
+	}
+
+	for (i = NV50_SU_MAX_ADDR - 1; i >= 0; --i) {
+		if (pc->r_addr[i].index >= 0) /* occupied for TGSI */
+			continue;
+		if (pc->r_addr[i].rhw < 0) { /* unused */
+			a = &pc->r_addr[i];
+			continue;
+		}
+		if (!a && pc->r_addr[i].acc != pc->insn_cur)
+			a = &pc->r_addr[i];
+
+		if (ref->hw - pc->r_addr[i].rhw < 128) {
+		/* alloc'd & suitable */
+			pc->r_addr[i].acc = pc->insn_cur;
+			return &pc->r_addr[i];
+		}
+	}
+	assert(a);
+	emit_set_addr(pc, a, ref->hw * 4);
+
+	a->rhw = ref->hw % 128;
+	a->acc = pc->insn_cur;
+	return a;
+}
 
 #define INTERP_LINEAR		0
-#define INTERP_FLAT			1
+#define INTERP_FLAT		1
 #define INTERP_PERSPECTIVE	2
 #define INTERP_CENTROID		4
 
@@ -457,6 +563,16 @@ emit_interp(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *iv,
 	emit(pc, e);
 }
 
+static INLINE void
+set_addr(struct nv50_program_exec *e, struct nv50_reg *a)
+{
+	assert(!(e->inst[0] & 0x0c000000));
+	assert(!(e->inst[1] & 0x00000004));
+
+	e->inst[0] |= (a->hw & 3) << 26;
+	e->inst[1] |= (a->hw >> 2) << 2;
+}
+
 static void
 set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s,
 	 struct nv50_program_exec *e)
@@ -467,6 +583,14 @@ set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s,
 	e->param.shift = s;
 	e->param.mask = m << (s % 32);
 
+	if (src->hw > 127)
+		set_addr(e, alloc_addr(pc, src));
+	else
+	if (src->acc < 0) {
+		assert(src->type == P_CONST);
+		set_addr(e, pc->addr[src->index]);
+	}
+
 	e->inst[1] |= (((src->type == P_IMMD) ? 0 : 1) << 22);
 }
 
@@ -475,11 +599,13 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
 {
 	struct nv50_program_exec *e = exec(pc);
 
-	e->inst[0] |= 0x10000000;
+	e->inst[0] = 0x10000000;
+	if (!pc->allow32)
+		set_long(pc, e);
 
 	set_dst(pc, dst, e);
 
-	if (pc->allow32 && dst->type != P_RESULT && src->type == P_IMMD) {
+	if (!is_long(e) && src->type == P_IMMD) {
 		set_immd(pc, src, e);
 		/*XXX: 32-bit, but steals part of "half" reg space - need to
 		 *     catch and handle this case if/when we do half-regs
@@ -543,6 +669,22 @@ check_swap_src_0_1(struct nv50_pc *pc,
 }
 
 static void
+set_src_0_restricted(struct nv50_pc *pc, struct nv50_reg *src,
+		     struct nv50_program_exec *e)
+{
+	struct nv50_reg *temp;
+
+	if (src->type != P_TEMP) {
+		temp = temp_temp(pc);
+		emit_mov(pc, temp, src);
+		src = temp;
+	}
+
+	alloc_reg(pc, src);
+	e->inst[0] |= (src->hw << 9);
+}
+
+static void
 set_src_0(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e)
 {
 	if (src->type == P_ATTR) {
@@ -583,7 +725,7 @@ set_src_1(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e)
 	}
 
 	alloc_reg(pc, src);
-	e->inst[0] |= (src->hw << 16);
+	e->inst[0] |= ((src->hw & 127) << 16);
 }
 
 static void
@@ -611,7 +753,7 @@ set_src_2(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e)
 	}
 
 	alloc_reg(pc, src);
-	e->inst[1] |= (src->hw << 14);
+	e->inst[1] |= ((src->hw & 127) << 14);
 }
 
 static void
@@ -674,6 +816,22 @@ emit_add(struct nv50_pc *pc, struct nv50_reg *dst,
 }
 
 static void
+emit_arl(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src,
+	 uint8_t s)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	set_long(pc, e);
+	e->inst[1] |= 0xc0000000;
+
+	e->inst[0] |= dst->hw << 2;
+	e->inst[0] |= s << 16; /* shift left */
+	set_src_0_restricted(pc, src, e);
+
+	emit(pc, e);
+}
+
+static void
 emit_minmax(struct nv50_pc *pc, unsigned sub, struct nv50_reg *dst,
 	    struct nv50_reg *src0, struct nv50_reg *src1)
 {
@@ -744,7 +902,11 @@ emit_flop(struct nv50_pc *pc, unsigned sub,
 	}
 
 	set_dst(pc, dst, e);
-	set_src_0(pc, src, e);
+
+	if (sub == 0 || sub == 2)
+		set_src_0_restricted(pc, src, e);
+	else
+		set_src_0(pc, src, e);
 
 	emit(pc, e);
 }
@@ -786,16 +948,20 @@ emit_precossin(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
 #define CVTOP_SAT	0x08
 #define CVTOP_ABS	0x10
 
+/* 0x04 == 32 bit dst */
+/* 0x40 == dst is float */
+/* 0x80 == src is float */
 #define CVT_F32_F32 0xc4
 #define CVT_F32_S32 0x44
 #define CVT_F32_U32 0x64
 #define CVT_S32_F32 0x8c
 #define CVT_S32_S32 0x0c
-#define CVT_F32_F32_ROP 0xcc
+#define CVT_NEG     0x20
+#define CVT_RI      0x08
 
 static void
 emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src,
-	 int wp, unsigned cop, unsigned fmt)
+	 int wp, unsigned cvn, unsigned fmt)
 {
 	struct nv50_program_exec *e;
 
@@ -803,8 +969,8 @@ emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src,
 	set_long(pc, e);
 
 	e->inst[0] |= 0xa0000000;
-	e->inst[1] |= 0x00004000;
-	e->inst[1] |= (cop << 16);
+	e->inst[1] |= 0x00004000; /* 32 bit src */
+	e->inst[1] |= (cvn << 16);
 	e->inst[1] |= (fmt << 24);
 	set_src_0(pc, src, e);
 
@@ -821,53 +987,85 @@ emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src,
 	emit(pc, e);
 }
 
+/* nv50 Condition codes:
+ *  0x1 = LT
+ *  0x2 = EQ
+ *  0x3 = LE
+ *  0x4 = GT
+ *  0x5 = NE
+ *  0x6 = GE
+ *  0x7 = set condition code ? (used before bra.lt/le/gt/ge)
+ *  0x8 = unordered bit (allows NaN)
+ */
 static void
-emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst,
+emit_set(struct nv50_pc *pc, unsigned ccode, struct nv50_reg *dst, int wp,
 	 struct nv50_reg *src0, struct nv50_reg *src1)
 {
+	static const unsigned cc_swapped[8] = { 0, 4, 2, 6, 1, 5, 3, 7 };
+
 	struct nv50_program_exec *e = exec(pc);
-	unsigned inv_cop[8] = { 0, 4, 2, 6, 1, 5, 3, 7 };
 	struct nv50_reg *rdst;
 
-	assert(c_op <= 7);
+	assert(ccode < 16);
 	if (check_swap_src_0_1(pc, &src0, &src1))
-		c_op = inv_cop[c_op];
+		ccode = cc_swapped[ccode & 7] | (ccode & 8);
 
 	rdst = dst;
-	if (dst->type != P_TEMP)
+	if (dst && dst->type != P_TEMP)
 		dst = alloc_temp(pc, NULL);
 
 	/* set.u32 */
 	set_long(pc, e);
 	e->inst[0] |= 0xb0000000;
-	e->inst[1] |= (3 << 29);
-	e->inst[1] |= (c_op << 14);
-	/*XXX: breaks things, .u32 by default?
-	 *     decuda will disasm as .u16 and use .lo/.hi regs, but this
-	 *     doesn't seem to match what the hw actually does.
-	inst[1] |= 0x04000000; << breaks things.. .u32 by default?
+	e->inst[1] |= 0x60000000 | (ccode << 14);
+
+	/* XXX: decuda will disasm as .u16 and use .lo/.hi regs, but
+	 * that doesn't seem to match what the hw actually does
+	e->inst[1] |= 0x04000000; << breaks things, u32 by default ?
 	 */
-	set_dst(pc, dst, e);
+
+	if (wp >= 0)
+		set_pred_wr(pc, 1, wp, e);
+	if (dst)
+		set_dst(pc, dst, e);
+	else {
+		e->inst[0] |= 0x000001fc;
+		e->inst[1] |= 0x00000008;
+	}
+
 	set_src_0(pc, src0, e);
 	set_src_1(pc, src1, e);
-	emit(pc, e);
 
-	/* cvt.f32.u32 */
-	e = exec(pc);
-	e->inst[0] = 0xa0000001;
-	e->inst[1] = 0x64014780;
-	set_dst(pc, rdst, e);
-	set_src_0(pc, dst, e);
 	emit(pc, e);
+	pc->if_cond = pc->p->exec_tail; /* record for OPCODE_IF */
 
-	if (dst != rdst)
+	/* cvt.f32.u32/s32 (?) if we didn't only write the predicate */
+	if (rdst)
+		emit_cvt(pc, rdst, dst, -1, CVTOP_ABS | CVTOP_RN, CVT_F32_S32);
+	if (rdst && rdst != dst)
 		free_temp(pc, dst);
 }
 
+static INLINE unsigned
+map_tgsi_setop_cc(unsigned op)
+{
+	switch (op) {
+	case TGSI_OPCODE_SLT: return 0x1;
+	case TGSI_OPCODE_SGE: return 0x6;
+	case TGSI_OPCODE_SEQ: return 0x2;
+	case TGSI_OPCODE_SGT: return 0x4;
+	case TGSI_OPCODE_SLE: return 0x3;
+	case TGSI_OPCODE_SNE: return 0xd;
+	default:
+		assert(0);
+		return 0;
+	}
+}
+
 static INLINE void
 emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
 {
-	emit_cvt(pc, dst, src, -1, CVTOP_FLOOR, CVT_F32_F32_ROP);
+	emit_cvt(pc, dst, src, -1, CVTOP_FLOOR, CVT_F32_F32 | CVT_RI);
 }
 
 static void
@@ -890,6 +1088,12 @@ emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
 	emit_cvt(pc, dst, src, -1, CVTOP_ABS, CVT_F32_F32);
 }
 
+static INLINE void
+emit_sat(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	emit_cvt(pc, dst, src, -1, CVTOP_SAT, CVT_F32_F32);
+}
+
 static void
 emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask,
 	 struct nv50_reg **src)
@@ -944,20 +1148,10 @@ emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask,
 	FREE(one);
 }
 
-static void
+static INLINE void
 emit_neg(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
 {
-	struct nv50_program_exec *e = exec(pc);
-
-	set_long(pc, e);
-	e->inst[0] |= 0xa0000000; /* delta */
-	e->inst[1] |= (7 << 29); /* delta */
-	e->inst[1] |= 0x04000000; /* negate arg0? probably not */
-	e->inst[1] |= (1 << 14); /* src .f32 */
-	set_dst(pc, dst, e);
-	set_src_0(pc, src, e);
-
-	emit(pc, e);
+	emit_cvt(pc, dst, src, -1, CVTOP_RN, CVT_F32_F32 | CVT_NEG);
 }
 
 static void
@@ -965,22 +1159,18 @@ emit_kil(struct nv50_pc *pc, struct nv50_reg *src)
 {
 	struct nv50_program_exec *e;
 	const int r_pred = 1;
+	unsigned cvn = CVT_F32_F32;
 
-	/* Sets predicate reg ? */
-	e = exec(pc);
-	e->inst[0] = 0xa00001fd;
-	e->inst[1] = 0xc4014788;
-	set_src_0(pc, src, e);
-	set_pred_wr(pc, 1, r_pred, e);
 	if (src->neg)
-		e->inst[1] |= 0x20000000;
-	emit(pc, e);
+		cvn |= CVT_NEG;
+	/* write predicate reg */
+	emit_cvt(pc, NULL, src, r_pred, CVTOP_RN, cvn);
 
-	/* This is probably KILP */
+	/* conditional discard */
 	e = exec(pc);
-	e->inst[0] = 0x000001fe;
+	e->inst[0] = 0x00000002;
 	set_long(pc, e);
-	set_pred(pc, 1 /* LT? */, r_pred, e);
+	set_pred(pc, 0x1 /* LT */, r_pred, e);
 	emit(pc, e);
 }
 
@@ -1073,10 +1263,11 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask,
 	emit(pc, e);
 
 #if 1
-	if (mask & 1) emit_mov(pc, dst[0], t[0]);
-	if (mask & 2) emit_mov(pc, dst[1], t[1]);
-	if (mask & 4) emit_mov(pc, dst[2], t[2]);
-	if (mask & 8) emit_mov(pc, dst[3], t[3]);
+	c = 0;
+	if (mask & 1) emit_mov(pc, dst[0], t[c++]);
+	if (mask & 2) emit_mov(pc, dst[1], t[c++]);
+	if (mask & 4) emit_mov(pc, dst[2], t[c++]);
+	if (mask & 8) emit_mov(pc, dst[3], t[c]);
 
 	free_temp4(pc, t);
 #else
@@ -1093,6 +1284,75 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask,
 }
 
 static void
+emit_branch(struct nv50_pc *pc, int pred, unsigned cc,
+	    struct nv50_program_exec **join)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	if (join) {
+		set_long(pc, e);
+		e->inst[0] |= 0xa0000002;
+		emit(pc, e);
+		*join = e;
+		e = exec(pc);
+	}
+
+	set_long(pc, e);
+	e->inst[0] |= 0x10000002;
+	if (pred >= 0)
+		set_pred(pc, cc, pred, e);
+	emit(pc, e);
+}
+
+static void
+emit_nop(struct nv50_pc *pc)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] = 0xf0000000;
+	set_long(pc, e);
+	e->inst[1] = 0xe0000000;
+	emit(pc, e);
+}
+
+static void
+emit_ddx(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	assert(src->type == P_TEMP);
+
+	e->inst[0] = 0xc0140000;
+	e->inst[1] = 0x89800000;
+	set_long(pc, e);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+	set_src_2(pc, src, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_ddy(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	assert(src->type == P_TEMP);
+
+	if (!src->neg) /* ! double negation */
+		emit_neg(pc, src, src);
+
+	e->inst[0] = 0xc0150000;
+	e->inst[1] = 0x8a400000;
+	set_long(pc, e);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+	set_src_2(pc, src, e);
+
+	emit(pc, e);
+}
+
+static void
 convert_to_long(struct nv50_pc *pc, struct nv50_program_exec *e)
 {
 	unsigned q = 0, m = ~0;
@@ -1140,10 +1400,14 @@ convert_to_long(struct nv50_pc *pc, struct nv50_program_exec *e)
 	e->inst[1] |= q;
 }
 
+/* Some operations support an optional negation flag. */
 static boolean
 negate_supported(const struct tgsi_full_instruction *insn, int i)
 {
+	int s;
+
 	switch (insn->Instruction.Opcode) {
+	case TGSI_OPCODE_DDY:
 	case TGSI_OPCODE_DP3:
 	case TGSI_OPCODE_DP4:
 	case TGSI_OPCODE_MUL:
@@ -1151,12 +1415,93 @@ negate_supported(const struct tgsi_full_instruction *insn, int i)
 	case TGSI_OPCODE_ADD:
 	case TGSI_OPCODE_SUB:
 	case TGSI_OPCODE_MAD:
-		return TRUE;
+		break;
 	case TGSI_OPCODE_POW:
-		return (i == 1) ? TRUE : FALSE;
+		if (i == 1)
+			break;
+		return FALSE;
 	default:
 		return FALSE;
 	}
+
+	/* Watch out for possible multiple uses of an nv50_reg, we
+	 * can't use nv50_reg::neg in these cases.
+	 */
+	for (s = 0; s < insn->Instruction.NumSrcRegs; ++s) {
+		if (s == i)
+			continue;
+		if ((insn->FullSrcRegisters[s].SrcRegister.Index ==
+		     insn->FullSrcRegisters[i].SrcRegister.Index) &&
+		    (insn->FullSrcRegisters[s].SrcRegister.File ==
+		     insn->FullSrcRegisters[i].SrcRegister.File))
+			return FALSE;
+	}
+
+	return TRUE;
+}
+
+/* Return a read mask for source registers deduced from opcode & write mask. */
+static unsigned
+nv50_tgsi_src_mask(const struct tgsi_full_instruction *insn, int c)
+{
+	unsigned x, mask = insn->FullDstRegisters[0].DstRegister.WriteMask;
+
+	switch (insn->Instruction.Opcode) {
+	case TGSI_OPCODE_COS:
+	case TGSI_OPCODE_SIN:
+		return (mask & 0x8) | ((mask & 0x7) ? 0x1 : 0x0);
+	case TGSI_OPCODE_DP3:
+		return 0x7;
+	case TGSI_OPCODE_DP4:
+	case TGSI_OPCODE_DPH:
+	case TGSI_OPCODE_KIL: /* WriteMask ignored */
+		return 0xf;
+	case TGSI_OPCODE_DST:
+		return mask & (c ? 0xa : 0x6);
+	case TGSI_OPCODE_EX2:
+	case TGSI_OPCODE_LG2:
+	case TGSI_OPCODE_POW:
+	case TGSI_OPCODE_RCP:
+	case TGSI_OPCODE_RSQ:
+	case TGSI_OPCODE_SCS:
+		return 0x1;
+	case TGSI_OPCODE_LIT:
+		return 0xb;
+	case TGSI_OPCODE_TEX:
+	case TGSI_OPCODE_TXP:
+	{
+		const struct tgsi_instruction_ext_texture *tex;
+
+		assert(insn->Instruction.Extended);
+		tex = &insn->InstructionExtTexture;
+
+		mask = 0x7;
+		if (insn->Instruction.Opcode == TGSI_OPCODE_TXP)
+			mask |= 0x8;
+
+		switch (tex->Texture) {
+		case TGSI_TEXTURE_1D:
+			mask &= 0x9;
+			break;
+		case TGSI_TEXTURE_2D:
+			mask &= 0xb;
+			break;
+		default:
+			break;
+		}
+	}
+		return mask;
+	case TGSI_OPCODE_XPD:
+		x = 0;
+		if (mask & 1) x |= 0x6;
+		if (mask & 2) x |= 0x5;
+		if (mask & 4) x |= 0x3;
+		return x;
+	default:
+		break;
+	}
+
+	return mask;
 }
 
 static struct nv50_reg *
@@ -1167,6 +1512,16 @@ tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst)
 		return &pc->temp[dst->DstRegister.Index * 4 + c];
 	case TGSI_FILE_OUTPUT:
 		return &pc->result[dst->DstRegister.Index * 4 + c];
+	case TGSI_FILE_ADDRESS:
+	{
+		struct nv50_reg *r = pc->addr[dst->DstRegister.Index * 4 + c];
+		if (!r) {
+			r = alloc_addr(pc, NULL);
+			pc->addr[dst->DstRegister.Index * 4 + c] = r;
+		}
+		assert(r);
+		return r;
+	}
 	case TGSI_FILE_NULL:
 		return NULL;
 	default:
@@ -1182,7 +1537,10 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src,
 {
 	struct nv50_reg *r = NULL;
 	struct nv50_reg *temp;
-	unsigned sgn, c;
+	unsigned sgn, c, swz;
+
+	if (src->SrcRegister.File != TGSI_FILE_CONSTANT)
+		assert(!src->SrcRegister.Indirect);
 
 	sgn = tgsi_util_get_full_src_register_sign_mode(src, chan);
 
@@ -1200,13 +1558,29 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src,
 			r = &pc->temp[src->SrcRegister.Index * 4 + c];
 			break;
 		case TGSI_FILE_CONSTANT:
-			r = &pc->param[src->SrcRegister.Index * 4 + c];
+			if (!src->SrcRegister.Indirect) {
+				r = &pc->param[src->SrcRegister.Index * 4 + c];
+				break;
+			}
+			/* Indicate indirection by setting r->acc < 0 and
+			 * use the index field to select the address reg.
+			 */
+			r = MALLOC_STRUCT(nv50_reg);
+			swz = tgsi_util_get_src_register_swizzle(
+						 &src->SrcRegisterInd, 0);
+			ctor_reg(r, P_CONST,
+				 src->SrcRegisterInd.Index * 4 + swz, c);
+			r->acc = -1;
 			break;
 		case TGSI_FILE_IMMEDIATE:
 			r = &pc->immd[src->SrcRegister.Index * 4 + c];
 			break;
 		case TGSI_FILE_SAMPLER:
 			break;
+		case TGSI_FILE_ADDRESS:
+			r = pc->addr[src->SrcRegister.Index * 4 + c];
+			assert(r);
+			break;
 		default:
 			assert(0);
 			break;
@@ -1258,93 +1632,175 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src,
 	return r;
 }
 
-/* returns TRUE if instruction can overwrite sources before they're read */
+/* return TRUE for ops that produce only a single result */
 static boolean
-direct2dest_op(const struct tgsi_full_instruction *insn)
+is_scalar_op(unsigned op)
 {
-	if (insn->Instruction.Saturate)
-		return FALSE;
-
-	switch (insn->Instruction.Opcode) {
+	switch (op) {
 	case TGSI_OPCODE_COS:
+	case TGSI_OPCODE_DP2:
 	case TGSI_OPCODE_DP3:
 	case TGSI_OPCODE_DP4:
 	case TGSI_OPCODE_DPH:
-	case TGSI_OPCODE_KIL:
-	case TGSI_OPCODE_LIT:
+	case TGSI_OPCODE_EX2:
+	case TGSI_OPCODE_LG2:
 	case TGSI_OPCODE_POW:
 	case TGSI_OPCODE_RCP:
 	case TGSI_OPCODE_RSQ:
-	case TGSI_OPCODE_SCS:
 	case TGSI_OPCODE_SIN:
+		/*
+	case TGSI_OPCODE_KIL:
+	case TGSI_OPCODE_LIT:
+	case TGSI_OPCODE_SCS:
+		*/
+		return TRUE;
+	default:
+		return FALSE;
+	}
+}
+
+/* Returns a bitmask indicating which dst components depend
+ * on source s, component c (reverse of nv50_tgsi_src_mask).
+ */
+static unsigned
+nv50_tgsi_dst_revdep(unsigned op, int s, int c)
+{
+	if (is_scalar_op(op))
+		return 0x1;
+
+	switch (op) {
+	case TGSI_OPCODE_DST:
+		return (1 << c) & (s ? 0xa : 0x6);
+	case TGSI_OPCODE_XPD:
+		switch (c) {
+		case 0: return 0x6;
+		case 1: return 0x5;
+		case 2: return 0x3;
+		case 3: return 0x0;
+		default:
+			assert(0);
+			return 0x0;
+		}
+	case TGSI_OPCODE_LIT:
+	case TGSI_OPCODE_SCS:
 	case TGSI_OPCODE_TEX:
 	case TGSI_OPCODE_TXP:
-		return FALSE;
+		/* these take care of dangerous swizzles themselves */
+		return 0x0;
+	case TGSI_OPCODE_IF:
+	case TGSI_OPCODE_KIL:
+		/* don't call this function for these ops */
+		assert(0);
+		return 0;
 	default:
-		return TRUE;
+		/* linear vector instruction */
+		return (1 << c);
 	}
 }
 
+static INLINE boolean
+has_pred(struct nv50_program_exec *e, unsigned cc)
+{
+	if (!is_long(e) || is_immd(e))
+		return FALSE;
+	return ((e->inst[1] & 0x780) == (cc << 7));
+}
+
+/* on ENDIF see if we can do "@p0.neu single_op" instead of:
+ *        join_at ENDIF
+ *        @p0.eq bra ENDIF
+ *        single_op
+ * ENDIF: nop.join
+ */
 static boolean
-nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
+nv50_kill_branch(struct nv50_pc *pc)
 {
-	const struct tgsi_full_instruction *inst = &tok->FullInstruction;
-	struct nv50_reg *rdst[4], *dst[4], *src[3][4], *temp;
+	int lvl = pc->if_lvl;
+
+	if (pc->if_insn[lvl]->next != pc->p->exec_tail)
+		return FALSE;
+
+	/* if ccode == 'true', the BRA is from an ELSE and the predicate
+	 * reg may no longer be valid, since we currently always use $p0
+	 */
+	if (has_pred(pc->if_insn[lvl], 0xf))
+		return FALSE;
+	assert(pc->if_insn[lvl] && pc->br_join[lvl]);
+
+	/* We'll use the exec allocated for JOIN_AT (as we can't easily
+	 * update prev's next); if exec_tail is BRK, update the pointer.
+	 */
+	if (pc->loop_lvl && pc->br_loop[pc->loop_lvl - 1] == pc->p->exec_tail)
+		pc->br_loop[pc->loop_lvl - 1] = pc->br_join[lvl];
+
+	pc->p->exec_size -= 4; /* remove JOIN_AT and BRA */
+
+	*pc->br_join[lvl] = *pc->p->exec_tail;
+
+	FREE(pc->if_insn[lvl]);
+	FREE(pc->p->exec_tail);
+
+	pc->p->exec_tail = pc->br_join[lvl];
+	pc->p->exec_tail->next = NULL;
+	set_pred(pc, 0xd, 0, pc->p->exec_tail);
+
+	return TRUE;
+}
+
+static boolean
+nv50_program_tx_insn(struct nv50_pc *pc,
+		     const struct tgsi_full_instruction *inst)
+{
+	struct nv50_reg *rdst[4], *dst[4], *brdc, *src[3][4], *temp;
 	unsigned mask, sat, unit;
-	boolean assimilate = FALSE;
 	int i, c;
 
 	mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
 	sat = inst->Instruction.Saturate == TGSI_SAT_ZERO_ONE;
 
+	memset(src, 0, sizeof(src));
+
 	for (c = 0; c < 4; c++) {
-		if (mask & (1 << c))
+		if ((mask & (1 << c)) && !pc->r_dst[c])
 			dst[c] = tgsi_dst(pc, c, &inst->FullDstRegisters[0]);
 		else
-			dst[c] = NULL;
-		rdst[c] = NULL;
-		src[0][c] = NULL;
-		src[1][c] = NULL;
-		src[2][c] = NULL;
+			dst[c] = pc->r_dst[c];
+		rdst[c] = dst[c];
 	}
 
 	for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
 		const struct tgsi_full_src_register *fs = &inst->FullSrcRegisters[i];
+		unsigned src_mask;
+		boolean neg_supp;
+
+		src_mask = nv50_tgsi_src_mask(inst, i);
+		neg_supp = negate_supported(inst, i);
 
 		if (fs->SrcRegister.File == TGSI_FILE_SAMPLER)
 			unit = fs->SrcRegister.Index;
 
 		for (c = 0; c < 4; c++)
-			src[i][c] = tgsi_src(pc, c, fs,
-					     negate_supported(inst, i));
+			if (src_mask & (1 << c))
+				src[i][c] = tgsi_src(pc, c, fs, neg_supp);
 	}
 
-	if (sat) {
-		for (c = 0; c < 4; c++) {
-			rdst[c] = dst[c];
-			dst[c] = temp_temp(pc);
-		}
+	brdc = temp = pc->r_brdc;
+	if (brdc && brdc->type != P_TEMP) {
+		temp = temp_temp(pc);
+		if (sat)
+			brdc = temp;
 	} else
-	if (direct2dest_op(inst)) {
+	if (sat) {
 		for (c = 0; c < 4; c++) {
-			if (!dst[c] || dst[c]->type != P_TEMP)
-				continue;
-
-			for (i = c + 1; i < 4; i++) {
-				if (dst[c] == src[0][i] ||
-				    dst[c] == src[1][i] ||
-				    dst[c] == src[2][i])
-					break;
-			}
-			if (i == 4)
+			if (!(mask & (1 << c)) || dst[c]->type == P_TEMP)
 				continue;
-
-			assimilate = TRUE;
 			rdst[c] = dst[c];
-			dst[c] = alloc_temp(pc, NULL);
+			dst[c] = temp_temp(pc);
 		}
 	}
 
+	assert(brdc || !is_scalar_op(inst->Instruction.Opcode));
+
 	switch (inst->Instruction.Opcode) {
 	case TGSI_OPCODE_ABS:
 		for (c = 0; c < 4; c++) {
@@ -1360,74 +1816,127 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 			emit_add(pc, dst[c], src[0][c], src[1][c]);
 		}
 		break;
-	case TGSI_OPCODE_COS:
+	case TGSI_OPCODE_ARL:
+		assert(src[0][0]);
 		temp = temp_temp(pc);
+		emit_cvt(pc, temp, src[0][0], -1, CVTOP_FLOOR, CVT_S32_F32);
+		emit_arl(pc, dst[0], temp, 4);
+		break;
+	case TGSI_OPCODE_BGNLOOP:
+		pc->loop_pos[pc->loop_lvl++] = pc->p->exec_size;
+		terminate_mbb(pc);
+		break;
+	case TGSI_OPCODE_BRK:
+		emit_branch(pc, -1, 0, NULL);
+		assert(pc->loop_lvl > 0);
+		pc->br_loop[pc->loop_lvl - 1] = pc->p->exec_tail;
+		break;
+	case TGSI_OPCODE_CEIL:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_cvt(pc, dst[c], src[0][c], -1,
+				 CVTOP_CEIL, CVT_F32_F32 | CVT_RI);
+		}
+		break;
+	case TGSI_OPCODE_CMP:
+		pc->allow32 = FALSE;
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_cvt(pc, NULL, src[0][c], 1, CVTOP_RN, CVT_F32_F32);
+			emit_mov(pc, dst[c], src[1][c]);
+			set_pred(pc, 0x1, 1, pc->p->exec_tail); /* @SF */
+			emit_mov(pc, dst[c], src[2][c]);
+			set_pred(pc, 0x6, 1, pc->p->exec_tail); /* @NSF */
+		}
+		break;
+	case TGSI_OPCODE_COS:
+		if (mask & 8) {
+			emit_precossin(pc, temp, src[0][3]);
+			emit_flop(pc, 5, dst[3], temp);
+			if (!(mask &= 7))
+				break;
+			if (temp == dst[3])
+				temp = brdc = temp_temp(pc);
+		}
 		emit_precossin(pc, temp, src[0][0]);
-		emit_flop(pc, 5, temp, temp);
+		emit_flop(pc, 5, brdc, temp);
+		break;
+	case TGSI_OPCODE_DDX:
 		for (c = 0; c < 4; c++) {
 			if (!(mask & (1 << c)))
 				continue;
-			emit_mov(pc, dst[c], temp);
+			emit_ddx(pc, dst[c], src[0][c]);
 		}
 		break;
-	case TGSI_OPCODE_DP3:
-		temp = temp_temp(pc);
-		emit_mul(pc, temp, src[0][0], src[1][0]);
-		emit_mad(pc, temp, src[0][1], src[1][1], temp);
-		emit_mad(pc, temp, src[0][2], src[1][2], temp);
+	case TGSI_OPCODE_DDY:
 		for (c = 0; c < 4; c++) {
 			if (!(mask & (1 << c)))
 				continue;
-			emit_mov(pc, dst[c], temp);
+			emit_ddy(pc, dst[c], src[0][c]);
 		}
 		break;
+	case TGSI_OPCODE_DP3:
+		emit_mul(pc, temp, src[0][0], src[1][0]);
+		emit_mad(pc, temp, src[0][1], src[1][1], temp);
+		emit_mad(pc, brdc, src[0][2], src[1][2], temp);
+		break;
 	case TGSI_OPCODE_DP4:
-		temp = temp_temp(pc);
 		emit_mul(pc, temp, src[0][0], src[1][0]);
 		emit_mad(pc, temp, src[0][1], src[1][1], temp);
 		emit_mad(pc, temp, src[0][2], src[1][2], temp);
-		emit_mad(pc, temp, src[0][3], src[1][3], temp);
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_mov(pc, dst[c], temp);
-		}
+		emit_mad(pc, brdc, src[0][3], src[1][3], temp);
 		break;
 	case TGSI_OPCODE_DPH:
-		temp = temp_temp(pc);
 		emit_mul(pc, temp, src[0][0], src[1][0]);
 		emit_mad(pc, temp, src[0][1], src[1][1], temp);
 		emit_mad(pc, temp, src[0][2], src[1][2], temp);
-		emit_add(pc, temp, src[1][3], temp);
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_mov(pc, dst[c], temp);
-		}
+		emit_add(pc, brdc, src[1][3], temp);
 		break;
 	case TGSI_OPCODE_DST:
-	{
-		struct nv50_reg *one = alloc_immd(pc, 1.0);
-		if (mask & (1 << 0))
-			emit_mov(pc, dst[0], one);
 		if (mask & (1 << 1))
 			emit_mul(pc, dst[1], src[0][1], src[1][1]);
 		if (mask & (1 << 2))
 			emit_mov(pc, dst[2], src[0][2]);
 		if (mask & (1 << 3))
 			emit_mov(pc, dst[3], src[1][3]);
-		FREE(one);
-	}
+		if (mask & (1 << 0))
+			emit_mov_immdval(pc, dst[0], 1.0f);
+		break;
+	case TGSI_OPCODE_ELSE:
+		emit_branch(pc, -1, 0, NULL);
+		pc->if_insn[--pc->if_lvl]->param.index = pc->p->exec_size;
+		pc->if_insn[pc->if_lvl++] = pc->p->exec_tail;
+		terminate_mbb(pc);
+		break;
+	case TGSI_OPCODE_ENDIF:
+		pc->if_insn[--pc->if_lvl]->param.index = pc->p->exec_size;
+
+		/* try to replace branch over 1 insn with a predicated insn */
+		if (nv50_kill_branch(pc) == TRUE)
+			break;
+
+		if (pc->br_join[pc->if_lvl]) {
+			pc->br_join[pc->if_lvl]->param.index = pc->p->exec_size;
+			pc->br_join[pc->if_lvl] = NULL;
+		}
+		terminate_mbb(pc);
+		/* emit a NOP as join point, we could set it on the next
+		 * one, but would have to make sure it is long and !immd
+		 */
+		emit_nop(pc);
+		pc->p->exec_tail->inst[1] |= 2;
+		break;
+	case TGSI_OPCODE_ENDLOOP:
+		emit_branch(pc, -1, 0, NULL);
+		pc->p->exec_tail->param.index = pc->loop_pos[--pc->loop_lvl];
+		pc->br_loop[pc->loop_lvl]->param.index = pc->p->exec_size;
+		terminate_mbb(pc);
 		break;
 	case TGSI_OPCODE_EX2:
-		temp = temp_temp(pc);
 		emit_preex2(pc, temp, src[0][0]);
-		emit_flop(pc, 6, temp, temp);
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_mov(pc, dst[c], temp);
-		}
+		emit_flop(pc, 6, brdc, temp);
 		break;
 	case TGSI_OPCODE_FLR:
 		for (c = 0; c < 4; c++) {
@@ -1445,24 +1954,25 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 			emit_sub(pc, dst[c], src[0][c], temp);
 		}
 		break;
+	case TGSI_OPCODE_IF:
+		/* emitting a join_at may not be necessary */
+		assert(pc->if_lvl < MAX_IF_DEPTH);
+		set_pred_wr(pc, 1, 0, pc->if_cond);
+		emit_branch(pc, 0, 2, &pc->br_join[pc->if_lvl]);
+		pc->if_insn[pc->if_lvl++] = pc->p->exec_tail;
+		terminate_mbb(pc);
+		break;
 	case TGSI_OPCODE_KIL:
 		emit_kil(pc, src[0][0]);
 		emit_kil(pc, src[0][1]);
 		emit_kil(pc, src[0][2]);
 		emit_kil(pc, src[0][3]);
-		pc->p->cfg.fp.regs[2] |= 0x00100000;
 		break;
 	case TGSI_OPCODE_LIT:
 		emit_lit(pc, &dst[0], mask, &src[0][0]);
 		break;
 	case TGSI_OPCODE_LG2:
-		temp = temp_temp(pc);
-		emit_flop(pc, 3, temp, src[0][0]);
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_mov(pc, dst[c], temp);
-		}
+		emit_flop(pc, 3, brdc, src[0][0]);
 		break;
 	case TGSI_OPCODE_LRP:
 		temp = temp_temp(pc);
@@ -1510,31 +2020,18 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 		}
 		break;
 	case TGSI_OPCODE_POW:
-		temp = temp_temp(pc);
-		emit_pow(pc, temp, src[0][0], src[1][0]);
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_mov(pc, dst[c], temp);
-		}
+		emit_pow(pc, brdc, src[0][0], src[1][0]);
 		break;
 	case TGSI_OPCODE_RCP:
-		for (c = 3; c >= 0; c--) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_flop(pc, 0, dst[c], src[0][0]);
-		}
+		emit_flop(pc, 0, brdc, src[0][0]);
 		break;
 	case TGSI_OPCODE_RSQ:
-		for (c = 3; c >= 0; c--) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_flop(pc, 2, dst[c], src[0][0]);
-		}
+		emit_flop(pc, 2, brdc, src[0][0]);
 		break;
 	case TGSI_OPCODE_SCS:
 		temp = temp_temp(pc);
-		emit_precossin(pc, temp, src[0][0]);
+		if (mask & 3)
+			emit_precossin(pc, temp, src[0][0]);
 		if (mask & (1 << 0))
 			emit_flop(pc, 5, dst[0], temp);
 		if (mask & (1 << 1))
@@ -1544,28 +2041,29 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 		if (mask & (1 << 3))
 			emit_mov_immdval(pc, dst[3], 1.0);
 		break;
-	case TGSI_OPCODE_SGE:
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_set(pc, 6, dst[c], src[0][c], src[1][c]);
-		}
-		break;
 	case TGSI_OPCODE_SIN:
-		temp = temp_temp(pc);
-		emit_precossin(pc, temp, src[0][0]);
-		emit_flop(pc, 4, temp, temp);
-		for (c = 0; c < 4; c++) {
-			if (!(mask & (1 << c)))
-				continue;
-			emit_mov(pc, dst[c], temp);
+		if (mask & 8) {
+			emit_precossin(pc, temp, src[0][3]);
+			emit_flop(pc, 4, dst[3], temp);
+			if (!(mask &= 7))
+				break;
+			if (temp == dst[3])
+				temp = brdc = temp_temp(pc);
 		}
+		emit_precossin(pc, temp, src[0][0]);
+		emit_flop(pc, 4, brdc, temp);
 		break;
 	case TGSI_OPCODE_SLT:
+	case TGSI_OPCODE_SGE:
+	case TGSI_OPCODE_SEQ:
+	case TGSI_OPCODE_SGT:
+	case TGSI_OPCODE_SLE:
+	case TGSI_OPCODE_SNE:
+		i = map_tgsi_setop_cc(inst->Instruction.Opcode);
 		for (c = 0; c < 4; c++) {
 			if (!(mask & (1 << c)))
 				continue;
-			emit_set(pc, 1, dst[c], src[0][c], src[1][c]);
+			emit_set(pc, i, dst[c], -1, src[0][c], src[1][c]);
 		}
 		break;
 	case TGSI_OPCODE_SUB:
@@ -1583,6 +2081,14 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 		emit_tex(pc, dst, mask, src[0], unit,
 			 inst->InstructionExtTexture.Texture, TRUE);
 		break;
+	case TGSI_OPCODE_TRUNC:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_cvt(pc, dst[c], src[0][c], -1,
+				 CVTOP_TRUNC, CVT_F32_F32 | CVT_RI);
+		}
+		break;
 	case TGSI_OPCODE_XPD:
 		temp = temp_temp(pc);
 		if (mask & (1 << 0)) {
@@ -1607,28 +2113,34 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 		return FALSE;
 	}
 
+	if (brdc) {
+		if (sat)
+			emit_sat(pc, brdc, brdc);
+		for (c = 0; c < 4; c++)
+			if ((mask & (1 << c)) && dst[c] != brdc)
+				emit_mov(pc, dst[c], brdc);
+	} else
 	if (sat) {
 		for (c = 0; c < 4; c++) {
 			if (!(mask & (1 << c)))
 				continue;
-			emit_cvt(pc, rdst[c], dst[c], -1, CVTOP_SAT,
-				 CVT_F32_F32);
+			/* in this case we saturate later */
+			if (dst[c]->type == P_TEMP && dst[c]->index < 0)
+				continue;
+			emit_sat(pc, rdst[c], dst[c]);
 		}
-	} else if (assimilate) {
-		for (c = 0; c < 4; c++)
-			if (rdst[c])
-				assimilate_temp(pc, rdst[c], dst[c]);
 	}
 
 	for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
 		for (c = 0; c < 4; c++) {
 			if (!src[i][c])
 				continue;
+			src[i][c]->neg = 0;
 			if (src[i][c]->index == -1 && src[i][c]->type == P_IMMD)
 				FREE(src[i][c]);
 			else
-			if (src[i][c]->acc == pc->insn_cur)
-				release_hw(pc, src[i][c]);
+			if (src[i][c]->acc < 0 && src[i][c]->type == P_CONST)
+				FREE(src[i][c]); /* indirect constant */
 		}
 	}
 
@@ -1636,180 +2148,271 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 	return TRUE;
 }
 
-/* Adjust a bitmask that indicates what components of a source are used,
- * we use this in tx_prep so we only load interpolants that are needed.
- */
-static void
-insn_adjust_mask(const struct tgsi_full_instruction *insn, unsigned *mask)
-{
-	const struct tgsi_instruction_ext_texture *tex;
-
-	switch (insn->Instruction.Opcode) {
-	case TGSI_OPCODE_DP3:
-		*mask = 0x7;
-		break;
-	case TGSI_OPCODE_DP4:
-	case TGSI_OPCODE_DPH:
-		*mask = 0xF;
-		break;
-	case TGSI_OPCODE_LIT:
-		*mask = 0xB;
-		break;
-	case TGSI_OPCODE_RCP:
-	case TGSI_OPCODE_RSQ:
-		*mask = 0x1;
-		break;
-	case TGSI_OPCODE_TEX:
-	case TGSI_OPCODE_TXP:
-		assert(insn->Instruction.Extended);
-		tex = &insn->InstructionExtTexture;
-
-		*mask = 0x7;
-		if (tex->Texture == TGSI_TEXTURE_1D)
-			*mask = 0x1;
-		else
-		if (tex->Texture == TGSI_TEXTURE_2D)
-			*mask = 0x3;
-
-		if (insn->Instruction.Opcode == TGSI_OPCODE_TXP)
-			*mask |= 0x8;
-		break;
-	default:
-		break;
-	}
-}
-
 static void
-prep_inspect_insn(struct nv50_pc *pc, const union tgsi_full_token *tok,
-		  unsigned *r_usage[2])
+prep_inspect_insn(struct nv50_pc *pc, const struct tgsi_full_instruction *insn)
 {
-	const struct tgsi_full_instruction *insn;
+	struct nv50_reg *reg = NULL;
 	const struct tgsi_full_src_register *src;
 	const struct tgsi_dst_register *dst;
+	unsigned i, c, k, mask;
 
-	unsigned i, c, k, n, mask, *acc_p;
-
-	insn = &tok->FullInstruction;
 	dst = &insn->FullDstRegisters[0].DstRegister;
 	mask = dst->WriteMask;
 
-	if (!r_usage[0])
-		r_usage[0] = CALLOC(pc->temp_nr * 4, sizeof(unsigned));
-	if (!r_usage[1])
-		r_usage[1] = CALLOC(pc->attr_nr * 4, sizeof(unsigned));
+        if (dst->File == TGSI_FILE_TEMPORARY)
+                reg = pc->temp;
+        else
+        if (dst->File == TGSI_FILE_OUTPUT)
+                reg = pc->result;
 
-	if (dst->File == TGSI_FILE_TEMPORARY) {
+	if (reg) {
 		for (c = 0; c < 4; c++) {
 			if (!(mask & (1 << c)))
 				continue;
-			r_usage[0][dst->Index * 4 + c] = pc->insn_nr;
+			reg[dst->Index * 4 + c].acc = pc->insn_nr;
 		}
 	}
 
 	for (i = 0; i < insn->Instruction.NumSrcRegs; i++) {
 		src = &insn->FullSrcRegisters[i];
 
-		switch (src->SrcRegister.File) {
-		case TGSI_FILE_TEMPORARY:
-			acc_p = r_usage[0];
-			break;
-		case TGSI_FILE_INPUT:
-			acc_p = r_usage[1];
-			break;
-		default:
+		if (src->SrcRegister.File == TGSI_FILE_TEMPORARY)
+			reg = pc->temp;
+		else
+		if (src->SrcRegister.File == TGSI_FILE_INPUT)
+			reg = pc->attr;
+		else
 			continue;
-		}
 
-		insn_adjust_mask(insn, &mask);
+		mask = nv50_tgsi_src_mask(insn, i);
 
 		for (c = 0; c < 4; c++) {
 			if (!(mask & (1 << c)))
 				continue;
-
 			k = tgsi_util_get_full_src_register_extswizzle(src, c);
-			switch (k) {
-			case TGSI_EXTSWIZZLE_X:
-			case TGSI_EXTSWIZZLE_Y:
-			case TGSI_EXTSWIZZLE_Z:
-			case TGSI_EXTSWIZZLE_W:
-				n = src->SrcRegister.Index * 4 + k;
-				acc_p[n] = pc->insn_nr;
-				break;
-			default:
-				break;
-			}
+
+			if (k > TGSI_EXTSWIZZLE_W)
+				continue;
+
+			reg[src->SrcRegister.Index * 4 + k].acc = pc->insn_nr;
 		}
 	}
 }
 
+/* Returns a bitmask indicating which dst components need to be
+ * written to temporaries first to avoid 'corrupting' sources.
+ *
+ * m[i]   (out) indicate component to write in the i-th position
+ * rdep[c] (in) bitmasks of dst[i] that require dst[c] as source
+ */
 static unsigned
-load_fp_attrib(struct nv50_pc *pc, int i, unsigned *acc, int *mid,
-	       int *aid, int *p_oid)
+nv50_revdep_reorder(unsigned m[4], unsigned rdep[4])
 {
-	struct nv50_reg *iv;
-	int oid, c, n;
-	unsigned mask = 0;
+	unsigned i, c, x, unsafe;
+
+	for (c = 0; c < 4; c++)
+		m[c] = c;
+
+	/* Swap as long as a dst component written earlier is depended on
+	 * by one written later, but the next one isn't depended on by it.
+	 */
+	for (c = 0; c < 3; c++) {
+		if (rdep[m[c + 1]] & (1 << m[c]))
+			continue; /* if next one is depended on by us */
+		for (i = c + 1; i < 4; i++)
+			/* if we are depended on by a later one */
+			if (rdep[m[c]] & (1 << m[i]))
+				break;
+		if (i == 4)
+			continue;
+		/* now, swap */
+		x = m[c];
+		m[c] = m[c + 1];
+		m[c + 1] = x;
 
-	iv = (pc->interp_mode[i] & INTERP_CENTROID) ? pc->iv_c : pc->iv_p;
+		/* restart */
+		c = 0;
+	}
+
+	/* mark dependencies that could not be resolved by reordering */
+	for (i = 0; i < 3; ++i)
+		for (c = i + 1; c < 4; ++c)
+			if (rdep[m[i]] & (1 << m[c]))
+				unsafe |= (1 << i);
 
-	for (c = 0, n = i * 4; c < 4; c++, n++) {
-		oid = (*p_oid)++;
-		pc->attr[n].type = P_TEMP;
-		pc->attr[n].index = i;
+	/* NOTE: $unsafe is with respect to order, not component */
+	return unsafe;
+}
 
-		if (pc->attr[n].acc == acc[n])
+/* Select a suitable dst register for broadcasting scalar results,
+ * or return NULL if we have to allocate an extra TEMP.
+ *
+ * If e.g. only 1 component is written, we may also emit the final
+ * result to a write-only register.
+ */
+static struct nv50_reg *
+tgsi_broadcast_dst(struct nv50_pc *pc,
+		   const struct tgsi_full_dst_register *fd, unsigned mask)
+{
+	if (fd->DstRegister.File == TGSI_FILE_TEMPORARY) {
+		int c = ffs(~mask & fd->DstRegister.WriteMask);
+		if (c)
+			return tgsi_dst(pc, c - 1, fd);
+	} else {
+		int c = ffs(fd->DstRegister.WriteMask) - 1;
+		if ((1 << c) == fd->DstRegister.WriteMask)
+			return tgsi_dst(pc, c, fd);
+	}
+
+	return NULL;
+}
+
+/* Scan source swizzles and return a bitmask indicating dst regs that
+ * also occur among the src regs, and fill rdep for nv50_revdep_reoder.
+ */
+static unsigned
+nv50_tgsi_scan_swizzle(const struct tgsi_full_instruction *insn,
+		       unsigned rdep[4])
+{
+	const struct tgsi_full_dst_register *fd = &insn->FullDstRegisters[0];
+	const struct tgsi_full_src_register *fs;
+	unsigned i, deqs = 0;
+
+	for (i = 0; i < 4; ++i)
+		rdep[i] = 0;
+
+	for (i = 0; i < insn->Instruction.NumSrcRegs; i++) {
+		unsigned chn, mask = nv50_tgsi_src_mask(insn, i);
+		boolean neg_supp = negate_supported(insn, i);
+
+		fs = &insn->FullSrcRegisters[i];
+		if (fs->SrcRegister.File != fd->DstRegister.File ||
+		    fs->SrcRegister.Index != fd->DstRegister.Index)
 			continue;
-		mask |= (1 << c);
 
-		pc->attr[n].acc = acc[n];
-		pc->attr[n].rhw = pc->attr[n].hw = -1;
-		alloc_reg(pc, &pc->attr[n]);
+		for (chn = 0; chn < 4; ++chn) {
+			unsigned s, c;
 
-		pc->attr[n].rhw = (*aid)++;
-		emit_interp(pc, &pc->attr[n], iv, pc->interp_mode[i]);
+			if (!(mask & (1 << chn))) /* src is not read */
+				continue;
+			c = tgsi_util_get_full_src_register_extswizzle(fs, chn);
+			s = tgsi_util_get_full_src_register_sign_mode(fs, chn);
+
+			if (c > TGSI_EXTSWIZZLE_W ||
+			    !(fd->DstRegister.WriteMask & (1 << c)))
+				continue;
 
-		pc->p->cfg.fp.map[(*mid) / 4] |= oid << (8 * ((*mid) % 4));
-		(*mid)++;
-		pc->p->cfg.fp.regs[1] += 0x00010001;
+			/* no danger if src is copied to TEMP first */
+			if ((s != TGSI_UTIL_SIGN_KEEP) &&
+			    (s != TGSI_UTIL_SIGN_TOGGLE || !neg_supp))
+				continue;
+
+			rdep[c] |= nv50_tgsi_dst_revdep(
+				insn->Instruction.Opcode, i, chn);
+			deqs |= (1 << c);
+		}
 	}
 
-	return mask;
+	return deqs;
 }
 
 static boolean
-nv50_program_tx_prep(struct nv50_pc *pc)
+nv50_tgsi_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 {
-	struct tgsi_parse_context p;
-	boolean ret = FALSE;
-	unsigned i, c;
-	unsigned fcol, bcol, fcrd, depr;
+	struct tgsi_full_instruction insn = tok->FullInstruction;
+	const struct tgsi_full_dst_register *fd;
+	unsigned i, deqs, rdep[4], m[4];
+
+	fd = &tok->FullInstruction.FullDstRegisters[0];
+	deqs = nv50_tgsi_scan_swizzle(&insn, rdep);
+
+	if (is_scalar_op(insn.Instruction.Opcode)) {
+		pc->r_brdc = tgsi_broadcast_dst(pc, fd, deqs);
+		if (!pc->r_brdc)
+			pc->r_brdc = temp_temp(pc);
+		return nv50_program_tx_insn(pc, &insn);
+	}
+	pc->r_brdc = NULL;
+
+	if (!deqs)
+		return nv50_program_tx_insn(pc, &insn);
+
+	deqs = nv50_revdep_reorder(m, rdep);
+
+	for (i = 0; i < 4; ++i) {
+		assert(pc->r_dst[m[i]] == NULL);
+
+		insn.FullDstRegisters[0].DstRegister.WriteMask =
+			fd->DstRegister.WriteMask & (1 << m[i]);
+
+		if (!insn.FullDstRegisters[0].DstRegister.WriteMask)
+			continue;
 
-	/* count (centroid) perspective interpolations */
-	unsigned centroid_loads = 0;
-	unsigned perspect_loads = 0;
+		if (deqs & (1 << i))
+			pc->r_dst[m[i]] = alloc_temp(pc, NULL);
 
-	/* track register access for temps and attrs */
-	unsigned *r_usage[2];
-	r_usage[0] = NULL;
-	r_usage[1] = NULL;
+		if (!nv50_program_tx_insn(pc, &insn))
+			return FALSE;
+	}
 
-	depr = fcol = bcol = fcrd = 0xffff;
+	for (i = 0; i < 4; i++) {
+		struct nv50_reg *reg = pc->r_dst[i];
+		if (!reg)
+			continue;
+		pc->r_dst[i] = NULL;
 
-	if (pc->p->type == PIPE_SHADER_FRAGMENT) {
-		pc->p->cfg.fp.regs[0] = 0x01000404;
-		pc->p->cfg.fp.regs[1] = 0x00000400;
+		if (insn.Instruction.Saturate == TGSI_SAT_ZERO_ONE)
+			emit_sat(pc, tgsi_dst(pc, i, fd), reg);
+		else
+			emit_mov(pc, tgsi_dst(pc, i, fd), reg);
+		free_temp(pc, reg);
 	}
 
-	tgsi_parse_init(&p, pc->p->pipe.tokens);
-	while (!tgsi_parse_end_of_tokens(&p)) {
-		const union tgsi_full_token *tok = &p.FullToken;
+	return TRUE;
+}
+
+static void
+load_interpolant(struct nv50_pc *pc, struct nv50_reg *reg)
+{
+	struct nv50_reg *iv, **ppiv;
+	unsigned mode = pc->interp_mode[reg->index];
+
+	ppiv = (mode & INTERP_CENTROID) ? &pc->iv_c : &pc->iv_p;
+	iv = *ppiv;
+
+	if ((mode & INTERP_PERSPECTIVE) && !iv) {
+		iv = *ppiv = alloc_temp(pc, NULL);
+		iv->rhw = popcnt4(pc->p->cfg.regs[1] >> 24) - 1;
 
-		tgsi_parse_token(&p);
+		emit_interp(pc, iv, NULL, mode & INTERP_CENTROID);
+		emit_flop(pc, 0, iv, iv);
+
+		/* XXX: when loading interpolants dynamically, move these
+		 * to the program head, or make sure it can't be skipped.
+		 */
+	}
+
+	emit_interp(pc, reg, iv, mode);
+}
+
+static boolean
+nv50_program_tx_prep(struct nv50_pc *pc)
+{
+	struct tgsi_parse_context tp;
+	struct nv50_program *p = pc->p;
+	boolean ret = FALSE;
+	unsigned i, c, flat_nr = 0;
+
+	tgsi_parse_init(&tp, pc->p->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&tp)) {
+		const union tgsi_full_token *tok = &tp.FullToken;
+
+		tgsi_parse_token(&tp);
 		switch (tok->Token.Type) {
 		case TGSI_TOKEN_TYPE_IMMEDIATE:
 		{
 			const struct tgsi_full_immediate *imm =
-				&p.FullToken.FullImmediate;
+				&tp.FullToken.FullImmediate;
 
 			ctor_immd(pc, imm->u[0].Float,
 				      imm->u[1].Float,
@@ -1820,88 +2423,69 @@ nv50_program_tx_prep(struct nv50_pc *pc)
 		case TGSI_TOKEN_TYPE_DECLARATION:
 		{
 			const struct tgsi_full_declaration *d;
-			unsigned last, first, mode;
+			unsigned si, last, first, mode;
 
-			d = &p.FullToken.FullDeclaration;
+			d = &tp.FullToken.FullDeclaration;
 			first = d->DeclarationRange.First;
 			last = d->DeclarationRange.Last;
 
 			switch (d->Declaration.File) {
 			case TGSI_FILE_TEMPORARY:
-				if (pc->temp_nr < (last + 1))
-					pc->temp_nr = last + 1;
 				break;
 			case TGSI_FILE_OUTPUT:
-				if (pc->result_nr < (last + 1))
-					pc->result_nr = last + 1;
-
-				if (!d->Declaration.Semantic)
+				if (!d->Declaration.Semantic ||
+				    p->type == PIPE_SHADER_FRAGMENT)
 					break;
 
+				si = d->Semantic.SemanticIndex;
 				switch (d->Semantic.SemanticName) {
-				case TGSI_SEMANTIC_POSITION:
-					depr = first;
-					pc->p->cfg.fp.regs[2] |= 0x00000100;
-					pc->p->cfg.fp.regs[3] |= 0x00000011;
+				case TGSI_SEMANTIC_BCOLOR:
+					p->cfg.two_side[si].hw = first;
+					if (p->cfg.io_nr > first)
+						p->cfg.io_nr = first;
 					break;
+				case TGSI_SEMANTIC_PSIZE:
+					p->cfg.psiz = first;
+					if (p->cfg.io_nr > first)
+						p->cfg.io_nr = first;
+					break;
+					/*
+				case TGSI_SEMANTIC_CLIP_DISTANCE:
+					p->cfg.clpd = MIN2(p->cfg.clpd, first);
+					break;
+					*/
 				default:
 					break;
 				}
-
 				break;
 			case TGSI_FILE_INPUT:
 			{
-				if (pc->attr_nr < (last + 1))
-					pc->attr_nr = last + 1;
-
-				if (pc->p->type != PIPE_SHADER_FRAGMENT)
+				if (p->type != PIPE_SHADER_FRAGMENT)
 					break;
 
 				switch (d->Declaration.Interpolate) {
 				case TGSI_INTERPOLATE_CONSTANT:
 					mode = INTERP_FLAT;
+					flat_nr++;
 					break;
 				case TGSI_INTERPOLATE_PERSPECTIVE:
 					mode = INTERP_PERSPECTIVE;
+					p->cfg.regs[1] |= 0x08 << 24;
 					break;
 				default:
 					mode = INTERP_LINEAR;
 					break;
 				}
-
-				if (d->Declaration.Semantic) {
-					switch (d->Semantic.SemanticName) {
-					case TGSI_SEMANTIC_POSITION:
-						fcrd = first;
-						break;
-					case TGSI_SEMANTIC_COLOR:
-						fcol = first;
-						mode = INTERP_PERSPECTIVE;
-						break;
-					case TGSI_SEMANTIC_BCOLOR:
-						bcol = first;
-						mode = INTERP_PERSPECTIVE;
-						break;
-					}
-				}
-
-				if (d->Declaration.Centroid) {
+				if (d->Declaration.Centroid)
 					mode |= INTERP_CENTROID;
-					if (mode & INTERP_PERSPECTIVE)
-						centroid_loads++;
-				} else
-				if (mode & INTERP_PERSPECTIVE)
-					perspect_loads++;
 
 				assert(last < 32);
 				for (i = first; i <= last; i++)
 					pc->interp_mode[i] = mode;
 			}
 				break;
+			case TGSI_FILE_ADDRESS:
 			case TGSI_FILE_CONSTANT:
-				if (pc->param_nr < (last + 1))
-					pc->param_nr = last + 1;
-				break;
 			case TGSI_FILE_SAMPLER:
 				break;
 			default:
@@ -1913,182 +2497,157 @@ nv50_program_tx_prep(struct nv50_pc *pc)
 			break;
 		case TGSI_TOKEN_TYPE_INSTRUCTION:
 			pc->insn_nr++;
-			prep_inspect_insn(pc, tok, r_usage);
+			prep_inspect_insn(pc, &tok->FullInstruction);
 			break;
 		default:
 			break;
 		}
 	}
 
-	if (pc->temp_nr) {
-		pc->temp = CALLOC(pc->temp_nr * 4, sizeof(struct nv50_reg));
-		if (!pc->temp)
-			goto out_err;
+	if (p->type == PIPE_SHADER_VERTEX) {
+		int rid = 0;
 
-		for (i = 0; i < pc->temp_nr; i++) {
-			for (c = 0; c < 4; c++) {
-				pc->temp[i*4+c].type = P_TEMP;
-				pc->temp[i*4+c].hw = -1;
-				pc->temp[i*4+c].rhw = -1;
-				pc->temp[i*4+c].index = i;
-				pc->temp[i*4+c].acc = r_usage[0][i*4+c];
+		for (i = 0; i < pc->attr_nr * 4; ++i) {
+			if (pc->attr[i].acc) {
+				pc->attr[i].hw = rid++;
+				p->cfg.attr[i / 32] |= 1 << (i % 32);
 			}
 		}
-	}
 
-	if (pc->attr_nr) {
-		int oid = 4, mid = 4, aid = 0;
-		/* oid = VP output id
-		 * aid = FP attribute/interpolant id
-		 * mid = VP output mapping field ID
-		 */
+		for (i = 0, rid = 0; i < pc->result_nr; ++i) {
+			p->cfg.io[i].hw = rid;
+			p->cfg.io[i].id_vp = i;
 
-		pc->attr = CALLOC(pc->attr_nr * 4, sizeof(struct nv50_reg));
-		if (!pc->attr)
-			goto out_err;
-
-		if (pc->p->type == PIPE_SHADER_FRAGMENT) {
-			/* position should be loaded first */
-			if (fcrd != 0xffff) {
-				unsigned mask;
-				mid = 0;
-				mask = load_fp_attrib(pc, fcrd, r_usage[1],
-						      &mid, &aid, &oid);
-				oid = 0;
-				pc->p->cfg.fp.regs[1] |= (mask << 24);
-				pc->p->cfg.fp.map[0] = 0x04040404 * fcrd;
-			}
-			pc->p->cfg.fp.map[0] += 0x03020100;
-
-			/* should do MAD fcrd.xy, fcrd, SOME_CONST, fcrd */
-
-			if (perspect_loads) {
-				pc->iv_p = alloc_temp(pc, NULL);
-
-				if (!(pc->p->cfg.fp.regs[1] & 0x08000000)) {
-					pc->p->cfg.fp.regs[1] |= 0x08000000;
-					pc->iv_p->rhw = aid++;
-					emit_interp(pc, pc->iv_p, NULL,
-						    INTERP_LINEAR);
-					emit_flop(pc, 0, pc->iv_p, pc->iv_p);
-				} else {
-					pc->iv_p->rhw = aid - 1;
-					emit_flop(pc, 0, pc->iv_p,
-						  &pc->attr[fcrd * 4 + 3]);
-				}
-			}
-
-			if (centroid_loads) {
-				pc->iv_c = alloc_temp(pc, NULL);
-				pc->iv_c->rhw = pc->iv_p ? aid - 1 : aid++;
-				emit_interp(pc, pc->iv_c, NULL,
-					    INTERP_CENTROID);
-				emit_flop(pc, 0, pc->iv_c, pc->iv_c);
-				pc->p->cfg.fp.regs[1] |= 0x08000000;
+			for (c = 0; c < 4; ++c) {
+				int n = i * 4 + c;
+				if (!pc->result[n].acc)
+					continue;
+				pc->result[n].hw = rid++;
+				p->cfg.io[i].mask |= 1 << c;
 			}
+		}
 
-			for (c = 0; c < 4; c++) {
-				/* I don't know what these values do, but
-				 * let's set them like the blob does:
-				 */
-				if (fcol != 0xffff && r_usage[1][fcol * 4 + c])
-					pc->p->cfg.fp.regs[0] += 0x00010000;
-				if (bcol != 0xffff && r_usage[1][bcol * 4 + c])
-					pc->p->cfg.fp.regs[0] += 0x00010000;
-			}
+		for (c = 0; c < 2; ++c)
+			if (p->cfg.two_side[c].hw < 0x40)
+				p->cfg.two_side[c] = p->cfg.io[
+					p->cfg.two_side[c].hw];
 
-			for (i = 0; i < pc->attr_nr; i++)
-				load_fp_attrib(pc, i, r_usage[1],
-					       &mid, &aid, &oid);
+		if (p->cfg.psiz < 0x40)
+			p->cfg.psiz = p->cfg.io[p->cfg.psiz].hw;
+	} else
+	if (p->type == PIPE_SHADER_FRAGMENT) {
+		int rid, aid;
+		unsigned n = 0, m = pc->attr_nr - flat_nr;
 
-			if (pc->iv_p)
-				free_temp(pc, pc->iv_p);
-			if (pc->iv_c)
-				free_temp(pc, pc->iv_c);
+		int base = (TGSI_SEMANTIC_POSITION ==
+			    p->info.input_semantic_name[0]) ? 0 : 1;
 
-			pc->p->cfg.fp.high_map = (mid / 4);
-			pc->p->cfg.fp.high_map += ((mid % 4) ? 1 : 0);
-		} else {
-			/* vertex program */
-			for (i = 0; i < pc->attr_nr * 4; i++) {
-				pc->p->cfg.vp.attr[aid / 32] |=
-					(1 << (aid % 32));
-				pc->attr[i].type = P_ATTR;
-				pc->attr[i].hw = aid++;
-				pc->attr[i].index = i / 4;
+		/* non-flat interpolants have to be mapped to
+		 * the lower hardware IDs, so sort them:
+		 */
+		for (i = 0; i < pc->attr_nr; i++) {
+			if (pc->interp_mode[i] == INTERP_FLAT) {
+				p->cfg.io[m].id_vp = i + base;
+				p->cfg.io[m++].id_fp = i;
+			} else {
+				if (!(pc->interp_mode[i] & INTERP_PERSPECTIVE))
+					p->cfg.io[n].linear = TRUE;
+				p->cfg.io[n].id_vp = i + base;
+				p->cfg.io[n++].id_fp = i;
 			}
 		}
-	}
 
-	if (pc->result_nr) {
-		int rid = 0;
+		if (!base) /* set w-coordinate mask from perspective interp */
+			p->cfg.io[0].mask |= p->cfg.regs[1] >> 24;
 
-		pc->result = CALLOC(pc->result_nr * 4, sizeof(struct nv50_reg));
-		if (!pc->result)
-			goto out_err;
+		aid = popcnt4( /* if fcrd isn't contained in cfg.io */
+			base ? (p->cfg.regs[1] >> 24) : p->cfg.io[0].mask);
 
-		for (i = 0; i < pc->result_nr; i++) {
-			for (c = 0; c < 4; c++) {
-				if (pc->p->type == PIPE_SHADER_FRAGMENT) {
-					pc->result[i*4+c].type = P_TEMP;
-					pc->result[i*4+c].hw = -1;
-					pc->result[i*4+c].rhw = (i == depr) ?
-						-1 : rid++;
-				} else {
-					pc->result[i*4+c].type = P_RESULT;
-					pc->result[i*4+c].hw = rid++;
-				}
-				pc->result[i*4+c].index = i;
-			}
+		for (n = 0; n < pc->attr_nr; ++n) {
+			p->cfg.io[n].hw = rid = aid;
+			i = p->cfg.io[n].id_fp;
+
+			for (c = 0; c < 4; ++c) {
+				if (!pc->attr[i * 4 + c].acc)
+					continue;
+				pc->attr[i * 4 + c].rhw = rid++;
+				p->cfg.io[n].mask |= 1 << c;
 
-			if (pc->p->type == PIPE_SHADER_FRAGMENT &&
-			    depr != 0xffff) {
-				pc->result[depr * 4 + 2].rhw =
-					(pc->result_nr - 1) * 4;
+				load_interpolant(pc, &pc->attr[i * 4 + c]);
 			}
+			aid += popcnt4(p->cfg.io[n].mask);
 		}
-	}
 
-	if (pc->param_nr) {
-		int rid = 0;
+		if (!base)
+			p->cfg.regs[1] |= p->cfg.io[0].mask << 24;
 
-		pc->param = CALLOC(pc->param_nr * 4, sizeof(struct nv50_reg));
-		if (!pc->param)
-			goto out_err;
+		m = popcnt4(p->cfg.regs[1] >> 24);
 
-		for (i = 0; i < pc->param_nr; i++) {
-			for (c = 0; c < 4; c++) {
-				pc->param[i*4+c].type = P_CONST;
-				pc->param[i*4+c].hw = rid++;
-				pc->param[i*4+c].index = i;
+		/* set count of non-position inputs and of non-flat
+		 * non-position inputs for FP_INTERPOLANT_CTRL
+		 */
+		p->cfg.regs[1] |= aid - m;
+
+		if (flat_nr) {
+			i = p->cfg.io[pc->attr_nr - flat_nr].hw;
+			p->cfg.regs[1] |= (i - m) << 16;
+		} else
+			p->cfg.regs[1] |= p->cfg.regs[1] << 16;
+
+		/* mark color semantic for light-twoside */
+		n = 0x40;
+		for (i = 0; i < pc->attr_nr; i++) {
+			ubyte si, sn;
+
+			sn = p->info.input_semantic_name[p->cfg.io[i].id_fp];
+			si = p->info.input_semantic_index[p->cfg.io[i].id_fp];
+
+			if (sn == TGSI_SEMANTIC_COLOR) {
+				p->cfg.two_side[si] = p->cfg.io[i];
+
+				/* increase colour count */
+				p->cfg.regs[0] += popcnt4(
+					p->cfg.two_side[si].mask) << 16;
+
+				n = MIN2(n, p->cfg.io[i].hw - m);
 			}
 		}
+		if (n < 0x40)
+			p->cfg.regs[0] += n;
+
+		/* Initialize FP results:
+		 * FragDepth is always first TGSI and last hw output
+		 */
+		i = p->info.writes_z ? 4 : 0;
+		for (rid = 0; i < pc->result_nr * 4; i++)
+			pc->result[i].rhw = rid++;
+		if (p->info.writes_z)
+			pc->result[2].rhw = rid;
+
+		p->cfg.high_result = rid;
 	}
 
 	if (pc->immd_nr) {
 		int rid = 0;
 
-		pc->immd = CALLOC(pc->immd_nr * 4, sizeof(struct nv50_reg));
+		pc->immd = MALLOC(pc->immd_nr * 4 * sizeof(struct nv50_reg));
 		if (!pc->immd)
 			goto out_err;
 
 		for (i = 0; i < pc->immd_nr; i++) {
-			for (c = 0; c < 4; c++) {
-				pc->immd[i*4+c].type = P_IMMD;
-				pc->immd[i*4+c].hw = rid++;
-				pc->immd[i*4+c].index = i;
-			}
+			for (c = 0; c < 4; c++, rid++)
+				ctor_reg(&pc->immd[rid], P_IMMD, i, rid);
 		}
 	}
 
 	ret = TRUE;
 out_err:
-	if (r_usage[0])
-		FREE(r_usage[0]);
-	if (r_usage[1])
-		FREE(r_usage[1]);
+	if (pc->iv_p)
+		free_temp(pc, pc->iv_p);
+	if (pc->iv_c)
+		free_temp(pc, pc->iv_c);
 
-	tgsi_parse_free(&p);
+	tgsi_parse_free(&tp);
 	return ret;
 }
 
@@ -2110,18 +2669,175 @@ free_nv50_pc(struct nv50_pc *pc)
 }
 
 static boolean
+ctor_nv50_pc(struct nv50_pc *pc, struct nv50_program *p)
+{
+	int i, c;
+	unsigned rtype[2] = { P_ATTR, P_RESULT };
+
+	pc->p = p;
+	pc->temp_nr = p->info.file_max[TGSI_FILE_TEMPORARY] + 1;
+	pc->attr_nr = p->info.file_max[TGSI_FILE_INPUT] + 1;
+	pc->result_nr = p->info.file_max[TGSI_FILE_OUTPUT] + 1;
+	pc->param_nr = p->info.file_max[TGSI_FILE_CONSTANT] + 1;
+	pc->addr_nr = p->info.file_max[TGSI_FILE_ADDRESS] + 1;
+	assert(pc->addr_nr <= 2);
+
+	p->cfg.high_temp = 4;
+
+	p->cfg.two_side[0].hw = 0x40;
+	p->cfg.two_side[1].hw = 0x40;
+
+	switch (p->type) {
+	case PIPE_SHADER_VERTEX:
+		p->cfg.psiz = 0x40;
+		p->cfg.clpd = 0x40;
+		p->cfg.io_nr = pc->result_nr;
+		break;
+	case PIPE_SHADER_FRAGMENT:
+		rtype[0] = rtype[1] = P_TEMP;
+
+		p->cfg.regs[0] = 0x01000004;
+		p->cfg.io_nr = pc->attr_nr;
+
+		if (p->info.writes_z) {
+			p->cfg.regs[2] |= 0x00000100;
+			p->cfg.regs[3] |= 0x00000011;
+		}
+		if (p->info.uses_kill)
+			p->cfg.regs[2] |= 0x00100000;
+		break;
+	}
+
+	if (pc->temp_nr) {
+		pc->temp = MALLOC(pc->temp_nr * 4 * sizeof(struct nv50_reg));
+		if (!pc->temp)
+			return FALSE;
+
+		for (i = 0; i < pc->temp_nr * 4; ++i)
+			ctor_reg(&pc->temp[i], P_TEMP, i / 4, -1);
+	}
+
+	if (pc->attr_nr) {
+		pc->attr = MALLOC(pc->attr_nr * 4 * sizeof(struct nv50_reg));
+		if (!pc->attr)
+			return FALSE;
+
+		for (i = 0; i < pc->attr_nr * 4; ++i)
+			ctor_reg(&pc->attr[i], rtype[0], i / 4, -1);
+	}
+
+	if (pc->result_nr) {
+		unsigned nr = pc->result_nr * 4;
+
+		pc->result = MALLOC(nr * sizeof(struct nv50_reg));
+		if (!pc->result)
+			return FALSE;
+
+		for (i = 0; i < nr; ++i)
+			ctor_reg(&pc->result[i], rtype[1], i / 4, -1);
+	}
+
+	if (pc->param_nr) {
+		int rid = 0;
+
+		pc->param = MALLOC(pc->param_nr * 4 * sizeof(struct nv50_reg));
+		if (!pc->param)
+			return FALSE;
+
+		for (i = 0; i < pc->param_nr; ++i)
+			for (c = 0; c < 4; ++c, ++rid)
+				ctor_reg(&pc->param[rid], P_CONST, i, rid);
+	}
+
+	if (pc->addr_nr) {
+		pc->addr = CALLOC(pc->addr_nr * 4, sizeof(struct nv50_reg *));
+		if (!pc->addr)
+			return FALSE;
+	}
+	for (i = 0; i < NV50_SU_MAX_ADDR; ++i)
+		ctor_reg(&pc->r_addr[i], P_ADDR, -1, i + 1);
+
+	return TRUE;
+}
+
+static void
+nv50_fp_move_results(struct nv50_pc *pc)
+{
+	struct nv50_reg reg;
+	unsigned i;
+
+	ctor_reg(&reg, P_TEMP, -1, -1);
+
+	for (i = 0; i < pc->result_nr * 4; ++i) {
+		if (pc->result[i].rhw < 0 || pc->result[i].hw < 0)
+			continue;
+		if (pc->result[i].rhw != pc->result[i].hw) {
+			reg.hw = pc->result[i].rhw;
+			emit_mov(pc, &reg, &pc->result[i]);
+		}
+	}
+}
+
+static void
+nv50_program_fixup_insns(struct nv50_pc *pc)
+{
+	struct nv50_program_exec *e, *prev = NULL, **bra_list;
+	unsigned i, n, pos;
+
+	bra_list = CALLOC(pc->p->exec_size, sizeof(struct nv50_program_exec *));
+
+	/* Collect branch instructions, we need to adjust their offsets
+	 * when converting 32 bit instructions to 64 bit ones
+	 */
+	for (n = 0, e = pc->p->exec_head; e; e = e->next)
+		if (e->param.index >= 0 && !e->param.mask)
+			bra_list[n++] = e;
+
+	/* Make sure we don't have any single 32 bit instructions. */
+	for (e = pc->p->exec_head, pos = 0; e; e = e->next) {
+		pos += is_long(e) ? 2 : 1;
+
+		if ((pos & 1) && (!e->next || is_long(e->next))) {
+			for (i = 0; i < n; ++i)
+				if (bra_list[i]->param.index >= pos)
+					bra_list[i]->param.index += 1;
+			convert_to_long(pc, e);
+			++pos;
+		}
+		if (e->next)
+			prev = e;
+	}
+
+	assert(!is_immd(pc->p->exec_head));
+	assert(!is_immd(pc->p->exec_tail));
+
+	/* last instruction must be long so it can have the end bit set */
+	if (!is_long(pc->p->exec_tail)) {
+		convert_to_long(pc, pc->p->exec_tail);
+		if (prev)
+			convert_to_long(pc, prev);
+	}
+	assert(!(pc->p->exec_tail->inst[1] & 2));
+	/* set the end-bit */
+	pc->p->exec_tail->inst[1] |= 1;
+
+	FREE(bra_list);
+}
+
+static boolean
 nv50_program_tx(struct nv50_program *p)
 {
 	struct tgsi_parse_context parse;
 	struct nv50_pc *pc;
-	unsigned k;
 	boolean ret;
 
 	pc = CALLOC_STRUCT(nv50_pc);
 	if (!pc)
 		return FALSE;
-	pc->p = p;
-	pc->p->cfg.high_temp = 4;
+
+	ret = ctor_nv50_pc(pc, p);
+	if (ret == FALSE)
+		goto out_cleanup;
 
 	ret = nv50_program_tx_prep(pc);
 	if (ret == FALSE)
@@ -2141,7 +2857,7 @@ nv50_program_tx(struct nv50_program *p)
 		switch (tok->Token.Type) {
 		case TGSI_TOKEN_TYPE_INSTRUCTION:
 			++pc->insn_cur;
-			ret = nv50_program_tx_insn(pc, tok);
+			ret = nv50_tgsi_insn(pc, tok);
 			if (ret == FALSE)
 				goto out_err;
 			break;
@@ -2150,48 +2866,10 @@ nv50_program_tx(struct nv50_program *p)
 		}
 	}
 
-	if (p->type == PIPE_SHADER_FRAGMENT) {
-		struct nv50_reg out;
-
-		out.type = P_TEMP;
-		for (k = 0; k < pc->result_nr * 4; k++) {
-			if (pc->result[k].rhw == -1)
-				continue;
-			if (pc->result[k].hw != pc->result[k].rhw) {
-				out.hw = pc->result[k].rhw;
-				emit_mov(pc, &out, &pc->result[k]);
-			}
-			if (pc->p->cfg.high_result < (pc->result[k].rhw + 1))
-				pc->p->cfg.high_result = pc->result[k].rhw + 1;
-		}
-	}
-
-	/* look for single half instructions and make them long */
-	struct nv50_program_exec *e, *e_prev;
-
-	for (k = 0, e = pc->p->exec_head, e_prev = NULL; e; e = e->next) {
-		if (!is_long(e))
-			k++;
-
-		if (!e->next || is_long(e->next)) {
-			if (k & 1)
-				convert_to_long(pc, e);
-			k = 0;
-		}
-
-		if (e->next)
-			e_prev = e;
-	}
-
-	if (!is_long(pc->p->exec_tail)) {
-		/* this may occur if moving FP results */
-		assert(e_prev && !is_long(e_prev));
-		convert_to_long(pc, e_prev);
-		convert_to_long(pc, pc->p->exec_tail);
-	}
+	if (pc->p->type == PIPE_SHADER_FRAGMENT)
+		nv50_fp_move_results(pc);
 
-	assert(is_long(pc->p->exec_tail) && !is_immd(pc->p->exec_head));
-	pc->p->exec_tail->inst[1] |= 0x00000001;
+	nv50_program_fixup_insns(pc);
 
 	p->param_nr = pc->param_nr * 4;
 	p->immd_nr = pc->immd_nr * 4;
@@ -2258,30 +2936,19 @@ nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p)
 					 p->immd_nr, NV50_CB_PMISC);
 	}
 
-	if (!p->data[1] && p->param_nr) {
-		struct nouveau_resource *heap =
-			nv50->screen->parm_heap[p->type];
-
-		if (nouveau_resource_alloc(heap, p->param_nr, p, &p->data[1])) {
-			while (heap->next && heap->size < p->param_nr) {
-				struct nv50_program *evict = heap->next->priv;
-				nouveau_resource_free(&evict->data[1]);
-			}
-
-			if (nouveau_resource_alloc(heap, p->param_nr, p,
-						   &p->data[1]))
-				assert(0);
-		}
-	}
+	assert(p->param_nr <= 512);
 
 	if (p->param_nr) {
-		unsigned cbuf = NV50_CB_PVP;
+		unsigned cb;
 		float *map = pipe_buffer_map(pscreen, nv50->constbuf[p->type],
 					     PIPE_BUFFER_USAGE_CPU_READ);
-		if (p->type == PIPE_SHADER_FRAGMENT)
-			cbuf = NV50_CB_PFP;
-		nv50_program_upload_data(nv50, map, p->data[1]->start,
-					 p->param_nr, cbuf);
+
+		if (p->type == PIPE_SHADER_VERTEX)
+			cb = NV50_CB_PVP;
+		else
+			cb = NV50_CB_PFP;
+
+		nv50_program_upload_data(nv50, map, 0, p->param_nr, cb);
 		pipe_buffer_unmap(pscreen, nv50->constbuf[p->type]);
 	}
 }
@@ -2303,32 +2970,41 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p)
 		upload = TRUE;
 	}
 
-	if ((p->data[0] && p->data[0]->start != p->data_start[0]) ||
-		(p->data[1] && p->data[1]->start != p->data_start[1])) {
-		for (e = p->exec_head; e; e = e->next) {
-			unsigned ei, ci, bs;
+	if (p->data[0] && p->data[0]->start != p->data_start[0])
+		upload = TRUE;
 
-			if (e->param.index < 0)
-				continue;
-			bs = (e->inst[1] >> 22) & 0x07;
-			assert(bs < 2);
-			ei = e->param.shift >> 5;
-			ci = e->param.index + p->data[bs]->start;
+	if (!upload)
+		return;
+
+	for (e = p->exec_head; e; e = e->next) {
+		unsigned ei, ci, bs;
+
+		if (e->param.index < 0)
+			continue;
+
+		if (e->param.mask == 0) {
+			assert(!(e->param.index & 1));
+			/* seem to be 8 byte steps */
+			ei = (e->param.index >> 1) + 0 /* START_ID */;
 
-			e->inst[ei] &= ~e->param.mask;
-			e->inst[ei] |= (ci << e->param.shift);
+			e->inst[0] &= 0xf0000fff;
+			e->inst[0] |= ei << 12;
+			continue;
 		}
 
-		if (p->data[0])
-			p->data_start[0] = p->data[0]->start;
-		if (p->data[1])
-			p->data_start[1] = p->data[1]->start;
+		bs = (e->inst[1] >> 22) & 0x07;
+		assert(bs < 2);
+		ei = e->param.shift >> 5;
+		ci = e->param.index;
+		if (bs == 0)
+			ci += p->data[bs]->start;
 
-		upload = TRUE;
+		e->inst[ei] &= ~e->param.mask;
+		e->inst[ei] |= (ci << e->param.shift);
 	}
 
-	if (!upload)
-		return;
+	if (p->data[0])
+		p->data_start[0] = p->data[0]->start;
 
 #ifdef NV50_PROGRAM_DUMP
 	NOUVEAU_ERR("-------\n");
@@ -2402,8 +3078,8 @@ nv50_vertprog_validate(struct nv50_context *nv50)
 	so_reloc (so, p->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD |
 		      NOUVEAU_BO_LOW, 0, 0);
 	so_method(so, tesla, NV50TCL_VP_ATTR_EN_0, 2);
-	so_data  (so, p->cfg.vp.attr[0]);
-	so_data  (so, p->cfg.vp.attr[1]);
+	so_data  (so, p->cfg.attr[0]);
+	so_data  (so, p->cfg.attr[1]);
 	so_method(so, tesla, NV50TCL_VP_REG_ALLOC_RESULT, 1);
 	so_data  (so, p->cfg.high_result);
 	so_method(so, tesla, NV50TCL_VP_RESULT_MAP_SIZE, 2);
@@ -2421,7 +3097,6 @@ nv50_fragprog_validate(struct nv50_context *nv50)
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct nv50_program *p = nv50->fragprog;
 	struct nouveau_stateobj *so;
-	unsigned i;
 
 	if (!p->translated) {
 		nv50_program_validate(nv50, p);
@@ -2438,29 +3113,186 @@ nv50_fragprog_validate(struct nv50_context *nv50)
 		      NOUVEAU_BO_HIGH, 0, 0);
 	so_reloc (so, p->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD |
 		      NOUVEAU_BO_LOW, 0, 0);
-	so_method(so, tesla, NV50TCL_MAP_SEMANTIC_0, 4);
-	so_data  (so, p->cfg.fp.regs[0]); /* 0x01000404 / 0x00040404 */
-	so_data  (so, 0x00000004);
-	so_data  (so, 0x00000000);
-	so_data  (so, 0x00000000);
-	so_method(so, tesla, NV50TCL_VP_RESULT_MAP(0), p->cfg.fp.high_map);
-	for (i = 0; i < p->cfg.fp.high_map; i++)
-		so_data(so, p->cfg.fp.map[i]);
-	so_method(so, tesla, NV50TCL_FP_INTERPOLANT_CTRL, 2);
-	so_data  (so, p->cfg.fp.regs[1]); /* 0x08040404 / 0x0f000401 */
+	so_method(so, tesla, NV50TCL_FP_REG_ALLOC_TEMP, 1);
 	so_data  (so, p->cfg.high_temp);
 	so_method(so, tesla, NV50TCL_FP_RESULT_COUNT, 1);
 	so_data  (so, p->cfg.high_result);
 	so_method(so, tesla, NV50TCL_FP_CTRL_UNK19A8, 1);
-	so_data  (so, p->cfg.fp.regs[2]);
+	so_data  (so, p->cfg.regs[2]);
 	so_method(so, tesla, NV50TCL_FP_CTRL_UNK196C, 1);
-	so_data  (so, p->cfg.fp.regs[3]);
+	so_data  (so, p->cfg.regs[3]);
 	so_method(so, tesla, NV50TCL_FP_START_ID, 1);
 	so_data  (so, 0); /* program start offset */
 	so_ref(so, &nv50->state.fragprog);
 	so_ref(NULL, &so);
 }
 
+static void
+nv50_pntc_replace(struct nv50_context *nv50, uint32_t pntc[8], unsigned base)
+{
+	struct nv50_program *fp = nv50->fragprog;
+	struct nv50_program *vp = nv50->vertprog;
+	unsigned i, c, m = base;
+
+	/* XXX: This can't work correctly in all cases yet, we either
+	 * have to create TGSI_SEMANTIC_PNTC or sprite_coord_mode has
+	 * to be per FP input instead of per VP output
+	 */
+	memset(pntc, 0, 8 * sizeof(uint32_t));
+
+	for (i = 0; i < fp->cfg.io_nr; i++) {
+		uint8_t sn, si;
+		uint8_t j = fp->cfg.io[i].id_vp, k = fp->cfg.io[i].id_fp;
+		unsigned n = popcnt4(fp->cfg.io[i].mask);
+
+		if (fp->info.input_semantic_name[k] != TGSI_SEMANTIC_GENERIC) {
+			m += n;
+			continue;
+		}
+
+		sn = vp->info.input_semantic_name[j];
+		si = vp->info.input_semantic_index[j];
+
+		if (j < fp->cfg.io_nr && sn == TGSI_SEMANTIC_GENERIC) {
+			ubyte mode =
+				nv50->rasterizer->pipe.sprite_coord_mode[si];
+
+			if (mode == PIPE_SPRITE_COORD_NONE) {
+				m += n;
+				continue;
+			}
+		}
+
+		/* this is either PointCoord or replaced by sprite coords */
+		for (c = 0; c < 4; c++) {
+			if (!(fp->cfg.io[i].mask & (1 << c)))
+				continue;
+			pntc[m / 8] |= (c + 1) << ((m % 8) * 4);
+			++m;
+		}
+	}
+}
+
+static int
+nv50_sreg4_map(uint32_t *p_map, int mid, uint32_t lin[4],
+	       struct nv50_sreg4 *fpi, struct nv50_sreg4 *vpo)
+{
+	int c;
+	uint8_t mv = vpo->mask, mf = fpi->mask, oid = vpo->hw;
+	uint8_t *map = (uint8_t *)p_map;
+
+	for (c = 0; c < 4; ++c) {
+		if (mf & 1) {
+			if (fpi->linear == TRUE)
+				lin[mid / 32] |= 1 << (mid % 32);
+			map[mid++] = (mv & 1) ? oid : ((c == 3) ? 0x41 : 0x40);
+		}
+
+		oid += mv & 1;
+		mf >>= 1;
+		mv >>= 1;
+	}
+
+	return mid;
+}
+
+void
+nv50_linkage_validate(struct nv50_context *nv50)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_program *vp = nv50->vertprog;
+	struct nv50_program *fp = nv50->fragprog;
+	struct nouveau_stateobj *so;
+	struct nv50_sreg4 dummy, *vpo;
+	int i, n, c, m = 0;
+	uint32_t map[16], lin[4], reg[5], pcrd[8];
+
+	memset(map, 0, sizeof(map));
+	memset(lin, 0, sizeof(lin));
+
+	reg[1] = 0x00000004; /* low and high clip distance map ids */
+	reg[2] = 0x00000000; /* layer index map id (disabled, GP only) */
+	reg[3] = 0x00000000; /* point size map id & enable */
+	reg[0] = fp->cfg.regs[0]; /* colour semantic reg */
+	reg[4] = fp->cfg.regs[1]; /* interpolant info */
+
+	dummy.linear = FALSE;
+	dummy.mask = 0xf; /* map all components of HPOS */
+	m = nv50_sreg4_map(map, m, lin, &dummy, &vp->cfg.io[0]);
+
+	dummy.mask = 0x0;
+
+	if (vp->cfg.clpd < 0x40) {
+		for (c = 0; c < vp->cfg.clpd_nr; ++c)
+			map[m++] = vp->cfg.clpd + c;
+		reg[1] = (m << 8);
+	}
+
+	reg[0] |= m << 8; /* adjust BFC0 id */
+
+	/* if light_twoside is active, it seems FFC0_ID == BFC0_ID is bad */
+	if (nv50->rasterizer->pipe.light_twoside) {
+		vpo = &vp->cfg.two_side[0];
+
+		m = nv50_sreg4_map(map, m, lin, &fp->cfg.two_side[0], &vpo[0]);
+		m = nv50_sreg4_map(map, m, lin, &fp->cfg.two_side[1], &vpo[1]);
+	}
+
+	reg[0] += m - 4; /* adjust FFC0 id */
+	reg[4] |= m << 8; /* set mid where 'normal' FP inputs start */
+
+	i = 0;
+	if (fp->info.input_semantic_name[0] == TGSI_SEMANTIC_POSITION)
+		i = 1;
+	for (; i < fp->cfg.io_nr; i++) {
+		ubyte sn = fp->info.input_semantic_name[fp->cfg.io[i].id_fp];
+		ubyte si = fp->info.input_semantic_index[fp->cfg.io[i].id_fp];
+
+		n = fp->cfg.io[i].id_vp;
+		if (n >= vp->cfg.io_nr ||
+		    vp->info.output_semantic_name[n] != sn ||
+		    vp->info.output_semantic_index[n] != si)
+			vpo = &dummy;
+		else
+			vpo = &vp->cfg.io[n];
+
+		m = nv50_sreg4_map(map, m, lin, &fp->cfg.io[i], vpo);
+	}
+
+	if (nv50->rasterizer->pipe.point_size_per_vertex) {
+		map[m / 4] |= vp->cfg.psiz << ((m % 4) * 8);
+		reg[3] = (m++ << 4) | 1;
+	}
+
+	/* now fill the stateobj */
+	so = so_new(64, 0);
+
+	n = (m + 3) / 4;
+	so_method(so, tesla, NV50TCL_VP_RESULT_MAP_SIZE, 1);
+	so_data  (so, m);
+	so_method(so, tesla, NV50TCL_VP_RESULT_MAP(0), n);
+	so_datap (so, map, n);
+
+	so_method(so, tesla, NV50TCL_MAP_SEMANTIC_0, 4);
+	so_datap (so, reg, 4);
+
+	so_method(so, tesla, NV50TCL_FP_INTERPOLANT_CTRL, 1);
+	so_data  (so, reg[4]);
+
+	so_method(so, tesla, 0x1540, 4);
+	so_datap (so, lin, 4);
+
+	if (nv50->rasterizer->pipe.point_sprite) {
+		nv50_pntc_replace(nv50, pcrd, (reg[4] >> 8) & 0xff);
+
+		so_method(so, tesla, NV50TCL_POINT_COORD_REPLACE_MAP(0), 8);
+		so_datap (so, pcrd, 8);
+	}
+
+        so_ref(so, &nv50->state.programs);
+        so_ref(NULL, &so);
+}
+
 void
 nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p)
 {
@@ -2476,7 +3308,6 @@ nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p)
 	nouveau_bo_ref(NULL, &p->bo);
 
 	nouveau_resource_free(&p->data[0]);
-	nouveau_resource_free(&p->data[1]);
 
 	p->translated = 0;
 }
diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h
index 096e0476aa..d78dee083f 100644
--- a/src/gallium/drivers/nv50/nv50_program.h
+++ b/src/gallium/drivers/nv50/nv50_program.h
@@ -15,6 +15,15 @@ struct nv50_program_exec {
 	} param;
 };
 
+struct nv50_sreg4 {
+	uint8_t hw;
+	uint8_t id_vp;
+	uint8_t id_fp;
+
+	uint8_t mask;
+	boolean linear;
+};
+
 struct nv50_program {
 	struct pipe_shader_state pipe;
 	struct tgsi_shader_info info;
@@ -24,8 +33,8 @@ struct nv50_program {
 	struct nv50_program_exec *exec_head;
 	struct nv50_program_exec *exec_tail;
 	unsigned exec_size;
-	struct nouveau_resource *data[2];
-	unsigned data_start[2];
+	struct nouveau_resource *data[1];
+	unsigned data_start[1];
 
 	struct nouveau_bo *bo;
 
@@ -36,14 +45,20 @@ struct nv50_program {
 	struct {
 		unsigned high_temp;
 		unsigned high_result;
-		struct {
-			unsigned attr[2];
-		} vp;
-		struct {
-			unsigned regs[4];
-			unsigned map[5];
-			unsigned high_map;
-		} fp;
+
+		uint32_t attr[2];
+		uint32_t regs[4];
+
+		/* for VPs, io_nr doesn't count 'private' results (PSIZ etc.) */
+		unsigned io_nr;
+		struct nv50_sreg4 io[PIPE_MAX_SHADER_OUTPUTS];
+
+		/* FP colour inputs, VP/GP back colour outputs */
+		struct nv50_sreg4 two_side[2];
+
+		/* VP only */
+		uint8_t clpd, clpd_nr;
+		uint8_t psiz;
 	} cfg;
 };
 
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index c7f80a2203..63dce0f4c2 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -35,6 +35,7 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen,
 {
 	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
 		switch (format) {
+		case PIPE_FORMAT_X8R8G8B8_UNORM:
 		case PIPE_FORMAT_A8R8G8B8_UNORM:
 		case PIPE_FORMAT_R5G6B5_UNORM:
 			return TRUE;
@@ -55,6 +56,7 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen,
 	} else {
 		switch (format) {
 		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_X8R8G8B8_UNORM:
 		case PIPE_FORMAT_A1R5G5B5_UNORM:
 		case PIPE_FORMAT_A4R4G4B4_UNORM:
 		case PIPE_FORMAT_R5G6B5_UNORM:
@@ -87,12 +89,10 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 1;
 	case PIPE_CAP_GLSL:
 		return 0;
-	case PIPE_CAP_S3TC:
-		return 1;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 1;
 	case PIPE_CAP_POINT_SPRITE:
-		return 0;
+		return 1;
 	case PIPE_CAP_MAX_RENDER_TARGETS:
 		return 8;
 	case PIPE_CAP_OCCLUSION_QUERY:
@@ -218,7 +218,16 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		tesla_class = NV54TCL;
 		break;
 	case 0xa0:
-		tesla_class = NVA0TCL;
+		switch (chipset) {
+		case 0xa0:
+		case 0xaa:
+		case 0xac:
+			tesla_class = NVA0TCL;
+			break;
+		default:
+			tesla_class = 0x8597;
+			break;
+		}
 		break;
 	default:
 		NOUVEAU_ERR("Not a known NV50 chipset: NV%02x\n", chipset);
@@ -226,12 +235,6 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		return NULL;
 	}
 
-	if (tesla_class == 0) {
-		NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset);
-		nv50_screen_destroy(pscreen);
-		return NULL;
-	}
-
 	ret = nouveau_grobj_alloc(chan, 0xbeef5097, tesla_class,
 		&screen->tesla);
 	if (ret) {
@@ -301,7 +304,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	so_data  (so, 8);
 
 	/* constant buffers for immediates and VP/FP parameters */
-	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 128*4*4,
+	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, (32 * 4) * 4,
 			     &screen->constbuf_misc[0]);
 	if (ret) {
 		nv50_screen_destroy(pscreen);
@@ -309,7 +312,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	}
 
 	for (i = 0; i < 2; i++) {
-		ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 128*4*4,
+		ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, (128 * 4) * 4,
 				     &screen->constbuf_parm[i]);
 		if (ret) {
 			nv50_screen_destroy(pscreen);
@@ -318,8 +321,8 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	}
 
 	if (nouveau_resource_init(&screen->immd_heap[0], 0, 128) ||
-		nouveau_resource_init(&screen->parm_heap[0], 0, 128) ||
-		nouveau_resource_init(&screen->parm_heap[1], 0, 128))
+	    nouveau_resource_init(&screen->parm_heap[0], 0, 512) ||
+	    nouveau_resource_init(&screen->parm_heap[1], 0, 512))
 	{
 		NOUVEAU_ERR("Error initialising constant buffers.\n");
 		nv50_screen_destroy(pscreen);
@@ -340,7 +343,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
 	so_reloc (so, screen->constbuf_misc[0], 0, NOUVEAU_BO_VRAM |
 		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_PMISC << 16) | 0x00000800);
+	so_data  (so, (NV50_CB_PMISC << 16) | 0x00000200);
 	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
 	so_data  (so, 0x00000001 | (NV50_CB_PMISC << 12));
 	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
@@ -364,48 +367,31 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
 	so_data  (so, 0x00000131 | (NV50_CB_PFP << 12));
 
-	/* Texture sampler/image unit setup - we abuse the constant buffer
-	 * upload mechanism for the moment to upload data to the tex config
-	 * blocks.  At some point we *may* want to go the NVIDIA way of doing
-	 * things?
-	 */
-	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 32*8*4, &screen->tic);
+	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 64*8*4, &screen->tic);
 	if (ret) {
 		nv50_screen_destroy(pscreen);
 		return NULL;
 	}
 
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_TIC << 16) | 0x0800);
 	so_method(so, screen->tesla, NV50TCL_TIC_ADDRESS_HIGH, 3);
 	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
 		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
 	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
 		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, 0x00000800);
+	so_data  (so, 0x000007ff);
 
-	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 32*8*4, &screen->tsc);
+	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 64*8*4, &screen->tsc);
 	if (ret) {
 		nv50_screen_destroy(pscreen);
 		return NULL;
 	}
 
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_TSC << 16) | 0x0800);
 	so_method(so, screen->tesla, NV50TCL_TSC_ADDRESS_HIGH, 3);
 	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
 		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
 	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
 		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, 0x00000800);
+	so_data  (so, 0x00000000);
 
 
 	/* Vertex array limits - max them out */
diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c
index 4283808ed9..ffaa5e29d1 100644
--- a/src/gallium/drivers/nv50/nv50_state.c
+++ b/src/gallium/drivers/nv50/nv50_state.c
@@ -146,6 +146,7 @@ nv50_sampler_state_create(struct pipe_context *pipe,
 		  (wrap_mode(cso->wrap_r) << 6));
 
 	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_ANISO:
 	case PIPE_TEX_FILTER_LINEAR:
 		tsc[1] |= NV50TSC_1_1_MAGF_LINEAR;
 		break;
@@ -156,6 +157,7 @@ nv50_sampler_state_create(struct pipe_context *pipe,
 	}
 
 	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_ANISO:
 	case PIPE_TEX_FILTER_LINEAR:
 		tsc[1] |= NV50TSC_1_1_MINF_LINEAR;
 		break;
@@ -183,21 +185,15 @@ nv50_sampler_state_create(struct pipe_context *pipe,
 	else
 	if (cso->max_anisotropy >= 12.0)
 		tsc[0] |= (6 << 20);
-	else
-	if (cso->max_anisotropy >= 10.0)
-		tsc[0] |= (5 << 20);
-	else
-	if (cso->max_anisotropy >= 8.0)
-		tsc[0] |= (4 << 20);
-	else
-	if (cso->max_anisotropy >= 6.0)
-		tsc[0] |= (3 << 20);
-	else
-	if (cso->max_anisotropy >= 4.0)
-		tsc[0] |= (2 << 20);
-	else
-	if (cso->max_anisotropy >= 2.0)
-		tsc[0] |= (1 << 20);
+	else {
+		tsc[0] |= (int)(cso->max_anisotropy * 0.5f) << 20;
+
+		if (cso->max_anisotropy >= 4.0)
+			tsc[1] |= NV50TSC_1_1_UNKN_ANISO_35;
+		else
+		if (cso->max_anisotropy >= 2.0)
+			tsc[1] |= NV50TSC_1_1_UNKN_ANISO_15;
+	}
 
 	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
 		tsc[0] |= (1 << 8);
@@ -276,6 +272,9 @@ nv50_rasterizer_state_create(struct pipe_context *pipe,
 	so_method(so, tesla, 0x1684, 1);
 	so_data  (so, cso->flatshade_first ? 0 : 1);
 
+	so_method(so, tesla, NV50TCL_VERTEX_TWO_SIDE_ENABLE, 1);
+	so_data  (so, cso->light_twoside);
+
 	so_method(so, tesla, NV50TCL_LINE_WIDTH, 1);
 	so_data  (so, fui(cso->line_width));
 	so_method(so, tesla, NV50TCL_LINE_SMOOTH_ENABLE, 1);
@@ -294,6 +293,9 @@ nv50_rasterizer_state_create(struct pipe_context *pipe,
 	so_method(so, tesla, NV50TCL_POINT_SIZE, 1);
 	so_data  (so, fui(cso->point_size));
 
+	so_method(so, tesla, NV50TCL_POINT_SPRITE_ENABLE, 1);
+	so_data  (so, cso->point_sprite);
+
 	so_method(so, tesla, NV50TCL_POLYGON_MODE_FRONT, 3);
 	if (cso->front_winding == PIPE_WINDING_CCW) {
 		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
index 344c2cf6dd..012911f41b 100644
--- a/src/gallium/drivers/nv50/nv50_state_validate.c
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -57,6 +57,9 @@ nv50_state_validate_fb(struct nv50_context *nv50)
 		case PIPE_FORMAT_A8R8G8B8_UNORM:
 			so_data(so, NV50TCL_RT_FORMAT_A8R8G8B8_UNORM);
 			break;
+		case PIPE_FORMAT_X8R8G8B8_UNORM:
+			so_data(so, NV50TCL_RT_FORMAT_X8R8G8B8_UNORM);
+			break;
 		case PIPE_FORMAT_R5G6B5_UNORM:
 			so_data(so, NV50TCL_RT_FORMAT_R5G6B5_UNORM);
 			break;
@@ -66,7 +69,8 @@ nv50_state_validate_fb(struct nv50_context *nv50)
 			so_data(so, NV50TCL_RT_FORMAT_X8R8G8B8_UNORM);
 			break;
 		}
-		so_data(so, bo->tile_mode << 4);
+		so_data(so, nv50_miptree(pt)->
+				level[fb->cbufs[i]->level].tile_mode << 4);
 		so_data(so, 0x00000000);
 
 		so_method(so, tesla, 0x1224, 1);
@@ -110,7 +114,8 @@ nv50_state_validate_fb(struct nv50_context *nv50)
 			so_data(so, NV50TCL_ZETA_FORMAT_S8Z24_UNORM);
 			break;
 		}
-		so_data(so, bo->tile_mode << 4);
+		so_data(so, nv50_miptree(pt)->
+				level[fb->zsbuf->level].tile_mode << 4);
 		so_data(so, 0x00000000);
 
 		so_method(so, tesla, 0x1538, 1);
@@ -187,6 +192,8 @@ nv50_state_emit(struct nv50_context *nv50)
 		so_emit(chan, nv50->state.vertprog);
 	if (nv50->state.dirty & NV50_NEW_FRAGPROG)
 		so_emit(chan, nv50->state.fragprog);
+	if (nv50->state.dirty & (NV50_NEW_FRAGPROG | NV50_NEW_VERTPROG))
+		so_emit(chan, nv50->state.programs);
 	if (nv50->state.dirty & NV50_NEW_RASTERIZER)
 		so_emit(chan, nv50->state.rast);
 	if (nv50->state.dirty & NV50_NEW_BLEND_COLOUR)
@@ -208,6 +215,15 @@ nv50_state_emit(struct nv50_context *nv50)
 			so_emit(chan, nv50->state.vtxattr);
 	}
 	nv50->state.dirty = 0;
+}
+
+void
+nv50_state_flush_notify(struct nouveau_channel *chan)
+{
+	struct nv50_context *nv50 = chan->user_private;
+
+	if (nv50->state.tic_upload && !(nv50->dirty & NV50_NEW_TEXTURE))
+		so_emit(chan, nv50->state.tic_upload);
 
 	so_emit_reloc_markers(chan, nv50->state.fb);
 	so_emit_reloc_markers(chan, nv50->state.vertprog);
@@ -220,6 +236,7 @@ boolean
 nv50_state_validate(struct nv50_context *nv50)
 {
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_grobj *eng2d = nv50->screen->eng2d;
 	struct nouveau_stateobj *so;
 	unsigned i;
 
@@ -238,6 +255,9 @@ nv50_state_validate(struct nv50_context *nv50)
 	if (nv50->dirty & (NV50_NEW_FRAGPROG | NV50_NEW_FRAGPROG_CB))
 		nv50_fragprog_validate(nv50);
 
+	if (nv50->dirty & (NV50_NEW_FRAGPROG | NV50_NEW_VERTPROG))
+		nv50_linkage_validate(nv50);
+
 	if (nv50->dirty & NV50_NEW_RASTERIZER)
 		so_ref(nv50->rasterizer->so, &nv50->state.rast);
 
@@ -299,7 +319,7 @@ scissor_uptodate:
 			goto viewport_uptodate;
 		nv50->state.viewport_bypass = bypass;
 
-		so = so_new(12, 0);
+		so = so_new(14, 0);
 		if (!bypass) {
 			so_method(so, tesla, NV50TCL_VIEWPORT_TRANSLATE(0), 3);
 			so_data  (so, fui(nv50->viewport.translate[0]));
@@ -312,12 +332,21 @@ scissor_uptodate:
 
 			so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1);
 			so_data  (so, 1);
+			/* 0x0000 = remove whole primitive only (xyz)
+			 * 0x1018 = remove whole primitive only (xy), clamp z
+			 * 0x1080 = clip primitive (xyz)
+			 * 0x1098 = clip primitive (xy), clamp z
+			 */
+			so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1);
+			so_data  (so, 0x1080);
 			/* no idea what 0f90 does */
 			so_method(so, tesla, 0x0f90, 1);
 			so_data  (so, 0);
 		} else {
 			so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1);
 			so_data  (so, 0);
+			so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1);
+			so_data  (so, 0x0000);
 			so_method(so, tesla, 0x0f90, 1);
 			so_data  (so, 1);
 		}
@@ -329,15 +358,25 @@ scissor_uptodate:
 viewport_uptodate:
 
 	if (nv50->dirty & NV50_NEW_SAMPLER) {
-		int i;
-
-		so = so_new(nv50->sampler_nr * 8 + 3, 0);
-		so_method(so, tesla, NV50TCL_CB_ADDR, 1);
-		so_data  (so, NV50_CB_TSC);
-		so_method(so, tesla, NV50TCL_CB_DATA(0) | 0x40000000,
-			nv50->sampler_nr * 8);
-		for (i = 0; i < nv50->sampler_nr; i++)
+		unsigned i;
+
+		so = so_new(nv50->sampler_nr * 9 + 23 + 4, 2);
+
+		nv50_so_init_sifc(nv50, so, nv50->screen->tsc, NOUVEAU_BO_VRAM,
+				  nv50->sampler_nr * 8 * 4);
+
+		for (i = 0; i < nv50->sampler_nr; i++) {
+			if (!nv50->sampler[i])
+				continue;
+			so_method(so, eng2d, NV50_2D_SIFC_DATA | (2 << 29), 8);
 			so_datap (so, nv50->sampler[i]->tsc, 8);
+		}
+
+		so_method(so, tesla, 0x1440, 1); /* sync SIFC */
+		so_data  (so, 0);
+		so_method(so, tesla, 0x1334, 1); /* flush TSC */
+		so_data  (so, 0);
+
 		so_ref(so, &nv50->state.tsc_upload);
 		so_ref(NULL, &so);
 	}
@@ -355,3 +394,33 @@ viewport_uptodate:
 	return TRUE;
 }
 
+void nv50_so_init_sifc(struct nv50_context *nv50,
+		       struct nouveau_stateobj *so,
+		       struct nouveau_bo *bo, unsigned reloc, unsigned size)
+{
+	struct nouveau_grobj *eng2d = nv50->screen->eng2d;
+
+	so_method(so, eng2d, NV50_2D_DST_FORMAT, 2);
+	so_data  (so, NV50_2D_DST_FORMAT_R8_UNORM);
+	so_data  (so, 1);
+	so_method(so, eng2d, NV50_2D_DST_PITCH, 5);
+	so_data  (so, 262144);
+	so_data  (so, 65536);
+	so_data  (so, 1);
+	so_reloc (so, bo, 0, reloc | NOUVEAU_BO_WR | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, bo, 0, reloc | NOUVEAU_BO_WR | NOUVEAU_BO_LOW, 0, 0);
+	so_method(so, eng2d, NV50_2D_SIFC_UNK0800, 2);
+	so_data  (so, 0);
+	so_data  (so, NV50_2D_SIFC_FORMAT_R8_UNORM);
+	so_method(so, eng2d, NV50_2D_SIFC_WIDTH, 10);
+	so_data  (so, size);
+	so_data  (so, 1);
+	so_data  (so, 0);
+	so_data  (so, 1);
+	so_data  (so, 0);
+	so_data  (so, 1);
+	so_data  (so, 0);
+	so_data  (so, 0);
+	so_data  (so, 0);
+	so_data  (so, 0);
+}
diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c
index b266324f58..6bf6f773b0 100644
--- a/src/gallium/drivers/nv50/nv50_surface.c
+++ b/src/gallium/drivers/nv50/nv50_surface.c
@@ -60,13 +60,13 @@ nv50_surface_set(struct nv50_screen *screen, struct pipe_surface *ps, int dst)
  	format = nv50_format(ps->format);
  	if (format < 0)
  		return 1;
-  
+
  	if (!bo->tile_flags) {
  		BEGIN_RING(chan, eng2d, mthd, 2);
  		OUT_RING  (chan, format);
  		OUT_RING  (chan, 1);
  		BEGIN_RING(chan, eng2d, mthd + 0x14, 5);
- 		OUT_RING  (chan, mt->level[0].pitch);
+		OUT_RING  (chan, mt->level[ps->level].pitch);
  		OUT_RING  (chan, ps->width);
  		OUT_RING  (chan, ps->height);
  		OUT_RELOCh(chan, bo, ps->offset, flags);
@@ -75,7 +75,7 @@ nv50_surface_set(struct nv50_screen *screen, struct pipe_surface *ps, int dst)
  		BEGIN_RING(chan, eng2d, mthd, 5);
  		OUT_RING  (chan, format);
  		OUT_RING  (chan, 0);
- 		OUT_RING  (chan, bo->tile_mode << 4);
+		OUT_RING  (chan, mt->level[ps->level].tile_mode << 4);
  		OUT_RING  (chan, 1);
  		OUT_RING  (chan, 0);
  		BEGIN_RING(chan, eng2d, mthd + 0x18, 4);
diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c
index 033cb50c11..e12a6ad648 100644
--- a/src/gallium/drivers/nv50/nv50_tex.c
+++ b/src/gallium/drivers/nv50/nv50_tex.c
@@ -25,99 +25,60 @@
 
 #include "nouveau/nouveau_stateobj.h"
 
+#define _(pf, tt, r, g, b, a, tf)                       	\
+{                                                       	\
+	PIPE_FORMAT_##pf,					\
+	NV50TIC_0_0_MAPR_##r | NV50TIC_0_0_TYPER_##tt |		\
+	NV50TIC_0_0_MAPG_##g | NV50TIC_0_0_TYPEG_##tt |		\
+	NV50TIC_0_0_MAPB_##b | NV50TIC_0_0_TYPEB_##tt |		\
+	NV50TIC_0_0_MAPA_##a | NV50TIC_0_0_TYPEA_##tt |		\
+	NV50TIC_0_0_FMT_##tf					\
+}
+
+struct nv50_texture_format {
+	enum pipe_format pf;
+	uint32_t hw;
+};
+
+#define NV50_TEX_FORMAT_LIST_SIZE \
+	(sizeof(nv50_tex_format_list) / sizeof(struct nv50_texture_format))
+
+static const struct nv50_texture_format nv50_tex_format_list[] =
+{
+	_(A8R8G8B8_UNORM, UNORM, C2, C1, C0, C3,  8_8_8_8),
+	_(X8R8G8B8_UNORM, UNORM, C2, C1, C0, ONE, 8_8_8_8),
+	_(A1R5G5B5_UNORM, UNORM, C2, C1, C0, C3,  1_5_5_5),
+	_(A4R4G4B4_UNORM, UNORM, C2, C1, C0, C3,  4_4_4_4),
+
+	_(R5G6B5_UNORM, UNORM, C2, C1, C0, ONE, 5_6_5),
+
+	_(L8_UNORM, UNORM, C0, C0, C0, ONE, 8),
+	_(A8_UNORM, UNORM, ZERO, ZERO, ZERO, C0, 8),
+	_(I8_UNORM, UNORM, C0, C0, C0, C0, 8),
+
+	_(A8L8_UNORM, UNORM, C0, C0, C0, C1, 8_8),
+
+	_(DXT1_RGB, UNORM, C0, C1, C2, ONE, DXT1),
+	_(DXT1_RGBA, UNORM, C0, C1, C2, C3, DXT1),
+	_(DXT3_RGBA, UNORM, C0, C1, C2, C3, DXT3),
+	_(DXT5_RGBA, UNORM, C0, C1, C2, C3, DXT5)
+};
+
+#undef _
+
 static int
 nv50_tex_construct(struct nv50_context *nv50, struct nouveau_stateobj *so,
 		   struct nv50_miptree *mt, int unit)
 {
-	switch (mt->base.base.format) {
-	case PIPE_FORMAT_A8R8G8B8_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C2 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_8_8_8_8);
-		break;
-	case PIPE_FORMAT_A1R5G5B5_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C2 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_1_5_5_5);
-		break;
-	case PIPE_FORMAT_A4R4G4B4_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C2 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_4_4_4_4);
-		break;
-	case PIPE_FORMAT_R5G6B5_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C2 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_5_6_5);
-		break;
-	case PIPE_FORMAT_L8_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_8);
-		break;
-	case PIPE_FORMAT_A8_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_C0 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_ZERO | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_ZERO | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_ZERO | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_8);
-		break;
-	case PIPE_FORMAT_I8_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_C0 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_8);
-		break;
-	case PIPE_FORMAT_A8L8_UNORM:
-		so_data(so, NV50TIC_0_0_MAPA_C1 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_8_8);
-		break;
-	case PIPE_FORMAT_DXT1_RGB:
-		so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_DXT1);
-		break;
-	case PIPE_FORMAT_DXT1_RGBA:
-		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_DXT1);
-		break;
-	case PIPE_FORMAT_DXT3_RGBA:
-		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_DXT3);
-		break;
-	case PIPE_FORMAT_DXT5_RGBA:
-		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
-			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
-			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
-			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
-			    NV50TIC_0_0_FMT_DXT5);
-		break;
-	default:
-		return 1;
-	}
+	unsigned i;
 
+	for (i = 0; i < NV50_TEX_FORMAT_LIST_SIZE; i++)
+		if (nv50_tex_format_list[i].pf == mt->base.base.format)
+			break;
+	if (i == NV50_TEX_FORMAT_LIST_SIZE)
+                return 1;
+
+	so_data (so, nv50_tex_format_list[i].hw);
 	so_reloc(so, mt->base.bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW |
 		     NOUVEAU_BO_RD, 0, 0);
 	if (nv50->sampler[unit]->normalized)
@@ -137,20 +98,24 @@ nv50_tex_construct(struct nv50_context *nv50, struct nouveau_stateobj *so,
 void
 nv50_tex_validate(struct nv50_context *nv50)
 {
+	struct nouveau_grobj *eng2d = nv50->screen->eng2d;
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct nouveau_stateobj *so;
-	int unit, push;
+	unsigned i, unit, push;
+
+	push = MAX2(nv50->miptree_nr, nv50->state.miptree_nr) * 2 + 23 + 6;
+	so = so_new(nv50->miptree_nr * 9 + push, nv50->miptree_nr + 2);
 
-	push  = nv50->miptree_nr * 9 + 2;
-	push += MAX2(nv50->miptree_nr, nv50->state.miptree_nr) * 2;
+	nv50_so_init_sifc(nv50, so, nv50->screen->tic, NOUVEAU_BO_VRAM,
+			  nv50->miptree_nr * 8 * 4);
 
-	so = so_new(push, nv50->miptree_nr * 2);
-	so_method(so, tesla, NV50TCL_CB_ADDR, 1);
-	so_data  (so, NV50_CB_TIC);
-	for (unit = 0; unit < nv50->miptree_nr; unit++) {
+	for (i = 0, unit = 0; unit < nv50->miptree_nr; ++unit) {
 		struct nv50_miptree *mt = nv50->miptree[unit];
 
-		so_method(so, tesla, NV50TCL_CB_DATA(0) | 0x40000000, 8);
+		if (!mt)
+			continue;
+
+		so_method(so, eng2d, NV50_2D_SIFC_DATA | (2 << 29), 8);
 		if (nv50_tex_construct(nv50, so, mt, unit)) {
 			NOUVEAU_ERR("failed tex validate\n");
 			so_ref(NULL, &so);
@@ -158,17 +123,25 @@ nv50_tex_validate(struct nv50_context *nv50)
 		}
 
 		so_method(so, tesla, NV50TCL_SET_SAMPLER_TEX, 1);
-		so_data  (so, (unit << NV50TCL_SET_SAMPLER_TEX_TIC_SHIFT) |
-			(unit << NV50TCL_SET_SAMPLER_TEX_SAMPLER_SHIFT) |
-			NV50TCL_SET_SAMPLER_TEX_VALID);
+		so_data  (so, (i++ << NV50TCL_SET_SAMPLER_TEX_TIC_SHIFT) |
+			  (unit << NV50TCL_SET_SAMPLER_TEX_SAMPLER_SHIFT) |
+			  NV50TCL_SET_SAMPLER_TEX_VALID);
 	}
 
 	for (; unit < nv50->state.miptree_nr; unit++) {
 		so_method(so, tesla, NV50TCL_SET_SAMPLER_TEX, 1);
 		so_data  (so,
-			(unit << NV50TCL_SET_SAMPLER_TEX_SAMPLER_SHIFT) | 0);
+			  (unit << NV50TCL_SET_SAMPLER_TEX_SAMPLER_SHIFT) | 0);
 	}
 
+	/* not sure if the following really do what I think: */
+	so_method(so, tesla, 0x1440, 1); /* sync SIFC */
+	so_data  (so, 0);
+	so_method(so, tesla, 0x1330, 1); /* flush TIC */
+	so_data  (so, 0);
+	so_method(so, tesla, 0x1338, 1); /* flush texture caches */
+	so_data  (so, 0x20);
+
 	so_ref(so, &nv50->state.tic_upload);
 	so_ref(NULL, &so);
 	nv50->state.miptree_nr = nv50->miptree_nr;
diff --git a/src/gallium/drivers/nv50/nv50_texture.h b/src/gallium/drivers/nv50/nv50_texture.h
index 207fb039f7..13f74c11c6 100644
--- a/src/gallium/drivers/nv50/nv50_texture.h
+++ b/src/gallium/drivers/nv50/nv50_texture.h
@@ -133,6 +133,8 @@
 #define NV50TSC_1_1_MIPF_NEAREST                                 0x00000080
 #define NV50TSC_1_1_MIPF_LINEAR                                  0x000000c0
 #define NV50TSC_1_1_LOD_BIAS_MASK                                0x01fff000
+#define NV50TSC_1_1_UNKN_ANISO_15                                0x10000000
+#define NV50TSC_1_1_UNKN_ANISO_35                                0x18000000
 
 #define NV50TSC_1_2_MIN_LOD_MASK                                 0x00000f00
 #define NV50TSC_1_2_MAX_LOD_MASK                                 0x00f00000
diff --git a/src/gallium/drivers/nv50/nv50_transfer.c b/src/gallium/drivers/nv50/nv50_transfer.c
index e9c3562194..9c289026bb 100644
--- a/src/gallium/drivers/nv50/nv50_transfer.c
+++ b/src/gallium/drivers/nv50/nv50_transfer.c
@@ -89,14 +89,14 @@ nv50_transfer_rect_m2mf(struct pipe_screen *pscreen,
 		if (src_bo->tile_flags) {
 			BEGIN_RING(chan, m2mf,
 				NV50_MEMORY_TO_MEMORY_FORMAT_TILING_POSITION_IN, 1);
-			OUT_RING  (chan, (sy << 16) | sx);
+			OUT_RING  (chan, (sy << 16) | (sx * cpp));
 		} else {
 			src_offset += (line_count * src_pitch);
 		}
 		if (dst_bo->tile_flags) {
 			BEGIN_RING(chan, m2mf,
 				NV50_MEMORY_TO_MEMORY_FORMAT_TILING_POSITION_OUT, 1);
-			OUT_RING  (chan, (dy << 16) | dx);
+			OUT_RING  (chan, (dy << 16) | (dx * cpp));
 		} else {
 			dst_offset += (line_count * dst_pitch);
 		}
@@ -161,7 +161,7 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
-	if (usage != PIPE_TRANSFER_WRITE) {
+	if (usage & PIPE_TRANSFER_READ) {
 		nv50_transfer_rect_m2mf(pscreen, mt->base.bo, tx->level_offset,
 					tx->level_pitch, tx->level_tiling,
 					x, y,
@@ -183,7 +183,7 @@ nv50_transfer_del(struct pipe_transfer *ptx)
 	struct nv50_transfer *tx = (struct nv50_transfer *)ptx;
 	struct nv50_miptree *mt = nv50_miptree(ptx->texture);
 
-	if (ptx->usage != PIPE_TRANSFER_READ) {
+	if (ptx->usage & PIPE_TRANSFER_WRITE) {
 		struct pipe_screen *pscreen = ptx->texture->screen;
 		nv50_transfer_rect_m2mf(pscreen, tx->bo, 0, tx->base.stride,
 					tx->bo->tile_mode, 0, 0,
diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c
index eeed148c7b..db54380241 100644
--- a/src/gallium/drivers/nv50/nv50_vbo.c
+++ b/src/gallium/drivers/nv50/nv50_vbo.c
@@ -26,6 +26,18 @@
 
 #include "nv50_context.h"
 
+static boolean
+nv50_push_elements_u08(struct nv50_context *, uint8_t *, unsigned);
+
+static boolean
+nv50_push_elements_u16(struct nv50_context *, uint16_t *, unsigned);
+
+static boolean
+nv50_push_elements_u32(struct nv50_context *, uint32_t *, unsigned);
+
+static boolean
+nv50_push_arrays(struct nv50_context *, unsigned, unsigned);
+
 static INLINE unsigned
 nv50_prim(unsigned mode)
 {
@@ -132,6 +144,7 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start,
 	struct nv50_context *nv50 = nv50_context(pipe);
 	struct nouveau_channel *chan = nv50->screen->tesla->channel;
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	boolean ret;
 
 	nv50_state_validate(nv50);
 
@@ -139,24 +152,25 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start,
 	OUT_RING  (chan, 0);
 	BEGIN_RING(chan, tesla, 0x142c, 1);
 	OUT_RING  (chan, 0);
-	BEGIN_RING(chan, tesla, 0x1440, 1);
-	OUT_RING  (chan, 0);
-	BEGIN_RING(chan, tesla, 0x1334, 1);
-	OUT_RING  (chan, 0);
 
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
 	OUT_RING  (chan, nv50_prim(mode));
-	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
-	OUT_RING  (chan, start);
-	OUT_RING  (chan, count);
+
+	if (nv50->vbo_fifo)
+		ret = nv50_push_arrays(nv50, start, count);
+	else {
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
+		OUT_RING  (chan, start);
+		OUT_RING  (chan, count);
+		ret = TRUE;
+	}
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 	OUT_RING  (chan, 0);
 
-	pipe->flush(pipe, 0, NULL);
-	return TRUE;
+	return ret;
 }
 
-static INLINE void
+static INLINE boolean
 nv50_draw_elements_inline_u08(struct nv50_context *nv50, uint8_t *map,
 			      unsigned start, unsigned count)
 {
@@ -165,6 +179,9 @@ nv50_draw_elements_inline_u08(struct nv50_context *nv50, uint8_t *map,
 
 	map += start;
 
+	if (nv50->vbo_fifo)
+		return nv50_push_elements_u08(nv50, map, count);
+
 	if (count & 1) {
 		BEGIN_RING(chan, tesla, 0x15e8, 1);
 		OUT_RING  (chan, map[0]);
@@ -183,9 +200,10 @@ nv50_draw_elements_inline_u08(struct nv50_context *nv50, uint8_t *map,
 		count -= nr;
 		map += nr;
 	}
+	return TRUE;
 }
 
-static INLINE void
+static INLINE boolean
 nv50_draw_elements_inline_u16(struct nv50_context *nv50, uint16_t *map,
 			      unsigned start, unsigned count)
 {
@@ -194,6 +212,9 @@ nv50_draw_elements_inline_u16(struct nv50_context *nv50, uint16_t *map,
 
 	map += start;
 
+	if (nv50->vbo_fifo)
+		return nv50_push_elements_u16(nv50, map, count);
+
 	if (count & 1) {
 		BEGIN_RING(chan, tesla, 0x15e8, 1);
 		OUT_RING  (chan, map[0]);
@@ -212,9 +233,10 @@ nv50_draw_elements_inline_u16(struct nv50_context *nv50, uint16_t *map,
 		count -= nr;
 		map += nr;
 	}
+	return TRUE;
 }
 
-static INLINE void
+static INLINE boolean
 nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint32_t *map,
 			      unsigned start, unsigned count)
 {
@@ -223,6 +245,9 @@ nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint32_t *map,
 
 	map += start;
 
+	if (nv50->vbo_fifo)
+		return nv50_push_elements_u32(nv50, map, count);
+
 	while (count) {
 		unsigned nr = count > 2047 ? 2047 : count;
 
@@ -232,6 +257,7 @@ nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint32_t *map,
 		count -= nr;
 		map += nr;
 	}
+	return TRUE;
 }
 
 boolean
@@ -244,6 +270,7 @@ nv50_draw_elements(struct pipe_context *pipe,
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct pipe_screen *pscreen = pipe->screen;
 	void *map;
+	boolean ret;
 	
 	map = pipe_buffer_map(pscreen, indexBuffer, PIPE_BUFFER_USAGE_CPU_READ);
 
@@ -258,23 +285,25 @@ nv50_draw_elements(struct pipe_context *pipe,
 	OUT_RING  (chan, nv50_prim(mode));
 	switch (indexSize) {
 	case 1:
-		nv50_draw_elements_inline_u08(nv50, map, start, count);
+		ret = nv50_draw_elements_inline_u08(nv50, map, start, count);
 		break;
 	case 2:
-		nv50_draw_elements_inline_u16(nv50, map, start, count);
+		ret = nv50_draw_elements_inline_u16(nv50, map, start, count);
 		break;
 	case 4:
-		nv50_draw_elements_inline_u32(nv50, map, start, count);
+		ret = nv50_draw_elements_inline_u32(nv50, map, start, count);
 		break;
 	default:
 		assert(0);
+		ret = FALSE;
+		break;
 	}
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 	OUT_RING  (chan, 0);
 
 	pipe_buffer_unmap(pscreen, indexBuffer);
-	pipe->flush(pipe, 0, NULL);
-	return TRUE;
+
+	return ret;
 }
 
 static INLINE boolean
@@ -341,17 +370,24 @@ nv50_vbo_validate(struct nv50_context *nv50)
 {
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct nouveau_stateobj *vtxbuf, *vtxfmt, *vtxattr;
-	unsigned i;
+	unsigned i, n_ve;
 
 	/* don't validate if Gallium took away our buffers */
 	if (nv50->vtxbuf_nr == 0)
 		return;
+	nv50->vbo_fifo = 0;
+
+	for (i = 0; i < nv50->vtxbuf_nr; ++i)
+		if (nv50->vtxbuf[i].stride &&
+		    !(nv50->vtxbuf[i].buffer->usage & PIPE_BUFFER_USAGE_VERTEX))
+			nv50->vbo_fifo = 0xffff;
+
+	n_ve = MAX2(nv50->vtxelt_nr, nv50->state.vtxelt_nr);
 
 	vtxattr = NULL;
-	vtxbuf = so_new(nv50->vtxelt_nr * 7, nv50->vtxelt_nr * 4);
-	vtxfmt = so_new(nv50->vtxelt_nr + 1, 0);
-	so_method(vtxfmt, tesla, NV50TCL_VERTEX_ARRAY_ATTRIB(0),
-		nv50->vtxelt_nr);
+	vtxbuf = so_new(n_ve * 7, nv50->vtxelt_nr * 4);
+	vtxfmt = so_new(n_ve + 1, 0);
+	so_method(vtxfmt, tesla, NV50TCL_VERTEX_ARRAY_ATTRIB(0), n_ve);
 
 	for (i = 0; i < nv50->vtxelt_nr; i++) {
 		struct pipe_vertex_element *ve = &nv50->vtxelt[i];
@@ -367,10 +403,19 @@ nv50_vbo_validate(struct nv50_context *nv50)
 			so_method(vtxbuf, tesla,
 				  NV50TCL_VERTEX_ARRAY_FORMAT(i), 1);
 			so_data  (vtxbuf, 0);
+
+			nv50->vbo_fifo &= ~(1 << i);
 			continue;
 		}
 		so_data(vtxfmt, hw | i);
 
+		if (nv50->vbo_fifo) {
+			so_method(vtxbuf, tesla,
+				  NV50TCL_VERTEX_ARRAY_FORMAT(i), 1);
+			so_data  (vtxbuf, 0);
+			continue;
+		}
+
 		so_method(vtxbuf, tesla, NV50TCL_VERTEX_ARRAY_FORMAT(i), 3);
 		so_data  (vtxbuf, 0x20000000 | vb->stride);
 		so_reloc (vtxbuf, bo, vb->buffer_offset +
@@ -389,6 +434,13 @@ nv50_vbo_validate(struct nv50_context *nv50)
 			  NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD |
 			  NOUVEAU_BO_LOW, 0, 0);
 	}
+	for (; i < n_ve; ++i) {
+		so_data  (vtxfmt, 0x7e080010);
+
+		so_method(vtxbuf, tesla, NV50TCL_VERTEX_ARRAY_FORMAT(i), 1);
+		so_data  (vtxbuf, 0);
+	}
+	nv50->state.vtxelt_nr = nv50->vtxelt_nr;
 
 	so_ref (vtxfmt, &nv50->state.vtxfmt);
 	so_ref (vtxbuf, &nv50->state.vtxbuf);
@@ -398,3 +450,320 @@ nv50_vbo_validate(struct nv50_context *nv50)
 	so_ref (NULL, &vtxattr);
 }
 
+typedef void (*pfn_push)(struct nouveau_channel *, void *);
+
+struct nv50_vbo_emitctx
+{
+	pfn_push push[16];
+	void *map[16];
+	unsigned stride[16];
+	unsigned nr_ve;
+	unsigned vtx_dwords;
+	unsigned vtx_max;
+};
+
+static INLINE void
+emit_vtx_next(struct nouveau_channel *chan, struct nv50_vbo_emitctx *emit)
+{
+	unsigned i;
+
+	for (i = 0; i < emit->nr_ve; ++i) {
+		emit->push[i](chan, emit->map[i]);
+		emit->map[i] += emit->stride[i];
+	}
+}
+
+static INLINE void
+emit_vtx(struct nouveau_channel *chan, struct nv50_vbo_emitctx *emit,
+	 uint32_t vi)
+{
+	unsigned i;
+
+	for (i = 0; i < emit->nr_ve; ++i)
+		emit->push[i](chan, emit->map[i] + emit->stride[i] * vi);
+}
+
+static INLINE boolean
+nv50_map_vbufs(struct nv50_context *nv50)
+{
+	int i;
+
+	for (i = 0; i < nv50->vtxbuf_nr; ++i) {
+		struct pipe_vertex_buffer *vb = &nv50->vtxbuf[i];
+		unsigned size, delta;
+
+		if (nouveau_bo(vb->buffer)->map)
+			continue;
+
+		size = vb->stride * (vb->max_index + 1);
+		delta = vb->buffer_offset;
+
+		if (!size)
+			size = vb->buffer->size - vb->buffer_offset;
+
+		if (nouveau_bo_map_range(nouveau_bo(vb->buffer),
+					 delta, size, NOUVEAU_BO_RD))
+			break;
+	}
+
+	if (i == nv50->vtxbuf_nr)
+		return TRUE;
+	for (; i >= 0; --i)
+		nouveau_bo_unmap(nouveau_bo(nv50->vtxbuf[i].buffer));
+	return FALSE;
+}
+
+static INLINE void
+nv50_unmap_vbufs(struct nv50_context *nv50)
+{
+        unsigned i;
+
+        for (i = 0; i < nv50->vtxbuf_nr; ++i)
+                if (nouveau_bo(nv50->vtxbuf[i].buffer)->map)
+                        nouveau_bo_unmap(nouveau_bo(nv50->vtxbuf[i].buffer));
+}
+
+static void
+emit_b32_1(struct nouveau_channel *chan, void *data)
+{
+	uint32_t *v = data;
+
+	OUT_RING(chan, v[0]);
+}
+
+static void
+emit_b32_2(struct nouveau_channel *chan, void *data)
+{
+	uint32_t *v = data;
+
+	OUT_RING(chan, v[0]);
+	OUT_RING(chan, v[1]);
+}
+
+static void
+emit_b32_3(struct nouveau_channel *chan, void *data)
+{
+	uint32_t *v = data;
+
+	OUT_RING(chan, v[0]);
+	OUT_RING(chan, v[1]);
+	OUT_RING(chan, v[2]);
+}
+
+static void
+emit_b32_4(struct nouveau_channel *chan, void *data)
+{
+	uint32_t *v = data;
+
+	OUT_RING(chan, v[0]);
+	OUT_RING(chan, v[1]);
+	OUT_RING(chan, v[2]);
+	OUT_RING(chan, v[3]);
+}
+
+static void
+emit_b16_1(struct nouveau_channel *chan, void *data)
+{
+	uint16_t *v = data;
+
+	OUT_RING(chan, v[0]);
+}
+
+static void
+emit_b16_3(struct nouveau_channel *chan, void *data)
+{
+	uint16_t *v = data;
+
+	OUT_RING(chan, (v[1] << 16) | v[0]);
+	OUT_RING(chan, v[2]);
+}
+
+static void
+emit_b08_1(struct nouveau_channel *chan, void *data)
+{
+	uint8_t *v = data;
+
+	OUT_RING(chan, v[0]);
+}
+
+static void
+emit_b08_3(struct nouveau_channel *chan, void *data)
+{
+	uint8_t *v = data;
+
+	OUT_RING(chan, (v[2] << 16) | (v[1] << 8) | v[0]);
+}
+
+static boolean
+emit_prepare(struct nv50_context *nv50, struct nv50_vbo_emitctx *emit,
+	     unsigned start)
+{
+	unsigned i;
+
+	if (nv50_map_vbufs(nv50) == FALSE)
+		return FALSE;
+
+	emit->nr_ve = 0;
+	emit->vtx_dwords = 0;
+
+	for (i = 0; i < nv50->vtxelt_nr; ++i) {
+		struct pipe_vertex_element *ve;
+		struct pipe_vertex_buffer *vb;
+		unsigned n, type, size;
+
+		ve = &nv50->vtxelt[i];
+		vb = &nv50->vtxbuf[ve->vertex_buffer_index];
+		if (!(nv50->vbo_fifo & (1 << i)))
+			continue;
+		n = emit->nr_ve++;
+
+		emit->stride[n] = vb->stride;
+		emit->map[n] = nouveau_bo(vb->buffer)->map +
+			(start * vb->stride + ve->src_offset);
+
+		type = pf_type(ve->src_format);
+		size = pf_size_x(ve->src_format) << pf_exp2(ve->src_format);
+
+		assert(ve->nr_components > 0 && ve->nr_components <= 4);
+
+		/* It shouldn't be necessary to push the implicit 1s
+		 * for case 3 and size 8 cases 1, 2, 3.
+		 */
+		switch (size) {
+		default:
+			NOUVEAU_ERR("unsupported vtxelt size: %u\n", size);
+			return FALSE;
+		case 32:
+			switch (ve->nr_components) {
+			case 1: emit->push[n] = emit_b32_1; break;
+			case 2: emit->push[n] = emit_b32_2; break;
+			case 3: emit->push[n] = emit_b32_3; break;
+			case 4: emit->push[n] = emit_b32_4; break;
+			}
+			emit->vtx_dwords += ve->nr_components;
+			break;
+		case 16:
+			switch (ve->nr_components) {
+			case 1: emit->push[n] = emit_b16_1; break;
+			case 2: emit->push[n] = emit_b32_1; break;
+			case 3: emit->push[n] = emit_b16_3; break;
+			case 4: emit->push[n] = emit_b32_2; break;
+			}
+			emit->vtx_dwords += (ve->nr_components + 1) >> 1;
+			break;
+		case 8:
+			switch (ve->nr_components) {
+			case 1: emit->push[n] = emit_b08_1; break;
+			case 2: emit->push[n] = emit_b16_1; break;
+			case 3: emit->push[n] = emit_b08_3; break;
+			case 4: emit->push[n] = emit_b32_1; break;
+			}
+			emit->vtx_dwords += 1;
+			break;
+		}
+	}
+
+	emit->vtx_max = 512 / emit->vtx_dwords;
+
+	return TRUE;
+}
+
+static boolean
+nv50_push_arrays(struct nv50_context *nv50, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv50->screen->base.channel;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_vbo_emitctx emit;
+
+	if (emit_prepare(nv50, &emit, start) == FALSE)
+		return FALSE;
+
+	while (count) {
+		unsigned i, dw, nr = MIN2(count, emit.vtx_max);
+	        dw = nr * emit.vtx_dwords;
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		for (i = 0; i < nr; ++i)
+			emit_vtx_next(chan, &emit);
+
+		count -= nr;
+	}
+	nv50_unmap_vbufs(nv50);
+
+	return TRUE;
+}
+
+static boolean
+nv50_push_elements_u32(struct nv50_context *nv50, uint32_t *map, unsigned count)
+{
+	struct nouveau_channel *chan = nv50->screen->base.channel;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_vbo_emitctx emit;
+
+	if (emit_prepare(nv50, &emit, 0) == FALSE)
+		return FALSE;
+
+	while (count) {
+		unsigned i, dw, nr = MIN2(count, emit.vtx_max);
+	        dw = nr * emit.vtx_dwords;
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		for (i = 0; i < nr; ++i)
+			emit_vtx(chan, &emit, *map++);
+
+		count -= nr;
+	}
+	nv50_unmap_vbufs(nv50);
+
+	return TRUE;
+}
+
+static boolean
+nv50_push_elements_u16(struct nv50_context *nv50, uint16_t *map, unsigned count)
+{
+	struct nouveau_channel *chan = nv50->screen->base.channel;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_vbo_emitctx emit;
+
+	if (emit_prepare(nv50, &emit, 0) == FALSE)
+		return FALSE;
+
+	while (count) {
+		unsigned i, dw, nr = MIN2(count, emit.vtx_max);
+	        dw = nr * emit.vtx_dwords;
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		for (i = 0; i < nr; ++i)
+			emit_vtx(chan, &emit, *map++);
+
+		count -= nr;
+	}
+	nv50_unmap_vbufs(nv50);
+
+	return TRUE;
+}
+
+static boolean
+nv50_push_elements_u08(struct nv50_context *nv50, uint8_t *map, unsigned count)
+{
+	struct nouveau_channel *chan = nv50->screen->base.channel;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_vbo_emitctx emit;
+
+	if (emit_prepare(nv50, &emit, 0) == FALSE)
+		return FALSE;
+
+	while (count) {
+		unsigned i, dw, nr = MIN2(count, emit.vtx_max);
+	        dw = nr * emit.vtx_dwords;
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		for (i = 0; i < nr; ++i)
+			emit_vtx(chan, &emit, *map++);
+
+		count -= nr;
+	}
+	nv50_unmap_vbufs(nv50);
+
+	return TRUE;
+}
diff --git a/src/gallium/drivers/r300/Makefile b/src/gallium/drivers/r300/Makefile
index d7a2c8c462..f73d80de88 100644
--- a/src/gallium/drivers/r300/Makefile
+++ b/src/gallium/drivers/r300/Makefile
@@ -4,11 +4,10 @@ include $(TOP)/configs/current
 LIBNAME = r300
 
 C_SOURCES = \
-	r3xx_fs.c \
-	r5xx_fs.c \
 	r300_chipset.c \
 	r300_clear.c \
 	r300_context.c \
+	r300_debug.c \
 	r300_emit.c \
 	r300_flush.c \
 	r300_fs.c \
@@ -19,14 +18,11 @@ C_SOURCES = \
 	r300_state_derived.c \
 	r300_state_invariant.c \
 	r300_vs.c \
-	r300_surface.c \
 	r300_texture.c \
 	r300_tgsi_to_rc.c
 
 LIBRARY_INCLUDES = \
-	-I$(TOP)/src/mesa/drivers/dri/r300/compiler \
-	-I$(TOP)/src/mesa \
-	-I$(TOP)/include
+	-I$(TOP)/src/mesa/drivers/dri/r300/compiler
 
 COMPILER_ARCHIVE = $(TOP)/src/mesa/drivers/dri/r300/compiler/libr300compiler.a
 
diff --git a/src/gallium/drivers/r300/SConscript b/src/gallium/drivers/r300/SConscript
index 493d7b28bc..b4c8ba2015 100644
--- a/src/gallium/drivers/r300/SConscript
+++ b/src/gallium/drivers/r300/SConscript
@@ -1,6 +1,10 @@
 Import('*')
 
+r300compiler = SConscript('#/src/mesa/drivers/dri/r300/compiler/SConscript')
+
 env = env.Clone()
+# add the paths for r300compiler
+env.Append(CPPPATH = ['#/src/mesa/drivers/dri/r300/compiler', '#/include', '#/src/mesa'])
 
 r300 = env.ConvenienceLibrary(
     target = 'r300',
@@ -23,7 +27,8 @@ r300 = env.ConvenienceLibrary(
         'r300_vs.c',
         'r300_surface.c',
         'r300_texture.c',
-    ])
+        'r300_tgsi_to_rc.c',
+    ] + r300compiler) + r300compiler
 
 Export('r300')
 
diff --git a/src/gallium/drivers/r300/r300_chipset.c b/src/gallium/drivers/r300/r300_chipset.c
index d138866d33..51fdb82ff3 100644
--- a/src/gallium/drivers/r300/r300_chipset.c
+++ b/src/gallium/drivers/r300/r300_chipset.c
@@ -21,6 +21,7 @@
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
 #include "r300_chipset.h"
+
 #include "util/u_debug.h"
 
 /* r300_chipset: A file all to itself for deducing the various properties of
@@ -31,7 +32,7 @@ void r300_parse_chipset(struct r300_capabilities* caps)
 {
     /* Reasonable defaults */
     caps->num_vert_fpus = 4;
-    caps->has_tcl = getenv("RADEON_NO_TCL") ? FALSE : TRUE;
+    caps->has_tcl = debug_get_bool_option("RADEON_NO_TCL", FALSE) ? FALSE : TRUE;
     caps->is_r500 = FALSE;
     caps->high_second_pipe = FALSE;
 
diff --git a/src/gallium/drivers/r300/r300_chipset.h b/src/gallium/drivers/r300/r300_chipset.h
index 322d4a57e4..0633a8b8a7 100644
--- a/src/gallium/drivers/r300/r300_chipset.h
+++ b/src/gallium/drivers/r300/r300_chipset.h
@@ -33,9 +33,11 @@ struct r300_capabilities {
     /* Chipset family */
     int family;
     /* The number of vertex floating-point units */
-    int num_vert_fpus;
+    unsigned num_vert_fpus;
     /* The number of fragment pipes */
-    int num_frag_pipes;
+    unsigned num_frag_pipes;
+    /* The number of z pipes */
+    unsigned num_z_pipes;
     /* Whether or not TCL is physically present */
     boolean has_tcl;
     /* Whether or not this is an RV515 or newer; R500s have many differences
diff --git a/src/gallium/drivers/r300/r300_clear.c b/src/gallium/drivers/r300/r300_clear.c
index 8b9cb819ae..02d6d504fc 100644
--- a/src/gallium/drivers/r300/r300_clear.c
+++ b/src/gallium/drivers/r300/r300_clear.c
@@ -21,6 +21,9 @@
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
 #include "r300_clear.h"
+#include "r300_context.h"
+
+#include "util/u_clear.h"
 
 /* Clears currently bound buffers. */
 void r300_clear(struct pipe_context* pipe,
diff --git a/src/gallium/drivers/r300/r300_clear.h b/src/gallium/drivers/r300/r300_clear.h
index cd5900565e..b8fcdf273c 100644
--- a/src/gallium/drivers/r300/r300_clear.h
+++ b/src/gallium/drivers/r300/r300_clear.h
@@ -23,9 +23,7 @@
 #ifndef R300_CLEAR_H
 #define R300_CLEAR_H
 
-#include "util/u_clear.h"
-
-#include "r300_context.h"
+struct pipe_context;
 
 void r300_clear(struct pipe_context* pipe,
                 unsigned buffers,
diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index da67bc29b8..c34fbb1123 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -20,87 +20,52 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
-#include "r300_context.h"
-
-static boolean r300_draw_range_elements(struct pipe_context* pipe,
-                                        struct pipe_buffer* indexBuffer,
-                                        unsigned indexSize,
-                                        unsigned minIndex,
-                                        unsigned maxIndex,
-                                        unsigned mode,
-                                        unsigned start,
-                                        unsigned count)
-{
-    struct r300_context* r300 = r300_context(pipe);
-    int i;
-
-    for (i = 0; i < r300->vertex_buffer_count; i++) {
-        void* buf = pipe_buffer_map(pipe->screen,
-                                    r300->vertex_buffers[i].buffer,
-                                    PIPE_BUFFER_USAGE_CPU_READ);
-        draw_set_mapped_vertex_buffer(r300->draw, i, buf);
-    }
-
-    if (indexBuffer) {
-        void* indices = pipe_buffer_map(pipe->screen, indexBuffer,
-                                        PIPE_BUFFER_USAGE_CPU_READ);
-        draw_set_mapped_element_buffer_range(r300->draw, indexSize,
-                                             minIndex, maxIndex, indices);
-    } else {
-        draw_set_mapped_element_buffer(r300->draw, 0, NULL);
-    }
+#include "draw/draw_context.h"
 
-    draw_set_mapped_constant_buffer(r300->draw,
-            r300->shader_constants[PIPE_SHADER_VERTEX].constants,
-            r300->shader_constants[PIPE_SHADER_VERTEX].count *
-                (sizeof(float) * 4));
+#include "pipe/p_inlines.h"
 
-    draw_arrays(r300->draw, mode, start, count);
+#include "tgsi/tgsi_scan.h"
 
-    for (i = 0; i < r300->vertex_buffer_count; i++) {
-        pipe_buffer_unmap(pipe->screen, r300->vertex_buffers[i].buffer);
-        draw_set_mapped_vertex_buffer(r300->draw, i, NULL);
-    }
-
-    if (indexBuffer) {
-        pipe_buffer_unmap(pipe->screen, indexBuffer);
-        draw_set_mapped_element_buffer_range(r300->draw, 0, start,
-                                             start + count - 1, NULL);
-    }
+#include "util/u_hash_table.h"
+#include "util/u_memory.h"
+#include "util/u_simple_list.h"
 
-    return TRUE;
-}
-
-static boolean r300_draw_elements(struct pipe_context* pipe,
-                                  struct pipe_buffer* indexBuffer,
-                                  unsigned indexSize, unsigned mode,
-                                  unsigned start, unsigned count)
+#include "r300_clear.h"
+#include "r300_context.h"
+#include "r300_flush.h"
+#include "r300_query.h"
+#include "r300_render.h"
+#include "r300_screen.h"
+#include "r300_state_derived.h"
+#include "r300_state_invariant.h"
+#include "r300_winsys.h"
+
+static enum pipe_error r300_clear_hash_table(void* key, void* value,
+                                             void* data)
 {
-    return r300_draw_range_elements(pipe, indexBuffer, indexSize, 0, ~0,
-                                    mode, start, count);
+    FREE(key);
+    FREE(value);
+    return PIPE_OK;
 }
 
-static boolean r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
-                                unsigned start, unsigned count)
+static void r300_destroy_context(struct pipe_context* context)
 {
-    return r300_draw_elements(pipe, NULL, 0, mode, start, count);
-}
-
-static void r300_destroy_context(struct pipe_context* context) {
     struct r300_context* r300 = r300_context(context);
     struct r300_query* query, * temp;
 
+    util_hash_table_foreach(r300->shader_hash_table, r300_clear_hash_table,
+        NULL);
+    util_hash_table_destroy(r300->shader_hash_table);
+
     draw_destroy(r300->draw);
 
     /* Free the OQ BO. */
     context->screen->buffer_destroy(r300->oqbo);
 
     /* If there are any queries pending or not destroyed, remove them now. */
-    if (r300->query_list) {
-        foreach_s(query, temp, r300->query_list) {
-            remove_from_list(query);
-            FREE(query);
-        }
+    foreach_s(query, temp, &r300->query_list) {
+        remove_from_list(query);
+        FREE(query);
     }
 
     FREE(r300->blend_color_state);
@@ -133,6 +98,13 @@ r300_is_buffer_referenced( struct pipe_context *pipe,
    return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
 }
 
+static void r300_flush_cb(void *data)
+{
+    struct r300_context* const cs_context_copy = data;
+
+    cs_context_copy->context.flush(&cs_context_copy->context, 0, NULL);
+}
+
 struct pipe_context* r300_create_context(struct pipe_screen* screen,
                                          struct r300_winsys* r300_winsys)
 {
@@ -144,7 +116,9 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     r300->winsys = r300_winsys;
 
     r300->context.winsys = (struct pipe_winsys*)r300_winsys;
-    r300->context.screen = r300_screen(screen);
+    r300->context.screen = screen;
+
+    r300_init_debug(r300);
 
     r300->context.destroy = r300_destroy_context;
 
@@ -152,11 +126,14 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
 
     r300->context.draw_arrays = r300_draw_arrays;
     r300->context.draw_elements = r300_draw_elements;
-    r300->context.draw_range_elements = r300_draw_range_elements;
+    r300->context.draw_range_elements = r300_swtcl_draw_range_elements;
 
     r300->context.is_texture_referenced = r300_is_texture_referenced;
     r300->context.is_buffer_referenced = r300_is_buffer_referenced;
 
+    r300->shader_hash_table = util_hash_table_create(r300_shader_key_hash,
+        r300_shader_key_compare);
+
     r300->blend_color_state = CALLOC_STRUCT(r300_blend_color_state);
     r300->rs_block = CALLOC_STRUCT(r300_rs_block);
     r300->scissor_state = CALLOC_STRUCT(r300_scissor_state);
@@ -180,13 +157,15 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
 
     r300_init_query_functions(r300);
 
-    r300_init_surface_functions(r300);
+    /* r300_init_surface_functions(r300); */
 
     r300_init_state_functions(r300);
 
     r300_emit_invariant_state(r300);
+
+    r300->winsys->set_flush_cb(r300->winsys, r300_flush_cb, r300);
     r300->dirty_state = R300_NEW_KITCHEN_SINK;
     r300->dirty_hw++;
-
+    make_empty_list(&r300->query_list);
     return &r300->context;
 }
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index f78492d4aa..30b80fa9db 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -23,21 +23,10 @@
 #ifndef R300_CONTEXT_H
 #define R300_CONTEXT_H
 
-#include "draw/draw_context.h"
 #include "draw/draw_vertex.h"
 
 #include "pipe/p_context.h"
 
-#include "tgsi/tgsi_scan.h"
-
-#include "util/u_memory.h"
-#include "util/u_simple_list.h"
-
-#include "r300_clear.h"
-#include "r300_query.h"
-#include "r300_screen.h"
-#include "r300_winsys.h"
-
 struct r300_fragment_shader;
 struct r300_vertex_shader;
 
@@ -62,7 +51,6 @@ struct r300_dsa_state {
     uint32_t z_buffer_control;  /* R300_ZB_CNTL: 0x4f00 */
     uint32_t z_stencil_control; /* R300_ZB_ZSTENCILCNTL: 0x4f04 */
     uint32_t stencil_ref_mask;  /* R300_ZB_STENCILREFMASK: 0x4f08 */
-    uint32_t z_buffer_top;      /* R300_ZB_ZTOP: 0x4f14 */
     uint32_t stencil_ref_bf;    /* R500_ZB_STENCILREFMASK_BF: 0x4fd4 */
 };
 
@@ -124,6 +112,10 @@ struct r300_viewport_state {
     uint32_t vte_control; /* R300_VAP_VTE_CNTL:      0x20b0 */
 };
 
+struct r300_ztop_state {
+    uint32_t z_buffer_top;      /* R300_ZB_ZTOP: 0x4f14 */
+};
+
 #define R300_NEW_BLEND           0x00000001
 #define R300_NEW_BLEND_COLOR     0x00000002
 #define R300_NEW_CLIP            0x00000004
@@ -141,7 +133,8 @@ struct r300_viewport_state {
 #define R300_NEW_VERTEX_FORMAT   0x04000000
 #define R300_NEW_VERTEX_SHADER   0x08000000
 #define R300_NEW_VIEWPORT        0x10000000
-#define R300_NEW_KITCHEN_SINK    0x1fffffff
+#define R300_NEW_QUERY           0x20000000
+#define R300_NEW_KITCHEN_SINK    0x3fffffff
 
 /* The next several objects are not pure Radeon state; they inherit from
  * various Gallium classes. */
@@ -172,6 +165,10 @@ struct r300_query {
     unsigned int count;
     /* The offset of this query into the query buffer, in bytes. */
     unsigned offset;
+    /* if we've flushed the query */
+    boolean flushed;
+    /* if begin has been emitted */
+    boolean begin_emitted;
     /* Linked list members. */
     struct r300_query* prev;
     struct r300_query* next;
@@ -184,8 +181,15 @@ struct r300_texture {
     /* Offsets into the buffer. */
     unsigned offset[PIPE_MAX_TEXTURE_LEVELS];
 
-    /* Stride (pitch?) of this texture in bytes */
-    unsigned stride;
+    /**
+     * If non-zero, override the natural texture layout with
+     * a custom stride (in bytes).
+     *
+     * \note Mipmapping fails for textures with a non-natural layout!
+     *
+     * \sa r300_texture_get_stride
+     */
+    unsigned stride_override;
 
     /* Total size of this texture, in bytes. */
     unsigned size;
@@ -211,10 +215,7 @@ struct r300_vertex_format {
     int fs_tab[16];
 };
 
-static struct pipe_viewport_state r300_viewport_identity = {
-    .scale = {1.0, 1.0, 1.0, 1.0},
-    .translate = {0.0, 0.0, 0.0, 0.0},
-};
+extern struct pipe_viewport_state r300_viewport_identity;
 
 struct r300_context {
     /* Parent class */
@@ -233,7 +234,14 @@ struct r300_context {
     /* Occlusion query buffer. */
     struct pipe_buffer* oqbo;
     /* Query list. */
-    struct r300_query* query_list;
+    struct r300_query *query_current;
+    struct r300_query query_list;
+
+    /* Shader hash table. Used to store vertex formatting information, which
+     * depends on the combination of both currently loaded shaders. */
+    struct util_hash_table* shader_hash_table;
+    /* Vertex formatting information. */
+    struct r300_vertex_format* vertex_info;
 
     /* Various CSO state objects. */
     /* Blend state. */
@@ -265,16 +273,23 @@ struct r300_context {
     /* Vertex buffers for Gallium. */
     struct pipe_vertex_buffer vertex_buffers[PIPE_MAX_ATTRIBS];
     int vertex_buffer_count;
-    /* Vertex information. */
-    struct r300_vertex_format vertex_info;
+    /* Vertex elements for Gallium. */
+    struct pipe_vertex_element vertex_elements[PIPE_MAX_ATTRIBS];
+    int vertex_element_count;
     /* Vertex shader. */
     struct r300_vertex_shader* vs;
     /* Viewport state. */
     struct r300_viewport_state* viewport_state;
+    /* ZTOP state. */
+    struct r300_ztop_state ztop_state;
+
     /* Bitmask of dirty state objects. */
     uint32_t dirty_state;
     /* Flag indicating whether or not the HW is dirty. */
     uint32_t dirty_hw;
+
+    /** Combination of DBG_xxx flags */
+    unsigned debug;
 };
 
 /* Convenience cast wrapper. */
@@ -288,4 +303,42 @@ struct draw_stage* r300_draw_stage(struct r300_context* r300);
 void r300_init_state_functions(struct r300_context* r300);
 void r300_init_surface_functions(struct r300_context* r300);
 
+/* Debug functionality. */
+
+/**
+ * Debug flags to disable/enable certain groups of debugging outputs.
+ *
+ * \note These may be rather coarse, and the grouping may be impractical.
+ * If you find, while debugging the driver, that a different grouping
+ * of these flags would be beneficial, just feel free to change them
+ * but make sure to update the documentation in r300_debug.c to reflect
+ * those changes.
+ */
+/*@{*/
+#define DBG_HELP    0x0000001
+#define DBG_FP      0x0000002
+#define DBG_VP      0x0000004
+#define DBG_CS      0x0000008
+#define DBG_DRAW    0x0000010
+#define DBG_TEX     0x0000020
+#define DBG_FALL    0x0000040
+/*@}*/
+
+static INLINE boolean DBG_ON(struct r300_context * ctx, unsigned flags)
+{
+    return (ctx->debug & flags) ? true : false;
+}
+
+static INLINE void DBG(struct r300_context * ctx, unsigned flags, const char * fmt, ...)
+{
+    if (DBG_ON(ctx, flags)) {
+        va_list va;
+        va_start(va, fmt);
+        debug_vprintf(fmt, va);
+        va_end(va);
+    }
+}
+
+void r300_init_debug(struct r300_context * ctx);
+
 #endif /* R300_CONTEXT_H */
diff --git a/src/gallium/drivers/r300/r300_cs.h b/src/gallium/drivers/r300/r300_cs.h
index 71b142c0db..883f0a02dc 100644
--- a/src/gallium/drivers/r300/r300_cs.h
+++ b/src/gallium/drivers/r300/r300_cs.h
@@ -49,7 +49,8 @@
     (RADEON_CP_PACKET0 | ((count) << 16) | ((register) >> 2))
 
 #define CS_LOCALS(context) \
-    struct r300_winsys* cs_winsys = context->winsys; \
+    struct r300_context* const cs_context_copy = (context); \
+    struct r300_winsys* cs_winsys = cs_context_copy->winsys; \
     int cs_count = 0;
 
 #define CHECK_CS(size) \
@@ -58,7 +59,7 @@
 #define BEGIN_CS(size) do { \
     CHECK_CS(size); \
     if (VERY_VERBOSE_CS) { \
-        debug_printf("r300: BEGIN_CS, count %d, in %s (%s:%d)\n", \
+        DBG(cs_context_copy, DBG_CS, "r300: BEGIN_CS, count %d, in %s (%s:%d)\n", \
                 size, __FUNCTION__, __FILE__, __LINE__); \
     } \
     cs_winsys->begin_cs(cs_winsys, (size), \
@@ -67,47 +68,55 @@
 } while (0)
 
 #define OUT_CS(value) do { \
+    if (VERY_VERBOSE_CS || VERY_VERBOSE_REGISTERS) { \
+        DBG(cs_context_copy, DBG_CS, "r300: writing %08x\n", value); \
+    } \
     cs_winsys->write_cs_dword(cs_winsys, (value)); \
     cs_count--; \
 } while (0)
 
 #define OUT_CS_32F(value) do { \
+    if (VERY_VERBOSE_CS || VERY_VERBOSE_REGISTERS) { \
+        DBG(cs_context_copy, DBG_CS, "r300: writing %f\n", value); \
+    } \
     cs_winsys->write_cs_dword(cs_winsys, fui(value)); \
     cs_count--; \
 } while (0)
 
 #define OUT_CS_REG(register, value) do { \
     if (VERY_VERBOSE_REGISTERS) \
-        debug_printf("r300: writing 0x%08X to register 0x%04X\n", \
+        DBG(cs_context_copy, DBG_CS, "r300: writing 0x%08X to register 0x%04X\n", \
             value, register); \
     assert(register); \
-    OUT_CS(CP_PACKET0(register, 0)); \
-    OUT_CS(value); \
+    cs_winsys->write_cs_dword(cs_winsys, CP_PACKET0(register, 0)); \
+    cs_winsys->write_cs_dword(cs_winsys, value); \
+    cs_count -= 2; \
 } while (0)
 
 /* Note: This expects count to be the number of registers,
  * not the actual packet0 count! */
 #define OUT_CS_REG_SEQ(register, count) do { \
     if (VERY_VERBOSE_REGISTERS) \
-        debug_printf("r300: writing register sequence of %d to 0x%04X\n", \
+        DBG(cs_context_copy, DBG_CS, "r300: writing register sequence of %d to 0x%04X\n", \
             count, register); \
     assert(register); \
-    OUT_CS(CP_PACKET0(register, ((count) - 1))); \
+    cs_winsys->write_cs_dword(cs_winsys, CP_PACKET0((register), ((count) - 1))); \
+    cs_count--; \
 } while (0)
 
 #define OUT_CS_RELOC(bo, offset, rd, wd, flags) do { \
-    debug_printf("r300: writing relocation for buffer %p, offset %d, " \
+    DBG(cs_context_copy, DBG_CS, "r300: writing relocation for buffer %p, offset %d, " \
             "domains (%d, %d, %d)\n", \
         bo, offset, rd, wd, flags); \
     assert(bo); \
-    OUT_CS(offset); \
+    cs_winsys->write_cs_dword(cs_winsys, offset); \
     cs_winsys->write_cs_reloc(cs_winsys, bo, rd, wd, flags); \
-    cs_count -= 2; \
+    cs_count -= 3; \
 } while (0)
 
 #define END_CS do { \
     if (VERY_VERBOSE_CS) { \
-        debug_printf("r300: END_CS in %s (%s:%d)\n", __FUNCTION__, \
+        DBG(cs_context_copy, DBG_CS, "r300: END_CS in %s (%s:%d)\n", __FUNCTION__, \
                 __FILE__, __LINE__); \
     } \
     if (cs_count != 0) \
@@ -117,7 +126,7 @@
 
 #define FLUSH_CS do { \
     if (VERY_VERBOSE_CS) { \
-        debug_printf("r300: FLUSH_CS in %s (%s:%d)\n\n", __FUNCTION__, \
+        DBG(cs_context_copy, DBG_CS, "r300: FLUSH_CS in %s (%s:%d)\n\n", __FUNCTION__, \
                 __FILE__, __LINE__); \
     } \
     cs_winsys->flush_cs(cs_winsys); \
@@ -127,27 +136,29 @@
 
 #define OUT_CS_ONE_REG(register, count) do { \
     if (VERY_VERBOSE_REGISTERS) \
-        debug_printf("r300: writing data sequence of %d to 0x%04X\n", \
+        DBG(cs_context_copy, DBG_CS, "r300: writing data sequence of %d to 0x%04X\n", \
             count, register); \
     assert(register); \
-    OUT_CS(CP_PACKET0(register, ((count) - 1)) | RADEON_ONE_REG_WR); \
+    cs_winsys->write_cs_dword(cs_winsys, CP_PACKET0((register), ((count) - 1)) | RADEON_ONE_REG_WR); \
+    cs_count--; \
 } while (0)
 
 #define CP_PACKET3(op, count) \
     (RADEON_CP_PACKET3 | (op) | ((count) << 16))
 
 #define OUT_CS_PKT3(op, count) do { \
-    OUT_CS(CP_PACKET3(op, count)); \
+    cs_winsys->write_cs_dword(cs_winsys, CP_PACKET3(op, count)); \
+    cs_count--; \
 } while (0)
 
 #define OUT_CS_INDEX_RELOC(bo, offset, count, rd, wd, flags) do { \
-    debug_printf("r300: writing relocation for index buffer %p," \
+    DBG(cs_context_copy, DBG_CS, "r300: writing relocation for index buffer %p," \
             "offset %d\n", bo, offset); \
     assert(bo); \
-    OUT_CS(offset); \
-    OUT_CS(count); \
+    cs_winsys->write_cs_dword(cs_winsys, offset); \
+    cs_winsys->write_cs_dword(cs_winsys, count); \
     cs_winsys->write_cs_reloc(cs_winsys, bo, rd, wd, flags); \
-    cs_count -= 2; \
+    cs_count -= 4; \
 } while (0)
 
 #endif /* R300_CS_H */
diff --git a/src/gallium/drivers/r300/r300_debug.c b/src/gallium/drivers/r300/r300_debug.c
new file mode 100644
index 0000000000..421253ca72
--- /dev/null
+++ b/src/gallium/drivers/r300/r300_debug.c
@@ -0,0 +1,91 @@
+/*
+ * Copyright 2009 Nicolai Haehnle <nhaehnle@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE. */
+
+#include "r300_context.h"
+
+#include <ctype.h>
+
+
+struct debug_option {
+    const char * name;
+    unsigned flag;
+    const char * description;
+};
+
+static struct debug_option debug_options[] = {
+    { "help", DBG_HELP, "Helpful meta-information about the driver" },
+    { "fp", DBG_FP, "Fragment program handling" },
+    { "vp", DBG_VP, "Vertex program handling" },
+    { "cs", DBG_CS, "Command submissions" },
+    { "draw", DBG_DRAW, "Draw and emit" },
+    { "tex", DBG_TEX, "Textures" },
+    { "fall", DBG_FALL, "Fallbacks" },
+
+    { "all", ~0, "Convenience option that enables all debug flags" },
+
+    /* must be last */
+    { 0, 0, 0 }
+};
+
+void r300_init_debug(struct r300_context * ctx)
+{
+    const char * options = debug_get_option("RADEON_DEBUG", 0);
+    boolean printhint = false;
+    size_t length;
+    struct debug_option * opt;
+
+    if (options) {
+        while(*options) {
+            if (*options == ' ' || *options == ',') {
+                options++;
+                continue;
+            }
+
+            length = strcspn(options, " ,");
+
+            for(opt = debug_options; opt->name; ++opt) {
+                if (!strncmp(options, opt->name, length)) {
+                    ctx->debug |= opt->flag;
+                    break;
+                }
+            }
+
+            if (!opt->name) {
+                debug_printf("Unknown debug option: %s\n", options);
+                printhint = true;
+            }
+
+            options += length;
+        }
+
+        if (!ctx->debug)
+            printhint = true;
+    }
+
+    if (printhint || ctx->debug & DBG_HELP) {
+        debug_printf("You can enable debug output by setting the RADEON_DEBUG environment variable\n"
+                     "to a comma-separated list of debug options. Available options are:\n");
+        for(opt = debug_options; opt->name; ++opt) {
+            debug_printf("    %s: %s\n", opt->name, opt->description);
+        }
+    }
+}
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index bd4d59e6f1..258c38fefd 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -22,9 +22,15 @@
 
 /* r300_emit: Functions for emitting state. */
 
-#include "r300_emit.h"
+#include "util/u_math.h"
 
+#include "r300_context.h"
+#include "r300_cs.h"
+#include "r300_emit.h"
 #include "r300_fs.h"
+#include "r300_screen.h"
+#include "r300_state_derived.h"
+#include "r300_state_inlines.h"
 #include "r300_vs.h"
 
 void r300_emit_blend_state(struct r300_context* r300,
@@ -105,7 +111,7 @@ void r300_emit_dsa_state(struct r300_context* r300,
     OUT_CS(dsa->z_buffer_control);
     OUT_CS(dsa->z_stencil_control);
     OUT_CS(dsa->stencil_ref_mask);
-    OUT_CS_REG(R300_ZB_ZTOP, dsa->z_buffer_top);
+    OUT_CS_REG(R300_ZB_ZTOP, r300->ztop_state.z_buffer_top);
     if (r300screen->caps->is_r500) {
         /* OUT_CS_REG(R500_ZB_STENCILREFMASK_BF, dsa->stencil_ref_bf); */
     }
@@ -211,7 +217,7 @@ void r300_emit_fragment_program_code(struct r300_context* r300,
     }
 
     if (constants->Count) {
-        OUT_CS_ONE_REG(R300_PFS_PARAM_0_X, constants->Count * 4);
+        OUT_CS_REG_SEQ(R300_PFS_PARAM_0_X, constants->Count * 4);
         for(i = 0; i < constants->Count; ++i) {
             const float * data = get_shader_constant(r300, &constants->Constants[i], externals);
             OUT_CS(pack_float24(data[0]));
@@ -282,7 +288,7 @@ void r300_emit_fb_state(struct r300_context* r300,
     for (i = 0; i < fb->nr_cbufs; i++) {
         tex = (struct r300_texture*)fb->cbufs[i]->texture;
         assert(tex && tex->buffer && "cbuf is marked, but NULL!");
-        pixpitch = tex->stride / tex->tex.block.size;
+        pixpitch = r300_texture_get_stride(tex, 0) / tex->tex.block.size;
 
         OUT_CS_REG_SEQ(R300_RB3D_COLOROFFSET0 + (4 * i), 1);
         OUT_CS_RELOC(tex->buffer, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
@@ -299,7 +305,7 @@ void r300_emit_fb_state(struct r300_context* r300,
     if (fb->zsbuf) {
         tex = (struct r300_texture*)fb->zsbuf->texture;
         assert(tex && tex->buffer && "zsbuf is marked, but NULL!");
-        pixpitch = tex->stride / tex->tex.block.size;
+        pixpitch = r300_texture_get_stride(tex, 0) / tex->tex.block.size;
 
         OUT_CS_REG_SEQ(R300_ZB_DEPTHOFFSET, 1);
         OUT_CS_RELOC(tex->buffer, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
@@ -319,31 +325,37 @@ void r300_emit_fb_state(struct r300_context* r300,
     END_CS;
 }
 
-void r300_emit_query_begin(struct r300_context* r300,
-                           struct r300_query* query)
+static void r300_emit_query_start(struct r300_context *r300)
 {
+    struct r300_capabilities *caps = r300_screen(r300->context.screen)->caps;
+    struct r300_query *query = r300->query_current;
     CS_LOCALS(r300);
 
+    if (!query)
+	return;
+
     /* XXX This will almost certainly not return good results
      * for overlapping queries. */
-    BEGIN_CS(2);
+    BEGIN_CS(4);
+    if (caps->family == CHIP_FAMILY_RV530) {
+        OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_ALL);
+    } else {
+        OUT_CS_REG(R300_SU_REG_DEST, R300_RASTER_PIPE_SELECT_ALL);
+    }
     OUT_CS_REG(R300_ZB_ZPASS_DATA, 0);
     END_CS;
+    query->begin_emitted = TRUE;
 }
 
-void r300_emit_query_end(struct r300_context* r300,
-                         struct r300_query* query)
+
+static void r300_emit_query_finish(struct r300_context *r300,
+                                   struct r300_query *query)
 {
     struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps;
     CS_LOCALS(r300);
 
-    if (!r300->winsys->add_buffer(r300->winsys, r300->oqbo,
-                0, RADEON_GEM_DOMAIN_GTT)) {
-        debug_printf("r300: There wasn't room for the OQ buffer!?"
-                " Oh noes!\n");
-    }
-
     assert(caps->num_frag_pipes);
+
     BEGIN_CS(6 * caps->num_frag_pipes + 2);
     /* I'm not so sure I like this switch, but it's hard to be elegant
      * when there's so many special cases...
@@ -380,6 +392,7 @@ void r300_emit_query_end(struct r300_context* r300,
             OUT_CS_REG_SEQ(R300_ZB_ZPASS_ADDR, 1);
             OUT_CS_RELOC(r300->oqbo, query->offset + (sizeof(uint32_t) * 0),
                     0, RADEON_GEM_DOMAIN_GTT, 0);
+            break;
         default:
             debug_printf("r300: Implementation error: Chipset reports %d"
                     " pixel pipes!\n", caps->num_frag_pipes);
@@ -389,7 +402,55 @@ void r300_emit_query_end(struct r300_context* r300,
     /* And, finally, reset it to normal... */
     OUT_CS_REG(R300_SU_REG_DEST, 0xF);
     END_CS;
+}
 
+static void rv530_emit_query_single(struct r300_context *r300,
+                                    struct r300_query *query)
+{
+    CS_LOCALS(r300);
+
+    BEGIN_CS(8);
+    OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_0);
+    OUT_CS_REG_SEQ(R300_ZB_ZPASS_ADDR, 1);
+    OUT_CS_RELOC(r300->oqbo, query->offset, 0, RADEON_GEM_DOMAIN_GTT, 0);
+    OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_ALL);
+    END_CS;
+}
+
+static void rv530_emit_query_double(struct r300_context *r300,
+                                    struct r300_query *query)
+{
+    CS_LOCALS(r300);
+
+    BEGIN_CS(14);
+    OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_0);
+    OUT_CS_REG_SEQ(R300_ZB_ZPASS_ADDR, 1);
+    OUT_CS_RELOC(r300->oqbo, query->offset, 0, RADEON_GEM_DOMAIN_GTT, 0);
+    OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_1);
+    OUT_CS_REG_SEQ(R300_ZB_ZPASS_ADDR, 1);
+    OUT_CS_RELOC(r300->oqbo, query->offset + sizeof(uint32_t), 0, RADEON_GEM_DOMAIN_GTT, 0);
+    OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_ALL);
+    END_CS;
+}
+
+void r300_emit_query_end(struct r300_context* r300)
+{
+    struct r300_capabilities *caps = r300_screen(r300->context.screen)->caps;
+    struct r300_query *query = r300->query_current;
+
+    if (!query)
+	return;
+
+    if (query->begin_emitted == FALSE)
+        return;
+
+    if (caps->family == CHIP_FAMILY_RV530) {
+        if (caps->num_z_pipes == 2)
+            rv530_emit_query_double(r300, query);
+        else
+            rv530_emit_query_single(r300, query);
+    } else 
+        r300_emit_query_finish(r300, query);
 }
 
 void r300_emit_rs_state(struct r300_context* r300, struct r300_rs_state* rs)
@@ -473,7 +534,8 @@ void r300_emit_texture(struct r300_context* r300,
     CS_LOCALS(r300);
 
     BEGIN_CS(16);
-    OUT_CS_REG(R300_TX_FILTER0_0 + (offset * 4), sampler->filter0);
+    OUT_CS_REG(R300_TX_FILTER0_0 + (offset * 4), sampler->filter0 |
+        (offset << 28));
     OUT_CS_REG(R300_TX_FILTER1_0 + (offset * 4), sampler->filter1);
     OUT_CS_REG(R300_TX_BORDER_COLOR_0 + (offset * 4), sampler->border_color);
 
@@ -490,9 +552,9 @@ void r300_emit_vertex_buffer(struct r300_context* r300)
 {
     CS_LOCALS(r300);
 
-    debug_printf("r300: Preparing vertex buffer %p for render, "
+    DBG(r300, DBG_DRAW, "r300: Preparing vertex buffer %p for render, "
             "vertex size %d\n", r300->vbo,
-            r300->vertex_info.vinfo.size);
+            r300->vertex_info->vinfo.size);
     /* Set the pointer to our vertex buffer. The emitted values are this:
      * PACKET3 [3D_LOAD_VBPNTR]
      * COUNT   [1]
@@ -503,8 +565,8 @@ void r300_emit_vertex_buffer(struct r300_context* r300)
     BEGIN_CS(7);
     OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, 3);
     OUT_CS(1);
-    OUT_CS(r300->vertex_info.vinfo.size |
-            (r300->vertex_info.vinfo.size << 8));
+    OUT_CS(r300->vertex_info->vinfo.size |
+            (r300->vertex_info->vinfo.size << 8));
     OUT_CS(r300->vbo_offset);
     OUT_CS_RELOC(r300->vbo, 0, RADEON_GEM_DOMAIN_GTT, 0, 0);
     END_CS;
@@ -516,30 +578,30 @@ void r300_emit_vertex_format_state(struct r300_context* r300)
     CS_LOCALS(r300);
 
     BEGIN_CS(26);
-    OUT_CS_REG(R300_VAP_VTX_SIZE, r300->vertex_info.vinfo.size);
+    OUT_CS_REG(R300_VAP_VTX_SIZE, r300->vertex_info->vinfo.size);
 
     OUT_CS_REG_SEQ(R300_VAP_VTX_STATE_CNTL, 2);
-    OUT_CS(r300->vertex_info.vinfo.hwfmt[0]);
-    OUT_CS(r300->vertex_info.vinfo.hwfmt[1]);
+    OUT_CS(r300->vertex_info->vinfo.hwfmt[0]);
+    OUT_CS(r300->vertex_info->vinfo.hwfmt[1]);
     OUT_CS_REG_SEQ(R300_VAP_OUTPUT_VTX_FMT_0, 2);
-    OUT_CS(r300->vertex_info.vinfo.hwfmt[2]);
-    OUT_CS(r300->vertex_info.vinfo.hwfmt[3]);
+    OUT_CS(r300->vertex_info->vinfo.hwfmt[2]);
+    OUT_CS(r300->vertex_info->vinfo.hwfmt[3]);
     /* for (i = 0; i < 4; i++) {
      *    debug_printf("hwfmt%d: 0x%08x\n", i,
-     *            r300->vertex_info.vinfo.hwfmt[i]);
+     *            r300->vertex_info->vinfo.hwfmt[i]);
      * } */
 
     OUT_CS_REG_SEQ(R300_VAP_PROG_STREAM_CNTL_0, 8);
     for (i = 0; i < 8; i++) {
-        OUT_CS(r300->vertex_info.vap_prog_stream_cntl[i]);
+        OUT_CS(r300->vertex_info->vap_prog_stream_cntl[i]);
         /* debug_printf("prog_stream_cntl%d: 0x%08x\n", i,
-         *        r300->vertex_info.vap_prog_stream_cntl[i]); */
+         *        r300->vertex_info->vap_prog_stream_cntl[i]); */
     }
     OUT_CS_REG_SEQ(R300_VAP_PROG_STREAM_CNTL_EXT_0, 8);
     for (i = 0; i < 8; i++) {
-        OUT_CS(r300->vertex_info.vap_prog_stream_cntl_ext[i]);
+        OUT_CS(r300->vertex_info->vap_prog_stream_cntl_ext[i]);
         /* debug_printf("prog_stream_cntl_ext%d: 0x%08x\n", i,
-         *        r300->vertex_info.vap_prog_stream_cntl_ext[i]); */
+         *        r300->vertex_info->vap_prog_stream_cntl_ext[i]); */
     }
     END_CS;
 }
@@ -656,6 +718,9 @@ void r300_emit_dirty_state(struct r300_context* r300)
 
     r300_update_derived_state(r300);
 
+    /* Clean out BOs. */
+    r300->winsys->reset_bos(r300->winsys);
+
     /* XXX check size */
 validate:
     /* Color buffers... */
@@ -681,7 +746,8 @@ validate:
     /* ...textures... */
     for (i = 0; i < r300->texture_count; i++) {
         tex = r300->textures[i];
-        assert(tex && tex->buffer && "texture is marked, but NULL!");
+        if (!tex)
+	    continue;
         if (!r300->winsys->add_buffer(r300->winsys, tex->buffer,
                     RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM, 0)) {
             r300->context.flush(&r300->context, 0, NULL);
@@ -715,6 +781,11 @@ validate:
         goto validate;
     }
 
+    if (r300->dirty_state & R300_NEW_QUERY) {
+        r300_emit_query_start(r300);
+        r300->dirty_state &= ~R300_NEW_QUERY;
+    }
+
     if (r300->dirty_state & R300_NEW_BLEND) {
         r300_emit_blend_state(r300, r300->blend_state);
         r300->dirty_state &= ~R300_NEW_BLEND;
@@ -768,12 +839,13 @@ validate:
     if (r300->dirty_state &
             (R300_ANY_NEW_SAMPLERS | R300_ANY_NEW_TEXTURES)) {
         for (i = 0; i < MIN2(r300->sampler_count, r300->texture_count); i++) {
-            if (r300->dirty_state &
-                    ((R300_NEW_SAMPLER << i) | (R300_NEW_TEXTURE << i))) {
-                r300_emit_texture(r300,
-                        r300->sampler_states[i],
-                        r300->textures[i],
-                        i);
+  	    if (r300->dirty_state &
+		((R300_NEW_SAMPLER << i) | (R300_NEW_TEXTURE << i))) {
+		if (r300->textures[i]) 
+		    r300_emit_texture(r300,
+				      r300->sampler_states[i],
+				      r300->textures[i],
+				      i);
                 r300->dirty_state &=
                     ~((R300_NEW_SAMPLER << i) | (R300_NEW_TEXTURE << i));
                 dirty_tex++;
diff --git a/src/gallium/drivers/r300/r300_emit.h b/src/gallium/drivers/r300/r300_emit.h
index 350691d592..02ac5bebbd 100644
--- a/src/gallium/drivers/r300/r300_emit.h
+++ b/src/gallium/drivers/r300/r300_emit.h
@@ -23,13 +23,6 @@
 #ifndef R300_EMIT_H
 #define R300_EMIT_H
 
-#include "util/u_math.h"
-
-#include "r300_context.h"
-#include "r300_cs.h"
-#include "r300_screen.h"
-#include "r300_state_inlines.h"
-
 struct rX00_fragment_program_code;
 struct r300_vertex_program_code;
 
@@ -56,6 +49,11 @@ void r500_emit_fragment_program_code(struct r300_context* r300,
 void r300_emit_fb_state(struct r300_context* r300,
                         struct pipe_framebuffer_state* fb);
 
+void r300_emit_query_begin(struct r300_context* r300,
+                           struct r300_query* query);
+
+void r300_emit_query_end(struct r300_context* r300);
+
 void r300_emit_rs_state(struct r300_context* r300, struct r300_rs_state* rs);
 
 void r300_emit_rs_block_state(struct r300_context* r300,
diff --git a/src/gallium/drivers/r300/r300_flush.c b/src/gallium/drivers/r300/r300_flush.c
index 0dff1c6f4f..14a08241fc 100644
--- a/src/gallium/drivers/r300/r300_flush.c
+++ b/src/gallium/drivers/r300/r300_flush.c
@@ -20,29 +20,48 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
+#include "draw/draw_context.h"
+#include "draw/draw_private.h"
+
+#include "util/u_simple_list.h"
+
+#include "r300_context.h"
+#include "r300_cs.h"
+#include "r300_emit.h"
 #include "r300_flush.h"
+#include "r300_state_invariant.h"
 
 static void r300_flush(struct pipe_context* pipe,
                        unsigned flags,
                        struct pipe_fence_handle** fence)
 {
-    struct r300_context* r300 = r300_context(pipe);
-    CS_LOCALS(r300);
+    struct r300_context *r300 = r300_context(pipe);
+    struct r300_query *query;
 
+    CS_LOCALS(r300);
     /* We probably need to flush Draw, but we may have been called from
-     * within Draw. This feels kludgy, but it might be the best thing. */
-    if (!r300->draw->flushing) {
+     * within Draw. This feels kludgy, but it might be the best thing.
+     *
+     * Of course, the best thing is to kill Draw with fire. :3 */
+    if (r300->draw && !r300->draw->flushing) {
         draw_flush(r300->draw);
     }
 
+    r300_emit_query_end(r300);
+
     if (r300->dirty_hw) {
         FLUSH_CS;
         r300_emit_invariant_state(r300);
         r300->dirty_state = R300_NEW_KITCHEN_SINK;
         r300->dirty_hw = 0;
     }
+    /* reset flushed query */
+    foreach(query, &r300->query_list) {
+        query->flushed = TRUE;
+    }
 }
 
+
 void r300_init_flush_functions(struct r300_context* r300)
 {
     r300->context.flush = r300_flush;
diff --git a/src/gallium/drivers/r300/r300_flush.h b/src/gallium/drivers/r300/r300_flush.h
index 9a83d89daa..0e9e6106bb 100644
--- a/src/gallium/drivers/r300/r300_flush.h
+++ b/src/gallium/drivers/r300/r300_flush.h
@@ -23,13 +23,6 @@
 #ifndef R300_FLUSH_H
 #define R300_FLUSH_H
 
-#include "draw/draw_private.h"
-
-#include "pipe/p_context.h"
-
-#include "r300_context.h"
-#include "r300_cs.h"
-
 void r300_init_flush_functions(struct r300_context* r300);
 
 #endif /* R300_FLUSH_H */
diff --git a/src/gallium/drivers/r300/r300_fs.c b/src/gallium/drivers/r300/r300_fs.c
index 36463b9a2e..29ddc84c41 100644
--- a/src/gallium/drivers/r300/r300_fs.c
+++ b/src/gallium/drivers/r300/r300_fs.c
@@ -21,10 +21,14 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
-#include "r300_fs.h"
+#include "tgsi/tgsi_dump.h"
 
+#include "r300_context.h"
+#include "r300_screen.h"
+#include "r300_fs.h"
 #include "r300_tgsi_to_rc.h"
 
+#include "radeon_code.h"
 #include "radeon_compiler.h"
 
 static void find_output_registers(struct r300_fragment_program_compiler * compiler,
@@ -96,7 +100,7 @@ void r300_translate_fragment_shader(struct r300_context* r300,
 
     memset(&compiler, 0, sizeof(compiler));
     rc_init(&compiler.Base);
-    compiler.Base.Debug = 1;
+    compiler.Base.Debug = DBG_ON(r300, DBG_FP);
 
     compiler.code = &fs->code;
     compiler.is_r500 = r300_screen(r300->context.screen)->caps->is_r500;
@@ -126,9 +130,9 @@ void r300_translate_fragment_shader(struct r300_context* r300,
     /* Invoke the compiler */
     r3xx_compile_fragment_program(&compiler);
     if (compiler.Base.Error) {
-        /* Todo: Fail gracefully */
-        fprintf(stderr, "r300 FP: Compiler error\n");
-        abort();
+        /* XXX failover maybe? */
+        DBG(r300, DBG_FP, "r300: Error compiling fragment program: %s\n",
+            compiler.Base.ErrorMsg);
     }
 
     /* And, finally... */
diff --git a/src/gallium/drivers/r300/r300_fs.h b/src/gallium/drivers/r300/r300_fs.h
index 9fab789402..e831c30301 100644
--- a/src/gallium/drivers/r300/r300_fs.h
+++ b/src/gallium/drivers/r300/r300_fs.h
@@ -24,11 +24,9 @@
 #ifndef R300_FS_H
 #define R300_FS_H
 
-#include "tgsi/tgsi_dump.h"
+#include "pipe/p_state.h"
 
-#include "r300_context.h"
-#include "r3xx_fs.h"
-#include "r5xx_fs.h"
+#include "tgsi/tgsi_scan.h"
 
 #include "radeon_code.h"
 
@@ -48,4 +46,10 @@ struct r300_fragment_shader {
 void r300_translate_fragment_shader(struct r300_context* r300,
                                     struct r300_fragment_shader* fs);
 
-    #endif /* R300_FS_H */
+static inline boolean r300_fragment_shader_writes_depth(struct r300_fragment_shader *fs)
+{
+    if (!fs)
+	return FALSE;
+    return (fs->code.writes_depth) ? TRUE : FALSE;
+}
+#endif /* R300_FS_H */
diff --git a/src/gallium/drivers/r300/r300_query.c b/src/gallium/drivers/r300/r300_query.c
index 1d5185b417..007f11efae 100644
--- a/src/gallium/drivers/r300/r300_query.c
+++ b/src/gallium/drivers/r300/r300_query.c
@@ -20,15 +20,23 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
+#include "util/u_memory.h"
+#include "util/u_simple_list.h"
+
+#include "r300_context.h"
+#include "r300_screen.h"
+#include "r300_cs.h"
+#include "r300_emit.h"
 #include "r300_query.h"
+#include "r300_reg.h"
 
-static struct pipe_query* r300_create_query(struct pipe_context* pipe,
+static struct pipe_query *r300_create_query(struct pipe_context *pipe,
                                             unsigned query_type)
 {
-    struct r300_context* r300 = r300_context(pipe);
-    struct r300_screen* r300screen = r300_screen(r300->context.screen);
-    unsigned query_size = r300screen->caps->num_frag_pipes * 4;
-    struct r300_query* q, * qptr;
+    struct r300_context *r300 = r300_context(pipe);
+    struct r300_screen *r300screen = r300_screen(r300->context.screen);
+    unsigned query_size;
+    struct r300_query *q, *qptr;
 
     q = CALLOC_STRUCT(r300_query);
 
@@ -37,13 +45,16 @@ static struct pipe_query* r300_create_query(struct pipe_context* pipe,
 
     q->active = FALSE;
 
-    if (!r300->query_list) {
-        r300->query_list = q;
-    } else if (!is_empty_list(r300->query_list)) {
-        qptr = last_elem(r300->query_list);
+    if (r300screen->caps->family == CHIP_FAMILY_RV530)
+	query_size = r300screen->caps->num_z_pipes * sizeof(uint32_t);
+    else
+	query_size = r300screen->caps->num_frag_pipes * sizeof(uint32_t);
+
+    if (!is_empty_list(&r300->query_list)) {
+        qptr = last_elem(&r300->query_list);
         q->offset = qptr->offset + query_size;
-        insert_at_tail(r300->query_list, q);
     }
+    insert_at_tail(&r300->query_list, q);
 
     /* XXX */
     if (q->offset >= 4096) {
@@ -69,24 +80,26 @@ static void r300_begin_query(struct pipe_context* pipe,
     struct r300_context* r300 = r300_context(pipe);
     struct r300_query* q = (struct r300_query*)query;
 
+    assert(r300->query_current == NULL);
+
     map = pipe->screen->buffer_map(pipe->screen, r300->oqbo,
             PIPE_BUFFER_USAGE_CPU_WRITE);
     map += q->offset / 4;
-    *map = ~0;
+    *map = ~0U;
     pipe->screen->buffer_unmap(pipe->screen, r300->oqbo);
 
-    r300_emit_dirty_state(r300);
-    r300_emit_query_begin(r300, q);
+    q->flushed = FALSE;
+    r300->query_current = q;
+    r300->dirty_state |= R300_NEW_QUERY;
 }
 
 static void r300_end_query(struct pipe_context* pipe,
-                           struct pipe_query* query)
+	                   struct pipe_query* query)
 {
     struct r300_context* r300 = r300_context(pipe);
-    struct r300_query* q = (struct r300_query*)query;
 
-    r300_emit_dirty_state(r300);
-    r300_emit_query_end(r300, q);
+    r300_emit_query_end(r300);
+    r300->query_current = NULL;
 }
 
 static boolean r300_get_query_result(struct pipe_context* pipe,
@@ -96,28 +109,30 @@ static boolean r300_get_query_result(struct pipe_context* pipe,
 {
     struct r300_context* r300 = r300_context(pipe);
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
-    struct r300_query* q = (struct r300_query*)query;
+    struct r300_query *q = (struct r300_query*)query;
     unsigned flags = PIPE_BUFFER_USAGE_CPU_READ;
     uint32_t* map;
-    uint32_t temp;
+    uint32_t temp = 0;
     unsigned i;
 
-    if (wait) {
+    if (q->flushed == FALSE)
         pipe->flush(pipe, 0, NULL);
-    } else {
+    if (!wait) {
         flags |= PIPE_BUFFER_USAGE_DONTBLOCK;
     }
 
     map = pipe->screen->buffer_map(pipe->screen, r300->oqbo, flags);
+    if (!map)
+        return FALSE;
     map += q->offset / 4;
     for (i = 0; i < r300screen->caps->num_frag_pipes; i++) {
-        if (*map == ~0) {
+        if (*map == ~0U) {
             /* Looks like our results aren't ready yet. */
             if (wait) {
                 debug_printf("r300: Despite waiting, OQ results haven't"
                         " come in yet.\n");
             }
-            temp = ~0;
+            temp = ~0U;
             break;
         }
         temp += *map;
@@ -125,7 +140,7 @@ static boolean r300_get_query_result(struct pipe_context* pipe,
     }
     pipe->screen->buffer_unmap(pipe->screen, r300->oqbo);
 
-    if (temp == ~0) {
+    if (temp == ~0U) {
         /* Our results haven't been written yet... */
         return FALSE;
     }
diff --git a/src/gallium/drivers/r300/r300_query.h b/src/gallium/drivers/r300/r300_query.h
index 4f50e8f844..48876da312 100644
--- a/src/gallium/drivers/r300/r300_query.h
+++ b/src/gallium/drivers/r300/r300_query.h
@@ -23,10 +23,6 @@
 #ifndef R300_QUERY_H
 #define R300_QUERY_H
 
-#include "r300_context.h"
-#include "r300_cs.h"
-#include "r300_reg.h"
-
 struct r300_context;
 
 static INLINE struct r300_query* r300_query(struct pipe_query* q)
diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h
index 03cd219cde..e920b2a5e7 100644
--- a/src/gallium/drivers/r300/r300_reg.h
+++ b/src/gallium/drivers/r300/r300_reg.h
@@ -348,6 +348,27 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_WRITE_ENA_W                         8
 #       define R300_SWIZZLE1_SHIFT                      16
 
+#       define R300_VAP_SWIZZLE_X001 \
+        ((R300_SWIZZLE_SELECT_X << R300_SWIZZLE_SELECT_X_SHIFT) | \
+         (R300_SWIZZLE_SELECT_FP_ZERO << R300_SWIZZLE_SELECT_Y_SHIFT) | \
+         (R300_SWIZZLE_SELECT_FP_ZERO << R300_SWIZZLE_SELECT_Z_SHIFT) | \
+         (R300_SWIZZLE_SELECT_FP_ONE << R300_SWIZZLE_SELECT_W_SHIFT) | \
+         (0xf << R300_WRITE_ENA_SHIFT))
+
+#       define R300_VAP_SWIZZLE_XY01 \
+        ((R300_SWIZZLE_SELECT_X << R300_SWIZZLE_SELECT_X_SHIFT) | \
+         (R300_SWIZZLE_SELECT_Y << R300_SWIZZLE_SELECT_Y_SHIFT) | \
+         (R300_SWIZZLE_SELECT_FP_ZERO << R300_SWIZZLE_SELECT_Z_SHIFT) | \
+         (R300_SWIZZLE_SELECT_FP_ONE << R300_SWIZZLE_SELECT_W_SHIFT) | \
+         (0xf << R300_WRITE_ENA_SHIFT))
+
+#       define R300_VAP_SWIZZLE_XYZ1 \
+        ((R300_SWIZZLE_SELECT_X << R300_SWIZZLE_SELECT_X_SHIFT) | \
+         (R300_SWIZZLE_SELECT_Y << R300_SWIZZLE_SELECT_Y_SHIFT) | \
+         (R300_SWIZZLE_SELECT_Z << R300_SWIZZLE_SELECT_Z_SHIFT) | \
+         (R300_SWIZZLE_SELECT_FP_ONE << R300_SWIZZLE_SELECT_W_SHIFT) | \
+         (0xf << R300_WRITE_ENA_SHIFT))
+
 #       define R300_VAP_SWIZZLE_XYZW \
         ((R300_SWIZZLE_SELECT_X << R300_SWIZZLE_SELECT_X_SHIFT) | \
          (R300_SWIZZLE_SELECT_Y << R300_SWIZZLE_SELECT_Y_SHIFT) | \
@@ -1172,6 +1193,13 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 /* SU Depth Offset value */
 #define R300_SU_DEPTH_OFFSET                0x42c4
 
+#define R300_SU_REG_DEST		    0x42c8
+#	define R300_RASTER_PIPE_SELECT_0	(1 << 0)
+#	define R300_RASTER_PIPE_SELECT_1	(1 << 1)
+#	define R300_RASTER_PIPE_SELECT_2	(1 << 2)
+#	define R300_RASTER_PIPE_SELECT_3	(1 << 3)
+#	define R300_RASTER_PIPE_SELECT_ALL	0xf
+
 
 /* BEGIN: Rasterization / Interpolators - many guesses */
 
@@ -1478,6 +1506,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_TX_PITCH_EN                  (1 << 31)
 #       define R300_TX_WIDTH(x)                  ((x) << 0)
 #       define R300_TX_HEIGHT(x)                 ((x) << 11)
+#       define R300_TX_DEPTH(x)                  ((x) << 22)
 #       define R300_TX_NUM_LEVELS(x)             ((x) << 26)
 
 #define R300_TX_FORMAT1_0                   0x44C0
@@ -2094,6 +2123,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R500_FG_ALPHA_VALUE                0x4be0
 #	define R500_FG_ALPHA_VALUE_MASK 0x0000ffff
 
+#define RV530_FG_ZBREG_DEST                 0x4be8
+#	define RV530_FG_ZBREG_DEST_PIPE_SELECT_0             (1 << 0)
+#	define RV530_FG_ZBREG_DEST_PIPE_SELECT_1             (1 << 1)
+#	define RV530_FG_ZBREG_DEST_PIPE_SELECT_ALL           (3 << 0)
 /* gap */
 
 /* Fragment program parameters in 7.16 floating point */
@@ -3312,10 +3345,6 @@ enum {
 
 #define R200_3D_DRAW_IMMD_2      0xC0003500
 
-/* XXX Oh look, stuff not brought over from docs yet */
-
-#define R300_SU_REG_DEST                    0x42C8
-
 #endif /* _R300_REG_H */
 
 /* *INDENT-ON* */
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index cd458d019a..6e2bcc62da 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -20,21 +20,218 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
-#include "draw/draw_pipe.h"
+#include "draw/draw_context.h"
 #include "draw/draw_vbuf.h"
+
+#include "pipe/p_inlines.h"
+
 #include "util/u_memory.h"
 
 #include "r300_cs.h"
 #include "r300_context.h"
+#include "r300_emit.h"
 #include "r300_reg.h"
+#include "r300_render.h"
 #include "r300_state_derived.h"
 
 /* r300_render: Vertex and index buffer primitive emission. */
+#define R300_MAX_VBO_SIZE  (1024 * 1024)
+
+static uint32_t r300_translate_primitive(unsigned prim)
+{
+    switch (prim) {
+        case PIPE_PRIM_POINTS:
+            return R300_VAP_VF_CNTL__PRIM_POINTS;
+        case PIPE_PRIM_LINES:
+            return R300_VAP_VF_CNTL__PRIM_LINES;
+        case PIPE_PRIM_LINE_LOOP:
+            return R300_VAP_VF_CNTL__PRIM_LINE_LOOP;
+        case PIPE_PRIM_LINE_STRIP:
+            return R300_VAP_VF_CNTL__PRIM_LINE_STRIP;
+        case PIPE_PRIM_TRIANGLES:
+            return R300_VAP_VF_CNTL__PRIM_TRIANGLES;
+        case PIPE_PRIM_TRIANGLE_STRIP:
+            return R300_VAP_VF_CNTL__PRIM_TRIANGLE_STRIP;
+        case PIPE_PRIM_TRIANGLE_FAN:
+            return R300_VAP_VF_CNTL__PRIM_TRIANGLE_FAN;
+        case PIPE_PRIM_QUADS:
+            return R300_VAP_VF_CNTL__PRIM_QUADS;
+        case PIPE_PRIM_QUAD_STRIP:
+            return R300_VAP_VF_CNTL__PRIM_QUAD_STRIP;
+        case PIPE_PRIM_POLYGON:
+            return R300_VAP_VF_CNTL__PRIM_POLYGON;
+        default:
+            return 0;
+    }
+}
+
+/* This is the fast-path drawing & emission for HW TCL. */
+boolean r300_draw_range_elements(struct pipe_context* pipe,
+                                 struct pipe_buffer* indexBuffer,
+                                 unsigned indexSize,
+                                 unsigned minIndex,
+                                 unsigned maxIndex,
+                                 unsigned mode,
+                                 unsigned start,
+                                 unsigned count)
+{
+    struct r300_context* r300 = r300_context(pipe);
+    CS_LOCALS(r300);
+    uint32_t prim = r300_translate_primitive(mode);
+    struct pipe_vertex_buffer* aos = r300->vertex_buffers;
+    unsigned aos_count = r300->vertex_buffer_count;
+    short* indices;
+    unsigned packet_size;
+    unsigned i;
+    bool invalid = FALSE;
+
+validate:
+    for (i = 0; i < aos_count; i++) {
+        if (!r300->winsys->add_buffer(r300->winsys, aos[i].buffer,
+                    RADEON_GEM_DOMAIN_GTT, 0)) {
+            pipe->flush(pipe, 0, NULL);
+            goto validate;
+        }
+    }
+    if (!r300->winsys->validate(r300->winsys)) {
+        pipe->flush(pipe, 0, NULL);
+        if (invalid) {
+            /* Well, hell. */
+            debug_printf("r300: Stuck in validation loop, gonna quit now.");
+            exit(1);
+        }
+        invalid = TRUE;
+        goto validate;
+    }
+
+    r300_emit_dirty_state(r300);
+
+    packet_size = (aos_count >> 1) * 3 + (aos_count & 1) * 2;
+
+    BEGIN_CS(3 + packet_size + (aos_count * 2));
+    OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, packet_size);
+    OUT_CS(aos_count);
+    for (i = 0; i < aos_count - 1; i += 2) {
+        OUT_CS(aos[i].stride |
+            (aos[i].stride << 8) |
+            (aos[i + 1].stride << 16) |
+            (aos[i + 1].stride << 24));
+        OUT_CS(aos[i].buffer_offset + start * 4 * aos[i].stride);
+        OUT_CS(aos[i + 1].buffer_offset + start * 4 * aos[i + 1].stride);
+    }
+    if (aos_count & 1) {
+        OUT_CS(aos[i].stride | (aos[i].stride << 8));
+        OUT_CS(aos[i].buffer_offset + start * 4 * aos[i].stride);
+    }
+    for (i = 0; i < aos_count; i++) {
+        OUT_CS_RELOC(aos[i].buffer, 0, RADEON_GEM_DOMAIN_GTT, 0, 0);
+    }
+    END_CS;
+
+    if (indexBuffer) {
+        indices = (short*)pipe_buffer_map(pipe->screen, indexBuffer,
+                                          PIPE_BUFFER_USAGE_CPU_READ);
+
+        /* Set the starting point. */
+        indices += start;
+
+        BEGIN_CS(2 + (count+1)/2);
+        OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, (count + 1)/2);
+        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) | prim);
+        for (i = 0; i < count - 1; i += 2) {
+            OUT_CS(indices[i + 1] << 16 | indices[i]);
+        }
+        if (count % 2) {
+            OUT_CS(indices[count - 1]);
+        }
+        END_CS;
+    } else {
+        BEGIN_CS(2);
+        OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0);
+        OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) |
+               prim);
+        END_CS;
+    }
+
+    return TRUE;
+}
+
+/* Simple helpers for context setup. Should probably be moved to util. */
+boolean r300_draw_elements(struct pipe_context* pipe,
+                           struct pipe_buffer* indexBuffer,
+                           unsigned indexSize, unsigned mode,
+                           unsigned start, unsigned count)
+{
+    return pipe->draw_range_elements(pipe, indexBuffer, indexSize, 0, ~0,
+                                     mode, start, count);
+}
+
+boolean r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
+                         unsigned start, unsigned count)
+{
+    return pipe->draw_elements(pipe, NULL, 0, mode, start, count);
+}
+
+/****************************************************************************
+ * The rest of this file is for SW TCL rendering only. Please be polite and *
+ * keep these functions separated so that they are easier to locate. ~C.    *
+ ***************************************************************************/
+
+/* Draw-based drawing for SW TCL chipsets. */
+boolean r300_swtcl_draw_range_elements(struct pipe_context* pipe,
+                                       struct pipe_buffer* indexBuffer,
+                                       unsigned indexSize,
+                                       unsigned minIndex,
+                                       unsigned maxIndex,
+                                       unsigned mode,
+                                       unsigned start,
+                                       unsigned count)
+{
+    struct r300_context* r300 = r300_context(pipe);
+    int i;
+
+    for (i = 0; i < r300->vertex_buffer_count; i++) {
+        void* buf = pipe_buffer_map(pipe->screen,
+                                    r300->vertex_buffers[i].buffer,
+                                    PIPE_BUFFER_USAGE_CPU_READ);
+        draw_set_mapped_vertex_buffer(r300->draw, i, buf);
+    }
+
+    if (indexBuffer) {
+        void* indices = pipe_buffer_map(pipe->screen, indexBuffer,
+                                        PIPE_BUFFER_USAGE_CPU_READ);
+        draw_set_mapped_element_buffer_range(r300->draw, indexSize,
+                                             minIndex, maxIndex, indices);
+    } else {
+        draw_set_mapped_element_buffer(r300->draw, 0, NULL);
+    }
+
+    draw_set_mapped_constant_buffer(r300->draw,
+            r300->shader_constants[PIPE_SHADER_VERTEX].constants,
+            r300->shader_constants[PIPE_SHADER_VERTEX].count *
+                (sizeof(float) * 4));
 
+    draw_arrays(r300->draw, mode, start, count);
+
+    for (i = 0; i < r300->vertex_buffer_count; i++) {
+        pipe_buffer_unmap(pipe->screen, r300->vertex_buffers[i].buffer);
+        draw_set_mapped_vertex_buffer(r300->draw, i, NULL);
+    }
+
+    if (indexBuffer) {
+        pipe_buffer_unmap(pipe->screen, indexBuffer);
+        draw_set_mapped_element_buffer_range(r300->draw, 0, start,
+                                             start + count - 1, NULL);
+    }
+
+    return TRUE;
+}
+
+/* Object for rendering using Draw. */
 struct r300_render {
     /* Parent class */
     struct vbuf_render base;
-    
+
     /* Pipe context */
     struct r300_context* r300;
 
@@ -45,7 +242,10 @@ struct r300_render {
 
     /* VBO */
     struct pipe_buffer* vbo;
-    size_t vbo_alloc_size;
+    size_t vbo_size;
+    size_t vbo_offset;
+    size_t vbo_max_used;
+    void * vbo_ptr;
 };
 
 static INLINE struct r300_render*
@@ -62,7 +262,7 @@ r300_render_get_vertex_info(struct vbuf_render* render)
 
     r300_update_derived_state(r300);
 
-    return &r300->vertex_info.vinfo;
+    return &r300->vertex_info->vinfo;
 }
 
 static boolean r300_render_allocate_vertices(struct vbuf_render* render,
@@ -74,19 +274,20 @@ static boolean r300_render_allocate_vertices(struct vbuf_render* render,
     struct pipe_screen* screen = r300->context.screen;
     size_t size = (size_t)vertex_size * (size_t)count;
 
-    if (r300render->vbo && (size > r300render->vbo_alloc_size)) {
-        pipe_buffer_reference(&r300render->vbo, NULL);
-    }
-    
-    if (!r300render->vbo) {
+    if (size + r300render->vbo_offset > r300render->vbo_size)
+    {
+        pipe_buffer_reference(&r300->vbo, NULL);
         r300render->vbo = pipe_buffer_create(screen,
                                              64,
                                              PIPE_BUFFER_USAGE_VERTEX,
-                                             size);
+                                             R300_MAX_VBO_SIZE);
+        r300render->vbo_offset = 0;
+        r300render->vbo_size = R300_MAX_VBO_SIZE;
     }
 
-    r300render->vbo_alloc_size = MAX2(size, r300render->vbo_alloc_size);
     r300render->vertex_size = vertex_size;
+    r300->vbo = r300render->vbo;
+    r300->vbo_offset = r300render->vbo_offset;
 
     return (r300render->vbo) ? TRUE : FALSE;
 }
@@ -96,8 +297,10 @@ static void* r300_render_map_vertices(struct vbuf_render* render)
     struct r300_render* r300render = r300_render(render);
     struct pipe_screen* screen = r300render->r300->context.screen;
 
-    return (unsigned char*)pipe_buffer_map(screen, r300render->vbo,
-                                           PIPE_BUFFER_USAGE_CPU_WRITE);
+    r300render->vbo_ptr = pipe_buffer_map(screen, r300render->vbo,
+                                          PIPE_BUFFER_USAGE_CPU_WRITE);
+
+    return (r300render->vbo_ptr + r300render->vbo_offset);
 }
 
 static void r300_render_unmap_vertices(struct vbuf_render* render,
@@ -106,7 +309,13 @@ static void r300_render_unmap_vertices(struct vbuf_render* render,
 {
     struct r300_render* r300render = r300_render(render);
     struct pipe_screen* screen = r300render->r300->context.screen;
+    CS_LOCALS(r300render->r300);
+    BEGIN_CS(2);
+    OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, max);
+    END_CS;
 
+    r300render->vbo_max_used = MAX2(r300render->vbo_max_used,
+                                    r300render->vertex_size * (max + 1));
     pipe_buffer_unmap(screen, r300render->vbo);
 }
 
@@ -114,65 +323,21 @@ static void r300_render_release_vertices(struct vbuf_render* render)
 {
     struct r300_render* r300render = r300_render(render);
 
-    pipe_buffer_reference(&r300render->vbo, NULL);
+    r300render->vbo_offset += r300render->vbo_max_used;
+    r300render->vbo_max_used = 0;
 }
 
 static boolean r300_render_set_primitive(struct vbuf_render* render,
                                                unsigned prim)
 {
     struct r300_render* r300render = r300_render(render);
-    r300render->prim = prim;
 
-    switch (prim) {
-        case PIPE_PRIM_POINTS:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_POINTS;
-            break;
-        case PIPE_PRIM_LINES:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_LINES;
-            break;
-        case PIPE_PRIM_LINE_LOOP:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_LINE_LOOP;
-            break;
-        case PIPE_PRIM_LINE_STRIP:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_LINE_STRIP;
-            break;
-        case PIPE_PRIM_TRIANGLES:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_TRIANGLES;
-            break;
-        case PIPE_PRIM_TRIANGLE_STRIP:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_TRIANGLE_STRIP;
-            break;
-        case PIPE_PRIM_TRIANGLE_FAN:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_TRIANGLE_FAN;
-            break;
-        case PIPE_PRIM_QUADS:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_QUADS;
-            break;
-        case PIPE_PRIM_QUAD_STRIP:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_QUAD_STRIP;
-            break;
-        case PIPE_PRIM_POLYGON:
-            r300render->hwprim = R300_VAP_VF_CNTL__PRIM_POLYGON;
-            break;
-        default:
-            return FALSE;
-            break;
-    }
+    r300render->prim = prim;
+    r300render->hwprim = r300_translate_primitive(prim);
 
     return TRUE;
 }
 
-static void prepare_render(struct r300_render* render, unsigned count)
-{
-    struct r300_context* r300 = render->r300;
-
-    CS_LOCALS(r300);
-
-    r300->vbo = render->vbo;
-
-    r300_emit_dirty_state(r300);
-}
-
 static void r300_render_draw_arrays(struct vbuf_render* render,
                                           unsigned start,
                                           unsigned count)
@@ -182,9 +347,9 @@ static void r300_render_draw_arrays(struct vbuf_render* render,
 
     CS_LOCALS(r300);
 
-    prepare_render(r300render, count);
+    r300_emit_dirty_state(r300);
 
-    debug_printf("r300: Doing vbuf render, count %d\n", count);
+    DBG(r300, DBG_DRAW, "r300: Doing vbuf render, count %d\n", count);
 
     BEGIN_CS(2);
     OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0);
@@ -199,39 +364,11 @@ static void r300_render_draw(struct vbuf_render* render,
 {
     struct r300_render* r300render = r300_render(render);
     struct r300_context* r300 = r300render->r300;
-    struct pipe_screen* screen = r300->context.screen;
-    struct pipe_buffer* index_buffer;
-    void* index_map;
     int i;
-    uint32_t index;
 
     CS_LOCALS(r300);
 
-    prepare_render(r300render, count);
-
-    /* Send our indices into an index buffer. */
-    index_buffer = pipe_buffer_create(screen, 64, PIPE_BUFFER_USAGE_VERTEX,
-                                      count * 2);
-    if (!index_buffer) {
-        return;
-    }
-
-/*
-    index_map = pipe_buffer_map(screen, index_buffer,
-                                PIPE_BUFFER_USAGE_CPU_WRITE);
-    memcpy(index_map, indices, count);
-    pipe_buffer_unmap(screen, index_buffer);
-
-    debug_printf("r300: Doing indexbuf render, count %d\n", count);
-
-    BEGIN_CS(8);
-    OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, 0);
-    OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) |
-           r300render->hwprim);
-    OUT_CS_PKT3(R300_PACKET3_INDX_BUFFER, 2);
-    OUT_CS(R300_INDX_BUFFER_ONE_REG_WR | (R300_VAP_PORT_IDX0 >> 2));
-    OUT_CS_INDEX_RELOC(index_buffer, 0, count, RADEON_GEM_DOMAIN_GTT, 0, 0);
-    END_CS; */
+    r300_emit_dirty_state(r300);
 
     BEGIN_CS(2 + (count+1)/2);
     OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, (count+1)/2);
@@ -271,6 +408,10 @@ static struct vbuf_render* r300_render_create(struct r300_context* r300)
     r300render->base.release_vertices = r300_render_release_vertices;
     r300render->base.destroy = r300_render_destroy;
 
+    r300render->vbo = NULL;
+    r300render->vbo_size = 0;
+    r300render->vbo_offset = 0;
+
     return &r300render->base;
 }
 
diff --git a/src/gallium/drivers/r300/r300_render.h b/src/gallium/drivers/r300/r300_render.h
new file mode 100644
index 0000000000..3d8f47ba75
--- /dev/null
+++ b/src/gallium/drivers/r300/r300_render.h
@@ -0,0 +1,52 @@
+/*
+ * Copyright 2009 Corbin Simpson <MostAwesomeDude@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE. */
+
+#ifndef R300_RENDER_H
+#define R300_RENDER_H
+
+boolean r300_draw_range_elements(struct pipe_context* pipe,
+                                 struct pipe_buffer* indexBuffer,
+                                 unsigned indexSize,
+                                 unsigned minIndex,
+                                 unsigned maxIndex,
+                                 unsigned mode,
+                                 unsigned start,
+                                 unsigned count);
+
+boolean r300_draw_elements(struct pipe_context* pipe,
+                           struct pipe_buffer* indexBuffer,
+                           unsigned indexSize, unsigned mode,
+                           unsigned start, unsigned count);
+
+boolean r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
+                         unsigned start, unsigned count);
+
+boolean r300_swtcl_draw_range_elements(struct pipe_context* pipe,
+                                       struct pipe_buffer* indexBuffer,
+                                       unsigned indexSize,
+                                       unsigned minIndex,
+                                       unsigned maxIndex,
+                                       unsigned mode,
+                                       unsigned start,
+                                       unsigned count);
+
+#endif /* R300_RENDER_H */
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 15740f6125..1d9f91d0f7 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -20,7 +20,14 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
+#include "pipe/p_inlines.h"
+#include "util/u_memory.h"
+#include "util/u_simple_screen.h"
+
+#include "r300_context.h"
 #include "r300_screen.h"
+#include "r300_texture.h"
+#include "r300_winsys.h"
 
 /* Return the identifier behind whom the brave coders responsible for this
  * amalgamation of code, sweat, and duct tape, routinely obscure their names.
@@ -73,13 +80,10 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
     struct r300_screen* r300screen = r300_screen(pscreen);
 
     switch (param) {
-        /* XXX cases marked "IN THEORY" are possible on the hardware,
-         * but haven't been implemented yet. */
         case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
             /* XXX I'm told this goes up to 16 */
             return 8;
         case PIPE_CAP_NPOT_TEXTURES:
-            /* IN THEORY */
             return 0;
         case PIPE_CAP_TWO_SIDED_STENCIL:
             if (r300screen->caps->is_r500) {
@@ -88,26 +92,32 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
                 return 0;
             }
         case PIPE_CAP_GLSL:
-            if (r300screen->caps->is_r500) {
-                return 1;
-            } else {
-                return 0;
-            }
-        case PIPE_CAP_S3TC:
+            /* I'll be frank. This is a lie.
+             *
+             * We don't truly support GLSL on any of this driver's chipsets.
+             * To be fair, no chipset supports the full GLSL specification
+             * to the best of our knowledge, but some of the less esoteric
+             * features are still missing here.
+             *
+             * Rather than cripple ourselves intentionally, I'm going to set
+             * this flag, and as Gallium's interface continues to change, I
+             * hope that this single monolithic GLSL enable can slowly get
+             * split down into many different pieces and the state tracker
+             * will handle fallbacks transparently, like it should.
+             *
+             * ~ C.
+             */
             return 1;
         case PIPE_CAP_ANISOTROPIC_FILTER:
             return 1;
         case PIPE_CAP_POINT_SPRITE:
-            /* IN THEORY */
-            return 0;
+            return 1;
         case PIPE_CAP_MAX_RENDER_TARGETS:
             return 4;
         case PIPE_CAP_OCCLUSION_QUERY:
-            /* IN THEORY */
-            return 0;
+            return 1;
         case PIPE_CAP_TEXTURE_SHADOW_MAP:
-            /* IN THEORY */
-            return 0;
+            return 1;
         case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
             if (r300screen->caps->is_r500) {
                 /* 13 == 4096x4096 */
@@ -142,10 +152,8 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
         case PIPE_CAP_TEXTURE_MIRROR_REPEAT:
             return 1;
         case PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS:
-            /* XXX guessing (what a terrible guess) */
-            return 2;
+            return 0;
         case PIPE_CAP_TGSI_CONT_SUPPORTED:
-            /* XXX */
             return 0;
         case PIPE_CAP_BLEND_EQUATION_SEPARATE:
             return 1;
@@ -186,16 +194,19 @@ static float r300_get_paramf(struct pipe_screen* pscreen, int param)
 static boolean check_tex_2d_format(enum pipe_format format, uint32_t usage,
                                    boolean is_r500)
 {
+    uint32_t retval = 0;
+
     switch (format) {
         /* Supported formats. */
         /* Colorbuffer */
         case PIPE_FORMAT_A4R4G4B4_UNORM:
         case PIPE_FORMAT_R5G6B5_UNORM:
         case PIPE_FORMAT_A1R5G5B5_UNORM:
-            return usage &
+            retval = usage &
                 (PIPE_TEXTURE_USAGE_RENDER_TARGET |
                  PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
                  PIPE_TEXTURE_USAGE_PRIMARY);
+            break;
 
         /* Texture */
         case PIPE_FORMAT_A8R8G8B8_SRGB:
@@ -205,31 +216,35 @@ static boolean check_tex_2d_format(enum pipe_format format, uint32_t usage,
         case PIPE_FORMAT_DXT3_RGBA:
         case PIPE_FORMAT_DXT5_RGBA:
         case PIPE_FORMAT_YCBCR:
-            return usage & PIPE_TEXTURE_USAGE_SAMPLER;
+            retval = usage & PIPE_TEXTURE_USAGE_SAMPLER;
+            break;
 
         /* Colorbuffer or texture */
         case PIPE_FORMAT_A8R8G8B8_UNORM:
+        case PIPE_FORMAT_X8R8G8B8_UNORM:
         case PIPE_FORMAT_R8G8B8A8_UNORM:
+        case PIPE_FORMAT_R8G8B8X8_UNORM:
         case PIPE_FORMAT_I8_UNORM:
-            return usage &
+            retval = usage &
                 (PIPE_TEXTURE_USAGE_RENDER_TARGET |
                  PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
                  PIPE_TEXTURE_USAGE_PRIMARY |
                  PIPE_TEXTURE_USAGE_SAMPLER);
+            break;
 
-        /* Z buffer */
+        /* Z buffer or texture */
         case PIPE_FORMAT_Z16_UNORM:
-            return usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL;
-
+        case PIPE_FORMAT_Z24X8_UNORM:
         /* Z buffer with stencil or texture */
         case PIPE_FORMAT_Z24S8_UNORM:
-            return usage &
+            retval = usage &
                 (PIPE_TEXTURE_USAGE_DEPTH_STENCIL |
                  PIPE_TEXTURE_USAGE_SAMPLER);
+            break;
 
         /* Definitely unsupported formats. */
         /* Non-usable Z buffer/stencil formats. */
-        case PIPE_FORMAT_Z24X8_UNORM:
+        case PIPE_FORMAT_Z32_UNORM:
         case PIPE_FORMAT_S8Z24_UNORM:
         case PIPE_FORMAT_X8Z24_UNORM:
             debug_printf("r300: Note: Got unsupported format: %s in %s\n",
@@ -239,7 +254,6 @@ static boolean check_tex_2d_format(enum pipe_format format, uint32_t usage,
         /* XXX These don't even exist
         case PIPE_FORMAT_A32R32G32B32:
         case PIPE_FORMAT_A16R16G16B16: */
-        /* XXX Insert YUV422 packed VYUY and YVYU here */
         /* XXX What the deuce is UV88? (r3xx accel page 14)
             debug_printf("r300: Warning: Got unimplemented format: %s in %s\n",
                 pf_name(format), __FUNCTION__);
@@ -263,7 +277,13 @@ static boolean check_tex_2d_format(enum pipe_format format, uint32_t usage,
             break;
     }
 
-    return FALSE;
+    /* If usage was a mask that contained multiple bits, and not all of them
+     * are supported, this will catch that and return FALSE.
+     * e.g. usage = 2 | 4; retval = 4; (retval >= usage) == FALSE
+     *
+     * This also returns FALSE for any unknown formats.
+     */
+    return (retval >= usage);
 }
 
 /* XXX moar targets */
@@ -302,7 +322,7 @@ r300_get_tex_transfer(struct pipe_screen *screen,
 {
     struct r300_texture *tex = (struct r300_texture *)texture;
     struct r300_transfer *trans;
-    unsigned offset;  /* in bytes */
+    unsigned offset = 0;  /* in bytes */
 
     /* XXX Add support for these things */
     if (texture->target == PIPE_TEXTURE_CUBE) {
@@ -323,13 +343,14 @@ r300_get_tex_transfer(struct pipe_screen *screen,
     if (trans) {
         pipe_texture_reference(&trans->transfer.texture, texture);
         trans->transfer.format = texture->format;
+        trans->transfer.x = x;
+        trans->transfer.y = y;
         trans->transfer.width = w;
         trans->transfer.height = h;
         trans->transfer.block = texture->block;
         trans->transfer.nblocksx = texture->nblocksx[level];
         trans->transfer.nblocksy = texture->nblocksy[level];
-        trans->transfer.stride = align(pf_get_stride(&trans->transfer.block,
-                                                     texture->width[level]), 32);
+        trans->transfer.stride = r300_texture_get_stride(tex, level);
         trans->transfer.usage = usage;
         trans->offset = offset;
     }
@@ -348,16 +369,9 @@ static void* r300_transfer_map(struct pipe_screen* screen,
 {
     struct r300_texture* tex = (struct r300_texture*)transfer->texture;
     char* map;
-    unsigned flags = 0;
 
-    if (transfer->usage != PIPE_TRANSFER_WRITE) {
-        flags |= PIPE_BUFFER_USAGE_CPU_READ;
-    }
-    if (transfer->usage != PIPE_TRANSFER_READ) {
-        flags |= PIPE_BUFFER_USAGE_CPU_WRITE;
-    }
-    
-    map = pipe_buffer_map(screen, tex->buffer, flags);
+    map = pipe_buffer_map(screen, tex->buffer,
+                          pipe_transfer_buffer_flags(transfer));
 
     if (!map) {
         return NULL;
@@ -393,6 +407,7 @@ struct pipe_screen* r300_create_screen(struct r300_winsys* r300_winsys)
 
     caps->pci_id = r300_winsys->pci_id;
     caps->num_frag_pipes = r300_winsys->gb_pipes;
+    caps->num_z_pipes = r300_winsys->z_pipes;
 
     r300_parse_chipset(caps);
 
diff --git a/src/gallium/drivers/r300/r300_screen.h b/src/gallium/drivers/r300/r300_screen.h
index 2a0e41fbc3..41df31f670 100644
--- a/src/gallium/drivers/r300/r300_screen.h
+++ b/src/gallium/drivers/r300/r300_screen.h
@@ -23,14 +23,9 @@
 #ifndef R300_SCREEN_H
 #define R300_SCREEN_H
 
-#include "pipe/p_inlines.h"
 #include "pipe/p_screen.h"
-#include "util/u_memory.h"
-#include "util/u_simple_screen.h"
 
 #include "r300_chipset.h"
-#include "r300_texture.h"
-#include "r300_winsys.h"
 
 struct r300_screen {
     /* Parent class */
diff --git a/src/gallium/drivers/r300/r300_shader_inlines.h b/src/gallium/drivers/r300/r300_shader_inlines.h
deleted file mode 100644
index a04f45b03e..0000000000
--- a/src/gallium/drivers/r300/r300_shader_inlines.h
+++ /dev/null
@@ -1,47 +0,0 @@
-/*
- * Copyright 2009 Corbin Simpson <MostAwesomeDude@gmail.com>
- *                Joakim Sindholt <opensource@zhasha.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#ifndef R300_SHADER_INLINES_H
-#define R300_SHADER_INLINES_H
-
-/* TGSI constants. TGSI is like XML: If it can't solve your problems, you're
- * not using enough of it. */
-static const struct tgsi_full_src_register r300_constant_zero = {
-    .SrcRegister.Extended = TRUE,
-    .SrcRegister.File = TGSI_FILE_NULL,
-    .SrcRegisterExtSwz.ExtSwizzleX = TGSI_EXTSWIZZLE_ZERO,
-    .SrcRegisterExtSwz.ExtSwizzleY = TGSI_EXTSWIZZLE_ZERO,
-    .SrcRegisterExtSwz.ExtSwizzleZ = TGSI_EXTSWIZZLE_ZERO,
-    .SrcRegisterExtSwz.ExtSwizzleW = TGSI_EXTSWIZZLE_ZERO,
-};
-
-static const struct tgsi_full_src_register r300_constant_one = {
-    .SrcRegister.Extended = TRUE,
-    .SrcRegister.File = TGSI_FILE_NULL,
-    .SrcRegisterExtSwz.ExtSwizzleX = TGSI_EXTSWIZZLE_ONE,
-    .SrcRegisterExtSwz.ExtSwizzleY = TGSI_EXTSWIZZLE_ONE,
-    .SrcRegisterExtSwz.ExtSwizzleZ = TGSI_EXTSWIZZLE_ONE,
-    .SrcRegisterExtSwz.ExtSwizzleW = TGSI_EXTSWIZZLE_ONE,
-};
-
-#endif /* R300_SHADER_INLINES_H */
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index c16cadd040..a3e1bc621a 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -20,16 +20,20 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
+#include "draw/draw_context.h"
+
 #include "util/u_math.h"
+#include "util/u_memory.h"
 #include "util/u_pack_color.h"
 
-#include "util/u_debug.h"
+#include "tgsi/tgsi_parse.h"
 
 #include "pipe/p_config.h"
 #include "pipe/internal/p_winsys_screen.h"
 
 #include "r300_context.h"
 #include "r300_reg.h"
+#include "r300_screen.h"
 #include "r300_state_inlines.h"
 #include "r300_fs.h"
 #include "r300_vs.h"
@@ -45,23 +49,46 @@ static void* r300_create_blend_state(struct pipe_context* pipe,
 {
     struct r300_blend_state* blend = CALLOC_STRUCT(r300_blend_state);
 
+    {
+	unsigned eqRGB = state->rgb_func;
+	unsigned srcRGB = state->rgb_src_factor;
+	unsigned dstRGB = state->rgb_dst_factor;
+
+	unsigned eqA = state->alpha_func;
+	unsigned srcA = state->alpha_src_factor;
+	unsigned dstA = state->alpha_dst_factor;
+
+	if (srcA != srcRGB ||
+	    dstA != dstRGB ||
+	    eqA != eqRGB) {
+	    blend->alpha_blend_control =
+		r300_translate_blend_function(eqA) |
+		(r300_translate_blend_factor(srcA) <<
+                    R300_SRC_BLEND_SHIFT) |
+                (r300_translate_blend_factor(dstA) <<
+		 R300_DST_BLEND_SHIFT);
+	    blend->blend_control |= R300_ALPHA_BLEND_ENABLE |
+		R300_SEPARATE_ALPHA_ENABLE;
+	} else {
+	    blend->alpha_blend_control = R300_COMB_FCN_ADD_CLAMP |
+		(R300_BLEND_GL_ONE << R300_SRC_BLEND_SHIFT) |
+		(R300_BLEND_GL_ZERO << R300_DST_BLEND_SHIFT);
+	}
+    }
     if (state->blend_enable) {
         /* XXX for now, always do separate alpha...
          * is it faster to do it with one reg? */
-        blend->blend_control = R300_ALPHA_BLEND_ENABLE |
-                R300_SEPARATE_ALPHA_ENABLE |
-                R300_READ_ENABLE |
+        blend->blend_control |= R300_READ_ENABLE |
                 r300_translate_blend_function(state->rgb_func) |
                 (r300_translate_blend_factor(state->rgb_src_factor) <<
                     R300_SRC_BLEND_SHIFT) |
                 (r300_translate_blend_factor(state->rgb_dst_factor) <<
                     R300_DST_BLEND_SHIFT);
-        blend->alpha_blend_control =
-                r300_translate_blend_function(state->alpha_func) |
-                (r300_translate_blend_factor(state->alpha_src_factor) <<
-                    R300_SRC_BLEND_SHIFT) |
-                (r300_translate_blend_factor(state->alpha_dst_factor) <<
-                    R300_DST_BLEND_SHIFT);
+    } else {
+	blend->blend_control = 
+	    R300_COMB_FCN_ADD_CLAMP |
+	    (R300_BLEND_GL_ONE << R300_SRC_BLEND_SHIFT) |
+	    (R300_BLEND_GL_ZERO << R300_DST_BLEND_SHIFT);
     }
 
     /* PIPE_LOGICOP_* don't need to be translated, fortunately. */
@@ -223,9 +250,6 @@ static void*
             R300_FG_ALPHA_FUNC_ENABLE;
         dsa->alpha_reference = CLAMP(state->alpha.ref_value * 1023.0f,
                                      0, 1023);
-    } else {
-        /* XXX need to fix this to be dynamically set
-        dsa->z_buffer_top = R300_ZTOP_ENABLE; */
     }
 
     return (void*)dsa;
@@ -308,7 +332,7 @@ static void r300_delete_fs_state(struct pipe_context* pipe, void* shader)
 {
     struct r300_fragment_shader* fs = (struct r300_fragment_shader*)shader;
     rc_constants_destroy(&fs->code.constants);
-    FREE(fs->state.tokens);
+    FREE((void*)fs->state.tokens);
     FREE(shader);
 }
 
@@ -429,6 +453,9 @@ static void r300_bind_rs_state(struct pipe_context* pipe, void* state)
 
     r300->rs_state = rs;
     r300->dirty_state |= R300_NEW_RASTERIZER;
+    r300->dirty_state |= R300_NEW_RS_BLOCK;
+    r300->dirty_state |= R300_NEW_SCISSOR;
+    r300->dirty_state |= R300_NEW_VIEWPORT;
 }
 
 /* Free rasterizer state. */
@@ -673,7 +700,7 @@ static void r300_delete_vs_state(struct pipe_context* pipe, void* shader)
 
         rc_constants_destroy(&vs->code.constants);
         draw_delete_vertex_shader(r300->draw, vs->draw);
-        FREE(vs->state.tokens);
+        FREE((void*)vs->state.tokens);
         FREE(shader);
     } else {
         draw_delete_vertex_shader(r300->draw,
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index c01e61a9b1..7d000e9e2d 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -20,14 +20,47 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
-#include "r300_state_derived.h"
+#include "draw/draw_context.h"
+
+#include "util/u_math.h"
+#include "util/u_memory.h"
 
+#include "r300_context.h"
 #include "r300_fs.h"
+#include "r300_screen.h"
+#include "r300_state_derived.h"
+#include "r300_state_inlines.h"
 #include "r300_vs.h"
 
 /* r300_state_derived: Various bits of state which are dependent upon
  * currently bound CSO data. */
 
+struct r300_shader_key {
+    struct r300_vertex_shader* vs;
+    struct r300_fragment_shader* fs;
+};
+
+struct r300_shader_derived_value {
+    struct r300_vertex_format* vformat;
+    struct r300_rs_block* rs_block;
+};
+
+unsigned r300_shader_key_hash(void* key) {
+    struct r300_shader_key* shader_key = (struct r300_shader_key*)key;
+    unsigned vs = (unsigned)shader_key->vs;
+    unsigned fs = (unsigned)shader_key->fs;
+
+    return (vs << 16) | (fs & 0xffff);
+}
+
+int r300_shader_key_compare(void* key1, void* key2) {
+    struct r300_shader_key* shader_key1 = (struct r300_shader_key*)key1;
+    struct r300_shader_key* shader_key2 = (struct r300_shader_key*)key2;
+
+    return (shader_key1->vs == shader_key2->vs) &&
+        (shader_key1->fs == shader_key2->fs);
+}
+
 /* Set up the vs_tab and routes. */
 static void r300_vs_tab_routes(struct r300_context* r300,
                                struct r300_vertex_format* vformat)
@@ -52,7 +85,7 @@ static void r300_vs_tab_routes(struct r300_context* r300,
     if (!r300screen->caps->has_tcl || !r300->rs_state->enable_vte)
     {
         for (i = 0; i < info->num_inputs; i++) {
-            switch (info->input_semantic_name[i]) {
+            switch (r300->vs->code.inputs[i]) {
                 case TGSI_SEMANTIC_POSITION:
                     pos = TRUE;
                     tab[i] = 0;
@@ -62,10 +95,12 @@ static void r300_vs_tab_routes(struct r300_context* r300,
                     cols++;
                     break;
                 case TGSI_SEMANTIC_PSIZE:
+                    assert(psize == FALSE);
                     psize = TRUE;
                     tab[i] = 15;
                     break;
                 case TGSI_SEMANTIC_FOG:
+                    assert(fog == FALSE);
                     fog = TRUE;
                     /* Fall through */
                 case TGSI_SEMANTIC_GENERIC:
@@ -124,7 +159,9 @@ static void r300_vs_tab_routes(struct r300_context* r300,
 
     vinfo->hwfmt[0] = 0x5555; /* XXX this is classic Mesa bonghits */
 
-    if (!pos) {
+    /* We need to add vertex position attribute only for SW TCL case,
+     * for HW TCL case it could be generated by vertex shader */
+    if (!pos && !r300screen->caps->has_tcl) {
         debug_printf("r300: Forcing vertex position attribute emit...\n");
         /* Make room for the position attribute
          * at the beginning of the tab. */
@@ -163,7 +200,7 @@ static void r300_vs_tab_routes(struct r300_context* r300,
         vinfo->hwfmt[3] |= (4 << (3 * i));
     }
 
-    for (i; i < texs; i++) {
+    for (; i < texs; i++) {
         draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE,
             draw_find_vs_output(r300->draw, TGSI_SEMANTIC_GENERIC, i));
         vinfo->hwfmt[1] |= (R300_INPUT_CNTL_TC0 << i);
@@ -187,21 +224,22 @@ static void r300_vertex_psc(struct r300_context* r300,
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
     struct vertex_info* vinfo = &vformat->vinfo;
     int* tab = vformat->vs_tab;
-    uint32_t temp;
-    int i, attrib_count;
+    uint16_t type, swizzle;
+    enum pipe_format format;
+    unsigned i, attrib_count;
 
     /* Vertex shaders have no semantics on their inputs,
      * so PSC should just route stuff based on their info,
      * and not on attrib information. */
     if (r300screen->caps->has_tcl) {
         attrib_count = r300->vs->info.num_inputs;
-        debug_printf("r300: routing %d attribs in psc for vs\n",
+        DBG(r300, DBG_DRAW, "r300: routing %d attribs in psc for vs\n",
                 attrib_count);
     } else {
         attrib_count = vinfo->num_attribs;
-        debug_printf("r300: attrib count: %d\n", attrib_count);
+        DBG(r300, DBG_DRAW, "r300: attrib count: %d\n", attrib_count);
         for (i = 0; i < attrib_count; i++) {
-            debug_printf("r300: attrib: offset %d, interp %d, size %d,"
+            DBG(r300, DBG_DRAW, "r300: attrib: offset %d, interp %d, size %d,"
                    " tab %d\n", vinfo->attrib[i].src_index,
                    vinfo->attrib[i].interp_mode, vinfo->attrib[i].emit,
                    tab[i]);
@@ -209,64 +247,43 @@ static void r300_vertex_psc(struct r300_context* r300,
     }
 
     for (i = 0; i < attrib_count; i++) {
-        /* Make sure we have a proper destination for our attribute */
+        /* Make sure we have a proper destination for our attribute. */
         assert(tab[i] != -1);
 
-        /* Add the attribute to the PSC table. */
-        temp = r300screen->caps->has_tcl ?
-            R300_DATA_TYPE_FLOAT_4 :
-            translate_vertex_data_type(vinfo->attrib[i].emit);
-        temp |= tab[i] << R300_DST_VEC_LOC_SHIFT;
+        format = draw_translate_vinfo_format(vinfo->attrib[i].emit);
+
+        /* Obtain the type of data in this attribute. */
+        type = r300_translate_vertex_data_type(format) |
+            tab[i] << R300_DST_VEC_LOC_SHIFT;
 
+        /* Obtain the swizzle for this attribute. Note that the default
+         * swizzle in the hardware is not XYZW! */
+        swizzle = r300_translate_vertex_data_swizzle(format);
+
+        /* Add the attribute to the PSC table. */
         if (i & 1) {
-            vformat->vap_prog_stream_cntl[i >> 1] &= 0x0000ffff;
-            vformat->vap_prog_stream_cntl[i >> 1] |= temp << 16;
+            vformat->vap_prog_stream_cntl[i >> 1] |= type << 16;
 
-            vformat->vap_prog_stream_cntl_ext[i >> 1] |=
-                (R300_VAP_SWIZZLE_XYZW << 16);
+            vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle << 16;
         } else {
-            vformat->vap_prog_stream_cntl[i >> 1] &= 0xffff0000;
-            vformat->vap_prog_stream_cntl[i >> 1] |= temp <<  0;
+            vformat->vap_prog_stream_cntl[i >> 1] |= type <<  0;
 
-            vformat->vap_prog_stream_cntl_ext[i >> 1] |=
-                (R300_VAP_SWIZZLE_XYZW <<  0);
+            vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle << 0;
         }
     }
 
     /* Set the last vector in the PSC. */
-    i--;
+    if (i) {
+        i -= 1;
+    }
     vformat->vap_prog_stream_cntl[i >> 1] |=
         (R300_LAST_VEC << (i & 1 ? 16 : 0));
 }
 
-/* Update the vertex format. */
-static void r300_update_vertex_format(struct r300_context* r300)
-{
-    struct r300_vertex_format vformat;
-    int i;
-
-    memset(&vformat, 0, sizeof(struct r300_vertex_format));
-    for (i = 0; i < 16; i++) {
-        vformat.vs_tab[i] = -1;
-        vformat.fs_tab[i] = -1;
-    }
-
-    r300_vs_tab_routes(r300, &vformat);
-
-    r300_vertex_psc(r300, &vformat);
-
-    if (memcmp(&r300->vertex_info, &vformat,
-                sizeof(struct r300_vertex_format))) {
-        memcpy(&r300->vertex_info, &vformat,
-                sizeof(struct r300_vertex_format));
-        r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
-    }
-}
-
 /* Set up the mappings from GB to US, for RS block. */
-static void r300_update_fs_tab(struct r300_context* r300)
+static void r300_update_fs_tab(struct r300_context* r300,
+                               struct r300_vertex_format* vformat)
 {
-    struct r300_vertex_format* vformat = &r300->vertex_info;
     struct tgsi_shader_info* info = &r300->fs->info;
     int i, cols = 0, texs = 0, cols_emitted = 0;
     int* tab = vformat->fs_tab;
@@ -299,18 +316,18 @@ static void r300_update_fs_tab(struct r300_context* r300)
     }
 
     /* Now that we know where everything is... */
-    debug_printf("r300: fp input count: %d\n", info->num_inputs);
+    DBG(r300, DBG_DRAW, "r300: fp input count: %d\n", info->num_inputs);
     for (i = 0; i < info->num_inputs; i++) {
         switch (tab[i]) {
             case INTERP_LINEAR:
-                debug_printf("r300: attrib: "
+                DBG(r300, DBG_DRAW, "r300: attrib: "
                         "stack offset %d, color,    tab %d\n",
                         i, cols_emitted);
                 tab[i] = cols_emitted;
                 cols_emitted++;
                 break;
             case INTERP_PERSPECTIVE:
-                debug_printf("r300: attrib: "
+                DBG(r300, DBG_DRAW, "r300: attrib: "
                         "stack offset %d, texcoord, tab %d\n",
                         i, cols + texs);
                 tab[i] = cols + texs;
@@ -328,53 +345,39 @@ static void r300_update_fs_tab(struct r300_context* r300)
 /* Set up the RS block. This is the part of the chipset that actually does
  * the rasterization of vertices into fragments. This is also the part of the
  * chipset that locks up if any part of it is even slightly wrong. */
-static void r300_update_rs_block(struct r300_context* r300)
+static void r300_update_rs_block(struct r300_context* r300,
+                                 struct r300_rs_block* rs)
 {
-    struct r300_rs_block* rs = r300->rs_block;
     struct tgsi_shader_info* info = &r300->fs->info;
-    int* tab = r300->vertex_info.fs_tab;
-    int col_count = 0, fp_offset = 0, i, memory_pos, tex_count = 0;
-
-    memset(rs, 0, sizeof(struct r300_rs_block));
+    int col_count = 0, fp_offset = 0, i, tex_count = 0;
+    int rs_tex_comp = 0;
 
     if (r300_screen(r300->context.screen)->caps->is_r500) {
         for (i = 0; i < info->num_inputs; i++) {
-            assert(tab[i] != -1);
-            memory_pos = tab[i] * 4;
             switch (info->input_semantic_name[i]) {
                 case TGSI_SEMANTIC_COLOR:
                     rs->ip[col_count] |=
-                        R500_RS_COL_PTR(memory_pos) |
+                        R500_RS_COL_PTR(col_count) |
                         R500_RS_COL_FMT(R300_RS_COL_FMT_RGBA);
                     col_count++;
                     break;
                 case TGSI_SEMANTIC_GENERIC:
                     rs->ip[tex_count] |=
-                        R500_RS_SEL_S(memory_pos) |
-                        R500_RS_SEL_T(memory_pos + 1) |
-                        R500_RS_SEL_R(memory_pos + 2) |
-                        R500_RS_SEL_Q(memory_pos + 3);
+                        R500_RS_SEL_S(rs_tex_comp) |
+                        R500_RS_SEL_T(rs_tex_comp + 1) |
+                        R500_RS_SEL_R(rs_tex_comp + 2) |
+                        R500_RS_SEL_Q(rs_tex_comp + 3);
                     tex_count++;
+                    rs_tex_comp += 4;
                     break;
                 default:
                     break;
             }
         }
 
-        if (col_count == 0) {
-            rs->ip[0] |= R500_RS_COL_FMT(R300_RS_COL_FMT_0001);
-        }
-
-        if (tex_count == 0) {
-            rs->ip[0] |=
-                R500_RS_SEL_S(R500_RS_IP_PTR_K0) |
-                R500_RS_SEL_T(R500_RS_IP_PTR_K0) |
-                R500_RS_SEL_R(R500_RS_IP_PTR_K0) |
-                R500_RS_SEL_Q(R500_RS_IP_PTR_K1);
-        }
-
         /* Rasterize at least one color, or bad things happen. */
         if ((col_count == 0) && (tex_count == 0)) {
+            rs->ip[0] |= R500_RS_COL_FMT(R300_RS_COL_FMT_0001);
             col_count++;
         }
 
@@ -391,23 +394,22 @@ static void r300_update_rs_block(struct r300_context* r300)
         }
     } else {
         for (i = 0; i < info->num_inputs; i++) {
-            assert(tab[i] != -1);
-            memory_pos = tab[i] * 4;
             switch (info->input_semantic_name[i]) {
                 case TGSI_SEMANTIC_COLOR:
                     rs->ip[col_count] |=
-                        R300_RS_COL_PTR(memory_pos) |
+                        R300_RS_COL_PTR(col_count) |
                         R300_RS_COL_FMT(R300_RS_COL_FMT_RGBA);
                     col_count++;
                     break;
                 case TGSI_SEMANTIC_GENERIC:
                     rs->ip[tex_count] |=
-                        R300_RS_TEX_PTR(memory_pos) |
+                        R300_RS_TEX_PTR(rs_tex_comp) |
                         R300_RS_SEL_S(R300_RS_SEL_C0) |
                         R300_RS_SEL_T(R300_RS_SEL_C1) |
                         R300_RS_SEL_R(R300_RS_SEL_C2) |
                         R300_RS_SEL_Q(R300_RS_SEL_C3);
                     tex_count++;
+                    rs_tex_comp+=4;
                     break;
                 default:
                     break;
@@ -444,21 +446,110 @@ static void r300_update_rs_block(struct r300_context* r300)
         }
     }
 
-    rs->count = (tex_count * 4) | (col_count << R300_IC_COUNT_SHIFT) |
+    rs->count = (rs_tex_comp) | (col_count << R300_IC_COUNT_SHIFT) |
         R300_HIRES_EN;
 
     rs->inst_count = MAX2(MAX2(col_count - 1, tex_count - 1), 0);
 }
 
+/* Update the vertex format. */
+static void r300_update_derived_shader_state(struct r300_context* r300)
+{
+    struct r300_shader_key* key;
+    struct r300_vertex_format* vformat;
+    struct r300_rs_block* rs_block;
+    struct r300_shader_derived_value* value;
+    int i;
+
+    /*
+    key = CALLOC_STRUCT(r300_shader_key);
+    key->vs = r300->vs;
+    key->fs = r300->fs;
+
+    value = (struct r300_shader_derived_value*)
+        util_hash_table_get(r300->shader_hash_table, (void*)key);
+    if (value) {
+        //vformat = value->vformat;
+        rs_block = value->rs_block;
+
+        FREE(key);
+    } else {
+        rs_block = CALLOC_STRUCT(r300_rs_block);
+        value = CALLOC_STRUCT(r300_shader_derived_value);
+
+        r300_update_rs_block(r300, rs_block);
+
+        //value->vformat = vformat;
+        value->rs_block = rs_block;
+        util_hash_table_set(r300->shader_hash_table,
+            (void*)key, (void*)value);
+    } */
+
+    /* XXX This will be refactored ASAP. */
+    vformat = CALLOC_STRUCT(r300_vertex_format);
+    rs_block = CALLOC_STRUCT(r300_rs_block);
+
+    for (i = 0; i < 16; i++) {
+        vformat->vs_tab[i] = -1;
+        vformat->fs_tab[i] = -1;
+    }
+
+    r300_vs_tab_routes(r300, vformat);
+    r300_vertex_psc(r300, vformat);
+    r300_update_fs_tab(r300, vformat);
+
+    r300_update_rs_block(r300, rs_block);
+
+    FREE(r300->vertex_info);
+    FREE(r300->rs_block);
+
+    r300->vertex_info = vformat;
+    r300->rs_block = rs_block;
+    r300->dirty_state |= (R300_NEW_VERTEX_FORMAT | R300_NEW_RS_BLOCK);
+}
+
+static void r300_update_ztop(struct r300_context* r300)
+{
+    r300->ztop_state.z_buffer_top = R300_ZTOP_ENABLE;
+
+    /* This is important enough that I felt it warranted a comment.
+     *
+     * According to the docs, these are the conditions where ZTOP must be
+     * disabled:
+     * 1) Alpha testing enabled
+     * 2) Texture kill instructions in fragment shader
+     * 3) Chroma key culling enabled
+     * 4) W-buffering enabled
+     *
+     * The docs claim that for the first three cases, if no ZS writes happen,
+     * then ZTOP can be used.
+     *
+     * Additionally, the following conditions require disabled ZTOP:
+     * ~) Depth writes in fragment shader
+     * ~) Outstanding occlusion queries
+     *
+     * ~C.
+     */
+    if (r300->dsa_state->alpha_function) {
+        r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE;
+    } else if (r300->fs->info.uses_kill) {
+        r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE;
+    } else if (r300_fragment_shader_writes_depth(r300->fs)) {
+        r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE;
+    } else if (r300->query_current) {
+        r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE;
+    }
+}
+
 void r300_update_derived_state(struct r300_context* r300)
 {
     if (r300->dirty_state &
-            (R300_NEW_FRAGMENT_SHADER | R300_NEW_VERTEX_SHADER)) {
-        r300_update_vertex_format(r300);
+        (R300_NEW_FRAGMENT_SHADER | R300_NEW_VERTEX_SHADER)) {
+        r300_update_derived_shader_state(r300);
     }
 
-    if (r300->dirty_state & R300_NEW_VERTEX_FORMAT) {
-        r300_update_fs_tab(r300);
-        r300_update_rs_block(r300);
+    if (r300->dirty_state &
+            (R300_NEW_DSA | R300_NEW_FRAGMENT_SHADER | R300_NEW_QUERY)) {
+        r300_update_ztop(r300);
     }
 }
diff --git a/src/gallium/drivers/r300/r300_state_derived.h b/src/gallium/drivers/r300/r300_state_derived.h
index 63ae8eb8d0..05ad535e2d 100644
--- a/src/gallium/drivers/r300/r300_state_derived.h
+++ b/src/gallium/drivers/r300/r300_state_derived.h
@@ -23,11 +23,11 @@
 #ifndef R300_STATE_DERIVED_H
 #define R300_STATE_DERIVED_H
 
-#include "draw/draw_vertex.h"
+struct r300_context;
 
-#include "r300_context.h"
-#include "r300_reg.h"
-#include "r300_state_inlines.h"
+unsigned r300_shader_key_hash(void* key);
+
+int r300_shader_key_compare(void* key1, void* key2);
 
 void r300_update_derived_state(struct r300_context* r300);
 
diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index 91b93fc367..ec11a41253 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -24,6 +24,8 @@
 #ifndef R300_STATE_INLINES_H
 #define R300_STATE_INLINES_H
 
+#include "draw/draw_vertex.h"
+
 #include "pipe/p_format.h"
 
 #include "r300_reg.h"
@@ -51,6 +53,7 @@ static INLINE uint32_t r300_translate_blend_function(int blend_func)
             return R300_COMB_FCN_MAX;
         default:
             debug_printf("r300: Unknown blend function %d\n", blend_func);
+            assert(0);
             break;
     }
     return 0;
@@ -98,6 +101,7 @@ static INLINE uint32_t r300_translate_blend_factor(int blend_fact)
         case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: */
         default:
             debug_printf("r300: Unknown blend factor %d\n", blend_fact);
+            assert(0);
             break;
     }
     return 0;
@@ -127,6 +131,7 @@ static INLINE uint32_t r300_translate_depth_stencil_function(int zs_func)
         default:
             debug_printf("r300: Unknown depth/stencil function %d\n",
                 zs_func);
+            assert(0);
             break;
     }
     return 0;
@@ -153,6 +158,7 @@ static INLINE uint32_t r300_translate_stencil_op(int s_op)
             return R300_ZS_INVERT;
         default:
             debug_printf("r300: Unknown stencil op %d", s_op);
+            assert(0);
             break;
     }
     return 0;
@@ -179,6 +185,7 @@ static INLINE uint32_t r300_translate_alpha_function(int alpha_func)
             return R300_FG_ALPHA_FUNC_ALWAYS;
         default:
             debug_printf("r300: Unknown alpha function %d", alpha_func);
+            assert(0);
             break;
     }
     return 0;
@@ -207,6 +214,7 @@ static INLINE uint32_t r300_translate_wrap(int wrap)
             return R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
         default:
             debug_printf("r300: Unknown texture wrap %d", wrap);
+            assert(0);
             return 0;
     }
 }
@@ -226,6 +234,7 @@ static INLINE uint32_t r300_translate_tex_filters(int min, int mag, int mip)
             break;
         default:
             debug_printf("r300: Unknown texture filter %d\n", min);
+            assert(0);
             break;
     }
     switch (mag) {
@@ -240,6 +249,7 @@ static INLINE uint32_t r300_translate_tex_filters(int min, int mag, int mip)
             break;
         default:
             debug_printf("r300: Unknown texture filter %d\n", mag);
+            assert(0);
             break;
     }
     switch (mip) {
@@ -254,6 +264,7 @@ static INLINE uint32_t r300_translate_tex_filters(int min, int mag, int mip)
             break;
         default:
             debug_printf("r300: Unknown texture filter %d\n", mip);
+            assert(0);
             break;
     }
 
@@ -277,6 +288,8 @@ static INLINE uint32_t r300_anisotropy(float max_aniso)
 
 /* Buffer formats. */
 
+/* Colorbuffer formats. This is the unswizzled format of the RB3D block's
+ * output. For the swizzling of the targets, check the shader's format. */
 static INLINE uint32_t r300_translate_colorformat(enum pipe_format format)
 {
     switch (format) {
@@ -292,7 +305,9 @@ static INLINE uint32_t r300_translate_colorformat(enum pipe_format format)
             return R300_COLOR_FORMAT_ARGB4444;
         /* 32-bit buffers */
         case PIPE_FORMAT_A8R8G8B8_UNORM:
-        case PIPE_FORMAT_Z24S8_UNORM:
+        case PIPE_FORMAT_X8R8G8B8_UNORM:
+        case PIPE_FORMAT_R8G8B8A8_UNORM:
+        case PIPE_FORMAT_R8G8B8X8_UNORM:
             return R300_COLOR_FORMAT_ARGB8888;
         /* XXX Not in pipe_format
         case PIPE_FORMAT_A32R32G32B32:
@@ -309,17 +324,21 @@ static INLINE uint32_t r300_translate_colorformat(enum pipe_format format)
             debug_printf("r300: Implementation error: "
                 "Got unsupported color format %s in %s\n",
                 pf_name(format), __FUNCTION__);
+            assert(0);
             break;
     }
     return 0;
 }
 
+/* Depthbuffer and stencilbuffer. Thankfully, we only support two flavors. */
 static INLINE uint32_t r300_translate_zsformat(enum pipe_format format)
 {
     switch (format) {
         /* 16-bit depth, no stencil */
         case PIPE_FORMAT_Z16_UNORM:
             return R300_DEPTHFORMAT_16BIT_INT_Z;
+        /* 24-bit depth, ignored stencil */
+        case PIPE_FORMAT_Z24X8_UNORM:
         /* 24-bit depth, 8-bit stencil */
         case PIPE_FORMAT_Z24S8_UNORM:
             return R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL;
@@ -327,25 +346,36 @@ static INLINE uint32_t r300_translate_zsformat(enum pipe_format format)
             debug_printf("r300: Implementation error: "
                 "Got unsupported ZS format %s in %s\n",
                 pf_name(format), __FUNCTION__);
+            assert(0);
             break;
     }
     return 0;
 }
 
-/* Translate pipe_format into US_OUT_FMT.
+/* Shader output formats. This is essentially the swizzle from the shader
+ * to the RB3D block.
+ *
  * Note that formats are stored from C3 to C0. */
 static INLINE uint32_t r300_translate_out_fmt(enum pipe_format format)
 {
     switch (format) {
         case PIPE_FORMAT_A8R8G8B8_UNORM:
+        case PIPE_FORMAT_X8R8G8B8_UNORM:
+        /* XXX */
         case PIPE_FORMAT_Z24S8_UNORM:
             return R300_US_OUT_FMT_C4_8 |
                 R300_C0_SEL_B | R300_C1_SEL_G |
                 R300_C2_SEL_R | R300_C3_SEL_A;
+        case PIPE_FORMAT_R8G8B8A8_UNORM:
+        case PIPE_FORMAT_R8G8B8X8_UNORM:
+            return R300_US_OUT_FMT_C4_8 |
+                R300_C0_SEL_A | R300_C1_SEL_B |
+                R300_C2_SEL_G | R300_C3_SEL_R;
         default:
             debug_printf("r300: Implementation error: "
                 "Got unsupported output format %s in %s\n",
                 pf_name(format), __FUNCTION__);
+            assert(0);
             return R300_US_OUT_FMT_UNUSED;
     }
     return 0;
@@ -372,31 +402,57 @@ static INLINE uint32_t r300_translate_gb_pipes(int pipe_count)
     return 0;
 }
 
-static INLINE uint32_t translate_vertex_data_type(int type) {
-    switch (type) {
-        case EMIT_1F:
-        case EMIT_1F_PSIZE:
+/* Translate pipe_formats into PSC vertex types. */
+static INLINE uint16_t
+r300_translate_vertex_data_type(enum pipe_format format) {
+    switch (format) {
+        case PIPE_FORMAT_R32_FLOAT:
             return R300_DATA_TYPE_FLOAT_1;
             break;
-        case EMIT_2F:
+        case PIPE_FORMAT_R32G32_FLOAT:
             return R300_DATA_TYPE_FLOAT_2;
             break;
-        case EMIT_3F:
+        case PIPE_FORMAT_R32G32B32_FLOAT:
             return R300_DATA_TYPE_FLOAT_3;
             break;
-        case EMIT_4F:
+        case PIPE_FORMAT_R32G32B32A32_FLOAT:
             return R300_DATA_TYPE_FLOAT_4;
             break;
-        case EMIT_4UB:
-            return R300_DATA_TYPE_BYTE;
+        case PIPE_FORMAT_R8G8B8A8_UNORM:
+            return R300_DATA_TYPE_BYTE |
+                R300_NORMALIZE;
             break;
         default:
             debug_printf("r300: Implementation error: "
-                    "Bad vertex data type!\n");
+                    "Bad vertex data format %s!\n", pf_name(format));
             assert(0);
             break;
     }
+    return 0;
+}
 
+static INLINE uint16_t
+r300_translate_vertex_data_swizzle(enum pipe_format format) {
+    switch (format) {
+        case PIPE_FORMAT_R32_FLOAT:
+            return R300_VAP_SWIZZLE_X001;
+            break;
+        case PIPE_FORMAT_R32G32_FLOAT:
+            return R300_VAP_SWIZZLE_XY01;
+            break;
+        case PIPE_FORMAT_R32G32B32_FLOAT:
+            return R300_VAP_SWIZZLE_XYZ1;
+            break;
+        case PIPE_FORMAT_R32G32B32A32_FLOAT:
+        case PIPE_FORMAT_R8G8B8A8_UNORM:
+            return R300_VAP_SWIZZLE_XYZW;
+            break;
+        default:
+            debug_printf("r300: Implementation error: "
+                    "Bad vertex data format %s!\n", pf_name(format));
+            assert(0);
+            break;
+    }
     return 0;
 }
 
diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c
index 7d822fec48..4865f16058 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.c
+++ b/src/gallium/drivers/r300/r300_state_invariant.c
@@ -21,8 +21,17 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
+#include "r300_context.h"
+#include "r300_cs.h"
+#include "r300_reg.h"
+#include "r300_screen.h"
 #include "r300_state_invariant.h"
 
+struct pipe_viewport_state r300_viewport_identity = {
+    .scale = {1.0, 1.0, 1.0, 1.0},
+    .translate = {0.0, 0.0, 0.0, 0.0},
+};
+
 /* Calculate and emit invariant state. This is data that the 3D engine
  * will probably want at the beginning of every CS, but it's not currently
  * handled by any CSO setup, and in addition it doesn't really change much.
diff --git a/src/gallium/drivers/r300/r300_state_invariant.h b/src/gallium/drivers/r300/r300_state_invariant.h
index 5bea6779fe..05cff0d6df 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.h
+++ b/src/gallium/drivers/r300/r300_state_invariant.h
@@ -23,11 +23,7 @@
 #ifndef R300_STATE_INVARIANT_H
 #define R300_STATE_INVARIANT_H
 
-#include "r300_chipset.h"
-#include "r300_context.h"
-#include "r300_cs.h"
-#include "r300_reg.h"
-#include "r300_state_inlines.h"
+struct r300_context;
 
 void r300_emit_invariant_state(struct r300_context* r300);
 
diff --git a/src/gallium/drivers/r300/r300_surface.c b/src/gallium/drivers/r300/r300_surface.c
deleted file mode 100644
index 96e6e4a77d..0000000000
--- a/src/gallium/drivers/r300/r300_surface.c
+++ /dev/null
@@ -1,372 +0,0 @@
-/*
- * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
- *                Joakim Sindholt <opensource@zhasha.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#include "r300_surface.h"
-
-static void r300_surface_setup(struct r300_context* r300,
-                               struct r300_texture* dest,
-                               unsigned x, unsigned y,
-                               unsigned w, unsigned h)
-{
-    struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps;
-    unsigned pixpitch = dest->stride / dest->tex.block.size;
-    CS_LOCALS(r300);
-
-    r300_emit_blend_state(r300, &blend_clear_state);
-    r300_emit_blend_color_state(r300, &blend_color_clear_state);
-    r300_emit_dsa_state(r300, &dsa_clear_state);
-    r300_emit_rs_state(r300, &rs_clear_state);
-
-    BEGIN_CS(26);
-
-    /* Viewport setup */
-    OUT_CS_REG_SEQ(R300_SE_VPORT_XSCALE, 6);
-    OUT_CS_32F((float)w);
-    OUT_CS_32F((float)x);
-    OUT_CS_32F((float)h);
-    OUT_CS_32F((float)y);
-    OUT_CS_32F(1.0);
-    OUT_CS_32F(0.0);
-
-    OUT_CS_REG(R300_VAP_VTE_CNTL, R300_VPORT_X_SCALE_ENA |
-            R300_VPORT_X_OFFSET_ENA |
-            R300_VPORT_Y_SCALE_ENA |
-            R300_VPORT_Y_OFFSET_ENA |
-            R300_VTX_XY_FMT | R300_VTX_Z_FMT);
-
-    /* Pixel scissors. */
-    OUT_CS_REG_SEQ(R300_SC_SCISSORS_TL, 2);
-    if (caps->is_r500) {
-        OUT_CS((x << R300_SCISSORS_X_SHIFT) | (y << R300_SCISSORS_Y_SHIFT));
-        OUT_CS(((w - 1) << R300_SCISSORS_X_SHIFT) | ((h - 1) << R300_SCISSORS_Y_SHIFT));
-    } else {
-        /* Non-R500 chipsets have an offset of 1440 in their scissors. */
-        OUT_CS(((x + 1440) << R300_SCISSORS_X_SHIFT) |
-                ((y + 1440) << R300_SCISSORS_Y_SHIFT));
-        OUT_CS((((w - 1) + 1440) << R300_SCISSORS_X_SHIFT) |
-                (((h - 1) + 1440) << R300_SCISSORS_Y_SHIFT));
-    }
-
-    /* Flush colorbuffer and blend caches. */
-    OUT_CS_REG(R300_RB3D_DSTCACHE_CTLSTAT,
-        R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_FLUSH_DIRTY_3D |
-        R300_RB3D_DSTCACHE_CTLSTAT_DC_FINISH_SIGNAL);
-    OUT_CS_REG(R300_ZB_ZCACHE_CTLSTAT,
-        R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_FLUSH_AND_FREE |
-        R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_FREE);
-
-    /* Setup colorbuffer. */
-    OUT_CS_REG_SEQ(R300_RB3D_COLOROFFSET0, 1);
-    OUT_CS_RELOC(dest->buffer, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
-    OUT_CS_REG_SEQ(R300_RB3D_COLORPITCH0, 1);
-    OUT_CS_RELOC(dest->buffer, pixpitch |
-                 r300_translate_colorformat(dest->tex.format), 0,
-                 RADEON_GEM_DOMAIN_VRAM, 0);
-    OUT_CS_REG(RB3D_COLOR_CHANNEL_MASK, 0xf);
-
-    END_CS;
-}
-
-/* Provides pipe_context's "surface_fill". Commonly used for clearing
- * buffers. */
-static void r300_surface_fill(struct pipe_context* pipe,
-                              struct pipe_surface* dest,
-                              unsigned x, unsigned y,
-                              unsigned w, unsigned h,
-                              unsigned color)
-{
-    int i;
-    float r, g, b, a, depth;
-    struct r300_context* r300 = r300_context(pipe);
-    struct r300_capabilities* caps = r300_screen(pipe->screen)->caps;
-    struct r300_texture* tex = (struct r300_texture*)dest->texture;
-    unsigned pixpitch = tex->stride / tex->tex.block.size;
-    boolean invalid = FALSE;
-    CS_LOCALS(r300);
-
-    a = (float)((color >> 24) & 0xff) / 255.0f;
-    r = (float)((color >> 16) & 0xff) / 255.0f;
-    g = (float)((color >>  8) & 0xff) / 255.0f;
-    b = (float)((color >>  0) & 0xff) / 255.0f;
-    debug_printf("r300: Filling surface %p at (%d,%d),"
-        " dimensions %dx%d (pixel pitch %d), color 0x%x\n",
-        dest, x, y, w, h, pixpitch, color);
-
-    /* Fallback? */
-    if (FALSE) {
-fallback:
-        debug_printf("r300: Falling back on surface clear...");
-        util_surface_fill(pipe, dest, x, y, w, h, color);
-        return;
-    }
-
-    /* Make sure our target BO is okay. */
-validate:
-    if (!r300->winsys->add_buffer(r300->winsys, tex->buffer,
-                0, RADEON_GEM_DOMAIN_VRAM)) {
-        r300->context.flush(&r300->context, 0, NULL);
-        goto validate;
-    }
-    if (!r300->winsys->validate(r300->winsys)) {
-        r300->context.flush(&r300->context, 0, NULL);
-        if (invalid) {
-            debug_printf("r300: Stuck in validation loop, gonna fallback.");
-            goto fallback;
-        }
-        invalid = TRUE;
-        goto validate;
-    }
-
-    r300_surface_setup(r300, tex, x, y, w, h);
-
-    /* Vertex shader setup */
-    if (caps->has_tcl) {
-        r300_emit_vertex_program_code(r300, &r300_passthrough_vertex_shader, 0);
-    } else {
-        BEGIN_CS(4);
-        OUT_CS_REG(R300_VAP_CNTL_STATUS,
-#ifdef PIPE_ARCH_BIG_ENDIAN
-                   R300_VC_32BIT_SWAP |
-#endif
-                   R300_VAP_TCL_BYPASS);
-        OUT_CS_REG(R300_VAP_CNTL, R300_PVS_NUM_SLOTS(5) |
-                R300_PVS_NUM_CNTLRS(5) |
-                R300_PVS_NUM_FPUS(caps->num_vert_fpus) |
-                R300_PVS_VF_MAX_VTX_NUM(12));
-        END_CS;
-    }
-
-    /* Fragment shader setup */
-    if (caps->is_r500) {
-        r500_emit_fragment_program_code(r300, &r5xx_passthrough_fragment_shader, 0);
-        r300_emit_rs_block_state(r300, &r5xx_rs_block_clear_state);
-    } else {
-        r300_emit_fragment_program_code(r300, &r3xx_passthrough_fragment_shader, 0);
-        r300_emit_rs_block_state(r300, &r3xx_rs_block_clear_state);
-    }
-
-    BEGIN_CS(26);
-
-    /* VAP stream control, mapping from input memory to PVS/RS memory */
-    if (caps->has_tcl) {
-        OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_0,
-            (R300_DATA_TYPE_FLOAT_4 << R300_DATA_TYPE_0_SHIFT) |
-            ((R300_LAST_VEC | (1 << R300_DST_VEC_LOC_SHIFT) |
-                R300_DATA_TYPE_FLOAT_4) << R300_DATA_TYPE_1_SHIFT));
-    } else {
-        OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_0,
-            (R300_DATA_TYPE_FLOAT_4 << R300_DATA_TYPE_0_SHIFT) |
-            ((R300_LAST_VEC | (2 << R300_DST_VEC_LOC_SHIFT) |
-                R300_DATA_TYPE_FLOAT_4) << R300_DATA_TYPE_1_SHIFT));
-    }
-    OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_EXT_0,
-            (R300_VAP_SWIZZLE_XYZW << R300_SWIZZLE0_SHIFT) |
-            (R300_VAP_SWIZZLE_XYZW << R300_SWIZZLE1_SHIFT));
-
-    /* VAP format controls */
-    OUT_CS_REG(R300_VAP_OUTPUT_VTX_FMT_0,
-            R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT |
-            R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT);
-    OUT_CS_REG(R300_VAP_OUTPUT_VTX_FMT_1, 0x0);
-
-    /* Disable textures */
-    OUT_CS_REG(R300_TX_ENABLE, 0x0);
-
-    /* The size of the point we're about to draw, in sixths of pixels */
-    OUT_CS_REG(R300_GA_POINT_SIZE,
-        ((h * 6)  & R300_POINTSIZE_Y_MASK) |
-        ((w * 6) << R300_POINTSIZE_X_SHIFT));
-
-    /* Vertex size. */
-    OUT_CS_REG(R300_VAP_VTX_SIZE, 0x8);
-
-    /* Packet3 with our point vertex */
-    OUT_CS_PKT3(R200_3D_DRAW_IMMD_2, 8);
-    OUT_CS(R300_PRIM_TYPE_POINT | R300_PRIM_WALK_RING |
-            (1 << R300_PRIM_NUM_VERTICES_SHIFT));
-    /* Position */
-    OUT_CS_32F(0.5);
-    OUT_CS_32F(0.5);
-    OUT_CS_32F(1.0);
-    OUT_CS_32F(1.0);
-    /* Color */
-    OUT_CS_32F(r);
-    OUT_CS_32F(g);
-    OUT_CS_32F(b);
-    OUT_CS_32F(a);
-
-    OUT_CS_REG(R300_RB3D_DSTCACHE_CTLSTAT, 0xA);
-
-    END_CS;
-
-    r300->dirty_hw++;
-}
-
-static void r300_surface_copy(struct pipe_context* pipe,
-                              struct pipe_surface* dest,
-                              unsigned destx, unsigned desty,
-                              struct pipe_surface* src,
-                              unsigned srcx, unsigned srcy,
-                              unsigned w, unsigned h)
-{
-    struct r300_context* r300 = r300_context(pipe);
-    struct r300_capabilities* caps = r300_screen(pipe->screen)->caps;
-    struct r300_texture* srctex = (struct r300_texture*)src->texture;
-    struct r300_texture* desttex = (struct r300_texture*)dest->texture;
-    unsigned pixpitch = srctex->stride / srctex->tex.block.size;
-    boolean invalid = FALSE;
-    float fsrcx = srcx, fsrcy = srcy, fdestx = destx, fdesty = desty;
-    CS_LOCALS(r300);
-
-    debug_printf("r300: Copying surface %p at (%d,%d) to %p at (%d, %d),"
-        " dimensions %dx%d (pixel pitch %d)\n",
-        src, srcx, srcy, dest, destx, desty, w, h, pixpitch);
-
-    if ((srctex->buffer == desttex->buffer) &&
-            ((destx < srcx + w) || (srcx < destx + w)) &&
-            ((desty < srcy + h) || (srcy < desty + h))) {
-fallback:
-        debug_printf("r300: Falling back on surface_copy\n");
-        util_surface_copy(pipe, FALSE, dest, destx, desty, src,
-                srcx, srcy, w, h);
-    }
-
-    /* Add our target BOs to the list. */
-validate:
-    if (!r300->winsys->add_buffer(r300->winsys, srctex->buffer,
-                RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM, 0)) {
-        r300->context.flush(&r300->context, 0, NULL);
-        goto validate;
-    }
-    if (!r300->winsys->add_buffer(r300->winsys, desttex->buffer,
-                0, RADEON_GEM_DOMAIN_VRAM)) {
-        r300->context.flush(&r300->context, 0, NULL);
-        goto validate;
-    }
-    if (!r300->winsys->validate(r300->winsys)) {
-        r300->context.flush(&r300->context, 0, NULL);
-        if (invalid) {
-            debug_printf("r300: Stuck in validation loop, gonna fallback.");
-            goto fallback;
-        }
-        invalid = TRUE;
-        goto validate;
-    }
-
-    r300_surface_setup(r300, desttex, destx, desty, w, h);
-
-    /* Setup the texture. */
-    r300_emit_texture(r300, &r300_sampler_copy_state, srctex, 0);
-
-    /* Flush and enable. */
-    r300_flush_textures(r300);
-
-    /* Vertex shader setup */
-    if (caps->has_tcl) {
-        r300_emit_vertex_program_code(r300, &r300_passthrough_vertex_shader, 0);
-    } else {
-        BEGIN_CS(4);
-        OUT_CS_REG(R300_VAP_CNTL_STATUS,
-#ifdef PIPE_ARCH_BIG_ENDIAN
-                   R300_VC_32BIT_SWAP |
-#endif
-                   R300_VAP_TCL_BYPASS);
-        OUT_CS_REG(R300_VAP_CNTL, R300_PVS_NUM_SLOTS(5) |
-                R300_PVS_NUM_CNTLRS(5) |
-                R300_PVS_NUM_FPUS(caps->num_vert_fpus) |
-                R300_PVS_VF_MAX_VTX_NUM(12));
-        END_CS;
-    }
-
-    /* Fragment shader setup */
-    if (caps->is_r500) {
-        r500_emit_fragment_program_code(r300, &r5xx_texture_fragment_shader, 0);
-        r300_emit_rs_block_state(r300, &r5xx_rs_block_copy_state);
-    } else {
-        r300_emit_fragment_program_code(r300, &r3xx_texture_fragment_shader, 0);
-        r300_emit_rs_block_state(r300, &r3xx_rs_block_copy_state);
-    }
-
-    BEGIN_CS(30);
-    /* VAP stream control, mapping from input memory to PVS/RS memory */
-    if (caps->has_tcl) {
-        OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_0,
-            (R300_DATA_TYPE_FLOAT_2 << R300_DATA_TYPE_0_SHIFT) |
-            ((R300_LAST_VEC | (1 << R300_DST_VEC_LOC_SHIFT) |
-                R300_DATA_TYPE_FLOAT_2) << R300_DATA_TYPE_1_SHIFT));
-    } else {
-        OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_0,
-            (R300_DATA_TYPE_FLOAT_2 << R300_DATA_TYPE_0_SHIFT) |
-            ((R300_LAST_VEC | (6 << R300_DST_VEC_LOC_SHIFT) |
-                R300_DATA_TYPE_FLOAT_2) << R300_DATA_TYPE_1_SHIFT));
-    }
-    OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_EXT_0,
-            (R300_VAP_SWIZZLE_XYZW << R300_SWIZZLE0_SHIFT) |
-            (R300_VAP_SWIZZLE_XYZW << R300_SWIZZLE1_SHIFT));
-
-    /* VAP format controls */
-    OUT_CS_REG(R300_VAP_OUTPUT_VTX_FMT_0,
-            R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT);
-    /* Two components of texture 0 */
-    OUT_CS_REG(R300_VAP_OUTPUT_VTX_FMT_1, 0x2);
-
-    /* Vertex size. */
-    OUT_CS_REG(R300_VAP_VTX_SIZE, 0x4);
-
-    /* Packet3 with our texcoords */
-    OUT_CS_PKT3(R200_3D_DRAW_IMMD_2, 16);
-    OUT_CS(R300_PRIM_TYPE_QUADS | R300_PRIM_WALK_RING |
-            (4 << R300_PRIM_NUM_VERTICES_SHIFT));
-    /* (x    , y    ) */
-    OUT_CS_32F(fdestx / dest->width);
-    OUT_CS_32F(fdesty / dest->height);
-    OUT_CS_32F(fsrcx  / src->width);
-    OUT_CS_32F(fsrcy  / src->height);
-    /* (x    , y + h) */
-    OUT_CS_32F(fdestx / dest->width);
-    OUT_CS_32F((fdesty + h) / dest->height);
-    OUT_CS_32F(fsrcx  / src->width);
-    OUT_CS_32F((fsrcy  + h) / src->height);
-    /* (x + w, y + h) */
-    OUT_CS_32F((fdestx + w) / dest->width);
-    OUT_CS_32F((fdesty + h) / dest->height);
-    OUT_CS_32F((fsrcx  + w) / src->width);
-    OUT_CS_32F((fsrcy  + h) / src->height);
-    /* (x + w, y    ) */
-    OUT_CS_32F((fdestx + w) / dest->width);
-    OUT_CS_32F(fdesty / dest->height);
-    OUT_CS_32F((fsrcx  + w) / src->width);
-    OUT_CS_32F(fsrcy  / src->height);
-
-    OUT_CS_REG(R300_RB3D_DSTCACHE_CTLSTAT, 0xA);
-
-    END_CS;
-
-    r300->dirty_hw++;
-}
-
-void r300_init_surface_functions(struct r300_context* r300)
-{
-    r300->context.surface_fill = r300_surface_fill;
-    r300->context.surface_copy = r300_surface_copy;
-}
diff --git a/src/gallium/drivers/r300/r300_surface.h b/src/gallium/drivers/r300/r300_surface.h
deleted file mode 100644
index f9e98b2ec9..0000000000
--- a/src/gallium/drivers/r300/r300_surface.h
+++ /dev/null
@@ -1,124 +0,0 @@
-/*
- * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#ifndef R300_SURFACE_H
-#define R300_SURFACE_H
-
-#include "pipe/p_context.h"
-#include "pipe/p_screen.h"
-
-#include "util/u_rect.h"
-
-#include "r300_context.h"
-#include "r300_cs.h"
-#include "r300_emit.h"
-#include "r300_fs.h"
-#include "r300_vs.h"
-#include "r300_state_inlines.h"
-
-static struct r300_blend_state blend_clear_state = {
-    .blend_control = 0x0,
-    .alpha_blend_control = 0x0,
-    .rop = 0x0,
-    .dither = 0x0,
-};
-
-static struct r300_blend_color_state blend_color_clear_state = {
-    .blend_color = 0x0,
-    .blend_color_red_alpha = 0x0,
-    .blend_color_green_blue = 0x0,
-};
-
-static struct r300_dsa_state dsa_clear_state = {
-    .alpha_function = 0x0,
-    .alpha_reference = 0x0,
-    .z_buffer_control = 0x0,
-    .z_stencil_control = 0x0,
-    .stencil_ref_mask = R300_STENCILWRITEMASK_MASK,
-    .z_buffer_top = R300_ZTOP_ENABLE,
-    .stencil_ref_bf = 0x0,
-};
-
-static struct r300_rs_state rs_clear_state = {
-    .point_minmax = 0x36000006,
-    .line_control = 0x00030006,
-    .depth_scale_front = 0x0,
-    .depth_offset_front = 0x0,
-    .depth_scale_back = 0x0,
-    .depth_offset_back = 0x0,
-    .polygon_offset_enable = 0x0,
-    .cull_mode = 0x0,
-    .line_stipple_config = 0x3BAAAAAB,
-    .line_stipple_value = 0x0,
-    .color_control = R300_SHADE_MODEL_FLAT,
-};
-
-static struct r300_rs_block r3xx_rs_block_clear_state = {
-    .ip[0] = R500_RS_SEL_S(R300_RS_SEL_C0) |
-        R500_RS_SEL_T(R300_RS_SEL_C0) |
-        R500_RS_SEL_R(R300_RS_SEL_C0) |
-        R500_RS_SEL_Q(R300_RS_SEL_K1),
-    .inst[0] = R300_RS_INST_COL_CN_WRITE,
-    .count = R300_IT_COUNT(0) | R300_IC_COUNT(1) | R300_HIRES_EN,
-    .inst_count = 0,
-};
-
-static struct r300_rs_block r5xx_rs_block_clear_state = {
-    .ip[0] = R500_RS_SEL_S(R500_RS_IP_PTR_K0) |
-        R500_RS_SEL_T(R500_RS_IP_PTR_K0) |
-        R500_RS_SEL_R(R500_RS_IP_PTR_K0) |
-        R500_RS_SEL_Q(R500_RS_IP_PTR_K1),
-    .inst[0] = R500_RS_INST_COL_CN_WRITE,
-    .count = R300_IT_COUNT(0) | R300_IC_COUNT(1) | R300_HIRES_EN,
-    .inst_count = 0,
-};
-
-/* The following state is used for surface_copy only. */
-
-static struct r300_rs_block r3xx_rs_block_copy_state = {
-    .ip[0] = R500_RS_SEL_S(R300_RS_SEL_K0) |
-        R500_RS_SEL_T(R300_RS_SEL_K0) |
-        R500_RS_SEL_R(R300_RS_SEL_K0) |
-        R500_RS_SEL_Q(R300_RS_SEL_K1),
-    .inst[0] = R300_RS_INST_COL_CN_WRITE,
-    .count = R300_IT_COUNT(2) | R300_IC_COUNT(0) | R300_HIRES_EN,
-    .inst_count = R300_RS_TX_OFFSET(0),
-};
-
-static struct r300_rs_block r5xx_rs_block_copy_state = {
-    .ip[0] = R500_RS_SEL_S(0) |
-        R500_RS_SEL_T(1) |
-        R500_RS_SEL_R(R500_RS_IP_PTR_K0) |
-        R500_RS_SEL_Q(R500_RS_IP_PTR_K1),
-    .inst[0] = R500_RS_INST_TEX_CN_WRITE,
-    .count = R300_IT_COUNT(2) | R300_IC_COUNT(0) | R300_HIRES_EN,
-    .inst_count = R300_RS_TX_OFFSET(0),
-};
-
-static struct r300_sampler_state r300_sampler_copy_state = {
-    .filter0 = R300_TX_WRAP_S(R300_TX_CLAMP) |
-        R300_TX_WRAP_T(R300_TX_CLAMP) |
-        R300_TX_MAG_FILTER_NEAREST |
-        R300_TX_MIN_FILTER_NEAREST,
-};
-
-#endif /* R300_SURFACE_H */
diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index 590052509c..3c8ff24e17 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -20,43 +20,71 @@
  * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  * USE OR OTHER DEALINGS IN THE SOFTWARE. */
 
+#include "pipe/p_screen.h"
+
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#include "r300_context.h"
 #include "r300_texture.h"
 
-static void r300_setup_texture_state(struct r300_texture* tex,
-                                     unsigned width,
-                                     unsigned height,
-                                     unsigned pitch,
-                                     unsigned levels)
+static void r300_setup_texture_state(struct r300_texture* tex)
 {
     struct r300_texture_state* state = &tex->state;
+    struct pipe_texture *pt = &tex->tex;
 
-    state->format0 = R300_TX_WIDTH((width - 1) & 0x7ff) |
-        R300_TX_HEIGHT((height - 1) & 0x7ff) |
-        R300_TX_NUM_LEVELS(levels) |
+    state->format0 = R300_TX_WIDTH((pt->width[0] - 1) & 0x7ff) |
+        R300_TX_HEIGHT((pt->height[0] - 1) & 0x7ff) |
+        R300_TX_DEPTH(util_logbase2(pt->depth[0]) & 0xf) |
+        R300_TX_NUM_LEVELS(pt->last_level) |
         R300_TX_PITCH_EN;
 
     /* XXX */
-    state->format1 = r300_translate_texformat(tex->tex.format);
+    state->format1 = r300_translate_texformat(pt->format);
+    if (pt->target == PIPE_TEXTURE_CUBE) {
+        state->format1 |= R300_TX_FORMAT_CUBIC_MAP;
+    }
+    if (pt->target == PIPE_TEXTURE_3D) {
+        state->format1 |= R300_TX_FORMAT_3D;
+    }
 
-    state->format2 = pitch - 1;
+    state->format2 = (r300_texture_get_stride(tex, 0) / pt->block.size) - 1;
 
-    /* Assume (somewhat foolishly) that oversized textures will
-     * not be permitted by the state tracker. */
-    if (width > 2048) {
+    /* Don't worry about accidentally setting this bit on non-r500;
+     * the kernel should catch it. */
+    if (pt->width[0] > 2048) {
         state->format2 |= R500_TXWIDTH_BIT11;
     }
-    if (height > 2048) {
+    if (pt->height[0] > 2048) {
         state->format2 |= R500_TXHEIGHT_BIT11;
     }
 
-    debug_printf("r300: Set texture state (%dx%d, pitch %d, %d levels)\n",
-            width, height, pitch, levels);
+    debug_printf("r300: Set texture state (%dx%d, %d levels)\n",
+		 pt->width[0], pt->height[0], pt->last_level);
+}
+
+/**
+ * Return the stride, in bytes, of the texture images of the given texture
+ * at the given level.
+ */
+unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level)
+{
+    if (tex->stride_override)
+        return tex->stride_override;
+
+    if (level > tex->tex.last_level) {
+        debug_printf("%s: level (%u) > last_level (%u)\n", __FUNCTION__,
+            level, tex->tex.last_level);
+        return 0;
+    }
+
+    return align(pf_get_stride(&tex->tex.block, tex->tex.width[level]), 32);
 }
 
 static void r300_setup_miptree(struct r300_texture* tex)
 {
     struct pipe_texture* base = &tex->tex;
-    int stride, size, offset;
+    int stride, size;
     int i;
 
     for (i = 0; i <= base->last_level; i++) {
@@ -69,12 +97,7 @@ static void r300_setup_miptree(struct r300_texture* tex)
         base->nblocksx[i] = pf_get_nblocksx(&base->block, base->width[i]);
         base->nblocksy[i] = pf_get_nblocksy(&base->block, base->height[i]);
 
-        /* Radeons enjoy things in multiples of 64.
-         *
-         * XXX
-         * POT, uncompressed, unmippmapped textures can be aligned to 32,
-         * instead of 64. */
-        stride = align(pf_get_stride(&base->block, base->width[i]), 32);
+        stride = r300_texture_get_stride(tex, i);
         size = stride * base->nblocksy[i] * base->depth[i];
 
         tex->offset[i] = align(tex->size, 32);
@@ -84,10 +107,6 @@ static void r300_setup_miptree(struct r300_texture* tex)
                 "(%dx%dx%d px, pitch %d bytes)\n",
                 i, base->width[i], base->height[i], base->depth[i],
                 stride);
-        /* Save stride of first level to the texture. */
-        if (i == 0) {
-            tex->stride = stride;
-        }
     }
 }
 
@@ -108,8 +127,7 @@ static struct pipe_texture*
 
     r300_setup_miptree(tex);
 
-    r300_setup_texture_state(tex, template->width[0], template->height[0],
-            template->width[0], template->last_level);
+    r300_setup_texture_state(tex);
 
     tex->buffer = screen->buffer_create(screen, 1024,
                                         PIPE_BUFFER_USAGE_PIXEL,
@@ -173,9 +191,7 @@ static struct pipe_texture*
 {
     struct r300_texture* tex;
 
-    /* XXX we should start doing mips now... */
     if (base->target != PIPE_TEXTURE_2D ||
-        base->last_level != 0 ||
         base->depth[0] != 1) {
         return NULL;
     }
@@ -189,17 +205,64 @@ static struct pipe_texture*
     pipe_reference_init(&tex->tex.reference, 1);
     tex->tex.screen = screen;
 
-    tex->stride = *stride;
+    tex->stride_override = *stride;
 
-    /* XXX */
-    r300_setup_texture_state(tex, tex->tex.width[0], tex->tex.height[0],
-            tex->stride, 0);
+    r300_setup_texture_state(tex);
 
     pipe_buffer_reference(&tex->buffer, buffer);
 
     return (struct pipe_texture*)tex;
 }
 
+static struct pipe_video_surface *
+r300_video_surface_create(struct pipe_screen *screen,
+                          enum pipe_video_chroma_format chroma_format,
+                          unsigned width, unsigned height)
+{
+    struct r300_video_surface *r300_vsfc;
+    struct pipe_texture template;
+
+    assert(screen);
+    assert(width && height);
+
+    r300_vsfc = CALLOC_STRUCT(r300_video_surface);
+    if (!r300_vsfc)
+       return NULL;
+
+    pipe_reference_init(&r300_vsfc->base.reference, 1);
+    r300_vsfc->base.screen = screen;
+    r300_vsfc->base.chroma_format = chroma_format;
+    r300_vsfc->base.width = width;
+    r300_vsfc->base.height = height;
+
+    memset(&template, 0, sizeof(struct pipe_texture));
+    template.target = PIPE_TEXTURE_2D;
+    template.format = PIPE_FORMAT_X8R8G8B8_UNORM;
+    template.last_level = 0;
+    template.width[0] = util_next_power_of_two(width);
+    template.height[0] = util_next_power_of_two(height);
+    template.depth[0] = 1;
+    pf_get_block(template.format, &template.block);
+    template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER |
+                         PIPE_TEXTURE_USAGE_RENDER_TARGET;
+
+    r300_vsfc->tex = screen->texture_create(screen, &template);
+    if (!r300_vsfc->tex)
+    {
+        FREE(r300_vsfc);
+        return NULL;
+    }
+
+    return &r300_vsfc->base;
+}
+
+static void r300_video_surface_destroy(struct pipe_video_surface *vsfc)
+{
+    struct r300_video_surface *r300_vsfc = r300_video_surface(vsfc);
+    pipe_texture_reference(&r300_vsfc->tex, NULL);
+    FREE(r300_vsfc);
+}
+
 void r300_init_screen_texture_functions(struct pipe_screen* screen)
 {
     screen->texture_create = r300_texture_create;
@@ -207,6 +270,9 @@ void r300_init_screen_texture_functions(struct pipe_screen* screen)
     screen->get_tex_surface = r300_get_tex_surface;
     screen->tex_surface_destroy = r300_tex_surface_destroy;
     screen->texture_blanket = r300_texture_blanket;
+
+    screen->video_surface_create = r300_video_surface_create;
+    screen->video_surface_destroy= r300_video_surface_destroy;
 }
 
 boolean r300_get_texture_buffer(struct pipe_texture* texture,
@@ -221,7 +287,7 @@ boolean r300_get_texture_buffer(struct pipe_texture* texture,
     pipe_buffer_reference(buffer, tex->buffer);
 
     if (stride) {
-        *stride = tex->stride;
+        *stride = r300_texture_get_stride(tex, 0);
     }
 
     return TRUE;
diff --git a/src/gallium/drivers/r300/r300_texture.h b/src/gallium/drivers/r300/r300_texture.h
index 3b56f0307c..55d1a0ac5c 100644
--- a/src/gallium/drivers/r300/r300_texture.h
+++ b/src/gallium/drivers/r300/r300_texture.h
@@ -23,15 +23,16 @@
 #ifndef R300_TEXTURE_H
 #define R300_TEXTURE_H
 
-#include "pipe/p_screen.h"
+#include "pipe/p_video_state.h"
 
-#include "util/u_math.h"
-
-#include "r300_context.h"
 #include "r300_reg.h"
 
+struct r300_texture;
+
 void r300_init_screen_texture_functions(struct pipe_screen* screen);
 
+unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level);
+
 /* Note the signature of R300_EASY_TX_FORMAT(A, R, G, B, FORMAT)... */
 static INLINE uint32_t r300_translate_texformat(enum pipe_format format)
 {
@@ -39,11 +40,23 @@ static INLINE uint32_t r300_translate_texformat(enum pipe_format format)
         /* X8 */
         case PIPE_FORMAT_I8_UNORM:
             return R300_EASY_TX_FORMAT(X, X, X, X, X8);
+        /* X16 */
+        case PIPE_FORMAT_R16_UNORM:
+            return R300_EASY_TX_FORMAT(X, X, X, X, X16);
+        case PIPE_FORMAT_R16_SNORM:
+            return R300_EASY_TX_FORMAT(X, X, X, X, X16) |
+                R300_TX_FORMAT_SIGNED;
+        case PIPE_FORMAT_Z16_UNORM:
+            return R300_EASY_TX_FORMAT(X, X, X, X, X16);
         /* W8Z8Y8X8 */
         case PIPE_FORMAT_A8R8G8B8_UNORM:
             return R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
         case PIPE_FORMAT_R8G8B8A8_UNORM:
             return R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8);
+        case PIPE_FORMAT_X8R8G8B8_UNORM:
+            return R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
+        case PIPE_FORMAT_R8G8B8X8_UNORM:
+            return R300_EASY_TX_FORMAT(Y, Z, ONE, X, W8Z8Y8X8);
         case PIPE_FORMAT_A8R8G8B8_SRGB:
             return R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8) |
                 R300_TX_FORMAT_GAMMA;
@@ -67,6 +80,7 @@ static INLINE uint32_t r300_translate_texformat(enum pipe_format format)
                 R300_TX_FORMAT_YUV_TO_RGB;
         /* W24_FP */
         case PIPE_FORMAT_Z24S8_UNORM:
+        case PIPE_FORMAT_Z24X8_UNORM:
             return R300_EASY_TX_FORMAT(X, X, X, X, W24_FP);
         default:
             debug_printf("r300: Implementation error: "
@@ -78,6 +92,18 @@ static INLINE uint32_t r300_translate_texformat(enum pipe_format format)
     return 0;
 }
 
+struct r300_video_surface
+{
+    struct pipe_video_surface   base;
+    struct pipe_texture         *tex;
+};
+
+static INLINE struct r300_video_surface *
+r300_video_surface(struct pipe_video_surface *pvs)
+{
+    return (struct r300_video_surface *)pvs;
+}
+
 #ifndef R300_WINSYS_H
 
 boolean r300_get_texture_buffer(struct pipe_texture* texture,
diff --git a/src/gallium/drivers/r300/r300_tgsi_to_rc.c b/src/gallium/drivers/r300/r300_tgsi_to_rc.c
index d68a104106..74d4fb5087 100644
--- a/src/gallium/drivers/r300/r300_tgsi_to_rc.c
+++ b/src/gallium/drivers/r300/r300_tgsi_to_rc.c
@@ -33,151 +33,150 @@
 static unsigned translate_opcode(unsigned opcode)
 {
     switch(opcode) {
-        case TGSI_OPCODE_ARL: return OPCODE_ARL;
-        case TGSI_OPCODE_MOV: return OPCODE_MOV;
-        case TGSI_OPCODE_LIT: return OPCODE_LIT;
-        case TGSI_OPCODE_RCP: return OPCODE_RCP;
-        case TGSI_OPCODE_RSQ: return OPCODE_RSQ;
-        case TGSI_OPCODE_EXP: return OPCODE_EXP;
-        case TGSI_OPCODE_LOG: return OPCODE_LOG;
-        case TGSI_OPCODE_MUL: return OPCODE_MUL;
-        case TGSI_OPCODE_ADD: return OPCODE_ADD;
-        case TGSI_OPCODE_DP3: return OPCODE_DP3;
-        case TGSI_OPCODE_DP4: return OPCODE_DP4;
-        case TGSI_OPCODE_DST: return OPCODE_DST;
-        case TGSI_OPCODE_MIN: return OPCODE_MIN;
-        case TGSI_OPCODE_MAX: return OPCODE_MAX;
-        case TGSI_OPCODE_SLT: return OPCODE_SLT;
-        case TGSI_OPCODE_SGE: return OPCODE_SGE;
-        case TGSI_OPCODE_MAD: return OPCODE_MAD;
-        case TGSI_OPCODE_SUB: return OPCODE_SUB;
-        case TGSI_OPCODE_LRP: return OPCODE_LRP;
-     /* case TGSI_OPCODE_CND: return OPCODE_CND; */
-        case TGSI_OPCODE_DP2A: return OPCODE_DP2A;
+        case TGSI_OPCODE_ARL: return RC_OPCODE_ARL;
+        case TGSI_OPCODE_MOV: return RC_OPCODE_MOV;
+        case TGSI_OPCODE_LIT: return RC_OPCODE_LIT;
+        case TGSI_OPCODE_RCP: return RC_OPCODE_RCP;
+        case TGSI_OPCODE_RSQ: return RC_OPCODE_RSQ;
+        case TGSI_OPCODE_EXP: return RC_OPCODE_EXP;
+        case TGSI_OPCODE_LOG: return RC_OPCODE_LOG;
+        case TGSI_OPCODE_MUL: return RC_OPCODE_MUL;
+        case TGSI_OPCODE_ADD: return RC_OPCODE_ADD;
+        case TGSI_OPCODE_DP3: return RC_OPCODE_DP3;
+        case TGSI_OPCODE_DP4: return RC_OPCODE_DP4;
+        case TGSI_OPCODE_DST: return RC_OPCODE_DST;
+        case TGSI_OPCODE_MIN: return RC_OPCODE_MIN;
+        case TGSI_OPCODE_MAX: return RC_OPCODE_MAX;
+        case TGSI_OPCODE_SLT: return RC_OPCODE_SLT;
+        case TGSI_OPCODE_SGE: return RC_OPCODE_SGE;
+        case TGSI_OPCODE_MAD: return RC_OPCODE_MAD;
+        case TGSI_OPCODE_SUB: return RC_OPCODE_SUB;
+        case TGSI_OPCODE_LRP: return RC_OPCODE_LRP;
+     /* case TGSI_OPCODE_CND: return RC_OPCODE_CND; */
+     /* case TGSI_OPCODE_CND0: return RC_OPCODE_CND0; */
+     /* case TGSI_OPCODE_DP2A: return RC_OPCODE_DP2A; */
                                         /* gap */
-        case TGSI_OPCODE_FRC: return OPCODE_FRC;
-     /* case TGSI_OPCODE_CLAMP: return OPCODE_CLAMP; */
-        case TGSI_OPCODE_FLR: return OPCODE_FLR;
-     /* case TGSI_OPCODE_ROUND: return OPCODE_ROUND; */
-        case TGSI_OPCODE_EX2: return OPCODE_EX2;
-        case TGSI_OPCODE_LG2: return OPCODE_LG2;
-        case TGSI_OPCODE_POW: return OPCODE_POW;
-        case TGSI_OPCODE_XPD: return OPCODE_XPD;
+        case TGSI_OPCODE_FRC: return RC_OPCODE_FRC;
+     /* case TGSI_OPCODE_CLAMP: return RC_OPCODE_CLAMP; */
+        case TGSI_OPCODE_FLR: return RC_OPCODE_FLR;
+     /* case TGSI_OPCODE_ROUND: return RC_OPCODE_ROUND; */
+        case TGSI_OPCODE_EX2: return RC_OPCODE_EX2;
+        case TGSI_OPCODE_LG2: return RC_OPCODE_LG2;
+        case TGSI_OPCODE_POW: return RC_OPCODE_POW;
+        case TGSI_OPCODE_XPD: return RC_OPCODE_XPD;
                                         /* gap */
-        case TGSI_OPCODE_ABS: return OPCODE_ABS;
-        case TGSI_OPCODE_RCC: return OPCODE_RCC;
-        case TGSI_OPCODE_DPH: return OPCODE_DPH;
-        case TGSI_OPCODE_COS: return OPCODE_COS;
-        case TGSI_OPCODE_DDX: return OPCODE_DDX;
-        case TGSI_OPCODE_DDY: return OPCODE_DDY;
-     /* case TGSI_OPCODE_KILP: return OPCODE_KILP; */
-        case TGSI_OPCODE_PK2H: return OPCODE_PK2H;
-        case TGSI_OPCODE_PK2US: return OPCODE_PK2US;
-        case TGSI_OPCODE_PK4B: return OPCODE_PK4B;
-        case TGSI_OPCODE_PK4UB: return OPCODE_PK4UB;
-        case TGSI_OPCODE_RFL: return OPCODE_RFL;
-        case TGSI_OPCODE_SEQ: return OPCODE_SEQ;
-        case TGSI_OPCODE_SFL: return OPCODE_SFL;
-        case TGSI_OPCODE_SGT: return OPCODE_SGT;
-        case TGSI_OPCODE_SIN: return OPCODE_SIN;
-        case TGSI_OPCODE_SLE: return OPCODE_SLE;
-        case TGSI_OPCODE_SNE: return OPCODE_SNE;
-        case TGSI_OPCODE_STR: return OPCODE_STR;
-        case TGSI_OPCODE_TEX: return OPCODE_TEX;
-        case TGSI_OPCODE_TXD: return OPCODE_TXD;
-        case TGSI_OPCODE_TXP: return OPCODE_TXP;
-        case TGSI_OPCODE_UP2H: return OPCODE_UP2H;
-        case TGSI_OPCODE_UP2US: return OPCODE_UP2US;
-        case TGSI_OPCODE_UP4B: return OPCODE_UP4B;
-        case TGSI_OPCODE_UP4UB: return OPCODE_UP4UB;
-        case TGSI_OPCODE_X2D: return OPCODE_X2D;
-        case TGSI_OPCODE_ARA: return OPCODE_ARA;
-        case TGSI_OPCODE_ARR: return OPCODE_ARR;
-        case TGSI_OPCODE_BRA: return OPCODE_BRA;
-        case TGSI_OPCODE_CAL: return OPCODE_CAL;
-        case TGSI_OPCODE_RET: return OPCODE_RET;
-        case TGSI_OPCODE_SSG: return OPCODE_SSG;
-        case TGSI_OPCODE_CMP: return OPCODE_CMP;
-        case TGSI_OPCODE_SCS: return OPCODE_SCS;
-        case TGSI_OPCODE_TXB: return OPCODE_TXB;
-     /* case TGSI_OPCODE_NRM: return OPCODE_NRM; */
-     /* case TGSI_OPCODE_DIV: return OPCODE_DIV; */
-        case TGSI_OPCODE_DP2: return OPCODE_DP2;
-        case TGSI_OPCODE_TXL: return OPCODE_TXL;
-        case TGSI_OPCODE_BRK: return OPCODE_BRK;
-        case TGSI_OPCODE_IF: return OPCODE_IF;
-     /* case TGSI_OPCODE_LOOP: return OPCODE_LOOP; */
-     /* case TGSI_OPCODE_REP: return OPCODE_REP; */
-        case TGSI_OPCODE_ELSE: return OPCODE_ELSE;
-        case TGSI_OPCODE_ENDIF: return OPCODE_ENDIF;
-        case TGSI_OPCODE_ENDLOOP: return OPCODE_ENDLOOP;
-     /* case TGSI_OPCODE_ENDREP: return OPCODE_ENDREP; */
-        case TGSI_OPCODE_PUSHA: return OPCODE_PUSHA;
-        case TGSI_OPCODE_POPA: return OPCODE_POPA;
-     /* case TGSI_OPCODE_CEIL: return OPCODE_CEIL; */
-     /* case TGSI_OPCODE_I2F: return OPCODE_I2F; */
-        case TGSI_OPCODE_NOT: return OPCODE_NOT;
-        case TGSI_OPCODE_TRUNC: return OPCODE_TRUNC;
-     /* case TGSI_OPCODE_SHL: return OPCODE_SHL; */
-     /* case TGSI_OPCODE_SHR: return OPCODE_SHR; */
-        case TGSI_OPCODE_AND: return OPCODE_AND;
-        case TGSI_OPCODE_OR: return OPCODE_OR;
-     /* case TGSI_OPCODE_MOD: return OPCODE_MOD; */
-        case TGSI_OPCODE_XOR: return OPCODE_XOR;
-     /* case TGSI_OPCODE_SAD: return OPCODE_SAD; */
-     /* case TGSI_OPCODE_TXF: return OPCODE_TXF; */
-     /* case TGSI_OPCODE_TXQ: return OPCODE_TXQ; */
-        case TGSI_OPCODE_CONT: return OPCODE_CONT;
-     /* case TGSI_OPCODE_EMIT: return OPCODE_EMIT; */
-     /* case TGSI_OPCODE_ENDPRIM: return OPCODE_ENDPRIM; */
-     /* case TGSI_OPCODE_BGNLOOP2: return OPCODE_BGNLOOP2; */
-        case TGSI_OPCODE_BGNSUB: return OPCODE_BGNSUB;
-     /* case TGSI_OPCODE_ENDLOOP2: return OPCODE_ENDLOOP2; */
-        case TGSI_OPCODE_ENDSUB: return OPCODE_ENDSUB;
-        case TGSI_OPCODE_NOISE1: return OPCODE_NOISE1;
-        case TGSI_OPCODE_NOISE2: return OPCODE_NOISE2;
-        case TGSI_OPCODE_NOISE3: return OPCODE_NOISE3;
-        case TGSI_OPCODE_NOISE4: return OPCODE_NOISE4;
-        case TGSI_OPCODE_NOP: return OPCODE_NOP;
+        case TGSI_OPCODE_ABS: return RC_OPCODE_ABS;
+     /* case TGSI_OPCODE_RCC: return RC_OPCODE_RCC; */
+        case TGSI_OPCODE_DPH: return RC_OPCODE_DPH;
+        case TGSI_OPCODE_COS: return RC_OPCODE_COS;
+        case TGSI_OPCODE_DDX: return RC_OPCODE_DDX;
+        case TGSI_OPCODE_DDY: return RC_OPCODE_DDY;
+     /* case TGSI_OPCODE_KILP: return RC_OPCODE_KILP; */
+     /* case TGSI_OPCODE_PK2H: return RC_OPCODE_PK2H; */
+     /* case TGSI_OPCODE_PK2US: return RC_OPCODE_PK2US; */
+     /* case TGSI_OPCODE_PK4B: return RC_OPCODE_PK4B; */
+     /* case TGSI_OPCODE_PK4UB: return RC_OPCODE_PK4UB; */
+     /* case TGSI_OPCODE_RFL: return RC_OPCODE_RFL; */
+        case TGSI_OPCODE_SEQ: return RC_OPCODE_SEQ;
+        case TGSI_OPCODE_SFL: return RC_OPCODE_SFL;
+        case TGSI_OPCODE_SGT: return RC_OPCODE_SGT;
+        case TGSI_OPCODE_SIN: return RC_OPCODE_SIN;
+        case TGSI_OPCODE_SLE: return RC_OPCODE_SLE;
+        case TGSI_OPCODE_SNE: return RC_OPCODE_SNE;
+     /* case TGSI_OPCODE_STR: return RC_OPCODE_STR; */
+        case TGSI_OPCODE_TEX: return RC_OPCODE_TEX;
+        case TGSI_OPCODE_TXD: return RC_OPCODE_TXD;
+        case TGSI_OPCODE_TXP: return RC_OPCODE_TXP;
+     /* case TGSI_OPCODE_UP2H: return RC_OPCODE_UP2H; */
+     /* case TGSI_OPCODE_UP2US: return RC_OPCODE_UP2US; */
+     /* case TGSI_OPCODE_UP4B: return RC_OPCODE_UP4B; */
+     /* case TGSI_OPCODE_UP4UB: return RC_OPCODE_UP4UB; */
+     /* case TGSI_OPCODE_X2D: return RC_OPCODE_X2D; */
+     /* case TGSI_OPCODE_ARA: return RC_OPCODE_ARA; */
+     /* case TGSI_OPCODE_ARR: return RC_OPCODE_ARR; */
+     /* case TGSI_OPCODE_BRA: return RC_OPCODE_BRA; */
+     /* case TGSI_OPCODE_CAL: return RC_OPCODE_CAL; */
+     /* case TGSI_OPCODE_RET: return RC_OPCODE_RET; */
+     /* case TGSI_OPCODE_SSG: return RC_OPCODE_SSG; */
+        case TGSI_OPCODE_CMP: return RC_OPCODE_CMP;
+        case TGSI_OPCODE_SCS: return RC_OPCODE_SCS;
+        case TGSI_OPCODE_TXB: return RC_OPCODE_TXB;
+     /* case TGSI_OPCODE_NRM: return RC_OPCODE_NRM; */
+     /* case TGSI_OPCODE_DIV: return RC_OPCODE_DIV; */
+     /* case TGSI_OPCODE_DP2: return RC_OPCODE_DP2; */
+        case TGSI_OPCODE_TXL: return RC_OPCODE_TXL;
+     /* case TGSI_OPCODE_BRK: return RC_OPCODE_BRK; */
+        case TGSI_OPCODE_IF: return RC_OPCODE_IF;
+     /* case TGSI_OPCODE_LOOP: return RC_OPCODE_LOOP; */
+     /* case TGSI_OPCODE_REP: return RC_OPCODE_REP; */
+        case TGSI_OPCODE_ELSE: return RC_OPCODE_ELSE;
+        case TGSI_OPCODE_ENDIF: return RC_OPCODE_ENDIF;
+     /* case TGSI_OPCODE_ENDLOOP: return RC_OPCODE_ENDLOOP; */
+     /* case TGSI_OPCODE_ENDREP: return RC_OPCODE_ENDREP; */
+     /* case TGSI_OPCODE_PUSHA: return RC_OPCODE_PUSHA; */
+     /* case TGSI_OPCODE_POPA: return RC_OPCODE_POPA; */
+     /* case TGSI_OPCODE_CEIL: return RC_OPCODE_CEIL; */
+     /* case TGSI_OPCODE_I2F: return RC_OPCODE_I2F; */
+     /* case TGSI_OPCODE_NOT: return RC_OPCODE_NOT; */
+     /* case TGSI_OPCODE_TRUNC: return RC_OPCODE_TRUNC; */
+     /* case TGSI_OPCODE_SHL: return RC_OPCODE_SHL; */
+     /* case TGSI_OPCODE_SHR: return RC_OPCODE_SHR; */
+     /* case TGSI_OPCODE_AND: return RC_OPCODE_AND; */
+     /* case TGSI_OPCODE_OR: return RC_OPCODE_OR; */
+     /* case TGSI_OPCODE_MOD: return RC_OPCODE_MOD; */
+     /* case TGSI_OPCODE_XOR: return RC_OPCODE_XOR; */
+     /* case TGSI_OPCODE_SAD: return RC_OPCODE_SAD; */
+     /* case TGSI_OPCODE_TXF: return RC_OPCODE_TXF; */
+     /* case TGSI_OPCODE_TXQ: return RC_OPCODE_TXQ; */
+     /* case TGSI_OPCODE_CONT: return RC_OPCODE_CONT; */
+     /* case TGSI_OPCODE_EMIT: return RC_OPCODE_EMIT; */
+     /* case TGSI_OPCODE_ENDPRIM: return RC_OPCODE_ENDPRIM; */
+     /* case TGSI_OPCODE_BGNLOOP2: return RC_OPCODE_BGNLOOP2; */
+     /* case TGSI_OPCODE_BGNSUB: return RC_OPCODE_BGNSUB; */
+     /* case TGSI_OPCODE_ENDLOOP2: return RC_OPCODE_ENDLOOP2; */
+     /* case TGSI_OPCODE_ENDSUB: return RC_OPCODE_ENDSUB; */
+     /* case TGSI_OPCODE_NOISE1: return RC_OPCODE_NOISE1; */
+     /* case TGSI_OPCODE_NOISE2: return RC_OPCODE_NOISE2; */
+     /* case TGSI_OPCODE_NOISE3: return RC_OPCODE_NOISE3; */
+     /* case TGSI_OPCODE_NOISE4: return RC_OPCODE_NOISE4; */
+        case TGSI_OPCODE_NOP: return RC_OPCODE_NOP;
                                         /* gap */
-        case TGSI_OPCODE_NRM4: return OPCODE_NRM4;
-     /* case TGSI_OPCODE_CALLNZ: return OPCODE_CALLNZ; */
-     /* case TGSI_OPCODE_IFC: return OPCODE_IFC; */
-     /* case TGSI_OPCODE_BREAKC: return OPCODE_BREAKC; */
-        case TGSI_OPCODE_KIL: return OPCODE_KIL;
-        case TGSI_OPCODE_END: return OPCODE_END;
-        case TGSI_OPCODE_SWZ: return OPCODE_SWZ;
+     /* case TGSI_OPCODE_NRM4: return RC_OPCODE_NRM4; */
+     /* case TGSI_OPCODE_CALLNZ: return RC_OPCODE_CALLNZ; */
+     /* case TGSI_OPCODE_IFC: return RC_OPCODE_IFC; */
+     /* case TGSI_OPCODE_BREAKC: return RC_OPCODE_BREAKC; */
+        case TGSI_OPCODE_KIL: return RC_OPCODE_KIL;
+        case TGSI_OPCODE_SWZ: return RC_OPCODE_SWZ;
     }
 
     fprintf(stderr, "Unknown opcode: %i\n", opcode);
-    abort();
+    return RC_OPCODE_ILLEGAL_OPCODE;
 }
 
 static unsigned translate_saturate(unsigned saturate)
 {
     switch(saturate) {
-        case TGSI_SAT_NONE: return SATURATE_OFF;
-        case TGSI_SAT_ZERO_ONE: return SATURATE_ZERO_ONE;
-        case TGSI_SAT_MINUS_PLUS_ONE: return SATURATE_PLUS_MINUS_ONE;
+        default:
+            fprintf(stderr, "Unknown saturate mode: %i\n", saturate);
+            /* fall-through */
+        case TGSI_SAT_NONE: return RC_SATURATE_NONE;
+        case TGSI_SAT_ZERO_ONE: return RC_SATURATE_ZERO_ONE;
     }
-
-    fprintf(stderr, "Unknown saturate mode: %i\n", saturate);
-    abort();
 }
 
 static unsigned translate_register_file(unsigned file)
 {
     switch(file) {
-        case TGSI_FILE_CONSTANT: return PROGRAM_CONSTANT;
-        case TGSI_FILE_IMMEDIATE: return PROGRAM_CONSTANT;
-        case TGSI_FILE_INPUT: return PROGRAM_INPUT;
-        case TGSI_FILE_OUTPUT: return PROGRAM_OUTPUT;
-        case TGSI_FILE_TEMPORARY: return PROGRAM_TEMPORARY;
-        case TGSI_FILE_ADDRESS: return PROGRAM_ADDRESS;
+        case TGSI_FILE_CONSTANT: return RC_FILE_CONSTANT;
+        case TGSI_FILE_IMMEDIATE: return RC_FILE_CONSTANT;
+        case TGSI_FILE_INPUT: return RC_FILE_INPUT;
+        case TGSI_FILE_OUTPUT: return RC_FILE_OUTPUT;
+        default:
+            fprintf(stderr, "Unhandled register file: %i\n", file);
+            /* fall-through */
+        case TGSI_FILE_TEMPORARY: return RC_FILE_TEMPORARY;
+        case TGSI_FILE_ADDRESS: return RC_FILE_ADDRESS;
     }
-
-    fprintf(stderr, "Unhandled register file: %i\n", file);
-    abort();
 }
 
 static int translate_register_index(
@@ -193,7 +192,7 @@ static int translate_register_index(
 
 static void transform_dstreg(
     struct tgsi_to_rc * ttr,
-    struct prog_dst_register * dst,
+    struct rc_dst_register * dst,
     struct tgsi_full_dst_register * src)
 {
     dst->File = translate_register_file(src->DstRegister.File);
@@ -204,7 +203,7 @@ static void transform_dstreg(
 
 static void transform_srcreg(
     struct tgsi_to_rc * ttr,
-    struct prog_src_register * dst,
+    struct rc_src_register * dst,
     struct tgsi_full_src_register * src)
 {
     dst->File = translate_register_file(src->SrcRegister.File);
@@ -220,61 +219,62 @@ static void transform_srcreg(
         (src->SrcRegisterExtSwz.NegateY << 1) |
         (src->SrcRegisterExtSwz.NegateZ << 2) |
         (src->SrcRegisterExtSwz.NegateW << 3);
-    dst->Negate ^= src->SrcRegister.Negate ? NEGATE_XYZW : 0;
+    dst->Negate ^= src->SrcRegister.Negate ? RC_MASK_XYZW : 0;
 }
 
 static void transform_texture(struct rc_instruction * dst, struct tgsi_instruction_ext_texture src)
 {
     switch(src.Texture) {
         case TGSI_TEXTURE_1D:
-            dst->I.TexSrcTarget = TEXTURE_1D_INDEX;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_1D;
             break;
         case TGSI_TEXTURE_2D:
-            dst->I.TexSrcTarget = TEXTURE_2D_INDEX;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_2D;
             break;
         case TGSI_TEXTURE_3D:
-            dst->I.TexSrcTarget = TEXTURE_3D_INDEX;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_3D;
             break;
         case TGSI_TEXTURE_CUBE:
-            dst->I.TexSrcTarget = TEXTURE_CUBE_INDEX;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_CUBE;
             break;
         case TGSI_TEXTURE_RECT:
-            dst->I.TexSrcTarget = TEXTURE_RECT_INDEX;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_RECT;
             break;
         case TGSI_TEXTURE_SHADOW1D:
-            dst->I.TexSrcTarget = TEXTURE_1D_INDEX;
-            dst->I.TexShadow = 1;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_1D;
+            dst->U.I.TexShadow = 1;
             break;
         case TGSI_TEXTURE_SHADOW2D:
-            dst->I.TexSrcTarget = TEXTURE_2D_INDEX;
-            dst->I.TexShadow = 1;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_2D;
+            dst->U.I.TexShadow = 1;
             break;
         case TGSI_TEXTURE_SHADOWRECT:
-            dst->I.TexSrcTarget = TEXTURE_RECT_INDEX;
-            dst->I.TexShadow = 1;
+            dst->U.I.TexSrcTarget = RC_TEXTURE_RECT;
+            dst->U.I.TexShadow = 1;
             break;
     }
 }
 
 static void transform_instruction(struct tgsi_to_rc * ttr, struct tgsi_full_instruction * src)
 {
+    struct rc_instruction * dst;
+    int i;
+
     if (src->Instruction.Opcode == TGSI_OPCODE_END)
         return;
 
-    struct rc_instruction * dst = rc_insert_new_instruction(ttr->compiler, ttr->compiler->Program.Instructions.Prev);
-    int i;
-
-    dst->I.Opcode = translate_opcode(src->Instruction.Opcode);
-    dst->I.SaturateMode = translate_saturate(src->Instruction.Saturate);
+    dst = rc_insert_new_instruction(ttr->compiler, ttr->compiler->Program.Instructions.Prev);
+    dst->U.I.Opcode = translate_opcode(src->Instruction.Opcode);
+    dst->U.I.SaturateMode = translate_saturate(src->Instruction.Saturate);
 
     if (src->Instruction.NumDstRegs)
-        transform_dstreg(ttr, &dst->I.DstReg, &src->FullDstRegisters[0]);
+        transform_dstreg(ttr, &dst->U.I.DstReg, &src->FullDstRegisters[0]);
 
     for(i = 0; i < src->Instruction.NumSrcRegs; ++i) {
         if (src->FullSrcRegisters[i].SrcRegister.File == TGSI_FILE_SAMPLER)
-            dst->I.TexSrcUnit = src->FullSrcRegisters[i].SrcRegister.Index;
+            dst->U.I.TexSrcUnit = src->FullSrcRegisters[i].SrcRegister.Index;
         else
-            transform_srcreg(ttr, &dst->I.SrcReg[i], &src->FullSrcRegisters[i]);
+            transform_srcreg(ttr, &dst->U.I.SrcReg[i], &src->FullSrcRegisters[i]);
     }
 
     /* Texturing. */
diff --git a/src/gallium/drivers/r300/r300_vs.c b/src/gallium/drivers/r300/r300_vs.c
index 2cb903bba2..eca85879a7 100644
--- a/src/gallium/drivers/r300/r300_vs.c
+++ b/src/gallium/drivers/r300/r300_vs.c
@@ -35,6 +35,8 @@ static void set_vertex_inputs_outputs(struct r300_vertex_program_compiler * c)
 {
     struct r300_vertex_shader * vs = c->UserData;
     struct tgsi_shader_info* info = &vs->info;
+    struct tgsi_parse_context parser;
+    struct tgsi_full_declaration * decl;
     boolean pointsize = false;
     int out_colors = 0;
     int colors = 0;
@@ -62,8 +64,6 @@ static void set_vertex_inputs_outputs(struct r300_vertex_program_compiler * c)
         }
     }
 
-    struct tgsi_parse_context parser;
-
     tgsi_parse_init(&parser, vs->state.tokens);
 
     while (!tgsi_parse_end_of_tokens(&parser)) {
@@ -72,7 +72,7 @@ static void set_vertex_inputs_outputs(struct r300_vertex_program_compiler * c)
         if (parser.FullToken.Token.Type != TGSI_TOKEN_TYPE_DECLARATION)
             continue;
 
-        struct tgsi_full_declaration * decl = &parser.FullToken.FullDeclaration;
+        decl = &parser.FullToken.FullDeclaration;
 
         if (decl->Declaration.File != TGSI_FILE_OUTPUT)
             continue;
@@ -116,7 +116,7 @@ void r300_translate_vertex_shader(struct r300_context* r300,
     /* Setup the compiler */
     rc_init(&compiler.Base);
 
-    compiler.Base.Debug = 1;
+    compiler.Base.Debug = DBG_ON(r300, DBG_VP);
     compiler.code = &vs->code;
     compiler.UserData = vs;
 
@@ -146,89 +146,3 @@ void r300_translate_vertex_shader(struct r300_context* r300,
     rc_destroy(&compiler.Base);
     vs->translated = TRUE;
 }
-
-
-/* XXX get these to r300_reg */
-#define R300_PVS_DST_OPCODE(x)   ((x) << 0)
-#   define R300_VE_DOT_PRODUCT            1
-#   define R300_VE_MULTIPLY               2
-#   define R300_VE_ADD                    3
-#   define R300_VE_MAXIMUM                7
-#   define R300_VE_SET_LESS_THAN          10
-#define R300_PVS_DST_MATH_INST     (1 << 6)
-#   define R300_ME_RECIP_DX               6
-#define R300_PVS_DST_MACRO_INST    (1 << 7)
-#   define R300_PVS_MACRO_OP_2CLK_MADD    0
-#define R300_PVS_DST_REG_TYPE(x) ((x) << 8)
-#   define R300_PVS_DST_REG_TEMPORARY     0
-#   define R300_PVS_DST_REG_A0            1
-#   define R300_PVS_DST_REG_OUT           2
-#   define R300_PVS_DST_REG_OUT_REPL_X    3
-#   define R300_PVS_DST_REG_ALT_TEMPORARY 4
-#   define R300_PVS_DST_REG_INPUT         5
-#define R300_PVS_DST_OFFSET(x)   ((x) << 13)
-#define R300_PVS_DST_WE(x)       ((x) << 20)
-#define R300_PVS_DST_WE_XYZW     (0xf << 20)
-
-#define R300_PVS_SRC_REG_TYPE(x) ((x) << 0)
-#   define R300_PVS_SRC_REG_TEMPORARY     0
-#   define R300_PVS_SRC_REG_INPUT         1
-#   define R300_PVS_SRC_REG_CONSTANT      2
-#   define R300_PVS_SRC_REG_ALT_TEMPORARY 3
-#define R300_PVS_SRC_OFFSET(x)   ((x) << 5)
-#define R300_PVS_SRC_SWIZZLE(x)  ((x) << 13)
-#   define R300_PVS_SRC_SELECT_X          0
-#   define R300_PVS_SRC_SELECT_Y          1
-#   define R300_PVS_SRC_SELECT_Z          2
-#   define R300_PVS_SRC_SELECT_W          3
-#   define R300_PVS_SRC_SELECT_FORCE_0    4
-#   define R300_PVS_SRC_SELECT_FORCE_1    5
-#   define R300_PVS_SRC_SWIZZLE_XYZW \
-    ((R300_PVS_SRC_SELECT_X | (R300_PVS_SRC_SELECT_Y << 3) | \
-     (R300_PVS_SRC_SELECT_Z << 6) | (R300_PVS_SRC_SELECT_W << 9)) << 13)
-#   define R300_PVS_SRC_SWIZZLE_ZERO \
-    ((R300_PVS_SRC_SELECT_FORCE_0 | (R300_PVS_SRC_SELECT_FORCE_0 << 3) | \
-     (R300_PVS_SRC_SELECT_FORCE_0 << 6) | \
-      (R300_PVS_SRC_SELECT_FORCE_0 << 9)) << 13)
-#   define R300_PVS_SRC_SWIZZLE_ONE \
-    ((R300_PVS_SRC_SELECT_FORCE_1 | (R300_PVS_SRC_SELECT_FORCE_1 << 3) | \
-     (R300_PVS_SRC_SELECT_FORCE_1 << 6) | \
-      (R300_PVS_SRC_SELECT_FORCE_1 << 9)) << 13)
-#define R300_PVS_MODIFIER_X        (1 << 25)
-#define R300_PVS_MODIFIER_Y        (1 << 26)
-#define R300_PVS_MODIFIER_Z        (1 << 27)
-#define R300_PVS_MODIFIER_W        (1 << 28)
-#define R300_PVS_NEGATE_XYZW \
-    (R300_PVS_MODIFIER_X | R300_PVS_MODIFIER_Y | \
-     R300_PVS_MODIFIER_Z | R300_PVS_MODIFIER_W)
-
-struct r300_vertex_program_code r300_passthrough_vertex_shader = {
-    .length = 8, /* two instructions */
-
-    /* MOV out[0], in[0] */
-    .body.d[0] = R300_PVS_DST_OPCODE(R300_VE_ADD) |
-        R300_PVS_DST_REG_TYPE(R300_PVS_DST_REG_OUT) |
-        R300_PVS_DST_OFFSET(0) | R300_PVS_DST_WE_XYZW,
-    .body.d[1] = R300_PVS_SRC_REG_TYPE(R300_PVS_SRC_REG_INPUT) |
-        R300_PVS_SRC_OFFSET(0) | R300_PVS_SRC_SWIZZLE_XYZW,
-    .body.d[2] = R300_PVS_SRC_SWIZZLE_ZERO,
-    .body.d[3] = 0x0,
-
-    /* MOV out[1], in[1] */
-    .body.d[4] = R300_PVS_DST_OPCODE(R300_VE_ADD) |
-        R300_PVS_DST_REG_TYPE(R300_PVS_DST_REG_OUT) |
-        R300_PVS_DST_OFFSET(1) | R300_PVS_DST_WE_XYZW,
-    .body.d[5] = R300_PVS_SRC_REG_TYPE(R300_PVS_SRC_REG_INPUT) |
-        R300_PVS_SRC_OFFSET(1) | R300_PVS_SRC_SWIZZLE_XYZW,
-    .body.d[6] = R300_PVS_SRC_SWIZZLE_ZERO,
-    .body.d[7] = 0x0,
-
-    .inputs[0] = 0,
-    .inputs[1] = 1,
-    .outputs[0] = 0,
-    .outputs[1] = 1,
-
-    .InputsRead = 3,
-    .OutputsWritten = 3
-};
-
diff --git a/src/gallium/drivers/r300/r300_winsys.h b/src/gallium/drivers/r300/r300_winsys.h
index f18ad75a47..864a6146b2 100644
--- a/src/gallium/drivers/r300/r300_winsys.h
+++ b/src/gallium/drivers/r300/r300_winsys.h
@@ -48,6 +48,9 @@ struct r300_winsys {
     /* GB pipe count */
     uint32_t gb_pipes;
 
+    /* Z pipe count (rv530 only) */
+    uint32_t z_pipes;
+
     /* GART size. */
     uint32_t gart_size;
 
@@ -92,6 +95,12 @@ struct r300_winsys {
 
     /* Flush the CS. */
     void (*flush_cs)(struct r300_winsys* winsys);
+
+    /* winsys flush - callback from winsys when flush required */
+    void (*set_flush_cb)(struct r300_winsys *winsys,
+			 void (*flush_cb)(void *), void *data);
+
+    void (*reset_bos)(struct r300_winsys *winsys);
 };
 
 struct pipe_context* r300_create_context(struct pipe_screen* screen,
diff --git a/src/gallium/drivers/r300/r3xx_fs.c b/src/gallium/drivers/r300/r3xx_fs.c
deleted file mode 100644
index c1c1194d58..0000000000
--- a/src/gallium/drivers/r300/r3xx_fs.c
+++ /dev/null
@@ -1,74 +0,0 @@
-/*
- * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
- *                Joakim Sindholt <opensource@zhasha.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#include "r3xx_fs.h"
-
-#include "r300_reg.h"
-
-struct rX00_fragment_program_code r3xx_passthrough_fragment_shader = {
-    .code.r300.alu.length = 1,
-    .code.r300.tex.length = 0,
-
-    .code.r300.config = 0,
-    .code.r300.pixsize = 0,
-    .code.r300.code_offset = 0,
-    .code.r300.code_addr[3] = R300_RGBA_OUT,
-
-    .code.r300.alu.inst[0].rgb_inst = R300_RGB_SWIZA(R300_ALU_ARGC_SRC0C_XYZ) |
-        R300_RGB_SWIZB(R300_ALU_ARGC_SRC0C_XYZ) |
-        R300_RGB_SWIZC(R300_ALU_ARGC_ZERO) |
-        R300_ALU_OUTC_CMP,
-    .code.r300.alu.inst[0].rgb_addr = R300_RGB_ADDR0(0) | R300_RGB_ADDR1(0) |
-        R300_RGB_ADDR2(0) | R300_ALU_DSTC_OUTPUT_XYZ,
-    .code.r300.alu.inst[0].alpha_inst = R300_ALPHA_SWIZA(R300_ALU_ARGA_SRC0A) |
-        R300_ALPHA_SWIZB(R300_ALU_ARGA_SRC0A) |
-        R300_ALPHA_SWIZC(R300_ALU_ARGA_ZERO) |
-        R300_ALU_OUTA_CMP,
-    .code.r300.alu.inst[0].alpha_addr = R300_ALPHA_ADDR0(0) |
-        R300_ALPHA_ADDR1(0) | R300_ALPHA_ADDR2(0) | R300_ALU_DSTA_OUTPUT,
-};
-
-struct rX00_fragment_program_code r3xx_texture_fragment_shader = {
-    .code.r300.alu.length = 1,
-    .code.r300.tex.length = 1,
-
-    .code.r300.config = R300_PFS_CNTL_FIRST_NODE_HAS_TEX,
-    .code.r300.pixsize = 0,
-    .code.r300.code_offset = 0,
-    .code.r300.code_addr[3] = R300_RGBA_OUT,
-
-    .code.r300.tex.inst[0] = R300_TEX_OP_LD << R300_TEX_INST_SHIFT,
-
-    .code.r300.alu.inst[0].rgb_inst = R300_RGB_SWIZA(R300_ALU_ARGC_SRC0C_XYZ) |
-        R300_RGB_SWIZB(R300_ALU_ARGC_SRC0C_XYZ) |
-        R300_RGB_SWIZC(R300_ALU_ARGC_ZERO) |
-        R300_ALU_OUTC_CMP,
-    .code.r300.alu.inst[0].rgb_addr = R300_RGB_ADDR0(0) | R300_RGB_ADDR1(0) |
-        R300_RGB_ADDR2(0) | R300_ALU_DSTC_OUTPUT_XYZ,
-    .code.r300.alu.inst[0].alpha_inst = R300_ALPHA_SWIZA(R300_ALU_ARGA_SRC0A) |
-        R300_ALPHA_SWIZB(R300_ALU_ARGA_SRC0A) |
-        R300_ALPHA_SWIZC(R300_ALU_ARGA_ZERO) |
-        R300_ALU_OUTA_CMP,
-    .code.r300.alu.inst[0].alpha_addr = R300_ALPHA_ADDR0(0) |
-        R300_ALPHA_ADDR1(0) | R300_ALPHA_ADDR2(0) | R300_ALU_DSTA_OUTPUT,
-};
diff --git a/src/gallium/drivers/r300/r3xx_fs.h b/src/gallium/drivers/r300/r3xx_fs.h
deleted file mode 100644
index 51cd245724..0000000000
--- a/src/gallium/drivers/r300/r3xx_fs.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/*
- * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
- *                Joakim Sindholt <opensource@zhasha.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#ifndef R3XX_FS_H
-#define R3XX_FS_H
-
-#include "radeon_code.h"
-
-struct rX00_fragment_program_code r3xx_passthrough_fragment_shader;
-struct rX00_fragment_program_code r3xx_texture_fragment_shader;
-
-#endif /* R3XX_FS_H */
diff --git a/src/gallium/drivers/r300/r5xx_fs.c b/src/gallium/drivers/r300/r5xx_fs.c
deleted file mode 100644
index f072deab0d..0000000000
--- a/src/gallium/drivers/r300/r5xx_fs.c
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
- *                Joakim Sindholt <opensource@zhasha.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#include "r5xx_fs.h"
-
-#include "r300_reg.h"
-
-/* XXX this all should find its way back to r300_reg */
-/* Swizzle tools */
-#define R500_SWIZZLE_ZERO 4
-#define R500_SWIZZLE_HALF 5
-#define R500_SWIZZLE_ONE 6
-#define R500_SWIZ_RGB_ZERO ((4 << 0) | (4 << 3) | (4 << 6))
-#define R500_SWIZ_RGB_ONE ((6 << 0) | (6 << 3) | (6 << 6))
-#define R500_SWIZ_RGB_RGB ((0 << 0) | (1 << 3) | (2 << 6))
-#define R500_SWIZ_MOD_NEG 1
-#define R500_SWIZ_MOD_ABS 2
-#define R500_SWIZ_MOD_NEG_ABS 3
-/* Swizzles for inst2 */
-#define R500_SWIZ_TEX_STRQ(x) ((x) << 8)
-#define R500_SWIZ_TEX_RGBA(x) ((x) << 24)
-/* Swizzles for inst3 */
-#define R500_SWIZ_RGB_A(x) ((x) << 2)
-#define R500_SWIZ_RGB_B(x) ((x) << 15)
-/* Swizzles for inst4 */
-#define R500_SWIZ_ALPHA_A(x) ((x) << 14)
-#define R500_SWIZ_ALPHA_B(x) ((x) << 21)
-/* Swizzle for inst5 */
-#define R500_SWIZ_RGBA_C(x) ((x) << 14)
-#define R500_SWIZ_ALPHA_C(x) ((x) << 27)
-/* Writemasks */
-#define R500_TEX_WMASK(x) ((x) << 11)
-#define R500_ALU_WMASK(x) ((x) << 11)
-#define R500_ALU_OMASK(x) ((x) << 15)
-#define R500_W_OMASK (1 << 31)
-
-struct rX00_fragment_program_code r5xx_passthrough_fragment_shader = {
-    .code.r500.max_temp_idx = 0,
-    .code.r500.inst_end = 0,
-
-    .code.r500.inst[0].inst0 = R500_INST_TYPE_OUT |
-        R500_INST_TEX_SEM_WAIT | R500_INST_LAST |
-        R500_INST_RGB_OMASK_RGB | R500_INST_ALPHA_OMASK |
-        R500_INST_RGB_CLAMP | R500_INST_ALPHA_CLAMP,
-    .code.r500.inst[0].inst1 =
-        R500_RGB_ADDR0(0) | R500_RGB_ADDR1(0) | R500_RGB_ADDR1_CONST |
-        R500_RGB_ADDR2(0) | R500_RGB_ADDR2_CONST,
-    .code.r500.inst[0].inst2 =
-        R500_ALPHA_ADDR0(0) | R500_ALPHA_ADDR1(0) | R500_ALPHA_ADDR1_CONST |
-        R500_ALPHA_ADDR2(0) | R500_ALPHA_ADDR2_CONST,
-    .code.r500.inst[0].inst3 =
-        R500_ALU_RGB_SEL_A_SRC0 | R500_ALU_RGB_R_SWIZ_A_R |
-        R500_ALU_RGB_G_SWIZ_A_G | R500_ALU_RGB_B_SWIZ_A_B |
-        R500_ALU_RGB_SEL_B_SRC0 | R500_ALU_RGB_R_SWIZ_B_R |
-        R500_ALU_RGB_B_SWIZ_B_G | R500_ALU_RGB_G_SWIZ_B_B,
-    .code.r500.inst[0].inst4 =
-        R500_ALPHA_OP_CMP | R500_ALPHA_SWIZ_A_A | R500_ALPHA_SWIZ_B_A,
-    .code.r500.inst[0].inst5 =
-        R500_ALU_RGBA_OP_CMP | R500_ALU_RGBA_R_SWIZ_0 |
-        R500_ALU_RGBA_G_SWIZ_0 | R500_ALU_RGBA_B_SWIZ_0 |
-        R500_ALU_RGBA_A_SWIZ_0,
-};
-
-struct rX00_fragment_program_code r5xx_texture_fragment_shader = {
-    .code.r500.max_temp_idx = 0,
-    .code.r500.inst_end = 1,
-
-    .code.r500.inst[0].inst0 = R500_INST_TYPE_TEX |
-        R500_INST_TEX_SEM_WAIT |
-        R500_INST_RGB_WMASK_RGB | R500_INST_ALPHA_WMASK |
-        R500_INST_RGB_CLAMP | R500_INST_ALPHA_CLAMP,
-    .code.r500.inst[0].inst1 = R500_TEX_ID(0) | R500_TEX_INST_LD |
-        R500_TEX_SEM_ACQUIRE | R500_TEX_IGNORE_UNCOVERED,
-    .code.r500.inst[0].inst2 = R500_TEX_SRC_ADDR(0) |
-        R500_TEX_SRC_S_SWIZ_R | R500_TEX_SRC_T_SWIZ_G |
-        R500_TEX_SRC_R_SWIZ_B | R500_TEX_SRC_Q_SWIZ_A |
-        R500_TEX_DST_ADDR(0) |
-        R500_TEX_DST_R_SWIZ_R | R500_TEX_DST_G_SWIZ_G |
-        R500_TEX_DST_B_SWIZ_B | R500_TEX_DST_A_SWIZ_A,
-    .code.r500.inst[0].inst3 = 0x0,
-    .code.r500.inst[0].inst4 = 0x0,
-    .code.r500.inst[0].inst5 = 0x0,
-
-    .code.r500.inst[1].inst0 = R500_INST_TYPE_OUT |
-        R500_INST_TEX_SEM_WAIT | R500_INST_LAST |
-        R500_INST_RGB_OMASK_RGB | R500_INST_ALPHA_OMASK |
-        R500_INST_RGB_CLAMP | R500_INST_ALPHA_CLAMP,
-    .code.r500.inst[1].inst1 =
-        R500_RGB_ADDR0(0) | R500_RGB_ADDR1(0) | R500_RGB_ADDR1_CONST |
-        R500_RGB_ADDR2(0) | R500_RGB_ADDR2_CONST,
-    .code.r500.inst[1].inst2 =
-        R500_ALPHA_ADDR0(0) | R500_ALPHA_ADDR1(0) | R500_ALPHA_ADDR1_CONST |
-        R500_ALPHA_ADDR2(0) | R500_ALPHA_ADDR2_CONST,
-    .code.r500.inst[1].inst3 =
-        R500_ALU_RGB_SEL_A_SRC0 | R500_ALU_RGB_R_SWIZ_A_R |
-        R500_ALU_RGB_G_SWIZ_A_G | R500_ALU_RGB_B_SWIZ_A_B |
-        R500_ALU_RGB_SEL_B_SRC0 | R500_ALU_RGB_R_SWIZ_B_R |
-        R500_ALU_RGB_B_SWIZ_B_G | R500_ALU_RGB_G_SWIZ_B_B,
-    .code.r500.inst[1].inst4 =
-        R500_ALPHA_OP_CMP | R500_ALPHA_SWIZ_A_A | R500_ALPHA_SWIZ_B_A,
-    .code.r500.inst[1].inst5 =
-        R500_ALU_RGBA_OP_CMP | R500_ALU_RGBA_R_SWIZ_0 |
-        R500_ALU_RGBA_G_SWIZ_0 | R500_ALU_RGBA_B_SWIZ_0 |
-        R500_ALU_RGBA_A_SWIZ_0,
-};
diff --git a/src/gallium/drivers/r300/r5xx_fs.h b/src/gallium/drivers/r300/r5xx_fs.h
deleted file mode 100644
index a4addde32b..0000000000
--- a/src/gallium/drivers/r300/r5xx_fs.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/*
- * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
- *                Joakim Sindholt <opensource@zhasha.com>
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE. */
-
-#ifndef R5XX_FS_H
-#define R5XX_FS_H
-
-#include "radeon_code.h"
-
-struct rX00_fragment_program_code r5xx_passthrough_fragment_shader;
-struct rX00_fragment_program_code r5xx_texture_fragment_shader;
-
-#endif /* R5XX_FS_H */
diff --git a/src/gallium/drivers/softpipe/Makefile b/src/gallium/drivers/softpipe/Makefile
index 516e3992fd..bcb887a0b2 100644
--- a/src/gallium/drivers/softpipe/Makefile
+++ b/src/gallium/drivers/softpipe/Makefile
@@ -6,26 +6,17 @@ LIBNAME = softpipe
 C_SOURCES = \
 	sp_fs_exec.c \
 	sp_fs_sse.c \
-	sp_fs_llvm.c \
 	sp_clear.c \
 	sp_flush.c \
 	sp_query.c \
 	sp_context.c \
 	sp_draw_arrays.c \
-	sp_prim_setup.c \
 	sp_prim_vbuf.c \
 	sp_quad_pipe.c \
-	sp_quad_alpha_test.c \
-	sp_quad_blend.c \
-	sp_quad_colormask.c \
-	sp_quad_coverage.c \
+	sp_quad_stipple.c \
 	sp_quad_depth_test.c \
-	sp_quad_earlyz.c \
 	sp_quad_fs.c \
-	sp_quad_occlusion.c \
-	sp_quad_output.c \
-	sp_quad_stencil.c \
-	sp_quad_stipple.c \
+	sp_quad_blend.c \
 	sp_screen.c \
         sp_setup.c \
 	sp_state_blend.c \
@@ -38,7 +29,9 @@ C_SOURCES = \
 	sp_state_vertex.c \
 	sp_texture.c \
 	sp_tex_sample.c \
+	sp_tex_tile_cache.c \
 	sp_tile_cache.c \
-	sp_surface.c 
+	sp_surface.c \
+	sp_video_context.c
 
 include ../../Makefile.template
diff --git a/src/gallium/drivers/softpipe/SConscript b/src/gallium/drivers/softpipe/SConscript
index f8720638a7..aac9edf44e 100644
--- a/src/gallium/drivers/softpipe/SConscript
+++ b/src/gallium/drivers/softpipe/SConscript
@@ -7,25 +7,16 @@ softpipe = env.ConvenienceLibrary(
 	source = [
 		'sp_fs_exec.c',
 		'sp_fs_sse.c',
-		'sp_fs_llvm.c',
 		'sp_clear.c',
 		'sp_context.c',
 		'sp_draw_arrays.c',
 		'sp_flush.c',
-		'sp_prim_setup.c',
 		'sp_prim_vbuf.c',
 		'sp_setup.c',
-		'sp_quad_alpha_test.c',
 		'sp_quad_blend.c',
 		'sp_quad_pipe.c',
-		'sp_quad_colormask.c',
-		'sp_quad_coverage.c',
 		'sp_quad_depth_test.c',
-		'sp_quad_earlyz.c',
 		'sp_quad_fs.c',
-		'sp_quad_occlusion.c',
-		'sp_quad_output.c',
-		'sp_quad_stencil.c',
 		'sp_quad_stipple.c',
 		'sp_query.c',
 		'sp_screen.c',
@@ -39,8 +30,10 @@ softpipe = env.ConvenienceLibrary(
 		'sp_state_vertex.c',
 		'sp_surface.c',
 		'sp_tex_sample.c',
+		'sp_tex_tile_cache.c',
 		'sp_texture.c',
 		'sp_tile_cache.c',
+		'sp_video_context.c',
 	])
 
-Export('softpipe')
\ No newline at end of file
+Export('softpipe')
diff --git a/src/gallium/drivers/softpipe/sp_clear.c b/src/gallium/drivers/softpipe/sp_clear.c
index d3af18e162..8fac8e6e05 100644
--- a/src/gallium/drivers/softpipe/sp_clear.c
+++ b/src/gallium/drivers/softpipe/sp_clear.c
@@ -36,8 +36,6 @@
 #include "util/u_pack_color.h"
 #include "sp_clear.h"
 #include "sp_context.h"
-#include "sp_surface.h"
-#include "sp_state.h"
 #include "sp_tile_cache.h"
 
 
diff --git a/src/gallium/drivers/softpipe/sp_clear.h b/src/gallium/drivers/softpipe/sp_clear.h
index 2e450672f5..9be3b86fe9 100644
--- a/src/gallium/drivers/softpipe/sp_clear.h
+++ b/src/gallium/drivers/softpipe/sp_clear.h
@@ -32,7 +32,6 @@
 #ifndef SP_CLEAR_H
 #define SP_CLEAR_H
 
-#include "pipe/p_state.h"
 struct pipe_context;
 
 extern void
diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c
index b4650c0dc5..94d000a5ac 100644
--- a/src/gallium/drivers/softpipe/sp_context.c
+++ b/src/gallium/drivers/softpipe/sp_context.c
@@ -31,17 +31,18 @@
  */
 
 #include "draw/draw_context.h"
+#include "draw/draw_vbuf.h"
 #include "pipe/p_defines.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "sp_clear.h"
 #include "sp_context.h"
 #include "sp_flush.h"
-#include "sp_prim_setup.h"
 #include "sp_prim_vbuf.h"
 #include "sp_state.h"
 #include "sp_surface.h"
 #include "sp_tile_cache.h"
+#include "sp_tex_tile_cache.h"
 #include "sp_texture.h"
 #include "sp_winsys.h"
 #include "sp_query.h"
@@ -72,18 +73,16 @@ softpipe_unmap_transfers(struct softpipe_context *sp)
 {
    uint i;
 
-   for (i = 0; i < sp->framebuffer.nr_cbufs; i++)
-      sp_flush_tile_cache(sp, sp->cbuf_cache[i]);
-   sp_flush_tile_cache(sp, sp->zsbuf_cache);
-
    for (i = 0; i < sp->framebuffer.nr_cbufs; i++) {
       sp_tile_cache_unmap_transfers(sp->cbuf_cache[i]);
    }
+
    sp_tile_cache_unmap_transfers(sp->zsbuf_cache);
 }
 
 
-static void softpipe_destroy( struct pipe_context *pipe )
+static void
+softpipe_destroy( struct pipe_context *pipe )
 {
    struct softpipe_context *softpipe = softpipe_context( pipe );
    uint i;
@@ -91,19 +90,9 @@ static void softpipe_destroy( struct pipe_context *pipe )
    if (softpipe->draw)
       draw_destroy( softpipe->draw );
 
-   for (i = 0; i < SP_NUM_QUAD_THREADS; i++) {
-      softpipe->quad[i].polygon_stipple->destroy( softpipe->quad[i].polygon_stipple );
-      softpipe->quad[i].earlyz->destroy( softpipe->quad[i].earlyz );
-      softpipe->quad[i].shade->destroy( softpipe->quad[i].shade );
-      softpipe->quad[i].alpha_test->destroy( softpipe->quad[i].alpha_test );
-      softpipe->quad[i].depth_test->destroy( softpipe->quad[i].depth_test );
-      softpipe->quad[i].stencil_test->destroy( softpipe->quad[i].stencil_test );
-      softpipe->quad[i].occlusion->destroy( softpipe->quad[i].occlusion );
-      softpipe->quad[i].coverage->destroy( softpipe->quad[i].coverage );
-      softpipe->quad[i].blend->destroy( softpipe->quad[i].blend );
-      softpipe->quad[i].colormask->destroy( softpipe->quad[i].colormask );
-      softpipe->quad[i].output->destroy( softpipe->quad[i].output );
-   }
+      softpipe->quad.shade->destroy( softpipe->quad.shade );
+      softpipe->quad.depth_test->destroy( softpipe->quad.depth_test );
+      softpipe->quad.blend->destroy( softpipe->quad.blend );
 
    for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
       sp_destroy_tile_cache(softpipe->cbuf_cache[i]);
@@ -113,7 +102,7 @@ static void softpipe_destroy( struct pipe_context *pipe )
    pipe_surface_reference(&softpipe->framebuffer.zsbuf, NULL);
 
    for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
-      sp_destroy_tile_cache(softpipe->tex_cache[i]);
+      sp_destroy_tex_tile_cache(softpipe->tex_cache[i]);
       pipe_texture_reference(&softpipe->texture[i], NULL);
    }
 
@@ -126,6 +115,15 @@ static void softpipe_destroy( struct pipe_context *pipe )
    FREE( softpipe );
 }
 
+
+/**
+ * if (the texture is being used as a framebuffer surface)
+ *    return PIPE_REFERENCED_FOR_WRITE
+ * else if (the texture is a bound texture source)
+ *    return PIPE_REFERENCED_FOR_READ  XXX not done yet
+ * else
+ *    return PIPE_UNREFERENCED
+ */
 static unsigned int
 softpipe_is_texture_referenced( struct pipe_context *pipe,
 				struct pipe_texture *texture,
@@ -134,15 +132,17 @@ softpipe_is_texture_referenced( struct pipe_context *pipe,
    struct softpipe_context *softpipe = softpipe_context( pipe );
    unsigned i;
 
-   if(softpipe->dirty_render_cache) {
+   if (softpipe->dirty_render_cache) {
       for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) {
-         if(softpipe->framebuffer.cbufs[i] && 
-            softpipe->framebuffer.cbufs[i]->texture == texture)
+         if (softpipe->framebuffer.cbufs[i] && 
+             softpipe->framebuffer.cbufs[i]->texture == texture) {
             return PIPE_REFERENCED_FOR_WRITE;
+         }
       }
-      if(softpipe->framebuffer.zsbuf && 
-         softpipe->framebuffer.zsbuf->texture == texture)
+      if (softpipe->framebuffer.zsbuf && 
+          softpipe->framebuffer.zsbuf->texture == texture) {
          return PIPE_REFERENCED_FOR_WRITE;
+      }
    }
    
    /* FIXME: we also need to do the same for the texture cache */
@@ -150,6 +150,7 @@ softpipe_is_texture_referenced( struct pipe_context *pipe,
    return PIPE_UNREFERENCED;
 }
 
+
 static unsigned int
 softpipe_is_buffer_referenced( struct pipe_context *pipe,
 			       struct pipe_buffer *buf)
@@ -157,6 +158,7 @@ softpipe_is_buffer_referenced( struct pipe_context *pipe,
    return PIPE_UNREFERENCED;
 }
 
+
 struct pipe_context *
 softpipe_create( struct pipe_screen *screen )
 {
@@ -227,7 +229,6 @@ softpipe_create( struct pipe_screen *screen )
    softpipe->pipe.is_buffer_referenced = softpipe_is_buffer_referenced;
 
    softpipe_init_query_funcs( softpipe );
-   softpipe_init_texture_funcs( softpipe );
 
    /*
     * Alloc caches for accessing drawing surfaces and textures.
@@ -238,41 +239,14 @@ softpipe_create( struct pipe_screen *screen )
    softpipe->zsbuf_cache = sp_create_tile_cache( screen );
 
    for (i = 0; i < PIPE_MAX_SAMPLERS; i++)
-      softpipe->tex_cache[i] = sp_create_tile_cache( screen );
+      softpipe->tex_cache[i] = sp_create_tex_tile_cache( screen );
 
 
    /* setup quad rendering stages */
-   for (i = 0; i < SP_NUM_QUAD_THREADS; i++) {
-      softpipe->quad[i].polygon_stipple = sp_quad_polygon_stipple_stage(softpipe);
-      softpipe->quad[i].earlyz = sp_quad_earlyz_stage(softpipe);
-      softpipe->quad[i].shade = sp_quad_shade_stage(softpipe);
-      softpipe->quad[i].alpha_test = sp_quad_alpha_test_stage(softpipe);
-      softpipe->quad[i].depth_test = sp_quad_depth_test_stage(softpipe);
-      softpipe->quad[i].stencil_test = sp_quad_stencil_test_stage(softpipe);
-      softpipe->quad[i].occlusion = sp_quad_occlusion_stage(softpipe);
-      softpipe->quad[i].coverage = sp_quad_coverage_stage(softpipe);
-      softpipe->quad[i].blend = sp_quad_blend_stage(softpipe);
-      softpipe->quad[i].colormask = sp_quad_colormask_stage(softpipe);
-      softpipe->quad[i].output = sp_quad_output_stage(softpipe);
-   }
-
-   /* vertex shader samplers */
-   for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
-      softpipe->tgsi.vert_samplers[i].base.get_samples = sp_get_samples_vertex;
-      softpipe->tgsi.vert_samplers[i].unit = i;
-      softpipe->tgsi.vert_samplers[i].sp = softpipe;
-      softpipe->tgsi.vert_samplers[i].cache = softpipe->tex_cache[i];
-      softpipe->tgsi.vert_samplers_list[i] = &softpipe->tgsi.vert_samplers[i];
-   }
+      softpipe->quad.shade = sp_quad_shade_stage(softpipe);
+      softpipe->quad.depth_test = sp_quad_depth_test_stage(softpipe);
+      softpipe->quad.blend = sp_quad_blend_stage(softpipe);
 
-   /* fragment shader samplers */
-   for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
-      softpipe->tgsi.frag_samplers[i].base.get_samples = sp_get_samples_fragment;
-      softpipe->tgsi.frag_samplers[i].unit = i;
-      softpipe->tgsi.frag_samplers[i].sp = softpipe;
-      softpipe->tgsi.frag_samplers[i].cache = softpipe->tex_cache[i];
-      softpipe->tgsi.frag_samplers_list[i] = &softpipe->tgsi.frag_samplers[i];
-   }
 
    /*
     * Create drawing context and plug our rendering stage into it.
@@ -286,30 +260,28 @@ softpipe_create( struct pipe_screen *screen )
                          (struct tgsi_sampler **)
                             softpipe->tgsi.vert_samplers_list);
 
-   softpipe->setup = sp_draw_render_stage(softpipe);
-   if (!softpipe->setup)
-      goto fail;
-
    if (debug_get_bool_option( "SP_NO_RAST", FALSE ))
       softpipe->no_rast = TRUE;
 
-   if (debug_get_bool_option( "SP_NO_VBUF", FALSE )) {
-      /* Deprecated path -- vbuf is the intended interface to the draw module:
-       */
-      draw_set_rasterize_stage(softpipe->draw, softpipe->setup);
-   }
-   else {
-      sp_init_vbuf(softpipe);
-   }
+   softpipe->vbuf_backend = sp_create_vbuf_backend(softpipe);
+   if (!softpipe->vbuf_backend)
+      goto fail;
+
+   softpipe->vbuf = draw_vbuf_stage(softpipe->draw, softpipe->vbuf_backend);
+   if (!softpipe->vbuf)
+      goto fail;
+
+   draw_set_rasterize_stage(softpipe->draw, softpipe->vbuf);
+   draw_set_render(softpipe->draw, softpipe->vbuf_backend);
+
+
 
    /* plug in AA line/point stages */
    draw_install_aaline_stage(softpipe->draw, &softpipe->pipe);
    draw_install_aapoint_stage(softpipe->draw, &softpipe->pipe);
 
-#if USE_DRAW_STAGE_PSTIPPLE
    /* Do polygon stipple w/ texture map + frag prog? */
    draw_install_pstipple_stage(softpipe->draw, &softpipe->pipe);
-#endif
 
    sp_init_surface_functions(softpipe);
 
diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h
index 7888c2f644..a735573d6f 100644
--- a/src/gallium/drivers/softpipe/sp_context.h
+++ b/src/gallium/drivers/softpipe/sp_context.h
@@ -36,24 +36,13 @@
 #include "draw/draw_vertex.h"
 
 #include "sp_quad_pipe.h"
-#include "sp_tex_sample.h"
 
 
-/**
- * This is a temporary variable for testing draw-stage polygon stipple.
- * If zero, do stipple in sp_quad_stipple.c
- */
-#define USE_DRAW_STAGE_PSTIPPLE 1
-
-/* Number of threads working on individual quads.
- * Setting to 1 disables this feature.
- */
-#define SP_NUM_QUAD_THREADS 1
-
 struct softpipe_vbuf_render;
 struct draw_context;
 struct draw_stage;
 struct softpipe_tile_cache;
+struct softpipe_tex_tile_cache;
 struct sp_fragment_shader;
 struct sp_vertex_shader;
 
@@ -62,12 +51,12 @@ struct softpipe_context {
    struct pipe_context pipe;  /**< base class */
 
    /** Constant state objects */
-   const struct pipe_blend_state *blend;
-   const struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS];
-   const struct pipe_depth_stencil_alpha_state *depth_stencil;
-   const struct pipe_rasterizer_state *rasterizer;
-   const struct sp_fragment_shader *fs;
-   const struct sp_vertex_shader *vs;
+   struct pipe_blend_state *blend;
+   struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS];
+   struct pipe_depth_stencil_alpha_state *depth_stencil;
+   struct pipe_rasterizer_state *rasterizer;
+   struct sp_fragment_shader *fs;
+   struct sp_vertex_shader *vs;
 
    /** Other rendering state */
    struct pipe_blend_color blend_color;
@@ -96,7 +85,7 @@ struct softpipe_context {
 
    /** Mapped vertex buffers */
    ubyte *mapped_vbuffer[PIPE_MAX_ATTRIBS];
-   
+
    /** Mapped constant buffers */
    void *mapped_constants[PIPE_SHADER_TYPES];
 
@@ -107,7 +96,15 @@ struct softpipe_context {
    /** Which vertex shader output slot contains point size */
    int psize_slot;
 
-   unsigned reduced_api_prim;  /**< PIPE_PRIM_POINTS, _LINES or _TRIANGLES */
+   /** The reduced version of the primitive supplied by the state tracker */
+   unsigned reduced_api_prim;
+
+   /**
+    * The reduced primitive after unfilled triangles, wide-line decomposition,
+    * etc, are taken into account.  This is the primitive type that's actually
+    * rasterized.
+    */
+   unsigned reduced_prim;
 
    /** Derived from scissor and surface bounds: */
    struct pipe_scissor_state cliprect;
@@ -116,41 +113,32 @@ struct softpipe_context {
 
    /** Software quad rendering pipeline */
    struct {
-      struct quad_stage *polygon_stipple;
-      struct quad_stage *earlyz;
       struct quad_stage *shade;
-      struct quad_stage *alpha_test;
-      struct quad_stage *stencil_test;
       struct quad_stage *depth_test;
-      struct quad_stage *occlusion;
-      struct quad_stage *coverage;
       struct quad_stage *blend;
-      struct quad_stage *colormask;
-      struct quad_stage *output;
-
       struct quad_stage *first; /**< points to one of the above stages */
-   } quad[SP_NUM_QUAD_THREADS];
+   } quad;
 
    /** TGSI exec things */
    struct {
-      struct sp_shader_sampler vert_samplers[PIPE_MAX_SAMPLERS];
-      struct sp_shader_sampler *vert_samplers_list[PIPE_MAX_SAMPLERS];
-      struct sp_shader_sampler frag_samplers[PIPE_MAX_SAMPLERS];
-      struct sp_shader_sampler *frag_samplers_list[PIPE_MAX_SAMPLERS];
+      struct sp_sampler_varient *vert_samplers_list[PIPE_MAX_SAMPLERS];
+      struct sp_sampler_varient *frag_samplers_list[PIPE_MAX_SAMPLERS];
    } tgsi;
 
    /** The primitive drawing context */
    struct draw_context *draw;
-   struct draw_stage *setup;
+
+   /** Draw module backend */
+   struct vbuf_render *vbuf_backend;
    struct draw_stage *vbuf;
-   struct softpipe_vbuf_render *vbuf_render;
 
    boolean dirty_render_cache;
-   
+
    struct softpipe_tile_cache *cbuf_cache[PIPE_MAX_COLOR_BUFS];
    struct softpipe_tile_cache *zsbuf_cache;
 
-   struct softpipe_tile_cache *tex_cache[PIPE_MAX_SAMPLERS];
+   unsigned tex_timestamp;
+   struct softpipe_tex_tile_cache *tex_cache[PIPE_MAX_SAMPLERS];
 
    unsigned use_sse : 1;
    unsigned dump_fs : 1;
@@ -164,5 +152,8 @@ softpipe_context( struct pipe_context *pipe )
    return (struct softpipe_context *)pipe;
 }
 
-#endif /* SP_CONTEXT_H */
+void
+softpipe_reset_sampler_varients(struct softpipe_context *softpipe);
+
 
+#endif /* SP_CONTEXT_H */
diff --git a/src/gallium/drivers/softpipe/sp_flush.c b/src/gallium/drivers/softpipe/sp_flush.c
index 4a14d49686..e38b767cf2 100644
--- a/src/gallium/drivers/softpipe/sp_flush.c
+++ b/src/gallium/drivers/softpipe/sp_flush.c
@@ -37,6 +37,7 @@
 #include "sp_surface.h"
 #include "sp_state.h"
 #include "sp_tile_cache.h"
+#include "sp_tex_tile_cache.h"
 #include "sp_winsys.h"
 
 
@@ -52,17 +53,19 @@ softpipe_flush( struct pipe_context *pipe,
 
    if (flags & PIPE_FLUSH_TEXTURE_CACHE) {
       for (i = 0; i < softpipe->num_textures; i++) {
-         sp_flush_tile_cache(softpipe, softpipe->tex_cache[i]);
+         sp_flush_tex_tile_cache(softpipe->tex_cache[i]);
       }
    }
 
-   if (flags & PIPE_FLUSH_RENDER_CACHE) {
+   if (flags & PIPE_FLUSH_SWAPBUFFERS) {
+      /* If this is a swapbuffers, just flush color buffers.
+       *
+       * The zbuffer changes are not discarded, but held in the cache
+       * in the hope that a later clear will wipe them out.
+       */
       for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++)
          if (softpipe->cbuf_cache[i])
-            sp_flush_tile_cache(softpipe, softpipe->cbuf_cache[i]);
-
-      if (softpipe->zsbuf_cache)
-         sp_flush_tile_cache(softpipe, softpipe->zsbuf_cache);
+            sp_flush_tile_cache(softpipe->cbuf_cache[i]);
 
       /* Need this call for hardware buffers before swapbuffers.
        *
@@ -71,7 +74,15 @@ softpipe_flush( struct pipe_context *pipe,
        * to unmap surfaces when flushing.
        */
       softpipe_unmap_transfers(softpipe);
-      
+   }
+   else if (flags & PIPE_FLUSH_RENDER_CACHE) {
+      for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++)
+         if (softpipe->cbuf_cache[i])
+            sp_flush_tile_cache(softpipe->cbuf_cache[i]);
+
+      if (softpipe->zsbuf_cache)
+         sp_flush_tile_cache(softpipe->zsbuf_cache);
+     
       softpipe->dirty_render_cache = FALSE;
    }
 
diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c
index 9ee86fe787..4076114d39 100644
--- a/src/gallium/drivers/softpipe/sp_fs_exec.c
+++ b/src/gallium/drivers/softpipe/sp_fs_exec.c
@@ -59,15 +59,34 @@ sp_exec_fragment_shader(const struct sp_fragment_shader *base)
 }
 
 
+static void
+exec_prepare( const struct sp_fragment_shader *base,
+	      struct tgsi_exec_machine *machine,
+	      struct tgsi_sampler **samplers )
+{
+   /*
+    * Bind tokens/shader to the interpreter's machine state.
+    * Avoid redundant binding.
+    */
+   if (machine->Tokens != base->shader.tokens) {
+      tgsi_exec_machine_bind_shader( machine,
+                                     base->shader.tokens,
+                                     PIPE_MAX_SAMPLERS,
+                                     samplers );
+   }
+}
+
+
+
 /**
  * Compute quad X,Y,Z,W for the four fragments in a quad.
  *
  * This should really be part of the compiled shader.
  */
-void
-sp_setup_pos_vector(const struct tgsi_interp_coef *coef,
-		    float x, float y,
-		    struct tgsi_exec_vector *quadpos)
+static void
+setup_pos_vector(const struct tgsi_interp_coef *coef,
+                 float x, float y,
+                 struct tgsi_exec_vector *quadpos)
 {
    uint chan;
    /* do X */
@@ -95,24 +114,6 @@ sp_setup_pos_vector(const struct tgsi_interp_coef *coef,
 }
 
 
-static void
-exec_prepare( const struct sp_fragment_shader *base,
-	      struct tgsi_exec_machine *machine,
-	      struct tgsi_sampler **samplers )
-{
-   /*
-    * Bind tokens/shader to the interpreter's machine state.
-    * Avoid redundant binding.
-    */
-   if (machine->Tokens != base->shader.tokens) {
-      tgsi_exec_machine_bind_shader( machine,
-                                     base->shader.tokens,
-                                     PIPE_MAX_SAMPLERS,
-                                     samplers );
-   }
-}
-
-
 /* TODO: hide the machine struct in here somewhere, remove from this
  * interface:
  */
@@ -122,11 +123,43 @@ exec_run( const struct sp_fragment_shader *base,
 	  struct quad_header *quad )
 {
    /* Compute X, Y, Z, W vals for this quad */
-   sp_setup_pos_vector(quad->posCoef, 
-		       (float)quad->input.x0, (float)quad->input.y0, 
-		       &machine->QuadPos);
+   setup_pos_vector(quad->posCoef, 
+                    (float)quad->input.x0, (float)quad->input.y0, 
+                    &machine->QuadPos);
    
-   return tgsi_exec_machine_run( machine );
+   quad->inout.mask &= tgsi_exec_machine_run( machine );
+   if (quad->inout.mask == 0)
+      return FALSE;
+
+   /* store outputs */
+   {
+      const ubyte *sem_name = base->info.output_semantic_name;
+      const ubyte *sem_index = base->info.output_semantic_index;
+      const uint n = base->info.num_outputs;
+      uint i;
+      for (i = 0; i < n; i++) {
+         switch (sem_name[i]) {
+         case TGSI_SEMANTIC_COLOR:
+            {
+               uint cbuf = sem_index[i];
+               memcpy(quad->output.color[cbuf],
+                      &machine->Outputs[i].xyzw[0].f[0],
+                      sizeof(quad->output.color[0]) );
+            }
+            break;
+         case TGSI_SEMANTIC_POSITION:
+            {
+               uint j;
+               for (j = 0; j < 4; j++) {
+                  quad->output.depth[j] = machine->Outputs[i].xyzw[2].f[j];
+               }
+            }
+            break;
+         }
+      }
+   }
+
+   return TRUE;
 }
 
 
diff --git a/src/gallium/drivers/softpipe/sp_fs_llvm.c b/src/gallium/drivers/softpipe/sp_fs_llvm.c
deleted file mode 100644
index 95c0d982d1..0000000000
--- a/src/gallium/drivers/softpipe/sp_fs_llvm.c
+++ /dev/null
@@ -1,205 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/**
- * Execute fragment shader using LLVM code generation.
- * Authors:
- *   Zack Rusin
- */
-
-#include "sp_context.h"
-#include "sp_state.h"
-#include "sp_fs.h"
-
-#include "pipe/p_state.h"
-#include "pipe/p_defines.h"
-#include "util/u_memory.h"
-#include "tgsi/tgsi_sse2.h"
-
-#if 0
-
-/**
- * Subclass of sp_fragment_shader
- */
-struct sp_llvm_fragment_shader
-{
-   struct sp_fragment_shader base;
-   struct gallivm_prog *llvm_prog;
-};
-
-
-static void
-shade_quad_llvm(struct quad_stage *qs,
-                struct quad_header *quad)
-{
-   struct quad_shade_stage *qss = quad_shade_stage(qs);
-   struct softpipe_context *softpipe = qs->softpipe;
-   float dests[4][16][4] ALIGN16_ATTRIB;
-   float inputs[4][16][4] ALIGN16_ATTRIB;
-   const float fx = (float) quad->x0;
-   const float fy = (float) quad->y0;
-   struct gallivm_prog *llvm = qss->llvm_prog;
-
-   inputs[0][0][0] = fx;
-   inputs[1][0][0] = fx + 1.0f;
-   inputs[2][0][0] = fx;
-   inputs[3][0][0] = fx + 1.0f;
-
-   inputs[0][0][1] = fy;
-   inputs[1][0][1] = fy;
-   inputs[2][0][1] = fy + 1.0f;
-   inputs[3][0][1] = fy + 1.0f;
-
-
-   gallivm_prog_inputs_interpolate(llvm, inputs, quad->coef);
-
-#if DLLVM
-   debug_printf("MASK = %d\n", quad->mask);
-   for (int i = 0; i < 4; ++i) {
-      for (int j = 0; j < 2; ++j) {
-         debug_printf("IN(%d,%d) [%f %f %f %f]\n", i, j, 
-                inputs[i][j][0], inputs[i][j][1], inputs[i][j][2], inputs[i][j][3]);
-      }
-   }
-#endif
-
-   quad->mask &=
-      gallivm_fragment_shader_exec(llvm, fx, fy, dests, inputs,
-                                   softpipe->mapped_constants[PIPE_SHADER_FRAGMENT],
-                                   qss->samplers);
-#if DLLVM
-   debug_printf("OUT LLVM = 1[%f %f %f %f], 2[%f %f %f %f]\n",
-          dests[0][0][0], dests[0][0][1], dests[0][0][2], dests[0][0][3], 
-          dests[0][1][0], dests[0][1][1], dests[0][1][2], dests[0][1][3]);
-#endif
-
-   /* store result color */
-   if (qss->colorOutSlot >= 0) {
-      unsigned i;
-      /* XXX need to handle multiple color outputs someday */
-      allvmrt(qss->stage.softpipe->fs->info.output_semantic_name[qss->colorOutSlot]
-             == TGSI_SEMANTIC_COLOR);
-      for (i = 0; i < QUAD_SIZE; ++i) {
-         quad->outputs.color[0][0][i] = dests[i][qss->colorOutSlot][0];
-         quad->outputs.color[0][1][i] = dests[i][qss->colorOutSlot][1];
-         quad->outputs.color[0][2][i] = dests[i][qss->colorOutSlot][2];
-         quad->outputs.color[0][3][i] = dests[i][qss->colorOutSlot][3];
-      }
-   }
-#if DLLVM
-   for (int i = 0; i < QUAD_SIZE; ++i) {
-      debug_printf("QLLVM%d(%d) [%f, %f, %f, %f]\n", i, qss->colorOutSlot,
-             quad->outputs.color[0][0][i],
-             quad->outputs.color[0][1][i],
-             quad->outputs.color[0][2][i],
-             quad->outputs.color[0][3][i]);
-   }
-#endif
-
-   /* store result Z */
-   if (qss->depthOutSlot >= 0) {
-      /* output[slot] is new Z */
-      uint i;
-      for (i = 0; i < 4; i++) {
-         quad->outputs.depth[i] = dests[i][0][2];
-      }
-   }
-   else {
-      /* copy input Z (which was interpolated by the executor) to output Z */
-      uint i;
-      for (i = 0; i < 4; i++) {
-         quad->outputs.depth[i] = inputs[i][0][2];
-      }
-   }
-#if DLLVM
-   debug_printf("D [%f, %f, %f, %f] mask = %d\n",
-             quad->outputs.depth[0],
-             quad->outputs.depth[1],
-             quad->outputs.depth[2],
-             quad->outputs.depth[3], quad->mask);
-#endif
-
-   /* shader may cull fragments */
-   if( quad->mask ) {
-      qs->next->run( qs->next, quad );
-   }
-}
-
-
-unsigned 
-run_llvm_fs( const struct sp_fragment_shader *base,
-	     struct foo *machine )
-{
-}
-
-
-void 
-delete_llvm_fs( struct sp_fragment_shader *base )
-{
-   FREE(base);
-}
-
-
-struct sp_fragment_shader *
-softpipe_create_fs_llvm(struct softpipe_context *softpipe,
-                        const struct pipe_shader_state *templ)
-{
-   struct sp_llvm_fragment_shader *shader = NULL;
-
-   /* LLVM fragment shaders currently disabled:
-    */
-   state = CALLOC_STRUCT(sp_llvm_shader_state);
-   if (!state)
-      return NULL;
-
-   state->llvm_prog = 0;
-
-   if (!gallivm_global_cpu_engine()) {
-      gallivm_cpu_engine_create(state->llvm_prog);
-   }
-   else
-      gallivm_cpu_jit_compile(gallivm_global_cpu_engine(), state->llvm_prog);
-   
-   if (shader) {
-      shader->base.run = run_llvm_fs;
-      shader->base.delete = delete_llvm_fs;
-   }
-
-   return shader;
-}
-
-
-#else
-
-struct sp_fragment_shader *
-softpipe_create_fs_llvm(struct softpipe_context *softpipe,
-		       const struct pipe_shader_state *templ)
-{
-   return NULL;
-}
-
-#endif
diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c
index 31ccc3bda9..f912950658 100644
--- a/src/gallium/drivers/softpipe/sp_fs_sse.c
+++ b/src/gallium/drivers/softpipe/sp_fs_sse.c
@@ -76,6 +76,43 @@ fs_sse_prepare( const struct sp_fragment_shader *base,
 }
 
 
+
+/**
+ * Compute quad X,Y,Z,W for the four fragments in a quad.
+ *
+ * This should really be part of the compiled shader.
+ */
+static void
+setup_pos_vector(const struct tgsi_interp_coef *coef,
+		    float x, float y,
+		    struct tgsi_exec_vector *quadpos)
+{
+   uint chan;
+   /* do X */
+   quadpos->xyzw[0].f[0] = x;
+   quadpos->xyzw[0].f[1] = x + 1;
+   quadpos->xyzw[0].f[2] = x;
+   quadpos->xyzw[0].f[3] = x + 1;
+
+   /* do Y */
+   quadpos->xyzw[1].f[0] = y;
+   quadpos->xyzw[1].f[1] = y;
+   quadpos->xyzw[1].f[2] = y + 1;
+   quadpos->xyzw[1].f[3] = y + 1;
+
+   /* do Z and W for all fragments in the quad */
+   for (chan = 2; chan < 4; chan++) {
+      const float dadx = coef->dadx[chan];
+      const float dady = coef->dady[chan];
+      const float a0 = coef->a0[chan] + dadx * x + dady * y;
+      quadpos->xyzw[chan].f[0] = a0;
+      quadpos->xyzw[chan].f[1] = a0 + dadx;
+      quadpos->xyzw[chan].f[2] = a0 + dady;
+      quadpos->xyzw[chan].f[3] = a0 + dadx + dady;
+   }
+}
+
+
 /* TODO: codegenerate the whole run function, skip this wrapper.
  * TODO: break dependency on tgsi_exec_machine struct
  * TODO: push Position calculation into the generated shader
@@ -89,9 +126,9 @@ fs_sse_run( const struct sp_fragment_shader *base,
    struct sp_sse_fragment_shader *shader = sp_sse_fragment_shader(base);
 
    /* Compute X, Y, Z, W vals for this quad -- place in temp[0] for now */
-   sp_setup_pos_vector(quad->posCoef, 
-		       (float)quad->input.x0, (float)quad->input.y0, 
-		       machine->Temps);
+   setup_pos_vector(quad->posCoef, 
+                    (float)quad->input.x0, (float)quad->input.y0, 
+                    machine->Temps);
 
    /* init kill mask */
    tgsi_set_kill_mask(machine, 0x0);
@@ -104,7 +141,39 @@ fs_sse_run( const struct sp_fragment_shader *base,
 		 /*, &machine->QuadPos*/
       );
 
-   return ~(machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0]);
+   quad->inout.mask &= ~(machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0]);
+   if (quad->inout.mask == 0)
+      return FALSE;
+
+   /* store outputs */
+   {
+      const ubyte *sem_name = base->info.output_semantic_name;
+      const ubyte *sem_index = base->info.output_semantic_index;
+      const uint n = base->info.num_outputs;
+      uint i;
+      for (i = 0; i < n; i++) {
+         switch (sem_name[i]) {
+         case TGSI_SEMANTIC_COLOR:
+            {
+               uint cbuf = sem_index[i];
+               memcpy(quad->output.color[cbuf],
+                      &machine->Outputs[i].xyzw[0].f[0],
+                      sizeof(quad->output.color[0]) );
+            }
+            break;
+         case TGSI_SEMANTIC_POSITION:
+            {
+               uint j;
+               for (j = 0; j < 4; j++) {
+                  quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j];
+               }
+            }
+            break;
+         }
+      }
+   }
+
+   return TRUE;
 }
 
 
diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.c b/src/gallium/drivers/softpipe/sp_prim_setup.c
deleted file mode 100644
index 038ff04d4f..0000000000
--- a/src/gallium/drivers/softpipe/sp_prim_setup.c
+++ /dev/null
@@ -1,190 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/**
- * \brief A draw stage that drives our triangle setup routines from
- * within the draw pipeline.  One of two ways to drive setup, the
- * other being in sp_prim_vbuf.c.
- *
- * \author  Keith Whitwell <keith@tungstengraphics.com>
- * \author  Brian Paul
- */
-
-
-#include "sp_context.h"
-#include "sp_setup.h"
-#include "sp_state.h"
-#include "sp_prim_setup.h"
-#include "draw/draw_pipe.h"
-#include "draw/draw_vertex.h"
-#include "util/u_memory.h"
-
-/**
- * Triangle setup info (derived from draw_stage).
- * Also used for line drawing (taking some liberties).
- */
-struct setup_stage {
-   struct draw_stage stage; /**< This must be first (base class) */
-
-   struct setup_context *setup;
-};
-
-
-
-/**
- * Basically a cast wrapper.
- */
-static INLINE struct setup_stage *setup_stage( struct draw_stage *stage )
-{
-   return (struct setup_stage *)stage;
-}
-
-
-typedef const float (*cptrf4)[4];
-
-static void
-do_tri(struct draw_stage *stage, struct prim_header *prim)
-{
-   struct setup_stage *setup = setup_stage( stage );
-   
-   setup_tri( setup->setup,
-              (cptrf4)prim->v[0]->data,
-              (cptrf4)prim->v[1]->data,
-              (cptrf4)prim->v[2]->data );
-}
-
-static void
-do_line(struct draw_stage *stage, struct prim_header *prim)
-{
-   struct setup_stage *setup = setup_stage( stage );
-
-   setup_line( setup->setup,
-               (cptrf4)prim->v[0]->data,
-               (cptrf4)prim->v[1]->data );
-}
-
-static void
-do_point(struct draw_stage *stage, struct prim_header *prim)
-{
-   struct setup_stage *setup = setup_stage( stage );
-
-   setup_point( setup->setup,
-                (cptrf4)prim->v[0]->data );
-}
-
-
-
-
-static void setup_begin( struct draw_stage *stage )
-{
-   struct setup_stage *setup = setup_stage(stage);
-
-   setup_prepare( setup->setup );
-
-   stage->point = do_point;
-   stage->line = do_line;
-   stage->tri = do_tri;
-}
-
-
-static void setup_first_point( struct draw_stage *stage,
-			       struct prim_header *header )
-{
-   setup_begin(stage);
-   stage->point( stage, header );
-}
-
-static void setup_first_line( struct draw_stage *stage,
-			       struct prim_header *header )
-{
-   setup_begin(stage);
-   stage->line( stage, header );
-}
-
-
-static void setup_first_tri( struct draw_stage *stage,
-			       struct prim_header *header )
-{
-   setup_begin(stage);
-   stage->tri( stage, header );
-}
-
-
-
-static void setup_flush( struct draw_stage *stage,
-			 unsigned flags )
-{
-   stage->point = setup_first_point;
-   stage->line = setup_first_line;
-   stage->tri = setup_first_tri;
-}
-
-
-static void reset_stipple_counter( struct draw_stage *stage )
-{
-}
-
-
-static void render_destroy( struct draw_stage *stage )
-{
-   struct setup_stage *ssetup = setup_stage(stage);
-   setup_destroy_context(ssetup->setup);
-   FREE( stage );
-}
-
-
-/**
- * Create a new primitive setup/render stage.
- */
-struct draw_stage *sp_draw_render_stage( struct softpipe_context *softpipe )
-{
-   struct setup_stage *sstage = CALLOC_STRUCT(setup_stage);
-
-   sstage->setup = setup_create_context(softpipe);
-   sstage->stage.draw = softpipe->draw;
-   sstage->stage.point = setup_first_point;
-   sstage->stage.line = setup_first_line;
-   sstage->stage.tri = setup_first_tri;
-   sstage->stage.flush = setup_flush;
-   sstage->stage.reset_stipple_counter = reset_stipple_counter;
-   sstage->stage.destroy = render_destroy;
-
-   return (struct draw_stage *)sstage;
-}
-
-struct setup_context *
-sp_draw_setup_context( struct draw_stage *stage )
-{
-   struct setup_stage *ssetup = setup_stage(stage);
-   return ssetup->setup;
-}
-
-void
-sp_draw_flush( struct draw_stage *stage )
-{
-   stage->flush( stage, 0 );
-}
diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.h b/src/gallium/drivers/softpipe/sp_prim_setup.h
deleted file mode 100644
index 49bdd98ed8..0000000000
--- a/src/gallium/drivers/softpipe/sp_prim_setup.h
+++ /dev/null
@@ -1,85 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef SP_PRIM_SETUP_H
-#define SP_PRIM_SETUP_H
-
-
-/**
- * vbuf is a special stage to gather the stream of triangles, lines, points
- * together and reconstruct vertex buffers for hardware upload.
- *
- * First attempt, work in progress.
- * 
- * TODO:
- *    - separate out vertex buffer building and primitive emit, ie >1 draw per vb.
- *    - tell vbuf stage how to build hw vertices directly
- *    - pass vbuf stage a buffer pointer for direct emit to agp/vram.
- *
- *
- *
- * Vertices are just an array of floats, with all the attributes
- * packed.  We currently assume a layout like:
- *
- * attr[0][0..3] - window position
- * attr[1..n][0..3] - remaining attributes.
- *
- * Attributes are assumed to be 4 floats wide but are packed so that
- * all the enabled attributes run contiguously.
- */
-
-
-struct draw_stage;
-struct softpipe_context;
-
-
-typedef void (*vbuf_draw_func)( struct pipe_context *pipe,
-                                unsigned prim,
-                                const ushort *elements,
-                                unsigned nr_elements,
-                                const void *vertex_buffer,
-                                unsigned nr_vertices );
-
-
-extern struct draw_stage *
-sp_draw_render_stage( struct softpipe_context *softpipe );
-
-extern struct setup_context *
-sp_draw_setup_context( struct draw_stage * );
-
-extern void
-sp_draw_flush( struct draw_stage * );
-
-
-extern struct draw_stage *
-sp_draw_vbuf_stage( struct draw_context *draw_context,
-                    struct pipe_context *pipe,
-                    vbuf_draw_func draw );
-
-
-#endif /* SP_PRIM_SETUP_H */
diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
index 42021789ea..5fbac06a53 100644
--- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c
+++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
@@ -37,13 +37,13 @@
 
 
 #include "sp_context.h"
+#include "sp_setup.h"
 #include "sp_state.h"
 #include "sp_prim_vbuf.h"
-#include "sp_prim_setup.h"
-#include "sp_setup.h"
 #include "draw/draw_context.h"
 #include "draw/draw_vbuf.h"
 #include "util/u_memory.h"
+#include "util/u_prim.h"
 
 
 #define SP_MAX_VBUF_INDEXES 1024
@@ -58,6 +58,8 @@ struct softpipe_vbuf_render
 {
    struct vbuf_render base;
    struct softpipe_context *softpipe;
+   struct setup_context *setup;
+
    uint prim;
    uint vertex_size;
    uint nr_vertices;
@@ -74,6 +76,11 @@ softpipe_vbuf_render(struct vbuf_render *vbr)
 }
 
 
+
+
+
+
+
 static const struct vertex_info *
 sp_vbuf_get_vertex_info(struct vbuf_render *vbr)
 {
@@ -104,36 +111,6 @@ sp_vbuf_allocate_vertices(struct vbuf_render *vbr,
 static void
 sp_vbuf_release_vertices(struct vbuf_render *vbr)
 {
-#if 0
-   {
-      struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
-      const struct vertex_info *info = 
-         softpipe_get_vbuf_vertex_info(cvbr->softpipe);
-      const float *vtx = (const float *) cvbr->vertex_buffer;
-      uint i, j;
-      debug_printf("%s (vtx_size = %u,  vtx_used = %u)\n",
-             __FUNCTION__, cvbr->vertex_size, cvbr->nr_vertices);
-      for (i = 0; i < cvbr->nr_vertices; i++) {
-         for (j = 0; j < info->num_attribs; j++) {
-            uint k;
-            switch (info->attrib[j].emit) {
-            case EMIT_4F:  k = 4;   break;
-            case EMIT_3F:  k = 3;   break;
-            case EMIT_2F:  k = 2;   break;
-            case EMIT_1F:  k = 1;   break;
-            default: assert(0);
-            }
-            debug_printf("Vert %u attr %u: ", i, j);
-            while (k-- > 0) {
-               debug_printf("%g ", vtx[0]);
-               vtx++;
-            }
-            debug_printf("\n");
-         }
-      }
-   }
-#endif
-
    /* keep the old allocation for next time */
 }
 
@@ -159,14 +136,11 @@ static boolean
 sp_vbuf_set_primitive(struct vbuf_render *vbr, unsigned prim)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
-
-   /* XXX: break this dependency - make setup_context live under
-    * softpipe, rename the old "setup" draw stage to something else.
-    */
-   struct setup_context *setup_ctx = sp_draw_setup_context(cvbr->softpipe->setup);
+   struct setup_context *setup_ctx = cvbr->setup;
    
-   setup_prepare( setup_ctx );
+   sp_setup_prepare( setup_ctx );
 
+   cvbr->softpipe->reduced_prim = u_reduced_prim(prim);
    cvbr->prim = prim;
    return TRUE;
 
@@ -191,25 +165,20 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
    struct softpipe_context *softpipe = cvbr->softpipe;
    const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float);
    const void *vertex_buffer = cvbr->vertex_buffer;
+   struct setup_context *setup_ctx = cvbr->setup;
    unsigned i;
 
-   /* XXX: break this dependency - make setup_context live under
-    * softpipe, rename the old "setup" draw stage to something else.
-    */
-   struct draw_stage *setup = softpipe->setup;
-   struct setup_context *setup_ctx = sp_draw_setup_context(setup);
-
    switch (cvbr->prim) {
    case PIPE_PRIM_POINTS:
       for (i = 0; i < nr; i++) {
-         setup_point( setup_ctx,
+         sp_setup_point( setup_ctx,
                       get_vert(vertex_buffer, indices[i-0], stride) );
       }
       break;
 
    case PIPE_PRIM_LINES:
       for (i = 1; i < nr; i += 2) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, indices[i-1], stride),
                      get_vert(vertex_buffer, indices[i-0], stride) );
       }
@@ -217,7 +186,7 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
 
    case PIPE_PRIM_LINE_STRIP:
       for (i = 1; i < nr; i ++) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, indices[i-1], stride),
                      get_vert(vertex_buffer, indices[i-0], stride) );
       }
@@ -225,27 +194,29 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
 
    case PIPE_PRIM_LINE_LOOP:
       for (i = 1; i < nr; i ++) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, indices[i-1], stride),
                      get_vert(vertex_buffer, indices[i-0], stride) );
       }
       if (nr) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, indices[nr-1], stride),
                      get_vert(vertex_buffer, indices[0], stride) );
       }
       break;
 
    case PIPE_PRIM_TRIANGLES:
-      for (i = 2; i < nr; i += 3) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 2; i < nr; i += 3) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-0], stride),
                        get_vert(vertex_buffer, indices[i-2], stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 2; i < nr; i += 3) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-2], stride),
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
@@ -254,15 +225,17 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
       break;
 
    case PIPE_PRIM_TRIANGLE_STRIP:
-      for (i = 2; i < nr; i += 1) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 2; i < nr; i += 1) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i+(i&1)-1], stride),
                        get_vert(vertex_buffer, indices[i-(i&1)], stride),
                        get_vert(vertex_buffer, indices[i-2], stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 2; i < nr; i += 1) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i+(i&1)-2], stride),
                        get_vert(vertex_buffer, indices[i-(i&1)-1], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
@@ -271,15 +244,17 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
       break;
 
    case PIPE_PRIM_TRIANGLE_FAN:
-      for (i = 2; i < nr; i += 1) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 2; i < nr; i += 1) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-0], stride),
                        get_vert(vertex_buffer, indices[0], stride),
                        get_vert(vertex_buffer, indices[i-1], stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 2; i < nr; i += 1) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[0], stride),
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
@@ -288,24 +263,26 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
       break;
 
    case PIPE_PRIM_QUADS:
-      for (i = 3; i < nr; i += 4) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 3; i < nr; i += 4) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-2], stride),
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-3], stride) );
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-0], stride),
                        get_vert(vertex_buffer, indices[i-3], stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 3; i < nr; i += 4) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-3], stride),
                        get_vert(vertex_buffer, indices[i-2], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
 
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-2], stride),
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
@@ -314,23 +291,25 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
       break;
 
    case PIPE_PRIM_QUAD_STRIP:
-      for (i = 3; i < nr; i += 2) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 3; i < nr; i += 2) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-0], stride),
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-3], stride));
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-2], stride),
                        get_vert(vertex_buffer, indices[i-0], stride),
                        get_vert(vertex_buffer, indices[i-3], stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 3; i < nr; i += 2) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-3], stride),
                        get_vert(vertex_buffer, indices[i-2], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, indices[i-1], stride),
                        get_vert(vertex_buffer, indices[i-3], stride),
                        get_vert(vertex_buffer, indices[i-0], stride) );
@@ -345,7 +324,7 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
        * flatshade_first state makes no difference.
        */
       for (i = 2; i < nr; i += 1) {
-         setup_tri( setup_ctx,
+         sp_setup_tri( setup_ctx,
                     get_vert(vertex_buffer, indices[i-0], stride),
                     get_vert(vertex_buffer, indices[i-1], stride),
                     get_vert(vertex_buffer, indices[0], stride) );
@@ -355,11 +334,6 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
    default:
       assert(0);
    }
-
-   /* XXX: why are we calling this???  If we had to call something, it
-    * would be a function in sp_setup.c:
-    */
-   sp_draw_flush( setup );
 }
 
 
@@ -372,28 +346,23 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
    struct softpipe_context *softpipe = cvbr->softpipe;
+   struct setup_context *setup_ctx = cvbr->setup;
    const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float);
    const void *vertex_buffer =
       (void *) get_vert(cvbr->vertex_buffer, start, stride);
    unsigned i;
 
-   /* XXX: break this dependency - make setup_context live under
-    * softpipe, rename the old "setup" draw stage to something else.
-    */
-   struct draw_stage *setup = softpipe->setup;
-   struct setup_context *setup_ctx = sp_draw_setup_context(setup);
-
    switch (cvbr->prim) {
    case PIPE_PRIM_POINTS:
       for (i = 0; i < nr; i++) {
-         setup_point( setup_ctx,
+         sp_setup_point( setup_ctx,
                       get_vert(vertex_buffer, i-0, stride) );
       }
       break;
 
    case PIPE_PRIM_LINES:
       for (i = 1; i < nr; i += 2) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, i-1, stride),
                      get_vert(vertex_buffer, i-0, stride) );
       }
@@ -401,7 +370,7 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
 
    case PIPE_PRIM_LINE_STRIP:
       for (i = 1; i < nr; i ++) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, i-1, stride),
                      get_vert(vertex_buffer, i-0, stride) );
       }
@@ -409,27 +378,29 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
 
    case PIPE_PRIM_LINE_LOOP:
       for (i = 1; i < nr; i ++) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, i-1, stride),
                      get_vert(vertex_buffer, i-0, stride) );
       }
       if (nr) {
-         setup_line( setup_ctx,
+         sp_setup_line( setup_ctx,
                      get_vert(vertex_buffer, nr-1, stride),
                      get_vert(vertex_buffer, 0, stride) );
       }
       break;
 
    case PIPE_PRIM_TRIANGLES:
-      for (i = 2; i < nr; i += 3) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 2; i < nr; i += 3) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-0, stride),
                        get_vert(vertex_buffer, i-2, stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 2; i < nr; i += 3) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-2, stride),
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-0, stride) );
@@ -438,15 +409,17 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
       break;
 
    case PIPE_PRIM_TRIANGLE_STRIP:
-      for (i = 2; i < nr; i++) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 2; i < nr; i++) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i+(i&1)-1, stride),
                        get_vert(vertex_buffer, i-(i&1), stride),
                        get_vert(vertex_buffer, i-2, stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 2; i < nr; i++) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i+(i&1)-2, stride),
                        get_vert(vertex_buffer, i-(i&1)-1, stride),
                        get_vert(vertex_buffer, i-0, stride) );
@@ -455,15 +428,17 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
       break;
 
    case PIPE_PRIM_TRIANGLE_FAN:
-      for (i = 2; i < nr; i += 1) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 2; i < nr; i += 1) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-0, stride),
                        get_vert(vertex_buffer, 0, stride),
                        get_vert(vertex_buffer, i-1, stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 2; i < nr; i += 1) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, 0, stride),
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-0, stride) );
@@ -472,23 +447,25 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
       break;
 
    case PIPE_PRIM_QUADS:
-      for (i = 3; i < nr; i += 4) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 3; i < nr; i += 4) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-2, stride),
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-3, stride) );
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-0, stride),
                        get_vert(vertex_buffer, i-3, stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 3; i < nr; i += 4) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-3, stride),
                        get_vert(vertex_buffer, i-2, stride),
                        get_vert(vertex_buffer, i-0, stride) );
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-2, stride),
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-0, stride) );
@@ -497,23 +474,25 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
       break;
 
    case PIPE_PRIM_QUAD_STRIP:
-      for (i = 3; i < nr; i += 2) {
-         if (softpipe->rasterizer->flatshade_first) {
-            setup_tri( setup_ctx,
+      if (softpipe->rasterizer->flatshade_first) {
+         for (i = 3; i < nr; i += 2) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-0, stride),
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-3, stride) );
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-2, stride),
                        get_vert(vertex_buffer, i-0, stride),
                        get_vert(vertex_buffer, i-3, stride) );
          }
-         else {
-            setup_tri( setup_ctx,
+      }
+      else {
+         for (i = 3; i < nr; i += 2) {
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-3, stride),
                        get_vert(vertex_buffer, i-2, stride),
                        get_vert(vertex_buffer, i-0, stride) );
-            setup_tri( setup_ctx,
+            sp_setup_tri( setup_ctx,
                        get_vert(vertex_buffer, i-1, stride),
                        get_vert(vertex_buffer, i-3, stride),
                        get_vert(vertex_buffer, i-0, stride) );
@@ -528,7 +507,7 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
        * flatshade_first state makes no difference.
        */
       for (i = 2; i < nr; i += 1) {
-         setup_tri( setup_ctx,
+         sp_setup_tri( setup_ctx,
                     get_vert(vertex_buffer, i-1, stride),
                     get_vert(vertex_buffer, i-0, stride),
                     get_vert(vertex_buffer, 0, stride) );
@@ -546,40 +525,38 @@ static void
 sp_vbuf_destroy(struct vbuf_render *vbr)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
-   cvbr->softpipe->vbuf_render = NULL;
+   sp_setup_destroy_context(cvbr->setup);
    FREE(cvbr);
 }
 
 
 /**
- * Initialize the post-transform vertex buffer information for the given
- * context.
+ * Create the post-transform vertex handler for the given context.
  */
-void
-sp_init_vbuf(struct softpipe_context *sp)
+struct vbuf_render *
+sp_create_vbuf_backend(struct softpipe_context *sp)
 {
-   assert(sp->draw);
+   struct softpipe_vbuf_render *cvbr = CALLOC_STRUCT(softpipe_vbuf_render);
 
-   sp->vbuf_render = CALLOC_STRUCT(softpipe_vbuf_render);
+   assert(sp->draw);
 
-   sp->vbuf_render->base.max_indices = SP_MAX_VBUF_INDEXES;
-   sp->vbuf_render->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE;
 
-   sp->vbuf_render->base.get_vertex_info = sp_vbuf_get_vertex_info;
-   sp->vbuf_render->base.allocate_vertices = sp_vbuf_allocate_vertices;
-   sp->vbuf_render->base.map_vertices = sp_vbuf_map_vertices;
-   sp->vbuf_render->base.unmap_vertices = sp_vbuf_unmap_vertices;
-   sp->vbuf_render->base.set_primitive = sp_vbuf_set_primitive;
-   sp->vbuf_render->base.draw = sp_vbuf_draw;
-   sp->vbuf_render->base.draw_arrays = sp_vbuf_draw_arrays;
-   sp->vbuf_render->base.release_vertices = sp_vbuf_release_vertices;
-   sp->vbuf_render->base.destroy = sp_vbuf_destroy;
+   cvbr->base.max_indices = SP_MAX_VBUF_INDEXES;
+   cvbr->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE;
 
-   sp->vbuf_render->softpipe = sp;
+   cvbr->base.get_vertex_info = sp_vbuf_get_vertex_info;
+   cvbr->base.allocate_vertices = sp_vbuf_allocate_vertices;
+   cvbr->base.map_vertices = sp_vbuf_map_vertices;
+   cvbr->base.unmap_vertices = sp_vbuf_unmap_vertices;
+   cvbr->base.set_primitive = sp_vbuf_set_primitive;
+   cvbr->base.draw = sp_vbuf_draw;
+   cvbr->base.draw_arrays = sp_vbuf_draw_arrays;
+   cvbr->base.release_vertices = sp_vbuf_release_vertices;
+   cvbr->base.destroy = sp_vbuf_destroy;
 
-   sp->vbuf = draw_vbuf_stage(sp->draw, &sp->vbuf_render->base);
+   cvbr->softpipe = sp;
 
-   draw_set_rasterize_stage(sp->draw, sp->vbuf);
+   cvbr->setup = sp_setup_create_context(cvbr->softpipe);
 
-   draw_set_render(sp->draw, &sp->vbuf_render->base);
+   return &cvbr->base;
 }
diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.h b/src/gallium/drivers/softpipe/sp_prim_vbuf.h
index 1de9cc2a89..ad01cc2f28 100644
--- a/src/gallium/drivers/softpipe/sp_prim_vbuf.h
+++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.h
@@ -31,8 +31,8 @@
 
 struct softpipe_context;
 
-extern void
-sp_init_vbuf(struct softpipe_context *softpipe);
+extern struct vbuf_render *
+sp_create_vbuf_backend(struct softpipe_context *softpipe);
 
 
 #endif /* SP_VBUF_H */
diff --git a/src/gallium/drivers/softpipe/sp_quad.h b/src/gallium/drivers/softpipe/sp_quad.h
index bd6c6cb912..a3236bd116 100644
--- a/src/gallium/drivers/softpipe/sp_quad.h
+++ b/src/gallium/drivers/softpipe/sp_quad.h
@@ -97,10 +97,10 @@ struct quad_header {
    struct quad_header_inout inout;
    struct quad_header_output output;
 
-   const struct tgsi_interp_coef *coef;
+   /* Redundant/duplicated:
+    */
    const struct tgsi_interp_coef *posCoef;
-
-   unsigned nr_attrs;
+   const struct tgsi_interp_coef *coef;
 };
 
 #endif /* SP_QUAD_H */
diff --git a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c b/src/gallium/drivers/softpipe/sp_quad_alpha_test.c
deleted file mode 100644
index 0845bae0e6..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c
+++ /dev/null
@@ -1,108 +0,0 @@
-
-/**
- * quad alpha test
- */
-
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_quad_pipe.h"
-#include "pipe/p_defines.h"
-#include "util/u_memory.h"
-
-
-static void
-alpha_test_quad(struct quad_stage *qs, struct quad_header *quad)
-{
-   struct softpipe_context *softpipe = qs->softpipe;
-   const float ref = softpipe->depth_stencil->alpha.ref_value;
-   unsigned passMask = 0x0, j;
-   const uint cbuf = 0; /* only output[0].alpha is tested */
-   const float *aaaa = quad->output.color[cbuf][3];
-
-   switch (softpipe->depth_stencil->alpha.func) {
-   case PIPE_FUNC_NEVER:
-      break;
-   case PIPE_FUNC_LESS:
-      /*
-       * If mask were an array [4] we could do this SIMD-style:
-       * passMask = (quad->outputs.color[0][3] <= vec4(ref));
-       */
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (aaaa[j] < ref) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_EQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (aaaa[j] == ref) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_LEQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (aaaa[j] <= ref) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_GREATER:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (aaaa[j] > ref) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_NOTEQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (aaaa[j] != ref) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_GEQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (aaaa[j] >= ref) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_ALWAYS:
-      passMask = MASK_ALL;
-      break;
-   default:
-      assert(0);
-   }
-
-   quad->inout.mask &= passMask;
-
-   if (quad->inout.mask)
-      qs->next->run(qs->next, quad);
-}
-
-
-static void alpha_test_begin(struct quad_stage *qs)
-{
-   qs->next->begin(qs->next);
-}
-
-
-static void alpha_test_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
-
-struct quad_stage *
-sp_quad_alpha_test_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
-
-   stage->softpipe = softpipe;
-   stage->begin = alpha_test_begin;
-   stage->run = alpha_test_quad;
-   stage->destroy = alpha_test_destroy;
-
-   return stage;
-}
diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c
index b1e18805c7..0ad0b98654 100644
--- a/src/gallium/drivers/softpipe/sp_quad_blend.c
+++ b/src/gallium/drivers/softpipe/sp_quad_blend.c
@@ -117,644 +117,865 @@ do { \
 
 
 static void
-logicop_quad(struct quad_stage *qs, struct quad_header *quad)
+logicop_quad(struct quad_stage *qs, 
+             float (*quadColor)[4],
+             float (*dest)[4])
 {
    struct softpipe_context *softpipe = qs->softpipe;
-   uint cbuf;
+   ubyte src[4][4], dst[4][4], res[4][4];
+   uint *src4 = (uint *) src;
+   uint *dst4 = (uint *) dst;
+   uint *res4 = (uint *) res;
+   uint j;
+
+
+   /* convert to ubyte */
+   for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */
+      dst[j][0] = float_to_ubyte(dest[j][0]); /* P0 */
+      dst[j][1] = float_to_ubyte(dest[j][1]); /* P1 */
+      dst[j][2] = float_to_ubyte(dest[j][2]); /* P2 */
+      dst[j][3] = float_to_ubyte(dest[j][3]); /* P3 */
+
+      src[j][0] = float_to_ubyte(quadColor[j][0]); /* P0 */
+      src[j][1] = float_to_ubyte(quadColor[j][1]); /* P1 */
+      src[j][2] = float_to_ubyte(quadColor[j][2]); /* P2 */
+      src[j][3] = float_to_ubyte(quadColor[j][3]); /* P3 */
+   }
 
-   /* loop over colorbuffer outputs */
-   for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
-      float dest[4][QUAD_SIZE];
-      ubyte src[4][4], dst[4][4], res[4][4];
-      uint *src4 = (uint *) src;
-      uint *dst4 = (uint *) dst;
-      uint *res4 = (uint *) res;
-      struct softpipe_cached_tile *
-         tile = sp_get_cached_tile(softpipe,
-                                   softpipe->cbuf_cache[cbuf],
-                                   quad->input.x0, quad->input.y0);
-      float (*quadColor)[4] = quad->output.color[cbuf];
-      uint i, j;
+   switch (softpipe->blend->logicop_func) {
+   case PIPE_LOGICOP_CLEAR:
+      for (j = 0; j < 4; j++)
+         res4[j] = 0;
+      break;
+   case PIPE_LOGICOP_NOR:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~(src4[j] | dst4[j]);
+      break;
+   case PIPE_LOGICOP_AND_INVERTED:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~src4[j] & dst4[j];
+      break;
+   case PIPE_LOGICOP_COPY_INVERTED:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~src4[j];
+      break;
+   case PIPE_LOGICOP_AND_REVERSE:
+      for (j = 0; j < 4; j++)
+         res4[j] = src4[j] & ~dst4[j];
+      break;
+   case PIPE_LOGICOP_INVERT:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~dst4[j];
+      break;
+   case PIPE_LOGICOP_XOR:
+      for (j = 0; j < 4; j++)
+         res4[j] = dst4[j] ^ src4[j];
+      break;
+   case PIPE_LOGICOP_NAND:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~(src4[j] & dst4[j]);
+      break;
+   case PIPE_LOGICOP_AND:
+      for (j = 0; j < 4; j++)
+         res4[j] = src4[j] & dst4[j];
+      break;
+   case PIPE_LOGICOP_EQUIV:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~(src4[j] ^ dst4[j]);
+      break;
+   case PIPE_LOGICOP_NOOP:
+      for (j = 0; j < 4; j++)
+         res4[j] = dst4[j];
+      break;
+   case PIPE_LOGICOP_OR_INVERTED:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~src4[j] | dst4[j];
+      break;
+   case PIPE_LOGICOP_COPY:
+      for (j = 0; j < 4; j++)
+         res4[j] = src4[j];
+      break;
+   case PIPE_LOGICOP_OR_REVERSE:
+      for (j = 0; j < 4; j++)
+         res4[j] = src4[j] | ~dst4[j];
+      break;
+   case PIPE_LOGICOP_OR:
+      for (j = 0; j < 4; j++)
+         res4[j] = src4[j] | dst4[j];
+      break;
+   case PIPE_LOGICOP_SET:
+      for (j = 0; j < 4; j++)
+         res4[j] = ~0;
+      break;
+   default:
+      assert(0);
+   }
 
-      /* get/swizzle dest colors */
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1);
-         int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1);
-         for (i = 0; i < 4; i++) {
-            dest[i][j] = tile->data.color[y][x][i];
-         }
-      }
+   for (j = 0; j < 4; j++) {
+      quadColor[j][0] = ubyte_to_float(res[j][0]);
+      quadColor[j][1] = ubyte_to_float(res[j][1]);
+      quadColor[j][2] = ubyte_to_float(res[j][2]);
+      quadColor[j][3] = ubyte_to_float(res[j][3]);
+   }
+}
 
-      /* convert to ubyte */
-      for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */
-         dst[j][0] = float_to_ubyte(dest[j][0]); /* P0 */
-         dst[j][1] = float_to_ubyte(dest[j][1]); /* P1 */
-         dst[j][2] = float_to_ubyte(dest[j][2]); /* P2 */
-         dst[j][3] = float_to_ubyte(dest[j][3]); /* P3 */
-
-         src[j][0] = float_to_ubyte(quadColor[j][0]); /* P0 */
-         src[j][1] = float_to_ubyte(quadColor[j][1]); /* P1 */
-         src[j][2] = float_to_ubyte(quadColor[j][2]); /* P2 */
-         src[j][3] = float_to_ubyte(quadColor[j][3]); /* P3 */
-      }
 
-      switch (softpipe->blend->logicop_func) {
-      case PIPE_LOGICOP_CLEAR:
-         for (j = 0; j < 4; j++)
-            res4[j] = 0;
-         break;
-      case PIPE_LOGICOP_NOR:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~(src4[j] | dst4[j]);
-         break;
-      case PIPE_LOGICOP_AND_INVERTED:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~src4[j] & dst4[j];
-         break;
-      case PIPE_LOGICOP_COPY_INVERTED:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~src4[j];
-         break;
-      case PIPE_LOGICOP_AND_REVERSE:
-         for (j = 0; j < 4; j++)
-            res4[j] = src4[j] & ~dst4[j];
-         break;
-      case PIPE_LOGICOP_INVERT:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~dst4[j];
-         break;
-      case PIPE_LOGICOP_XOR:
-         for (j = 0; j < 4; j++)
-            res4[j] = dst4[j] ^ src4[j];
-         break;
-      case PIPE_LOGICOP_NAND:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~(src4[j] & dst4[j]);
-         break;
-      case PIPE_LOGICOP_AND:
-         for (j = 0; j < 4; j++)
-            res4[j] = src4[j] & dst4[j];
-         break;
-      case PIPE_LOGICOP_EQUIV:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~(src4[j] ^ dst4[j]);
-         break;
-      case PIPE_LOGICOP_NOOP:
-         for (j = 0; j < 4; j++)
-            res4[j] = dst4[j];
-         break;
-      case PIPE_LOGICOP_OR_INVERTED:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~src4[j] | dst4[j];
-         break;
-      case PIPE_LOGICOP_COPY:
-         for (j = 0; j < 4; j++)
-            res4[j] = src4[j];
-         break;
-      case PIPE_LOGICOP_OR_REVERSE:
-         for (j = 0; j < 4; j++)
-            res4[j] = src4[j] | ~dst4[j];
-         break;
-      case PIPE_LOGICOP_OR:
-         for (j = 0; j < 4; j++)
-            res4[j] = src4[j] | dst4[j];
-         break;
-      case PIPE_LOGICOP_SET:
-         for (j = 0; j < 4; j++)
-            res4[j] = ~0;
-         break;
-      default:
-         assert(0);
-      }
 
-      for (j = 0; j < 4; j++) {
-         quadColor[j][0] = ubyte_to_float(res[j][0]);
-         quadColor[j][1] = ubyte_to_float(res[j][1]);
-         quadColor[j][2] = ubyte_to_float(res[j][2]);
-         quadColor[j][3] = ubyte_to_float(res[j][3]);
-      }
+static void
+blend_quad(struct quad_stage *qs, 
+           float (*quadColor)[4],
+           float (*dest)[4])
+{
+   static const float zero[4] = { 0, 0, 0, 0 };
+   static const float one[4] = { 1, 1, 1, 1 };
+   struct softpipe_context *softpipe = qs->softpipe;
+   float source[4][QUAD_SIZE];
+
+   /*
+    * Compute src/first term RGB
+    */
+   switch (softpipe->blend->rgb_src_factor) {
+   case PIPE_BLENDFACTOR_ONE:
+      VEC4_COPY(source[0], quadColor[0]); /* R */
+      VEC4_COPY(source[1], quadColor[1]); /* G */
+      VEC4_COPY(source[2], quadColor[2]); /* B */
+      break;
+   case PIPE_BLENDFACTOR_SRC_COLOR:
+      VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */
+      VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */
+      VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA:
+   {
+      const float *alpha = quadColor[3];
+      VEC4_MUL(source[0], quadColor[0], alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */
+      VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */
+      VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */
+      break;
+   case PIPE_BLENDFACTOR_DST_ALPHA:
+   {
+      const float *alpha = dest[3];
+      VEC4_MUL(source[0], quadColor[0], alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+   {
+      const float *alpha = quadColor[3];
+      float diff[4], temp[4];
+      VEC4_SUB(diff, one, dest[3]);
+      VEC4_MIN(temp, alpha, diff);
+      VEC4_MUL(source[0], quadColor[0], temp); /* R */
+      VEC4_MUL(source[1], quadColor[1], temp); /* G */
+      VEC4_MUL(source[2], quadColor[2], temp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+   {
+      float comp[4];
+      VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */
+      VEC4_MUL(source[0], quadColor[0], comp); /* R */
+      VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */
+      VEC4_MUL(source[1], quadColor[1], comp); /* G */
+      VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */
+      VEC4_MUL(source[2], quadColor[2], comp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_CONST_ALPHA:
+   {
+      float alpha[4];
+      VEC4_SCALAR(alpha, softpipe->blend_color.color[3]);
+      VEC4_MUL(source[0], quadColor[0], alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_SRC1_COLOR:
+      assert(0); /* to do */
+      break;
+   case PIPE_BLENDFACTOR_SRC1_ALPHA:
+      assert(0); /* to do */
+      break;
+   case PIPE_BLENDFACTOR_ZERO:
+      VEC4_COPY(source[0], zero); /* R */
+      VEC4_COPY(source[1], zero); /* G */
+      VEC4_COPY(source[2], zero); /* B */
+      break;
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+   {
+      float inv_comp[4];
+      VEC4_SUB(inv_comp, one, quadColor[0]); /* R */
+      VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */
+      VEC4_SUB(inv_comp, one, quadColor[1]); /* G */
+      VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */
+      VEC4_SUB(inv_comp, one, quadColor[2]); /* B */
+      VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+   {
+      float inv_alpha[4];
+      VEC4_SUB(inv_alpha, one, quadColor[3]);
+      VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+   {
+      float inv_alpha[4];
+      VEC4_SUB(inv_alpha, one, dest[3]);
+      VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+   {
+      float inv_comp[4];
+      VEC4_SUB(inv_comp, one, dest[0]); /* R */
+      VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */
+      VEC4_SUB(inv_comp, one, dest[1]); /* G */
+      VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */
+      VEC4_SUB(inv_comp, one, dest[2]); /* B */
+      VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+   {
+      float inv_comp[4];
+      /* R */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]);
+      VEC4_MUL(source[0], quadColor[0], inv_comp);
+      /* G */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]);
+      VEC4_MUL(source[1], quadColor[1], inv_comp);
+      /* B */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]);
+      VEC4_MUL(source[2], quadColor[2], inv_comp);
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+   {
+      float inv_alpha[4];
+      VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]);
+      VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+      assert(0); /* to do */
+      break;
+   case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
+      assert(0); /* to do */
+      break;
+   default:
+      assert(0);
+   }
+
+   /*
+    * Compute src/first term A
+    */
+   switch (softpipe->blend->alpha_src_factor) {
+   case PIPE_BLENDFACTOR_ONE:
+      VEC4_COPY(source[3], quadColor[3]); /* A */
+      break;
+   case PIPE_BLENDFACTOR_SRC_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_SRC_ALPHA:
+   {
+      const float *alpha = quadColor[3];
+      VEC4_MUL(source[3], quadColor[3], alpha); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_DST_ALPHA:
+      VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+      /* multiply alpha by 1.0 */
+      VEC4_COPY(source[3], quadColor[3]); /* A */
+      break;
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_CONST_ALPHA:
+   {
+      float comp[4];
+      VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */
+      VEC4_MUL(source[3], quadColor[3], comp); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_ZERO:
+      VEC4_COPY(source[3], zero); /* A */
+      break;
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+   {
+      float inv_alpha[4];
+      VEC4_SUB(inv_alpha, one, quadColor[3]);
+      VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+   {
+      float inv_alpha[4];
+      VEC4_SUB(inv_alpha, one, dest[3]);
+      VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+   {
+      float inv_comp[4];
+      /* A */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]);
+      VEC4_MUL(source[3], quadColor[3], inv_comp);
+   }
+   break;
+   default:
+      assert(0);
    }
 
-   /* pass quad to next stage */
-   qs->next->run(qs->next, quad);
-}
 
+   /*
+    * Compute dest/second term RGB
+    */
+   switch (softpipe->blend->rgb_dst_factor) {
+   case PIPE_BLENDFACTOR_ONE:
+      /* dest = dest * 1   NO-OP, leave dest as-is */
+      break;
+   case PIPE_BLENDFACTOR_SRC_COLOR:
+      VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */
+      VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */
+      VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA:
+      VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */
+      VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */
+      VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */
+      break;
+   case PIPE_BLENDFACTOR_DST_ALPHA:
+      VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */
+      VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */
+      VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */
+      break;
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      VEC4_MUL(dest[0], dest[0], dest[0]); /* R */
+      VEC4_MUL(dest[1], dest[1], dest[1]); /* G */
+      VEC4_MUL(dest[2], dest[2], dest[2]); /* B */
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+      assert(0); /* illegal */
+      break;
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+   {
+      float comp[4];
+      VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */
+      VEC4_MUL(dest[0], dest[0], comp); /* R */
+      VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */
+      VEC4_MUL(dest[1], dest[1], comp); /* G */
+      VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */
+      VEC4_MUL(dest[2], dest[2], comp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_CONST_ALPHA:
+   {
+      float comp[4];
+      VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */
+      VEC4_MUL(dest[0], dest[0], comp); /* R */
+      VEC4_MUL(dest[1], dest[1], comp); /* G */
+      VEC4_MUL(dest[2], dest[2], comp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_ZERO:
+      VEC4_COPY(dest[0], zero); /* R */
+      VEC4_COPY(dest[1], zero); /* G */
+      VEC4_COPY(dest[2], zero); /* B */
+      break;
+   case PIPE_BLENDFACTOR_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_SRC1_ALPHA:
+      /* XXX what are these? */
+      assert(0);
+      break;
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+   {
+      float inv_comp[4];
+      VEC4_SUB(inv_comp, one, quadColor[0]); /* R */
+      VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */
+      VEC4_SUB(inv_comp, one, quadColor[1]); /* G */
+      VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */
+      VEC4_SUB(inv_comp, one, quadColor[2]); /* B */
+      VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+   {
+      float one_minus_alpha[QUAD_SIZE];
+      VEC4_SUB(one_minus_alpha, one, quadColor[3]);
+      VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */
+      VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */
+      VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+   {
+      float inv_comp[4];
+      VEC4_SUB(inv_comp, one, dest[3]); /* A */
+      VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */
+      VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */
+      VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+   {
+      float inv_comp[4];
+      VEC4_SUB(inv_comp, one, dest[0]); /* R */
+      VEC4_MUL(dest[0], dest[0], inv_comp); /* R */
+      VEC4_SUB(inv_comp, one, dest[1]); /* G */
+      VEC4_MUL(dest[1], dest[1], inv_comp); /* G */
+      VEC4_SUB(inv_comp, one, dest[2]); /* B */
+      VEC4_MUL(dest[2], dest[2], inv_comp); /* B */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+   {
+      float inv_comp[4];
+      /* R */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]);
+      VEC4_MUL(dest[0], dest[0], inv_comp);
+      /* G */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]);
+      VEC4_MUL(dest[1], dest[1], inv_comp);
+      /* B */
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]);
+      VEC4_MUL(dest[2], dest[2], inv_comp);
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+   {
+      float inv_comp[4];
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]);
+      VEC4_MUL(dest[0], dest[0], inv_comp);
+      VEC4_MUL(dest[1], dest[1], inv_comp);
+      VEC4_MUL(dest[2], dest[2], inv_comp);
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
+      /* XXX what are these? */
+      assert(0);
+      break;
+   default:
+      assert(0);
+   }
 
+   /*
+    * Compute dest/second term A
+    */
+   switch (softpipe->blend->alpha_dst_factor) {
+   case PIPE_BLENDFACTOR_ONE:
+      /* dest = dest * 1   NO-OP, leave dest as-is */
+      break;
+   case PIPE_BLENDFACTOR_SRC_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_SRC_ALPHA:
+      VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */
+      break;
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_DST_ALPHA:
+      VEC4_MUL(dest[3], dest[3], dest[3]); /* A */
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+      assert(0); /* illegal */
+      break;
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_CONST_ALPHA:
+   {
+      float comp[4];
+      VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */
+      VEC4_MUL(dest[3], dest[3], comp); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_ZERO:
+      VEC4_COPY(dest[3], zero); /* A */
+      break;
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+   {
+      float one_minus_alpha[QUAD_SIZE];
+      VEC4_SUB(one_minus_alpha, one, quadColor[3]);
+      VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+   {
+      float inv_comp[4];
+      VEC4_SUB(inv_comp, one, dest[3]); /* A */
+      VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */
+   }
+   break;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+      /* fall-through */
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+   {
+      float inv_comp[4];
+      VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]);
+      VEC4_MUL(dest[3], dest[3], inv_comp);
+   }
+   break;
+   default:
+      assert(0);
+   }
 
+   /*
+    * Combine RGB terms
+    */
+   switch (softpipe->blend->rgb_func) {
+   case PIPE_BLEND_ADD:
+      VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */
+      VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */
+      VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */
+      break;
+   case PIPE_BLEND_SUBTRACT:
+      VEC4_SUB_SAT(quadColor[0], source[0], dest[0]); /* R */
+      VEC4_SUB_SAT(quadColor[1], source[1], dest[1]); /* G */
+      VEC4_SUB_SAT(quadColor[2], source[2], dest[2]); /* B */
+      break;
+   case PIPE_BLEND_REVERSE_SUBTRACT:
+      VEC4_SUB_SAT(quadColor[0], dest[0], source[0]); /* R */
+      VEC4_SUB_SAT(quadColor[1], dest[1], source[1]); /* G */
+      VEC4_SUB_SAT(quadColor[2], dest[2], source[2]); /* B */
+      break;
+   case PIPE_BLEND_MIN:
+      VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */
+      VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */
+      VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */
+      break;
+   case PIPE_BLEND_MAX:
+      VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */
+      VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */
+      VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */
+      break;
+   default:
+      assert(0);
+   }
+
+   /*
+    * Combine A terms
+    */
+   switch (softpipe->blend->alpha_func) {
+   case PIPE_BLEND_ADD:
+      VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */
+      break;
+   case PIPE_BLEND_SUBTRACT:
+      VEC4_SUB_SAT(quadColor[3], source[3], dest[3]); /* A */
+      break;
+   case PIPE_BLEND_REVERSE_SUBTRACT:
+      VEC4_SUB_SAT(quadColor[3], dest[3], source[3]); /* A */
+      break;
+   case PIPE_BLEND_MIN:
+      VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */
+      break;
+   case PIPE_BLEND_MAX:
+      VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */
+      break;
+   default:
+      assert(0);
+   }
+}
 
 static void
-blend_quad(struct quad_stage *qs, struct quad_header *quad)
+colormask_quad(struct quad_stage *qs,
+               float (*quadColor)[4],
+               float (*dest)[4])
 {
-   static const float zero[4] = { 0, 0, 0, 0 };
-   static const float one[4] = { 1, 1, 1, 1 };
+   struct softpipe_context *softpipe = qs->softpipe;
 
+   /* R */
+   if (!(softpipe->blend->colormask & PIPE_MASK_R))
+      COPY_4V(quadColor[0], dest[0]);
+
+   /* G */
+   if (!(softpipe->blend->colormask & PIPE_MASK_G))
+      COPY_4V(quadColor[1], dest[1]);
+
+   /* B */
+   if (!(softpipe->blend->colormask & PIPE_MASK_B))
+      COPY_4V(quadColor[2], dest[2]);
+
+   /* A */
+   if (!(softpipe->blend->colormask & PIPE_MASK_A))
+      COPY_4V(quadColor[3], dest[3]);
+}
+
+
+static void
+blend_fallback(struct quad_stage *qs, 
+               struct quad_header *quads[],
+               unsigned nr)
+{
    struct softpipe_context *softpipe = qs->softpipe;
-   uint cbuf;
+   const struct pipe_blend_state *blend = softpipe->blend;
+   unsigned cbuf;
+
+   for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) 
+   {
+      float dest[4][QUAD_SIZE];
+      struct softpipe_cached_tile *tile
+         = sp_get_cached_tile(softpipe->cbuf_cache[cbuf],
+                              quads[0]->input.x0, 
+                              quads[0]->input.y0);
+      uint q, i, j;
+
+      for (q = 0; q < nr; q++) {
+         struct quad_header *quad = quads[q];
+         float (*quadColor)[4] = quad->output.color[cbuf];
+         const int itx = (quad->input.x0 & (TILE_SIZE-1));
+         const int ity = (quad->input.y0 & (TILE_SIZE-1));
+
+         /* get/swizzle dest colors 
+          */
+         for (j = 0; j < QUAD_SIZE; j++) {
+            int x = itx + (j & 1);
+            int y = ity + (j >> 1);
+            for (i = 0; i < 4; i++) {
+               dest[i][j] = tile->data.color[y][x][i];
+            }
+         }
 
-   if (softpipe->blend->logicop_enable) {
-      logicop_quad(qs, quad);
-      return;
+
+         if (blend->logicop_enable) {
+            logicop_quad( qs, quadColor, dest );
+         }
+         else if (blend->blend_enable) {
+            blend_quad( qs, quadColor, dest );
+         }
+
+         if (blend->colormask != 0xf)
+            colormask_quad( qs, quadColor, dest );
+   
+         /* Output color values
+          */
+         for (j = 0; j < QUAD_SIZE; j++) {
+            if (quad->inout.mask & (1 << j)) {
+               int x = itx + (j & 1);
+               int y = ity + (j >> 1);
+               for (i = 0; i < 4; i++) { /* loop over color chans */
+                  tile->data.color[y][x][i] = quadColor[i][j];
+               }
+            }
+         }
+      }
    }
+}
 
-   /* loop over colorbuffer outputs */
-   for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
-      float source[4][QUAD_SIZE], dest[4][QUAD_SIZE];
-      struct softpipe_cached_tile *tile
-         = sp_get_cached_tile(softpipe,
-                              softpipe->cbuf_cache[cbuf],
-                              quad->input.x0, quad->input.y0);
-      float (*quadColor)[4] = quad->output.color[cbuf];
-      uint i, j;
 
+static void
+blend_single_add_src_alpha_inv_src_alpha(struct quad_stage *qs, 
+                                         struct quad_header *quads[],
+                                         unsigned nr)
+{
+   static const float one[4] = { 1, 1, 1, 1 };
+   float one_minus_alpha[QUAD_SIZE];
+   float dest[4][QUAD_SIZE];
+   float source[4][QUAD_SIZE];
+   uint i, j, q;
+
+   struct softpipe_cached_tile *tile
+      = sp_get_cached_tile(qs->softpipe->cbuf_cache[0],
+                           quads[0]->input.x0, 
+                           quads[0]->input.y0);
+
+   for (q = 0; q < nr; q++) {
+      struct quad_header *quad = quads[q];
+      float (*quadColor)[4] = quad->output.color[0];
+      const float *alpha = quadColor[3];
+      const int itx = (quad->input.x0 & (TILE_SIZE-1));
+      const int ity = (quad->input.y0 & (TILE_SIZE-1));
+      
       /* get/swizzle dest colors */
       for (j = 0; j < QUAD_SIZE; j++) {
-         int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1);
-         int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1);
+         int x = itx + (j & 1);
+         int y = ity + (j >> 1);
          for (i = 0; i < 4; i++) {
             dest[i][j] = tile->data.color[y][x][i];
          }
       }
 
-      /*
-       * Compute src/first term RGB
-       */
-      switch (softpipe->blend->rgb_src_factor) {
-      case PIPE_BLENDFACTOR_ONE:
-         VEC4_COPY(source[0], quadColor[0]); /* R */
-         VEC4_COPY(source[1], quadColor[1]); /* G */
-         VEC4_COPY(source[2], quadColor[2]); /* B */
-         break;
-      case PIPE_BLENDFACTOR_SRC_COLOR:
-         VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */
-         VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */
-         VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */
-         break;
-      case PIPE_BLENDFACTOR_SRC_ALPHA:
-         {
-            const float *alpha = quadColor[3];
-            VEC4_MUL(source[0], quadColor[0], alpha); /* R */
-            VEC4_MUL(source[1], quadColor[1], alpha); /* G */
-            VEC4_MUL(source[2], quadColor[2], alpha); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_DST_COLOR:
-         VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */
-         VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */
-         VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */
-         break;
-      case PIPE_BLENDFACTOR_DST_ALPHA:
-         {
-            const float *alpha = dest[3];
-            VEC4_MUL(source[0], quadColor[0], alpha); /* R */
-            VEC4_MUL(source[1], quadColor[1], alpha); /* G */
-            VEC4_MUL(source[2], quadColor[2], alpha); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
-         {
-            const float *alpha = quadColor[3];
-            float diff[4], temp[4];
-            VEC4_SUB(diff, one, dest[3]);
-            VEC4_MIN(temp, alpha, diff);
-            VEC4_MUL(source[0], quadColor[0], temp); /* R */
-            VEC4_MUL(source[1], quadColor[1], temp); /* G */
-            VEC4_MUL(source[2], quadColor[2], temp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_CONST_COLOR:
-         {
-            float comp[4];
-            VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */
-            VEC4_MUL(source[0], quadColor[0], comp); /* R */
-            VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */
-            VEC4_MUL(source[1], quadColor[1], comp); /* G */
-            VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */
-            VEC4_MUL(source[2], quadColor[2], comp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_CONST_ALPHA:
-         {
-            float alpha[4];
-            VEC4_SCALAR(alpha, softpipe->blend_color.color[3]);
-            VEC4_MUL(source[0], quadColor[0], alpha); /* R */
-            VEC4_MUL(source[1], quadColor[1], alpha); /* G */
-            VEC4_MUL(source[2], quadColor[2], alpha); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_SRC1_COLOR:
-         assert(0); /* to do */
-         break;
-      case PIPE_BLENDFACTOR_SRC1_ALPHA:
-         assert(0); /* to do */
-         break;
-      case PIPE_BLENDFACTOR_ZERO:
-         VEC4_COPY(source[0], zero); /* R */
-         VEC4_COPY(source[1], zero); /* G */
-         VEC4_COPY(source[2], zero); /* B */
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC_COLOR:
-         {
-            float inv_comp[4];
-            VEC4_SUB(inv_comp, one, quadColor[0]); /* R */
-            VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */
-            VEC4_SUB(inv_comp, one, quadColor[1]); /* G */
-            VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */
-            VEC4_SUB(inv_comp, one, quadColor[2]); /* B */
-            VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-         {
-            float inv_alpha[4];
-            VEC4_SUB(inv_alpha, one, quadColor[3]);
-            VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */
-            VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */
-            VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_DST_ALPHA:
-         {
-            float inv_alpha[4];
-            VEC4_SUB(inv_alpha, one, dest[3]);
-            VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */
-            VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */
-            VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_DST_COLOR:
-         {
-            float inv_comp[4];
-            VEC4_SUB(inv_comp, one, dest[0]); /* R */
-            VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */
-            VEC4_SUB(inv_comp, one, dest[1]); /* G */
-            VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */
-            VEC4_SUB(inv_comp, one, dest[2]); /* B */
-            VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_CONST_COLOR:
-         {
-            float inv_comp[4];
-            /* R */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]);
-            VEC4_MUL(source[0], quadColor[0], inv_comp);
-            /* G */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]);
-            VEC4_MUL(source[1], quadColor[1], inv_comp);
-            /* B */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]);
-            VEC4_MUL(source[2], quadColor[2], inv_comp);
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
-         {
-            float inv_alpha[4];
-            VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]);
-            VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */
-            VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */
-            VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */
+      VEC4_MUL(source[0], quadColor[0], alpha); /* R */
+      VEC4_MUL(source[1], quadColor[1], alpha); /* G */
+      VEC4_MUL(source[2], quadColor[2], alpha); /* B */
+      VEC4_MUL(source[3], quadColor[3], alpha); /* A */
+
+      VEC4_SUB(one_minus_alpha, one, alpha);
+      VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */
+      VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */
+      VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */
+      VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* B */
+
+      VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */
+      VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */
+      VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */
+      VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */
+
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (quad->inout.mask & (1 << j)) {
+            int x = itx + (j & 1);
+            int y = ity + (j >> 1);
+            for (i = 0; i < 4; i++) { /* loop over color chans */
+               tile->data.color[y][x][i] = quadColor[i][j];
+            }
          }
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
-         assert(0); /* to do */
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
-         assert(0); /* to do */
-         break;
-      default:
-         assert(0);
       }
+   }
+}
 
-      /*
-       * Compute src/first term A
-       */
-      switch (softpipe->blend->alpha_src_factor) {
-      case PIPE_BLENDFACTOR_ONE:
-         VEC4_COPY(source[3], quadColor[3]); /* A */
-         break;
-      case PIPE_BLENDFACTOR_SRC_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_SRC_ALPHA:
-         {
-            const float *alpha = quadColor[3];
-            VEC4_MUL(source[3], quadColor[3], alpha); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_DST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_DST_ALPHA:
-         VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */
-         break;
-      case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
-         /* multiply alpha by 1.0 */
-         VEC4_COPY(source[3], quadColor[3]); /* A */
-         break;
-      case PIPE_BLENDFACTOR_CONST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_CONST_ALPHA:
-         {
-            float comp[4];
-            VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */
-            VEC4_MUL(source[3], quadColor[3], comp); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_ZERO:
-         VEC4_COPY(source[3], zero); /* A */
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-         {
-            float inv_alpha[4];
-            VEC4_SUB(inv_alpha, one, quadColor[3]);
-            VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_DST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_INV_DST_ALPHA:
-         {
-            float inv_alpha[4];
-            VEC4_SUB(inv_alpha, one, dest[3]);
-            VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_CONST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
-         {
-            float inv_comp[4];
-            /* A */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]);
-            VEC4_MUL(source[3], quadColor[3], inv_comp);
+static void
+blend_single_add_one_one(struct quad_stage *qs, 
+                         struct quad_header *quads[],
+                         unsigned nr)
+{
+   float dest[4][QUAD_SIZE];
+   uint i, j, q;
+
+   struct softpipe_cached_tile *tile
+      = sp_get_cached_tile(qs->softpipe->cbuf_cache[0],
+                           quads[0]->input.x0, 
+                           quads[0]->input.y0);
+
+   for (q = 0; q < nr; q++) {
+      struct quad_header *quad = quads[q];
+      float (*quadColor)[4] = quad->output.color[0];
+      const int itx = (quad->input.x0 & (TILE_SIZE-1));
+      const int ity = (quad->input.y0 & (TILE_SIZE-1));
+      
+      /* get/swizzle dest colors */
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = itx + (j & 1);
+         int y = ity + (j >> 1);
+         for (i = 0; i < 4; i++) {
+            dest[i][j] = tile->data.color[y][x][i];
          }
-         break;
-      default:
-         assert(0);
       }
+     
+      VEC4_ADD_SAT(quadColor[0], quadColor[0], dest[0]); /* R */
+      VEC4_ADD_SAT(quadColor[1], quadColor[1], dest[1]); /* G */
+      VEC4_ADD_SAT(quadColor[2], quadColor[2], dest[2]); /* B */
+      VEC4_ADD_SAT(quadColor[3], quadColor[3], dest[3]); /* A */
 
-
-      /*
-       * Compute dest/second term RGB
-       */
-      switch (softpipe->blend->rgb_dst_factor) {
-      case PIPE_BLENDFACTOR_ONE:
-         /* dest = dest * 1   NO-OP, leave dest as-is */
-         break;
-      case PIPE_BLENDFACTOR_SRC_COLOR:
-         VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */
-         VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */
-         VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */
-         break;
-      case PIPE_BLENDFACTOR_SRC_ALPHA:
-         VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */
-         VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */
-         VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */
-         break;
-      case PIPE_BLENDFACTOR_DST_ALPHA:
-         VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */
-         VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */
-         VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */
-         break;
-      case PIPE_BLENDFACTOR_DST_COLOR:
-         VEC4_MUL(dest[0], dest[0], dest[0]); /* R */
-         VEC4_MUL(dest[1], dest[1], dest[1]); /* G */
-         VEC4_MUL(dest[2], dest[2], dest[2]); /* B */
-         break;
-      case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
-         assert(0); /* illegal */
-         break;
-      case PIPE_BLENDFACTOR_CONST_COLOR:
-         {
-            float comp[4];
-            VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */
-            VEC4_MUL(dest[0], dest[0], comp); /* R */
-            VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */
-            VEC4_MUL(dest[1], dest[1], comp); /* G */
-            VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */
-            VEC4_MUL(dest[2], dest[2], comp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_CONST_ALPHA:
-         {
-            float comp[4];
-            VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */
-            VEC4_MUL(dest[0], dest[0], comp); /* R */
-            VEC4_MUL(dest[1], dest[1], comp); /* G */
-            VEC4_MUL(dest[2], dest[2], comp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_ZERO:
-         VEC4_COPY(dest[0], zero); /* R */
-         VEC4_COPY(dest[1], zero); /* G */
-         VEC4_COPY(dest[2], zero); /* B */
-         break;
-      case PIPE_BLENDFACTOR_SRC1_COLOR:
-      case PIPE_BLENDFACTOR_SRC1_ALPHA:
-         /* XXX what are these? */
-         assert(0);
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC_COLOR:
-         {
-            float inv_comp[4];
-            VEC4_SUB(inv_comp, one, quadColor[0]); /* R */
-            VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */
-            VEC4_SUB(inv_comp, one, quadColor[1]); /* G */
-            VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */
-            VEC4_SUB(inv_comp, one, quadColor[2]); /* B */
-            VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-         {
-            float one_minus_alpha[QUAD_SIZE];
-            VEC4_SUB(one_minus_alpha, one, quadColor[3]);
-            VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */
-            VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */
-            VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_DST_ALPHA:
-         {
-            float inv_comp[4];
-            VEC4_SUB(inv_comp, one, dest[3]); /* A */
-            VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */
-            VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */
-            VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_DST_COLOR:
-         {
-            float inv_comp[4];
-            VEC4_SUB(inv_comp, one, dest[0]); /* R */
-            VEC4_MUL(dest[0], dest[0], inv_comp); /* R */
-            VEC4_SUB(inv_comp, one, dest[1]); /* G */
-            VEC4_MUL(dest[1], dest[1], inv_comp); /* G */
-            VEC4_SUB(inv_comp, one, dest[2]); /* B */
-            VEC4_MUL(dest[2], dest[2], inv_comp); /* B */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_CONST_COLOR:
-         {
-            float inv_comp[4];
-            /* R */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]);
-            VEC4_MUL(dest[0], dest[0], inv_comp);
-            /* G */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]);
-            VEC4_MUL(dest[1], dest[1], inv_comp);
-            /* B */
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]);
-            VEC4_MUL(dest[2], dest[2], inv_comp);
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
-         {
-            float inv_comp[4];
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]);
-            VEC4_MUL(dest[0], dest[0], inv_comp);
-            VEC4_MUL(dest[1], dest[1], inv_comp);
-            VEC4_MUL(dest[2], dest[2], inv_comp);
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (quad->inout.mask & (1 << j)) {
+            int x = itx + (j & 1);
+            int y = ity + (j >> 1);
+            for (i = 0; i < 4; i++) { /* loop over color chans */
+               tile->data.color[y][x][i] = quadColor[i][j];
+            }
          }
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
-      case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
-         /* XXX what are these? */
-         assert(0);
-         break;
-      default:
-         assert(0);
       }
+   }
+}
 
-      /*
-       * Compute dest/second term A
-       */
-      switch (softpipe->blend->alpha_dst_factor) {
-      case PIPE_BLENDFACTOR_ONE:
-         /* dest = dest * 1   NO-OP, leave dest as-is */
-         break;
-      case PIPE_BLENDFACTOR_SRC_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_SRC_ALPHA:
-         VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */
-         break;
-      case PIPE_BLENDFACTOR_DST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_DST_ALPHA:
-         VEC4_MUL(dest[3], dest[3], dest[3]); /* A */
-         break;
-      case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
-         assert(0); /* illegal */
-         break;
-      case PIPE_BLENDFACTOR_CONST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_CONST_ALPHA:
-         {
-            float comp[4];
-            VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */
-            VEC4_MUL(dest[3], dest[3], comp); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_ZERO:
-         VEC4_COPY(dest[3], zero); /* A */
-         break;
-      case PIPE_BLENDFACTOR_INV_SRC_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-         {
-            float one_minus_alpha[QUAD_SIZE];
-            VEC4_SUB(one_minus_alpha, one, quadColor[3]);
-            VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_DST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_INV_DST_ALPHA:
-         {
-            float inv_comp[4];
-            VEC4_SUB(inv_comp, one, dest[3]); /* A */
-            VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */
-         }
-         break;
-      case PIPE_BLENDFACTOR_INV_CONST_COLOR:
-         /* fall-through */
-      case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
-         {
-            float inv_comp[4];
-            VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]);
-            VEC4_MUL(dest[3], dest[3], inv_comp);
+
+static void
+single_output_color(struct quad_stage *qs, 
+                    struct quad_header *quads[],
+                    unsigned nr)
+{
+   uint i, j, q;
+
+   struct softpipe_cached_tile *tile
+      = sp_get_cached_tile(qs->softpipe->cbuf_cache[0],
+                           quads[0]->input.x0, 
+                           quads[0]->input.y0);
+
+   for (q = 0; q < nr; q++) {
+      struct quad_header *quad = quads[q];
+      float (*quadColor)[4] = quad->output.color[0];
+      const int itx = (quad->input.x0 & (TILE_SIZE-1));
+      const int ity = (quad->input.y0 & (TILE_SIZE-1));
+      
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (quad->inout.mask & (1 << j)) {
+            int x = itx + (j & 1);
+            int y = ity + (j >> 1);
+            for (i = 0; i < 4; i++) { /* loop over color chans */
+               tile->data.color[y][x][i] = quadColor[i][j];
+            }
          }
-         break;
-      default:
-         assert(0);
       }
+   }
+}
+
+static void
+blend_noop(struct quad_stage *qs, 
+           struct quad_header *quads[],
+           unsigned nr)
+{
+}
 
-      /*
-       * Combine RGB terms
-       */
-      switch (softpipe->blend->rgb_func) {
-      case PIPE_BLEND_ADD:
-         VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */
-         VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */
-         VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */
-         break;
-      case PIPE_BLEND_SUBTRACT:
-         VEC4_SUB_SAT(quadColor[0], source[0], dest[0]); /* R */
-         VEC4_SUB_SAT(quadColor[1], source[1], dest[1]); /* G */
-         VEC4_SUB_SAT(quadColor[2], source[2], dest[2]); /* B */
-         break;
-      case PIPE_BLEND_REVERSE_SUBTRACT:
-         VEC4_SUB_SAT(quadColor[0], dest[0], source[0]); /* R */
-         VEC4_SUB_SAT(quadColor[1], dest[1], source[1]); /* G */
-         VEC4_SUB_SAT(quadColor[2], dest[2], source[2]); /* B */
-         break;
-      case PIPE_BLEND_MIN:
-         VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */
-         VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */
-         VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */
-         break;
-      case PIPE_BLEND_MAX:
-         VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */
-         VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */
-         VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */
-         break;
-      default:
-         assert(0);
-      }
 
-      /*
-       * Combine A terms
-       */
-      switch (softpipe->blend->alpha_func) {
-      case PIPE_BLEND_ADD:
-         VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */
-         break;
-      case PIPE_BLEND_SUBTRACT:
-         VEC4_SUB_SAT(quadColor[3], source[3], dest[3]); /* A */
-         break;
-      case PIPE_BLEND_REVERSE_SUBTRACT:
-         VEC4_SUB_SAT(quadColor[3], dest[3], source[3]); /* A */
-         break;
-      case PIPE_BLEND_MIN:
-         VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */
-         break;
-      case PIPE_BLEND_MAX:
-         VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */
-         break;
-      default:
-         assert(0);
+static void
+choose_blend_quad(struct quad_stage *qs, 
+                  struct quad_header *quads[],
+                  unsigned nr)
+{
+   struct softpipe_context *softpipe = qs->softpipe;
+   const struct pipe_blend_state *blend = softpipe->blend;
+
+   qs->run = blend_fallback;
+   
+   if (softpipe->framebuffer.nr_cbufs == 0) {
+      qs->run = blend_noop;
+   }
+   else if (!softpipe->blend->logicop_enable &&
+            softpipe->blend->colormask == 0xf &&
+            softpipe->framebuffer.nr_cbufs == 1)
+   {
+      if (!blend->blend_enable) {
+         qs->run = single_output_color;
       }
+      else if (blend->rgb_src_factor == blend->alpha_src_factor &&
+               blend->rgb_dst_factor == blend->alpha_dst_factor &&
+               blend->rgb_func == blend->alpha_func)
+      {
+         if (blend->alpha_func == PIPE_BLEND_ADD) {
+            if (blend->rgb_src_factor == PIPE_BLENDFACTOR_ONE &&
+                blend->rgb_dst_factor == PIPE_BLENDFACTOR_ONE) {
+               qs->run = blend_single_add_one_one;
+            }
+            else if (blend->rgb_src_factor == PIPE_BLENDFACTOR_SRC_ALPHA &&
+                blend->rgb_dst_factor == PIPE_BLENDFACTOR_INV_SRC_ALPHA)
+               qs->run = blend_single_add_src_alpha_inv_src_alpha;
 
-   } /* cbuf loop */
+         }
+      }
+   }
 
-   /* pass blended quad to next stage */
-   qs->next->run(qs->next, quad);
+   qs->run(qs, quads, nr);
 }
 
 
 static void blend_begin(struct quad_stage *qs)
 {
-   qs->next->begin(qs->next);
+   qs->run = choose_blend_quad;
 }
 
 
@@ -770,7 +991,7 @@ struct quad_stage *sp_quad_blend_stage( struct softpipe_context *softpipe )
 
    stage->softpipe = softpipe;
    stage->begin = blend_begin;
-   stage->run = blend_quad;
+   stage->run = choose_blend_quad;
    stage->destroy = blend_destroy;
 
    return stage;
diff --git a/src/gallium/drivers/softpipe/sp_quad_bufloop.c b/src/gallium/drivers/softpipe/sp_quad_bufloop.c
deleted file mode 100644
index 953d8516b9..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_bufloop.c
+++ /dev/null
@@ -1,74 +0,0 @@
-
-#include "util/u_memory.h"
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_surface.h"
-#include "sp_quad_pipe.h"
-
-
-/**
- * Loop over colorbuffers, passing quad to next stage each time.
- */
-static void
-cbuf_loop_quad(struct quad_stage *qs, struct quad_header *quad)
-{
-   struct softpipe_context *softpipe = qs->softpipe;
-   float tmp[PIPE_MAX_COLOR_BUFS][4][QUAD_SIZE];
-   unsigned i;
-
-   assert(sizeof(quad->outputs.color) == sizeof(tmp));
-   assert(softpipe->framebuffer.nr_cbufs <= PIPE_MAX_COLOR_BUFS);
-
-   /* make copy of original colors since they can get modified
-    * by blending and masking.
-    * XXX we won't have to do this if the fragment program actually emits
-    * N separate colors and we're drawing to N color buffers (MRT).
-    * But if we emitted one color and glDrawBuffer(GL_FRONT_AND_BACK) is
-    * in effect, we need to save/restore colors like this.
-    */
-   memcpy(tmp, quad->outputs.color, sizeof(tmp));
-
-   for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) {
-      /* set current cbuffer */
-#if 0 /* obsolete & going away */
-      softpipe->current_cbuf = i;
-#endif
-
-      /* pass blended quad to next stage */
-      qs->next->run(qs->next, quad);
-
-      /* restore quad's colors for next buffer */
-      memcpy(quad->outputs.color, tmp, sizeof(tmp));
-   }
-}
-
-
-static void cbuf_loop_begin(struct quad_stage *qs)
-{
-   qs->next->begin(qs->next);
-}
-
-
-static void cbuf_loop_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
-
-/**
- * Create the colorbuffer loop stage.
- * This is used to implement multiple render targets and GL_FRONT_AND_BACK
- * rendering.
- */
-struct quad_stage *sp_quad_bufloop_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
-
-   stage->softpipe = softpipe;
-   stage->begin = cbuf_loop_begin;
-   stage->run = cbuf_loop_quad;
-   stage->destroy = cbuf_loop_destroy;
-
-   return stage;
-}
-
diff --git a/src/gallium/drivers/softpipe/sp_quad_colormask.c b/src/gallium/drivers/softpipe/sp_quad_colormask.c
deleted file mode 100644
index dc90e5d5e9..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_colormask.c
+++ /dev/null
@@ -1,116 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/**
- * \brief  quad colormask stage
- * \author Brian Paul
- */
-
-#include "pipe/p_defines.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_surface.h"
-#include "sp_quad_pipe.h"
-#include "sp_tile_cache.h"
-
-
-
-/**
- * XXX colormask could be rolled into blending...
- */
-static void
-colormask_quad(struct quad_stage *qs, struct quad_header *quad)
-{
-   struct softpipe_context *softpipe = qs->softpipe;
-   uint cbuf;
-
-   /* loop over colorbuffer outputs */
-   for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
-      float dest[4][QUAD_SIZE];
-      struct softpipe_cached_tile *tile
-         = sp_get_cached_tile(softpipe,
-                              softpipe->cbuf_cache[cbuf],
-                              quad->input.x0, quad->input.y0);
-      float (*quadColor)[4] = quad->output.color[cbuf];
-      uint i, j;
-
-      /* get/swizzle dest colors */
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1);
-         int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1);
-         for (i = 0; i < 4; i++) {
-            dest[i][j] = tile->data.color[y][x][i];
-         }
-      }
-
-      /* R */
-      if (!(softpipe->blend->colormask & PIPE_MASK_R))
-          COPY_4V(quadColor[0], dest[0]);
-
-      /* G */
-      if (!(softpipe->blend->colormask & PIPE_MASK_G))
-          COPY_4V(quadColor[1], dest[1]);
-
-      /* B */
-      if (!(softpipe->blend->colormask & PIPE_MASK_B))
-          COPY_4V(quadColor[2], dest[2]);
-
-      /* A */
-      if (!(softpipe->blend->colormask & PIPE_MASK_A))
-          COPY_4V(quadColor[3], dest[3]);
-   }
-
-   /* pass quad to next stage */
-   qs->next->run(qs->next, quad);
-}
-
-
-static void colormask_begin(struct quad_stage *qs)
-{
-   qs->next->begin(qs->next);
-}
-
-
-static void colormask_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
-
-struct quad_stage *sp_quad_colormask_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
-
-   stage->softpipe = softpipe;
-   stage->begin = colormask_begin;
-   stage->run = colormask_quad;
-   stage->destroy = colormask_destroy;
-
-   return stage;
-}
diff --git a/src/gallium/drivers/softpipe/sp_quad_coverage.c b/src/gallium/drivers/softpipe/sp_quad_coverage.c
deleted file mode 100644
index 4aeee85870..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_coverage.c
+++ /dev/null
@@ -1,94 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-/**
- * \brief  Apply AA coverage to quad alpha valus
- * \author  Brian Paul
- */
-
-
-#include "pipe/p_defines.h"
-#include "util/u_memory.h"
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_quad_pipe.h"
-
-
-/**
- * Multiply quad's alpha values by the fragment coverage.
- */
-static void
-coverage_quad(struct quad_stage *qs, struct quad_header *quad)
-{
-   struct softpipe_context *softpipe = qs->softpipe;
-   const uint prim = quad->input.prim;
-
-   if ((softpipe->rasterizer->poly_smooth && prim == QUAD_PRIM_TRI) ||
-       (softpipe->rasterizer->line_smooth && prim == QUAD_PRIM_LINE) ||
-       (softpipe->rasterizer->point_smooth && prim == QUAD_PRIM_POINT)) {
-      uint cbuf;
-
-      /* loop over colorbuffer outputs */
-      for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
-         float (*quadColor)[4] = quad->output.color[cbuf];
-         unsigned j;
-         for (j = 0; j < QUAD_SIZE; j++) {
-            assert(quad->input.coverage[j] >= 0.0);
-            assert(quad->input.coverage[j] <= 1.0);
-         quadColor[3][j] *= quad->input.coverage[j];
-         }
-      }
-   }
-
-   qs->next->run(qs->next, quad);
-}
-
-
-static void coverage_begin(struct quad_stage *qs)
-{
-   qs->next->begin(qs->next);
-}
-
-
-static void coverage_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
-
-struct quad_stage *sp_quad_coverage_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
-
-   stage->softpipe = softpipe;
-   stage->begin = coverage_begin;
-   stage->run = coverage_quad;
-   stage->destroy = coverage_destroy;
-
-   return stage;
-}
diff --git a/src/gallium/drivers/softpipe/sp_quad_depth_test.c b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
index d463930bae..0ca86c4e1c 100644
--- a/src/gallium/drivers/softpipe/sp_quad_depth_test.c
+++ b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
@@ -31,61 +31,109 @@
 
 #include "pipe/p_defines.h"
 #include "util/u_memory.h"
+#include "tgsi/tgsi_scan.h"
 #include "sp_context.h"
 #include "sp_quad.h"
 #include "sp_surface.h"
 #include "sp_quad_pipe.h"
 #include "sp_tile_cache.h"
+#include "sp_state.h"           /* for sp_fragment_shader */
 
 
-/**
- * Do depth testing for a quad.
- * Not static since it's used by the stencil code.
- */
+struct depth_data {
+   struct pipe_surface *ps;
+   enum pipe_format format;
+   unsigned bzzzz[QUAD_SIZE];  /**< Z values fetched from depth buffer */
+   unsigned qzzzz[QUAD_SIZE];  /**< Z values from the quad */
+   ubyte stencilVals[QUAD_SIZE];
+   struct softpipe_cached_tile *tile;
+};
 
-/*
- * To increase efficiency, we should probably have multiple versions
- * of this function that are specifically for Z16, Z32 and FP Z buffers.
- * Try to effectively do that with codegen...
- */
 
-void
-sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad)
+
+static void
+get_depth_stencil_values( struct depth_data *data,
+                          const struct quad_header *quad )
 {
-   struct softpipe_context *softpipe = qs->softpipe;
-   struct pipe_surface *ps = softpipe->framebuffer.zsbuf;
-   const enum pipe_format format = ps->format;
-   unsigned bzzzz[QUAD_SIZE];  /**< Z values fetched from depth buffer */
-   unsigned qzzzz[QUAD_SIZE];  /**< Z values from the quad */
-   unsigned zmask = 0;
    unsigned j;
-   struct softpipe_cached_tile *tile
-      = sp_get_cached_tile(softpipe, softpipe->zsbuf_cache, quad->input.x0, quad->input.y0);
+   const struct softpipe_cached_tile *tile = data->tile;
+
+   switch (data->format) {
+   case PIPE_FORMAT_Z16_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         data->bzzzz[j] = tile->data.depth16[y][x];
+      }
+      break;
+   case PIPE_FORMAT_Z32_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         data->bzzzz[j] = tile->data.depth32[y][x];
+      }
+      break;
+   case PIPE_FORMAT_X8Z24_UNORM:
+   case PIPE_FORMAT_S8Z24_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         data->bzzzz[j] = tile->data.depth32[y][x] & 0xffffff;
+         data->stencilVals[j] = tile->data.depth32[y][x] >> 24;
+      }
+   break;
+   case PIPE_FORMAT_Z24X8_UNORM:
+   case PIPE_FORMAT_Z24S8_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         data->bzzzz[j] = tile->data.depth32[y][x] >> 8;
+         data->stencilVals[j] = tile->data.depth32[y][x] & 0xff;
+      }
+      break;
+   default:
+      assert(0);
+   }
+}
 
-   assert(ps); /* shouldn't get here if there's no zbuffer */
+/* If the shader has not been run, interpolate the depth values
+ * ourselves.
+ */
+static void
+interpolate_quad_depth( struct quad_header *quad )
+{
+   const float fx = (float) quad->input.x0;
+   const float fy = (float) quad->input.y0;
+   const float dzdx = quad->posCoef->dadx[2];
+   const float dzdy = quad->posCoef->dady[2];
+   const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy;
 
-   /*
-    * Convert quad's float depth values to int depth values (qzzzz).
+   quad->output.depth[0] = z0;
+   quad->output.depth[1] = z0 + dzdx;
+   quad->output.depth[2] = z0 + dzdy;
+   quad->output.depth[3] = z0 + dzdx + dzdy;
+}
+
+
+static void
+convert_quad_depth( struct depth_data *data, 
+                    const struct quad_header *quad )
+{
+   unsigned j;
+
+   /* Convert quad's float depth values to int depth values (qzzzz).
     * If the Z buffer stores integer values, we _have_ to do the depth
     * compares with integers (not floats).  Otherwise, the float->int->float
     * conversion of Z values (which isn't an identity function) will cause
     * Z-fighting errors.
-    *
-    * Also, get the zbuffer values (bzzzz) from the cached tile.
     */
-   switch (format) {
+   switch (data->format) {
    case PIPE_FORMAT_Z16_UNORM:
       {
          float scale = 65535.0;
 
          for (j = 0; j < QUAD_SIZE; j++) {
-            qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
-         }
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            bzzzz[j] = tile->data.depth16[y][x];
+            data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
          }
       }
       break;
@@ -94,47 +142,247 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad)
          double scale = (double) (uint) ~0UL;
 
          for (j = 0; j < QUAD_SIZE; j++) {
-            qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
-         }
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            bzzzz[j] = tile->data.depth32[y][x];
+            data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
          }
       }
       break;
    case PIPE_FORMAT_X8Z24_UNORM:
-      /* fall-through */
    case PIPE_FORMAT_S8Z24_UNORM:
       {
          float scale = (float) ((1 << 24) - 1);
 
          for (j = 0; j < QUAD_SIZE; j++) {
-            qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
-         }
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            bzzzz[j] = tile->data.depth32[y][x] & 0xffffff;
+            data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
          }
       }
       break;
    case PIPE_FORMAT_Z24X8_UNORM:
-      /* fall-through */
    case PIPE_FORMAT_Z24S8_UNORM:
       {
          float scale = (float) ((1 << 24) - 1);
 
          for (j = 0; j < QUAD_SIZE; j++) {
-            qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
+            data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale);
          }
+      }
+      break;
+   default:
+      assert(0);
+   }
+}
 
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            bzzzz[j] = tile->data.depth32[y][x] >> 8;
+
+
+static void
+write_depth_stencil_values( struct depth_data *data,
+                            struct quad_header *quad )
+{
+   struct softpipe_cached_tile *tile = data->tile;
+   unsigned j;
+
+   /* put updated Z values back into cached tile */
+   switch (data->format) {
+   case PIPE_FORMAT_Z16_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         tile->data.depth16[y][x] = (ushort) data->bzzzz[j];
+      }
+      break;
+   case PIPE_FORMAT_X8Z24_UNORM:
+   case PIPE_FORMAT_Z32_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         tile->data.depth32[y][x] = data->bzzzz[j];
+      }
+      break;
+   case PIPE_FORMAT_S8Z24_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         tile->data.depth32[y][x] = (data->stencilVals[j] << 24) | data->bzzzz[j];
+      }
+      break;
+   case PIPE_FORMAT_Z24S8_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         tile->data.depth32[y][x] = (data->bzzzz[j] << 8) | data->stencilVals[j];
+      }
+      break;
+   case PIPE_FORMAT_Z24X8_UNORM:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         int x = quad->input.x0 % TILE_SIZE + (j & 1);
+         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
+         tile->data.depth32[y][x] = data->bzzzz[j] << 8;
+      }
+      break;
+   default:
+      assert(0);
+   }
+}
+
+
+
+
+/** Only 8-bit stencil supported */
+#define STENCIL_MAX 0xff
+
+
+/**
+ * Do the basic stencil test (compare stencil buffer values against the
+ * reference value.
+ *
+ * \param data->stencilVals  the stencil values from the stencil buffer
+ * \param func  the stencil func (PIPE_FUNC_x)
+ * \param ref  the stencil reference value
+ * \param valMask  the stencil value mask indicating which bits of the stencil
+ *                 values and ref value are to be used.
+ * \return mask indicating which pixels passed the stencil test
+ */
+static unsigned
+do_stencil_test(struct depth_data *data,
+                unsigned func,
+                unsigned ref, unsigned valMask)
+{
+   unsigned passMask = 0x0;
+   unsigned j;
+
+   ref &= valMask;
+
+   switch (func) {
+   case PIPE_FUNC_NEVER:
+      /* passMask = 0x0 */
+      break;
+   case PIPE_FUNC_LESS:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (ref < (data->stencilVals[j] & valMask)) {
+            passMask |= (1 << j);
+         }
+      }
+      break;
+   case PIPE_FUNC_EQUAL:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (ref == (data->stencilVals[j] & valMask)) {
+            passMask |= (1 << j);
+         }
+      }
+      break;
+   case PIPE_FUNC_LEQUAL:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (ref <= (data->stencilVals[j] & valMask)) {
+            passMask |= (1 << j);
+         }
+      }
+      break;
+   case PIPE_FUNC_GREATER:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (ref > (data->stencilVals[j] & valMask)) {
+            passMask |= (1 << j);
+         }
+      }
+      break;
+   case PIPE_FUNC_NOTEQUAL:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (ref != (data->stencilVals[j] & valMask)) {
+            passMask |= (1 << j);
+         }
+      }
+      break;
+   case PIPE_FUNC_GEQUAL:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (ref >= (data->stencilVals[j] & valMask)) {
+            passMask |= (1 << j);
+         }
+      }
+      break;
+   case PIPE_FUNC_ALWAYS:
+      passMask = MASK_ALL;
+      break;
+   default:
+      assert(0);
+   }
+
+   return passMask;
+}
+
+
+/**
+ * Apply the stencil operator to stencil values.
+ *
+ * \param data->stencilVals  the stencil buffer values (read and written)
+ * \param mask  indicates which pixels to update
+ * \param op  the stencil operator (PIPE_STENCIL_OP_x)
+ * \param ref  the stencil reference value
+ * \param wrtMask  writemask controlling which bits are changed in the
+ *                 stencil values
+ */
+static void
+apply_stencil_op(struct depth_data *data,
+                 unsigned mask, unsigned op, ubyte ref, ubyte wrtMask)
+{
+   unsigned j;
+   ubyte newstencil[QUAD_SIZE];
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      newstencil[j] = data->stencilVals[j];
+   }
+
+   switch (op) {
+   case PIPE_STENCIL_OP_KEEP:
+      /* no-op */
+      break;
+   case PIPE_STENCIL_OP_ZERO:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            newstencil[j] = 0;
+         }
+      }
+      break;
+   case PIPE_STENCIL_OP_REPLACE:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            newstencil[j] = ref;
+         }
+      }
+      break;
+   case PIPE_STENCIL_OP_INCR:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            if (data->stencilVals[j] < STENCIL_MAX) {
+               newstencil[j] = data->stencilVals[j] + 1;
+            }
+         }
+      }
+      break;
+   case PIPE_STENCIL_OP_DECR:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            if (data->stencilVals[j] > 0) {
+               newstencil[j] = data->stencilVals[j] - 1;
+            }
+         }
+      }
+      break;
+   case PIPE_STENCIL_OP_INCR_WRAP:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            newstencil[j] = data->stencilVals[j] + 1;
+         }
+      }
+      break;
+   case PIPE_STENCIL_OP_DECR_WRAP:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            newstencil[j] = data->stencilVals[j] - 1;
+         }
+      }
+      break;
+   case PIPE_STENCIL_OP_INVERT:
+      for (j = 0; j < QUAD_SIZE; j++) {
+         if (mask & (1 << j)) {
+            newstencil[j] = ~data->stencilVals[j];
          }
       }
       break;
@@ -142,6 +390,39 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad)
       assert(0);
    }
 
+   /*
+    * update the stencil values
+    */
+   if (wrtMask != STENCIL_MAX) {
+      /* apply bit-wise stencil buffer writemask */
+      for (j = 0; j < QUAD_SIZE; j++) {
+         data->stencilVals[j] = (wrtMask & newstencil[j]) | (~wrtMask & data->stencilVals[j]);
+      }
+   }
+   else {
+      for (j = 0; j < QUAD_SIZE; j++) {
+         data->stencilVals[j] = newstencil[j];
+      }
+   }
+}
+
+   
+
+/*
+ * To increase efficiency, we should probably have multiple versions
+ * of this function that are specifically for Z16, Z32 and FP Z buffers.
+ * Try to effectively do that with codegen...
+ */
+
+static boolean
+depth_test_quad(struct quad_stage *qs, 
+                struct depth_data *data,
+                struct quad_header *quad)
+{
+   struct softpipe_context *softpipe = qs->softpipe;
+   unsigned zmask = 0;
+   unsigned j;
+
    switch (softpipe->depth_stencil->depth.func) {
    case PIPE_FUNC_NEVER:
       /* zmask = 0 */
@@ -151,37 +432,37 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad)
        * Like this:  quad->mask &= (quad->outputs.depth < zzzz);
        */
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (qzzzz[j] < bzzzz[j]) 
+	 if (data->qzzzz[j] < data->bzzzz[j]) 
 	    zmask |= 1 << j;
       }
       break;
    case PIPE_FUNC_EQUAL:
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (qzzzz[j] == bzzzz[j]) 
+	 if (data->qzzzz[j] == data->bzzzz[j]) 
 	    zmask |= 1 << j;
       }
       break;
    case PIPE_FUNC_LEQUAL:
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (qzzzz[j] <= bzzzz[j]) 
+	 if (data->qzzzz[j] <= data->bzzzz[j]) 
 	    zmask |= (1 << j);
       }
       break;
    case PIPE_FUNC_GREATER:
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (qzzzz[j] > bzzzz[j]) 
+	 if (data->qzzzz[j] > data->bzzzz[j]) 
 	    zmask |= (1 << j);
       }
       break;
    case PIPE_FUNC_NOTEQUAL:
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (qzzzz[j] != bzzzz[j]) 
+	 if (data->qzzzz[j] != data->bzzzz[j]) 
 	    zmask |= (1 << j);
       }
       break;
    case PIPE_FUNC_GEQUAL:
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (qzzzz[j] >= bzzzz[j]) 
+	 if (data->qzzzz[j] >= data->bzzzz[j]) 
 	    zmask |= (1 << j);
       }
       break;
@@ -193,80 +474,480 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad)
    }
 
    quad->inout.mask &= zmask;
+   if (quad->inout.mask == 0)
+      return FALSE;
 
+   /* Update our internal copy only if writemask set.  Even if
+    * depth.writemask is FALSE, may still need to write out buffer
+    * data due to stencil changes.
+    */
    if (softpipe->depth_stencil->depth.writemask) {
-      
-      /* This is also efficient with sse / spe instructions: 
-       */
       for (j = 0; j < QUAD_SIZE; j++) {
-	 if (quad->inout.mask & (1 << j)) {
-	    bzzzz[j] = qzzzz[j];
-	 }
+         if (quad->inout.mask & (1 << j)) {
+            data->bzzzz[j] = data->qzzzz[j];
+         }
       }
+   }
 
-      /* put updated Z values back into cached tile */
-      switch (format) {
-      case PIPE_FORMAT_Z16_UNORM:
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            tile->data.depth16[y][x] = (ushort) bzzzz[j];
+   return TRUE;
+}
+
+
+
+/**
+ * Do stencil (and depth) testing.  Stenciling depends on the outcome of
+ * depth testing.
+ */
+static void
+depth_stencil_test_quad(struct quad_stage *qs, 
+                        struct depth_data *data,
+                        struct quad_header *quad)
+{
+   struct softpipe_context *softpipe = qs->softpipe;
+   unsigned func, zFailOp, zPassOp, failOp;
+   ubyte ref, wrtMask, valMask;
+   uint face = quad->input.facing;
+
+   if (!softpipe->depth_stencil->stencil[1].enabled) {
+      /* single-sided stencil test, use front (face=0) state */
+      face = 0;
+   }
+
+   /* choose front or back face function, operator, etc */
+   /* XXX we could do these initializations once per primitive */
+   func    = softpipe->depth_stencil->stencil[face].func;
+   failOp  = softpipe->depth_stencil->stencil[face].fail_op;
+   zFailOp = softpipe->depth_stencil->stencil[face].zfail_op;
+   zPassOp = softpipe->depth_stencil->stencil[face].zpass_op;
+   ref     = softpipe->depth_stencil->stencil[face].ref_value;
+   wrtMask = softpipe->depth_stencil->stencil[face].writemask;
+   valMask = softpipe->depth_stencil->stencil[face].valuemask;
+
+
+   /* do the stencil test first */
+   {
+      unsigned passMask, failMask;
+      passMask = do_stencil_test(data, func, ref, valMask);
+      failMask = quad->inout.mask & ~passMask;
+      quad->inout.mask &= passMask;
+
+      if (failOp != PIPE_STENCIL_OP_KEEP) {
+         apply_stencil_op(data, failMask, failOp, ref, wrtMask);
+      }
+   }
+
+   if (quad->inout.mask) {
+      /* now the pixels that passed the stencil test are depth tested */
+      if (softpipe->depth_stencil->depth.enabled) {
+         const unsigned origMask = quad->inout.mask;
+
+         depth_test_quad(qs, data, quad);  /* quad->mask is updated */
+
+         /* update stencil buffer values according to z pass/fail result */
+         if (zFailOp != PIPE_STENCIL_OP_KEEP) {
+            const unsigned zFailMask = origMask & ~quad->inout.mask;
+            apply_stencil_op(data, zFailMask, zFailOp, ref, wrtMask);
          }
-         break;
-      case PIPE_FORMAT_X8Z24_UNORM:
-         /* fall-through */
-         /* (yes, this falls through to a different case than above) */
-      case PIPE_FORMAT_Z32_UNORM:
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            tile->data.depth32[y][x] = bzzzz[j];
+
+         if (zPassOp != PIPE_STENCIL_OP_KEEP) {
+            const unsigned zPassMask = origMask & quad->inout.mask;
+            apply_stencil_op(data, zPassMask, zPassOp, ref, wrtMask);
          }
-         break;
-      case PIPE_FORMAT_S8Z24_UNORM:
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            uint s8z24 = tile->data.depth32[y][x];
-            s8z24 = (s8z24 & 0xff000000) | bzzzz[j];
-            tile->data.depth32[y][x] = s8z24;
+      }
+      else {
+         /* no depth test, apply Zpass operator to stencil buffer values */
+         apply_stencil_op(data, quad->inout.mask, zPassOp, ref, wrtMask);
+      }
+   }
+}
+
+
+#define ALPHATEST( FUNC, COMP )                                         \
+   static int                                                          \
+   alpha_test_quads_##FUNC( struct quad_stage *qs,                      \
+                           struct quad_header *quads[],                 \
+                           unsigned nr )                                \
+   {                                                                    \
+      const float ref = qs->softpipe->depth_stencil->alpha.ref_value;   \
+      const uint cbuf = 0; /* only output[0].alpha is tested */         \
+      unsigned pass_nr = 0;                                             \
+      unsigned i;                                                       \
+                                                                        \
+      for (i = 0; i < nr; i++) {                                        \
+         const float *aaaa = quads[i]->output.color[cbuf][3];           \
+         unsigned passMask = 0;                                         \
+                                                                        \
+         if (aaaa[0] COMP ref) passMask |= (1 << 0);                    \
+         if (aaaa[1] COMP ref) passMask |= (1 << 1);                    \
+         if (aaaa[2] COMP ref) passMask |= (1 << 2);                    \
+         if (aaaa[3] COMP ref) passMask |= (1 << 3);                    \
+                                                                        \
+         quads[i]->inout.mask &= passMask;                              \
+                                                                        \
+         if (quads[i]->inout.mask)                                      \
+            quads[pass_nr++] = quads[i];                                \
+      }                                                                 \
+                                                                        \
+      return pass_nr;                                                   \
+   }
+
+
+ALPHATEST( LESS,     < )
+ALPHATEST( EQUAL,    == )
+ALPHATEST( LEQUAL,   <= )
+ALPHATEST( GREATER,  > )
+ALPHATEST( NOTEQUAL, != )
+ALPHATEST( GEQUAL,   >= )
+
+
+/* XXX: Incorporate into shader using KILP.
+ */
+static int
+alpha_test_quads(struct quad_stage *qs, 
+                 struct quad_header *quads[], 
+                 unsigned nr)
+{
+   switch (qs->softpipe->depth_stencil->alpha.func) {
+   case PIPE_FUNC_LESS:
+      return alpha_test_quads_LESS( qs, quads, nr );
+   case PIPE_FUNC_EQUAL:
+      return alpha_test_quads_EQUAL( qs, quads, nr );
+      break;
+   case PIPE_FUNC_LEQUAL:
+      return alpha_test_quads_LEQUAL( qs, quads, nr );
+   case PIPE_FUNC_GREATER:
+      return alpha_test_quads_GREATER( qs, quads, nr );
+   case PIPE_FUNC_NOTEQUAL:
+      return alpha_test_quads_NOTEQUAL( qs, quads, nr );
+   case PIPE_FUNC_GEQUAL:
+      return alpha_test_quads_GEQUAL( qs, quads, nr );
+   case PIPE_FUNC_ALWAYS:
+      return nr;
+   case PIPE_FUNC_NEVER:
+   default:
+      return 0;
+   }
+}
+
+static unsigned mask_count[16] = 
+{
+   0,                           /* 0x0 */
+   1,                           /* 0x1 */
+   1,                           /* 0x2 */
+   2,                           /* 0x3 */
+   1,                           /* 0x4 */
+   2,                           /* 0x5 */
+   2,                           /* 0x6 */
+   3,                           /* 0x7 */
+   1,                           /* 0x8 */
+   2,                           /* 0x9 */
+   2,                           /* 0xa */
+   3,                           /* 0xb */
+   2,                           /* 0xc */
+   3,                           /* 0xd */
+   3,                           /* 0xe */
+   4,                           /* 0xf */
+};
+
+
+
+static void
+depth_test_quads_fallback(struct quad_stage *qs, 
+                          struct quad_header *quads[],
+                          unsigned nr)
+{
+   unsigned i, pass = 0;
+   const struct sp_fragment_shader *fs = qs->softpipe->fs;
+   boolean interp_depth = !fs->info.writes_z;
+   struct depth_data data;
+
+
+   if (qs->softpipe->depth_stencil->alpha.enabled) {
+      nr = alpha_test_quads(qs, quads, nr);
+   }
+
+   if (qs->softpipe->framebuffer.zsbuf && 
+       (qs->softpipe->depth_stencil->depth.enabled ||
+        qs->softpipe->depth_stencil->stencil[0].enabled)) {
+
+      data.ps = qs->softpipe->framebuffer.zsbuf;
+      data.format = data.ps->format;
+      data.tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, 
+                                     quads[0]->input.x0, 
+                                     quads[0]->input.y0);
+
+      for (i = 0; i < nr; i++) {
+         get_depth_stencil_values(&data, quads[i]);
+
+         if (qs->softpipe->depth_stencil->depth.enabled) {
+            if (interp_depth)
+               interpolate_quad_depth(quads[i]);
+
+            convert_quad_depth(&data, quads[i]);
          }
-         break;
-      case PIPE_FORMAT_Z24S8_UNORM:
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            uint z24s8 = tile->data.depth32[y][x];
-            z24s8 = (z24s8 & 0xff) | (bzzzz[j] << 8);
-            tile->data.depth32[y][x] = z24s8;
+
+         if (qs->softpipe->depth_stencil->stencil[0].enabled) {
+            depth_stencil_test_quad(qs, &data, quads[i]);
+            write_depth_stencil_values(&data, quads[i]);
          }
-         break;
-      case PIPE_FORMAT_Z24X8_UNORM:
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int x = quad->input.x0 % TILE_SIZE + (j & 1);
-            int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-            tile->data.depth32[y][x] = bzzzz[j] << 8;
+         else {
+            if (!depth_test_quad(qs, &data, quads[i]))
+               continue;
+
+            if (qs->softpipe->depth_stencil->depth.writemask)
+               write_depth_stencil_values(&data, quads[i]);
          }
-         break;
-      default:
-         assert(0);
+
+
+         quads[pass++] = quads[i];
+      }
+
+      nr = pass;
+   }
+
+   if (qs->softpipe->active_query_count) {
+      for (i = 0; i < nr; i++) 
+         qs->softpipe->occlusion_count += mask_count[quads[i]->inout.mask];
+   }
+
+   if (nr)
+      qs->next->run(qs->next, quads, nr);
+}
+
+/* XXX: this function assumes setup function actually emits linear
+ * spans of quads.  It seems a lot more natural to do (early)
+ * depth-testing on spans rather than quads.
+ */
+static void
+depth_interp_z16_less_write(struct quad_stage *qs, 
+                            struct quad_header *quads[],
+                            unsigned nr)
+{
+   unsigned i, pass = 0;
+   const unsigned ix = quads[0]->input.x0;
+   const unsigned iy = quads[0]->input.y0;
+   const float fx = (float) ix;
+   const float fy = (float) iy;
+   const float dzdx = quads[0]->posCoef->dadx[2];
+   const float dzdy = quads[0]->posCoef->dady[2];
+   const float z0 = quads[0]->posCoef->a0[2] + dzdx * fx + dzdy * fy;
+   struct softpipe_cached_tile *tile;
+   ushort (*depth16)[TILE_SIZE];
+   ushort idepth[4], depth_step;
+   const float scale = 65535.0;
+
+   idepth[0] = (ushort)((z0) * scale);
+   idepth[1] = (ushort)((z0 + dzdx) * scale);
+   idepth[2] = (ushort)((z0 + dzdy) * scale);
+   idepth[3] = (ushort)((z0 + dzdx + dzdy) * scale);
+
+   depth_step = (ushort)(dzdx * 2 * scale);
+
+   tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, ix, iy);
+
+   depth16 = (ushort (*)[TILE_SIZE])
+      &tile->data.depth16[iy % TILE_SIZE][ix % TILE_SIZE];
+
+   for (i = 0; i < nr; i++) {
+      unsigned outmask = quads[i]->inout.mask;
+      unsigned mask = 0;
+      
+      if ((outmask & 1) && idepth[0] < depth16[0][0]) {
+         depth16[0][0] = idepth[0];
+         mask |= (1 << 0);
+      }
+
+      if ((outmask & 2) && idepth[1] < depth16[0][1]) {
+         depth16[0][1] = idepth[1];
+         mask |= (1 << 1);
+      }
+
+      if ((outmask & 4) && idepth[2] < depth16[1][0]) {
+         depth16[1][0] = idepth[2];
+         mask |= (1 << 2);
+      }
+
+      if ((outmask & 8) && idepth[3] < depth16[1][1]) {
+         depth16[1][1] = idepth[3];
+         mask |= (1 << 3);
       }
+
+      idepth[0] += depth_step;
+      idepth[1] += depth_step;
+      idepth[2] += depth_step;
+      idepth[3] += depth_step;
+
+      depth16 = (ushort (*)[TILE_SIZE]) &depth16[0][2];
+
+      quads[i]->inout.mask = mask;
+      if (quads[i]->inout.mask)
+         quads[pass++] = quads[i];
    }
+
+   if (pass)
+      qs->next->run(qs->next, quads, pass);
+
 }
 
 
 static void
-depth_test_quad(struct quad_stage *qs, struct quad_header *quad)
+depth_interp_z16_lequal_write(struct quad_stage *qs, 
+                            struct quad_header *quads[],
+                            unsigned nr)
 {
-   sp_depth_test_quad(qs, quad);
+   unsigned i, pass = 0;
+   const unsigned ix = quads[0]->input.x0;
+   const unsigned iy = quads[0]->input.y0;
+   const float fx = (float) ix;
+   const float fy = (float) iy;
+   const float dzdx = quads[0]->posCoef->dadx[2];
+   const float dzdy = quads[0]->posCoef->dady[2];
+   const float z0 = quads[0]->posCoef->a0[2] + dzdx * fx + dzdy * fy;
+   struct softpipe_cached_tile *tile;
+   ushort (*depth16)[TILE_SIZE];
+   ushort idepth[4], depth_step;
+   const float scale = 65535.0;
+
+   idepth[0] = (ushort)((z0) * scale);
+   idepth[1] = (ushort)((z0 + dzdx) * scale);
+   idepth[2] = (ushort)((z0 + dzdy) * scale);
+   idepth[3] = (ushort)((z0 + dzdx + dzdy) * scale);
+
+   depth_step = (ushort)(dzdx * 2 * scale);
+
+   tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, ix, iy);
+
+   depth16 = (ushort (*)[TILE_SIZE])
+      &tile->data.depth16[iy % TILE_SIZE][ix % TILE_SIZE];
+
+   for (i = 0; i < nr; i++) {
+      unsigned outmask = quads[i]->inout.mask;
+      unsigned mask = 0;
+      
+      if ((outmask & 1) && idepth[0] <= depth16[0][0]) {
+         depth16[0][0] = idepth[0];
+         mask |= (1 << 0);
+      }
+
+      if ((outmask & 2) && idepth[1] <= depth16[0][1]) {
+         depth16[0][1] = idepth[1];
+         mask |= (1 << 1);
+      }
+
+      if ((outmask & 4) && idepth[2] <= depth16[1][0]) {
+         depth16[1][0] = idepth[2];
+         mask |= (1 << 2);
+      }
+
+      if ((outmask & 8) && idepth[3] <= depth16[1][1]) {
+         depth16[1][1] = idepth[3];
+         mask |= (1 << 3);
+      }
+
+      idepth[0] += depth_step;
+      idepth[1] += depth_step;
+      idepth[2] += depth_step;
+      idepth[3] += depth_step;
+
+      depth16 = (ushort (*)[TILE_SIZE]) &depth16[0][2];
+
+      quads[i]->inout.mask = mask;
+      if (quads[i]->inout.mask)
+         quads[pass++] = quads[i];
+   }
+
+   if (pass)
+      qs->next->run(qs->next, quads, pass);
 
-   if (quad->inout.mask)
-      qs->next->run(qs->next, quad);
 }
 
 
+
+
+
+static void
+depth_noop(struct quad_stage *qs, 
+           struct quad_header *quads[],
+           unsigned nr)
+{
+   qs->next->run(qs->next, quads, nr);
+}
+
+
+
+static void
+choose_depth_test(struct quad_stage *qs, 
+                  struct quad_header *quads[],
+                  unsigned nr)
+{
+   boolean interp_depth = !qs->softpipe->fs->info.writes_z;
+
+   boolean alpha = qs->softpipe->depth_stencil->alpha.enabled;
+
+   boolean depth = (qs->softpipe->framebuffer.zsbuf && 
+                    qs->softpipe->depth_stencil->depth.enabled);
+
+   unsigned depthfunc = qs->softpipe->depth_stencil->depth.func;
+
+   boolean stencil = qs->softpipe->depth_stencil->stencil[0].enabled;
+
+   boolean depthwrite = qs->softpipe->depth_stencil->depth.writemask;
+
+   boolean occlusion = qs->softpipe->active_query_count;
+
+
+   if (!alpha &&
+       !depth &&
+       !stencil) {
+      qs->run = depth_noop;
+   }
+   else if (!alpha && 
+            interp_depth && 
+            depth && 
+            depthwrite && 
+            !occlusion &&
+            !stencil) 
+   {
+      switch (depthfunc) {
+      case PIPE_FUNC_LESS:
+         switch (qs->softpipe->framebuffer.zsbuf->format) {
+         case PIPE_FORMAT_Z16_UNORM:
+            qs->run = depth_interp_z16_less_write;
+            break;
+         default:
+            qs->run = depth_test_quads_fallback;
+            break;
+         }
+         break;
+      case PIPE_FUNC_LEQUAL:
+         switch (qs->softpipe->framebuffer.zsbuf->format) {
+         case PIPE_FORMAT_Z16_UNORM:
+            qs->run = depth_interp_z16_lequal_write;
+            break;
+         default:
+            qs->run = depth_test_quads_fallback;
+            break;
+         }
+         break;
+      default:
+         qs->run = depth_test_quads_fallback;
+      }
+   }
+   else {
+      qs->run = depth_test_quads_fallback;
+   }
+
+
+   qs->run( qs, quads, nr );
+}
+
+
+
+
+
 static void depth_test_begin(struct quad_stage *qs)
 {
+   qs->run = choose_depth_test;
    qs->next->begin(qs->next);
 }
 
@@ -283,7 +964,7 @@ struct quad_stage *sp_quad_depth_test_stage( struct softpipe_context *softpipe )
 
    stage->softpipe = softpipe;
    stage->begin = depth_test_begin;
-   stage->run = depth_test_quad;
+   stage->run = choose_depth_test;
    stage->destroy = depth_test_destroy;
 
    return stage;
diff --git a/src/gallium/drivers/softpipe/sp_quad_earlyz.c b/src/gallium/drivers/softpipe/sp_quad_earlyz.c
deleted file mode 100644
index 496fd39ed1..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_earlyz.c
+++ /dev/null
@@ -1,88 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/**
- * \brief  Quad early-z testing
- */
-
-#include "pipe/p_defines.h"
-#include "util/u_memory.h"
-#include "sp_quad.h"
-#include "sp_quad_pipe.h"
-
-
-/**
- * All this stage does is compute the quad's Z values (which is normally
- * done by the shading stage).
- * The next stage will do the actual depth test.
- */
-static void
-earlyz_quad(
-   struct quad_stage    *qs,
-   struct quad_header   *quad )
-{
-   const float fx = (float) quad->input.x0;
-   const float fy = (float) quad->input.y0;
-   const float dzdx = quad->posCoef->dadx[2];
-   const float dzdy = quad->posCoef->dady[2];
-   const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy;
-
-   quad->output.depth[0] = z0;
-   quad->output.depth[1] = z0 + dzdx;
-   quad->output.depth[2] = z0 + dzdy;
-   quad->output.depth[3] = z0 + dzdx + dzdy;
-
-   qs->next->run( qs->next, quad );
-}
-
-static void
-earlyz_begin(
-   struct quad_stage *qs )
-{
-   qs->next->begin( qs->next );
-}
-
-static void
-earlyz_destroy(
-   struct quad_stage *qs )
-{
-   FREE( qs );
-}
-
-struct quad_stage *
-sp_quad_earlyz_stage(
-   struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT( quad_stage );
-
-   stage->softpipe = softpipe;
-   stage->begin = earlyz_begin;
-   stage->run = earlyz_quad;
-   stage->destroy = earlyz_destroy;
-
-   return stage;
-}
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index 28f8d1a60e..1e7533d0f9 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -68,72 +68,69 @@ quad_shade_stage(struct quad_stage *qs)
 /**
  * Execute fragment shader for the four fragments in the quad.
  */
-static void
+static INLINE boolean
 shade_quad(struct quad_stage *qs, struct quad_header *quad)
 {
    struct quad_shade_stage *qss = quad_shade_stage( qs );
    struct softpipe_context *softpipe = qs->softpipe;
    struct tgsi_exec_machine *machine = qss->machine;
-   boolean z_written;
-   
-   /* Consts do not require 16 byte alignment. */
-   machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT];
-
-   machine->InterpCoefs = quad->coef;
 
    /* run shader */
-   quad->inout.mask &= softpipe->fs->run( softpipe->fs, machine, quad );
-
-   /* store outputs */
-   z_written = FALSE;
-   {
-      const ubyte *sem_name = softpipe->fs->info.output_semantic_name;
-      const ubyte *sem_index = softpipe->fs->info.output_semantic_index;
-      const uint n = qss->stage.softpipe->fs->info.num_outputs;
-      uint i;
-      for (i = 0; i < n; i++) {
-         switch (sem_name[i]) {
-         case TGSI_SEMANTIC_COLOR:
-            {
-               uint cbuf = sem_index[i];
-               memcpy(quad->output.color[cbuf],
-                      &machine->Outputs[i].xyzw[0].f[0],
-                      sizeof(quad->output.color[0]) );
-            }
-            break;
-         case TGSI_SEMANTIC_POSITION:
-            {
-               uint j;
-               for (j = 0; j < 4; j++) {
-                  quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j];
-               }
-               z_written = TRUE;
-            }
-            break;
-         }
+   return softpipe->fs->run( softpipe->fs, machine, quad );
+}
+
+
+
+static void
+coverage_quad(struct quad_stage *qs, struct quad_header *quad)
+{
+   struct softpipe_context *softpipe = qs->softpipe;
+   uint cbuf;
+
+   /* loop over colorbuffer outputs */
+   for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
+      float (*quadColor)[4] = quad->output.color[cbuf];
+      unsigned j;
+      for (j = 0; j < QUAD_SIZE; j++) {
+         assert(quad->input.coverage[j] >= 0.0);
+         assert(quad->input.coverage[j] <= 1.0);
+         quadColor[3][j] *= quad->input.coverage[j];
       }
    }
+}
+
 
-   if (!z_written) {
-      /* compute Z values now, as in the quad earlyz stage */
-      /* XXX we should really only do this if the earlyz stage is not used */
-      const float fx = (float) quad->input.x0;
-      const float fy = (float) quad->input.y0;
-      const float dzdx = quad->posCoef->dadx[2];
-      const float dzdy = quad->posCoef->dady[2];
-      const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy;
-
-      quad->output.depth[0] = z0;
-      quad->output.depth[1] = z0 + dzdx;
-      quad->output.depth[2] = z0 + dzdy;
-      quad->output.depth[3] = z0 + dzdx + dzdy;
-   }
 
-   /* shader may cull fragments */
-   if (quad->inout.mask) {
-      qs->next->run( qs->next, quad );
+static void
+shade_quads(struct quad_stage *qs, 
+                 struct quad_header *quads[],
+                 unsigned nr)
+{
+   struct quad_shade_stage *qss = quad_shade_stage( qs );
+   struct softpipe_context *softpipe = qs->softpipe;
+   struct tgsi_exec_machine *machine = qss->machine;
+
+   unsigned i, pass = 0;
+   
+   machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT];
+   machine->InterpCoefs = quads[0]->coef;
+
+   for (i = 0; i < nr; i++) {
+      if (!shade_quad(qs, quads[i]))
+         continue;
+
+      if (/*do_coverage*/ 0)
+         coverage_quad( qs, quads[i] );
+
+      quads[pass++] = quads[i];
    }
+   
+   if (pass)
+      qs->next->run(qs->next, quads, pass);
 }
+   
+
+
 
 
 /**
@@ -174,7 +171,7 @@ sp_quad_shade_stage( struct softpipe_context *softpipe )
 
    qss->stage.softpipe = softpipe;
    qss->stage.begin = shade_begin;
-   qss->stage.run = shade_quad;
+   qss->stage.run = shade_quads;
    qss->stage.destroy = shade_destroy;
 
    qss->machine = tgsi_exec_machine_create();
diff --git a/src/gallium/drivers/softpipe/sp_quad_output.c b/src/gallium/drivers/softpipe/sp_quad_output.c
deleted file mode 100644
index 92d5f9f3c1..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_output.c
+++ /dev/null
@@ -1,103 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "util/u_memory.h"
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_surface.h"
-#include "sp_quad_pipe.h"
-#include "sp_tile_cache.h"
-
-
-/**
- * Last step of quad processing: write quad colors to the framebuffer,
- * taking mask into account.
- */
-static void
-output_quad(struct quad_stage *qs, struct quad_header *quad)
-{
-   /* in-tile pos: */
-   const int itx = quad->input.x0 % TILE_SIZE;
-   const int ity = quad->input.y0 % TILE_SIZE;
-
-   struct softpipe_context *softpipe = qs->softpipe;
-   uint cbuf;
-
-   /* loop over colorbuffer outputs */
-   for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
-      struct softpipe_cached_tile *tile
-         = sp_get_cached_tile(softpipe,
-                              softpipe->cbuf_cache[cbuf],
-                              quad->input.x0, quad->input.y0);
-      float (*quadColor)[4] = quad->output.color[cbuf];
-      int i, j;
-
-      /* get/swizzle dest colors */
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (quad->inout.mask & (1 << j)) {
-            int x = itx + (j & 1);
-            int y = ity + (j >> 1);
-            for (i = 0; i < 4; i++) { /* loop over color chans */
-               tile->data.color[y][x][i] = quadColor[i][j];
-            }
-            if (0) {
-               debug_printf("sp write pixel %d,%d: %g, %g, %g\n",
-                            quad->input.x0 + x,
-                            quad->input.y0 + y,
-                            quadColor[0][j],
-                            quadColor[1][j],
-                            quadColor[2][j]);
-            }
-         }
-      }
-   }
-}
-
-
-static void output_begin(struct quad_stage *qs)
-{
-   assert(qs->next == NULL);
-}
-
-
-static void output_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
-
-struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
-
-   stage->softpipe = softpipe;
-   stage->begin = output_begin;
-   stage->run = output_quad;
-   stage->destroy = output_destroy;
-
-   return stage;
-}
diff --git a/src/gallium/drivers/softpipe/sp_quad_pipe.c b/src/gallium/drivers/softpipe/sp_quad_pipe.c
index b5f69b7426..1b5bab4eca 100644
--- a/src/gallium/drivers/softpipe/sp_quad_pipe.c
+++ b/src/gallium/drivers/softpipe/sp_quad_pipe.c
@@ -31,88 +31,33 @@
 #include "pipe/p_shader_tokens.h"
 
 static void
-sp_push_quad_first(
-   struct softpipe_context *sp,
-   struct quad_stage *quad,
-   uint i )
+sp_push_quad_first( struct softpipe_context *sp,
+                    struct quad_stage *quad )
 {
-   quad->next = sp->quad[i].first;
-   sp->quad[i].first = quad;
+   quad->next = sp->quad.first;
+   sp->quad.first = quad;
 }
 
-static void
-sp_build_depth_stencil(
-   struct softpipe_context *sp,
-   uint i )
-{
-   if (sp->depth_stencil->stencil[0].enabled ||
-       sp->depth_stencil->stencil[1].enabled) {
-      sp_push_quad_first( sp, sp->quad[i].stencil_test, i );
-   }
-   else if (sp->depth_stencil->depth.enabled &&
-            sp->framebuffer.zsbuf) {
-      sp_push_quad_first( sp, sp->quad[i].depth_test, i );
-   }
-}
 
 void
 sp_build_quad_pipeline(struct softpipe_context *sp)
 {
-   uint i;
-
    boolean early_depth_test =
-               sp->depth_stencil->depth.enabled &&
-               sp->framebuffer.zsbuf &&
-               !sp->depth_stencil->alpha.enabled &&
-               !sp->fs->info.uses_kill &&
-               !sp->fs->info.writes_z;
-
-   /* build up the pipeline in reverse order... */
-   for (i = 0; i < SP_NUM_QUAD_THREADS; i++) {
-      sp->quad[i].first = sp->quad[i].output;
-
-      if (sp->blend->colormask != 0xf) {
-         sp_push_quad_first( sp, sp->quad[i].colormask, i );
-      }
+      sp->depth_stencil->depth.enabled &&
+      sp->framebuffer.zsbuf &&
+      !sp->depth_stencil->alpha.enabled &&
+      !sp->fs->info.uses_kill &&
+      !sp->fs->info.writes_z;
 
-      if (sp->blend->blend_enable ||
-          sp->blend->logicop_enable) {
-         sp_push_quad_first( sp, sp->quad[i].blend, i );
-      }
+   sp->quad.first = sp->quad.blend;
 
-      if (sp->active_query_count) {
-         sp_push_quad_first( sp, sp->quad[i].occlusion, i );
-      }
-
-      if (sp->rasterizer->poly_smooth ||
-          sp->rasterizer->line_smooth ||
-          sp->rasterizer->point_smooth) {
-         sp_push_quad_first( sp, sp->quad[i].coverage, i );
-      }
-
-      if (!early_depth_test) {
-         sp_build_depth_stencil( sp, i );
-      }
-
-      if (sp->depth_stencil->alpha.enabled) {
-         sp_push_quad_first( sp, sp->quad[i].alpha_test, i );
-      }
-
-      /* XXX always enable shader? */
-      if (1) {
-         sp_push_quad_first( sp, sp->quad[i].shade, i );
-      }
-
-      if (early_depth_test) {
-         sp_build_depth_stencil( sp, i );
-         sp_push_quad_first( sp, sp->quad[i].earlyz, i );
-      }
-
-#if !USE_DRAW_STAGE_PSTIPPLE
-      if (sp->rasterizer->poly_stipple_enable) {
-         sp_push_quad_first( sp, sp->quad[i].polygon_stipple, i );
-      }
-#endif
+   if (early_depth_test) {
+      sp_push_quad_first( sp, sp->quad.shade );
+      sp_push_quad_first( sp, sp->quad.depth_test );
+   }
+   else {
+      sp_push_quad_first( sp, sp->quad.depth_test );
+      sp_push_quad_first( sp, sp->quad.shade );
    }
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_quad_pipe.h b/src/gallium/drivers/softpipe/sp_quad_pipe.h
index 0e40586ffc..c0aa134831 100644
--- a/src/gallium/drivers/softpipe/sp_quad_pipe.h
+++ b/src/gallium/drivers/softpipe/sp_quad_pipe.h
@@ -49,7 +49,7 @@ struct quad_stage {
    void (*begin)(struct quad_stage *qs);
 
    /** the stage action */
-   void (*run)(struct quad_stage *qs, struct quad_header *quad);
+   void (*run)(struct quad_stage *qs, struct quad_header *quad[], unsigned nr);
 
    void (*destroy)(struct quad_stage *qs);
 };
@@ -69,6 +69,4 @@ struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe );
 
 void sp_build_quad_pipeline(struct softpipe_context *sp);
 
-void sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad);
-
 #endif /* SP_QUAD_PIPE_H */
diff --git a/src/gallium/drivers/softpipe/sp_quad_stencil.c b/src/gallium/drivers/softpipe/sp_quad_stencil.c
deleted file mode 100644
index 5e9d447737..0000000000
--- a/src/gallium/drivers/softpipe/sp_quad_stencil.c
+++ /dev/null
@@ -1,352 +0,0 @@
-
-/**
- * \brief Quad stencil testing
- */
-
-
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_surface.h"
-#include "sp_tile_cache.h"
-#include "sp_quad_pipe.h"
-#include "pipe/p_defines.h"
-#include "util/u_memory.h"
-
-
-/** Only 8-bit stencil supported */
-#define STENCIL_MAX 0xff
-
-
-/**
- * Do the basic stencil test (compare stencil buffer values against the
- * reference value.
- *
- * \param stencilVals  the stencil values from the stencil buffer
- * \param func  the stencil func (PIPE_FUNC_x)
- * \param ref  the stencil reference value
- * \param valMask  the stencil value mask indicating which bits of the stencil
- *                 values and ref value are to be used.
- * \return mask indicating which pixels passed the stencil test
- */
-static unsigned
-do_stencil_test(const ubyte stencilVals[QUAD_SIZE], unsigned func,
-                unsigned ref, unsigned valMask)
-{
-   unsigned passMask = 0x0;
-   unsigned j;
-
-   ref &= valMask;
-
-   switch (func) {
-   case PIPE_FUNC_NEVER:
-      /* passMask = 0x0 */
-      break;
-   case PIPE_FUNC_LESS:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (ref < (stencilVals[j] & valMask)) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_EQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (ref == (stencilVals[j] & valMask)) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_LEQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (ref <= (stencilVals[j] & valMask)) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_GREATER:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (ref > (stencilVals[j] & valMask)) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_NOTEQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (ref != (stencilVals[j] & valMask)) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_GEQUAL:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (ref >= (stencilVals[j] & valMask)) {
-            passMask |= (1 << j);
-         }
-      }
-      break;
-   case PIPE_FUNC_ALWAYS:
-      passMask = MASK_ALL;
-      break;
-   default:
-      assert(0);
-   }
-
-   return passMask;
-}
-
-
-/**
- * Apply the stencil operator to stencil values.
- *
- * \param stencilVals  the stencil buffer values (read and written)
- * \param mask  indicates which pixels to update
- * \param op  the stencil operator (PIPE_STENCIL_OP_x)
- * \param ref  the stencil reference value
- * \param wrtMask  writemask controlling which bits are changed in the
- *                 stencil values
- */
-static void
-apply_stencil_op(ubyte stencilVals[QUAD_SIZE],
-                 unsigned mask, unsigned op, ubyte ref, ubyte wrtMask)
-{
-   unsigned j;
-   ubyte newstencil[QUAD_SIZE];
-
-   for (j = 0; j < QUAD_SIZE; j++) {
-      newstencil[j] = stencilVals[j];
-   }
-
-   switch (op) {
-   case PIPE_STENCIL_OP_KEEP:
-      /* no-op */
-      break;
-   case PIPE_STENCIL_OP_ZERO:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            newstencil[j] = 0;
-         }
-      }
-      break;
-   case PIPE_STENCIL_OP_REPLACE:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            newstencil[j] = ref;
-         }
-      }
-      break;
-   case PIPE_STENCIL_OP_INCR:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            if (stencilVals[j] < STENCIL_MAX) {
-               newstencil[j] = stencilVals[j] + 1;
-            }
-         }
-      }
-      break;
-   case PIPE_STENCIL_OP_DECR:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            if (stencilVals[j] > 0) {
-               newstencil[j] = stencilVals[j] - 1;
-            }
-         }
-      }
-      break;
-   case PIPE_STENCIL_OP_INCR_WRAP:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            newstencil[j] = stencilVals[j] + 1;
-         }
-      }
-      break;
-   case PIPE_STENCIL_OP_DECR_WRAP:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            newstencil[j] = stencilVals[j] - 1;
-         }
-      }
-      break;
-   case PIPE_STENCIL_OP_INVERT:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         if (mask & (1 << j)) {
-            newstencil[j] = ~stencilVals[j];
-         }
-      }
-      break;
-   default:
-      assert(0);
-   }
-
-   /*
-    * update the stencil values
-    */
-   if (wrtMask != STENCIL_MAX) {
-      /* apply bit-wise stencil buffer writemask */
-      for (j = 0; j < QUAD_SIZE; j++) {
-         stencilVals[j] = (wrtMask & newstencil[j]) | (~wrtMask & stencilVals[j]);
-      }
-   }
-   else {
-      for (j = 0; j < QUAD_SIZE; j++) {
-         stencilVals[j] = newstencil[j];
-      }
-   }
-}
-
-
-/**
- * Do stencil (and depth) testing.  Stenciling depends on the outcome of
- * depth testing.
- */
-static void
-stencil_test_quad(struct quad_stage *qs, struct quad_header *quad)
-{
-   struct softpipe_context *softpipe = qs->softpipe;
-   struct pipe_surface *ps = softpipe->framebuffer.zsbuf;
-   unsigned func, zFailOp, zPassOp, failOp;
-   ubyte ref, wrtMask, valMask;
-   ubyte stencilVals[QUAD_SIZE];
-   struct softpipe_cached_tile *tile
-      = sp_get_cached_tile(softpipe, softpipe->zsbuf_cache, quad->input.x0, quad->input.y0);
-   uint j;
-   uint face = quad->input.facing;
-
-   if (!softpipe->depth_stencil->stencil[1].enabled) {
-      /* single-sided stencil test, use front (face=0) state */
-      face = 0;
-   }
-
-   /* choose front or back face function, operator, etc */
-   /* XXX we could do these initializations once per primitive */
-   func    = softpipe->depth_stencil->stencil[face].func;
-   failOp  = softpipe->depth_stencil->stencil[face].fail_op;
-   zFailOp = softpipe->depth_stencil->stencil[face].zfail_op;
-   zPassOp = softpipe->depth_stencil->stencil[face].zpass_op;
-   ref     = softpipe->depth_stencil->stencil[face].ref_value;
-   wrtMask = softpipe->depth_stencil->stencil[face].writemask;
-   valMask = softpipe->depth_stencil->stencil[face].valuemask;
-
-   assert(ps); /* shouldn't get here if there's no stencil buffer */
-
-   /* get stencil values from cached tile */
-   switch (ps->format) {
-   case PIPE_FORMAT_S8Z24_UNORM:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = quad->input.x0 % TILE_SIZE + (j & 1);
-         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-         stencilVals[j] = tile->data.depth32[y][x] >> 24;
-      }
-      break;
-   case PIPE_FORMAT_Z24S8_UNORM:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = quad->input.x0 % TILE_SIZE + (j & 1);
-         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-         stencilVals[j] = tile->data.depth32[y][x] & 0xff;
-      }
-      break;
-   case PIPE_FORMAT_S8_UNORM:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = quad->input.x0 % TILE_SIZE + (j & 1);
-         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-         stencilVals[j] = tile->data.stencil8[y][x];
-      }
-      break;
-   default:
-      assert(0);
-   }
-
-   /* do the stencil test first */
-   {
-      unsigned passMask, failMask;
-      passMask = do_stencil_test(stencilVals, func, ref, valMask);
-      failMask = quad->inout.mask & ~passMask;
-      quad->inout.mask &= passMask;
-
-      if (failOp != PIPE_STENCIL_OP_KEEP) {
-         apply_stencil_op(stencilVals, failMask, failOp, ref, wrtMask);
-      }
-   }
-
-   if (quad->inout.mask) {
-      /* now the pixels that passed the stencil test are depth tested */
-      if (softpipe->depth_stencil->depth.enabled) {
-         const unsigned origMask = quad->inout.mask;
-
-         sp_depth_test_quad(qs, quad);  /* quad->mask is updated */
-
-         /* update stencil buffer values according to z pass/fail result */
-         if (zFailOp != PIPE_STENCIL_OP_KEEP) {
-            const unsigned failMask = origMask & ~quad->inout.mask;
-            apply_stencil_op(stencilVals, failMask, zFailOp, ref, wrtMask);
-         }
-
-         if (zPassOp != PIPE_STENCIL_OP_KEEP) {
-            const unsigned passMask = origMask & quad->inout.mask;
-            apply_stencil_op(stencilVals, passMask, zPassOp, ref, wrtMask);
-         }
-      }
-      else {
-         /* no depth test, apply Zpass operator to stencil buffer values */
-         apply_stencil_op(stencilVals, quad->inout.mask, zPassOp, ref, wrtMask);
-      }
-
-   }
-
-   /* put new stencil values into cached tile */
-   switch (ps->format) {
-   case PIPE_FORMAT_S8Z24_UNORM:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = quad->input.x0 % TILE_SIZE + (j & 1);
-         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-         uint s8z24 = tile->data.depth32[y][x];
-         s8z24 = (stencilVals[j] << 24) | (s8z24 & 0xffffff);
-         tile->data.depth32[y][x] = s8z24;
-      }
-      break;
-   case PIPE_FORMAT_Z24S8_UNORM:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = quad->input.x0 % TILE_SIZE + (j & 1);
-         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-         uint z24s8 = tile->data.depth32[y][x];
-         z24s8 = (z24s8 & 0xffffff00) | stencilVals[j];
-         tile->data.depth32[y][x] = z24s8;
-      }
-      break;
-   case PIPE_FORMAT_S8_UNORM:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = quad->input.x0 % TILE_SIZE + (j & 1);
-         int y = quad->input.y0 % TILE_SIZE + (j >> 1);
-         tile->data.stencil8[y][x] = stencilVals[j];
-      }
-      break;
-   default:
-      assert(0);
-   }
-
-   if (quad->inout.mask)
-      qs->next->run(qs->next, quad);
-}
-
-
-static void stencil_begin(struct quad_stage *qs)
-{
-   qs->next->begin(qs->next);
-}
-
-
-static void stencil_destroy(struct quad_stage *qs)
-{
-   FREE( qs );
-}
-
-
-struct quad_stage *sp_quad_stencil_test_stage( struct softpipe_context *softpipe )
-{
-   struct quad_stage *stage = CALLOC_STRUCT(quad_stage);
-
-   stage->softpipe = softpipe;
-   stage->begin = stencil_begin;
-   stage->run = stencil_test_quad;
-   stage->destroy = stencil_destroy;
-
-   return stage;
-}
diff --git a/src/gallium/drivers/softpipe/sp_quad_stipple.c b/src/gallium/drivers/softpipe/sp_quad_stipple.c
index 07162db7b6..a0527a596a 100644
--- a/src/gallium/drivers/softpipe/sp_quad_stipple.c
+++ b/src/gallium/drivers/softpipe/sp_quad_stipple.c
@@ -14,14 +14,20 @@
  * Apply polygon stipple to quads produced by triangle rasterization
  */
 static void
-stipple_quad(struct quad_stage *qs, struct quad_header *quad)
+stipple_quad(struct quad_stage *qs, struct quad_header *quads[], unsigned nr)
 {
    static const uint bit31 = 1 << 31;
    static const uint bit30 = 1 << 30;
+   unsigned pass = nr;
+
+   struct softpipe_context *softpipe = qs->softpipe;
+   unsigned q;
+
+   pass = 0;
+
+   for (q = 0; q < nr; q++)  {
+      struct quad_header *quad = quads[q];
 
-   if (quad->input.prim == QUAD_PRIM_TRI) {
-      struct softpipe_context *softpipe = qs->softpipe;
-      /* need to invert Y to index into OpenGL's stipple pattern */
       const int col0 = quad->input.x0 % 32;
       const int y0 = quad->input.y0;
       const int y1 = y0 + 1;
@@ -41,13 +47,11 @@ stipple_quad(struct quad_stage *qs, struct quad_header *quad)
       if ((stipple1 & (bit30 >> col0)) == 0)
          quad->inout.mask &= ~MASK_BOTTOM_RIGHT;
 
-      if (!quad->inout.mask) {
-         /* all fragments failed stipple test, end of quad pipeline */
-         return;
-      }
+      if (quad->inout.mask)
+         quads[pass++] = quad;
    }
 
-   qs->next->run(qs->next, quad);
+   qs->next->run(qs->next, quads, pass);
 }
 
 
diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index 769425bd12..81fb7aa20c 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -40,7 +40,7 @@
 static const char *
 softpipe_get_vendor(struct pipe_screen *screen)
 {
-   return "Tungsten Graphics, Inc.";
+   return "VMware, Inc.";
 }
 
 
@@ -65,8 +65,6 @@ softpipe_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_GLSL:
       return 1;
-   case PIPE_CAP_S3TC:
-      return 0;
    case PIPE_CAP_ANISOTROPIC_FILTER:
       return 0;
    case PIPE_CAP_POINT_SPRITE:
@@ -137,10 +135,14 @@ softpipe_is_format_supported( struct pipe_screen *screen,
           target == PIPE_TEXTURE_CUBE);
 
    switch(format) {
+   case PIPE_FORMAT_L16_UNORM:
+   case PIPE_FORMAT_YCBCR_REV:
+   case PIPE_FORMAT_YCBCR:
    case PIPE_FORMAT_DXT1_RGB:
    case PIPE_FORMAT_DXT1_RGBA:
    case PIPE_FORMAT_DXT3_RGBA:
    case PIPE_FORMAT_DXT5_RGBA:
+   case PIPE_FORMAT_Z32_FLOAT:
       return FALSE;
    default:
       return TRUE;
diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c
index de3ae3c369..00fb52a64f 100644
--- a/src/gallium/drivers/softpipe/sp_setup.c
+++ b/src/gallium/drivers/softpipe/sp_setup.c
@@ -33,7 +33,6 @@
  */
 
 #include "sp_context.h"
-#include "sp_prim_setup.h"
 #include "sp_quad.h"
 #include "sp_quad_pipe.h"
 #include "sp_setup.h"
@@ -61,87 +60,9 @@ struct edge {
    int lines;		/**< number of lines on this edge */
 };
 
-#if SP_NUM_QUAD_THREADS > 1
 
-/* Set to 1 if you want other threads to be instantly
- * notified of pending jobs.
- */
-#define INSTANT_NOTEMPTY_NOTIFY 0
-
-struct thread_info
-{
-   struct setup_context *setup;
-   uint id;
-   pipe_thread handle;
-};
-
-struct quad_job;
-
-typedef void (* quad_job_routine)( struct setup_context *setup, uint thread, struct quad_job *job );
-
-struct quad_job
-{
-   struct quad_header_input input;
-   struct quad_header_inout inout;
-   quad_job_routine routine;
-};
-
-#define NUM_QUAD_JOBS 64
-
-struct quad_job_que
-{
-   struct quad_job jobs[NUM_QUAD_JOBS];
-   uint first;
-   uint last;
-   pipe_mutex que_mutex;
-   pipe_condvar que_notfull_condvar;
-   pipe_condvar que_notempty_condvar;
-   uint jobs_added;
-   uint jobs_done;
-   pipe_condvar que_done_condvar;
-};
-
-static void
-add_quad_job( struct quad_job_que *que, struct quad_header *quad, quad_job_routine routine )
-{
-#if INSTANT_NOTEMPTY_NOTIFY
-   boolean empty;
-#endif
-
-   /* Wait for empty slot, see if the que is empty.
-    */
-   pipe_mutex_lock( que->que_mutex );
-   while ((que->last + 1) % NUM_QUAD_JOBS == que->first) {
-#if !INSTANT_NOTEMPTY_NOTIFY
-      pipe_condvar_broadcast( que->que_notempty_condvar );
-#endif
-      pipe_condvar_wait( que->que_notfull_condvar, que->que_mutex );
-   }
-#if INSTANT_NOTEMPTY_NOTIFY
-   empty = que->last == que->first;
-#endif
-   que->jobs_added++;
-   pipe_mutex_unlock( que->que_mutex );
+#define MAX_QUADS 16
 
-   /* Submit new job.
-    */
-   que->jobs[que->last].input = quad->input;
-   que->jobs[que->last].inout = quad->inout;
-   que->jobs[que->last].routine = routine;
-   que->last = (que->last + 1) % NUM_QUAD_JOBS;
-
-#if INSTANT_NOTEMPTY_NOTIFY
-   /* If the que was empty, notify consumers there's a job to be done.
-    */
-   if (empty) {
-      pipe_mutex_lock( que->que_mutex );
-      pipe_condvar_broadcast( que->que_notempty_condvar );
-      pipe_mutex_unlock( que->que_mutex );
-   }
-#endif
-}
-
-#endif
 
 /**
  * Triangle setup info (derived from draw_stage).
@@ -164,22 +85,19 @@ struct setup_context {
    struct edge emaj;
 
    float oneoverarea;
+   int facing;
+
+   struct quad_header quad[MAX_QUADS];
+   struct quad_header *quad_ptrs[MAX_QUADS];
+   unsigned count;
 
    struct tgsi_interp_coef coef[PIPE_MAX_SHADER_INPUTS];
    struct tgsi_interp_coef posCoef;  /* For Z, W */
-   struct quad_header quad;
-
-#if SP_NUM_QUAD_THREADS > 1
-   struct quad_job_que que;
-   struct thread_info threads[SP_NUM_QUAD_THREADS];
-#endif
 
    struct {
       int left[2];   /**< [0] = row0, [1] = row1 */
       int right[2];
       int y;
-      unsigned y_flags;
-      unsigned mask;     /**< mask of MASK_BOTTOM/TOP_LEFT/RIGHT bits */
    } span;
 
 #if DEBUG_FRAGS
@@ -188,69 +106,9 @@ struct setup_context {
 #endif
 
    unsigned winding;		/* which winding to cull */
+   unsigned nr_vertex_attrs;
 };
 
-#if SP_NUM_QUAD_THREADS > 1
-
-static PIPE_THREAD_ROUTINE( quad_thread, param )
-{
-   struct thread_info *info = (struct thread_info *) param;
-   struct quad_job_que *que = &info->setup->que;
-
-   for (;;) {
-      struct quad_job job;
-      boolean full;
-
-      /* Wait for an available job.
-       */
-      pipe_mutex_lock( que->que_mutex );
-      while (que->last == que->first)
-         pipe_condvar_wait( que->que_notempty_condvar, que->que_mutex );
-
-      /* See if the que is full.
-       */
-      full = (que->last + 1) % NUM_QUAD_JOBS == que->first;
-
-      /* Take a job and remove it from que.
-       */
-      job = que->jobs[que->first];
-      que->first = (que->first + 1) % NUM_QUAD_JOBS;
-
-      /* Notify the producer if the que is not full.
-       */
-      if (full)
-         pipe_condvar_signal( que->que_notfull_condvar );
-      pipe_mutex_unlock( que->que_mutex );
-
-      job.routine( info->setup, info->id, &job );
-
-      /* Notify the producer if that's the last finished job.
-       */
-      pipe_mutex_lock( que->que_mutex );
-      que->jobs_done++;
-      if (que->jobs_added == que->jobs_done)
-         pipe_condvar_signal( que->que_done_condvar );
-      pipe_mutex_unlock( que->que_mutex );
-   }
-
-   return NULL;
-}
-
-#define WAIT_FOR_COMPLETION(setup) \
-   do {\
-      pipe_mutex_lock( setup->que.que_mutex );\
-      if (!INSTANT_NOTEMPTY_NOTIFY)\
-         pipe_condvar_broadcast( setup->que.que_notempty_condvar );\
-      while (setup->que.jobs_added != setup->que.jobs_done)\
-         pipe_condvar_wait( setup->que.que_done_condvar, setup->que.que_mutex );\
-      pipe_mutex_unlock( setup->que.que_mutex );\
-   } while (0)
-
-#else
-
-#define WAIT_FOR_COMPLETION(setup) ((void) 0)
-
-#endif
 
 
 
@@ -313,98 +171,18 @@ quad_clip( struct setup_context *setup, struct quad_header *quad )
  * Emit a quad (pass to next stage) with clipping.
  */
 static INLINE void
-clip_emit_quad( struct setup_context *setup, struct quad_header *quad, uint thread )
+clip_emit_quad( struct setup_context *setup, struct quad_header *quad )
 {
    quad_clip( setup, quad );
+
    if (quad->inout.mask) {
       struct softpipe_context *sp = setup->softpipe;
 
-      sp->quad[thread].first->run( sp->quad[thread].first, quad );
+      sp->quad.first->run( sp->quad.first, &quad, 1 );
    }
 }
 
-#if SP_NUM_QUAD_THREADS > 1
 
-static void
-clip_emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job )
-{
-   struct quad_header quad;
-
-   quad.input = job->input;
-   quad.inout = job->inout;
-   quad.coef = setup->quad.coef;
-   quad.posCoef = setup->quad.posCoef;
-   quad.nr_attrs = setup->quad.nr_attrs;
-   clip_emit_quad( setup, &quad, thread );
-}
-
-#define CLIP_EMIT_QUAD(setup) add_quad_job( &setup->que, &setup->quad, clip_emit_quad_job )
-
-#else
-
-#define CLIP_EMIT_QUAD(setup) clip_emit_quad( setup, &setup->quad, 0 )
-
-#endif
-
-/**
- * Emit a quad (pass to next stage).  No clipping is done.
- */
-static INLINE void
-emit_quad( struct setup_context *setup, struct quad_header *quad, uint thread )
-{
-   struct softpipe_context *sp = setup->softpipe;
-#if DEBUG_FRAGS
-   uint mask = quad->inout.mask;
-#endif
-
-#if DEBUG_FRAGS
-   if (mask & 1) setup->numFragsEmitted++;
-   if (mask & 2) setup->numFragsEmitted++;
-   if (mask & 4) setup->numFragsEmitted++;
-   if (mask & 8) setup->numFragsEmitted++;
-#endif
-   sp->quad[thread].first->run( sp->quad[thread].first, quad );
-#if DEBUG_FRAGS
-   mask = quad->inout.mask;
-   if (mask & 1) setup->numFragsWritten++;
-   if (mask & 2) setup->numFragsWritten++;
-   if (mask & 4) setup->numFragsWritten++;
-   if (mask & 8) setup->numFragsWritten++;
-#endif
-}
-
-#if SP_NUM_QUAD_THREADS > 1
-
-static void
-emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job )
-{
-   struct quad_header quad;
-
-   quad.input = job->input;
-   quad.inout = job->inout;
-   quad.coef = setup->quad.coef;
-   quad.posCoef = setup->quad.posCoef;
-   quad.nr_attrs = setup->quad.nr_attrs;
-   emit_quad( setup, &quad, thread );
-}
-
-#define EMIT_QUAD(setup,x,y,mask) do {\
-      setup->quad.input.x0 = x;\
-      setup->quad.input.y0 = y;\
-      setup->quad.inout.mask = mask;\
-      add_quad_job( &setup->que, &setup->quad, emit_quad_job );\
-   } while (0)
-
-#else
-
-#define EMIT_QUAD(setup,x,y,mask) do {\
-      setup->quad.input.x0 = x;\
-      setup->quad.input.y0 = y;\
-      setup->quad.inout.mask = mask;\
-      emit_quad( setup, &setup->quad, 0 );\
-   } while (0)
-
-#endif
 
 /**
  * Given an X or Y coordinate, return the block/quad coordinate that it
@@ -412,7 +190,12 @@ emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job )
  */
 static INLINE int block( int x )
 {
-   return x & ~1;
+   return x & ~(2-1);
+}
+
+static INLINE int block_x( int x )
+{
+   return x & ~(16-1);
 }
 
 
@@ -421,72 +204,63 @@ static INLINE int block( int x )
  */
 static void flush_spans( struct setup_context *setup )
 {
+   const int step = 16;
    const int xleft0 = setup->span.left[0];
    const int xleft1 = setup->span.left[1];
    const int xright0 = setup->span.right[0];
    const int xright1 = setup->span.right[1];
-   int minleft, maxright;
+   struct quad_stage *pipe = setup->softpipe->quad.first;
+
+
+   int minleft = block_x(MIN2(xleft0, xleft1));
+   int maxright = MAX2(xright0, xright1);
    int x;
 
-   switch (setup->span.y_flags) {
-   case 0x3:
-      /* both odd and even lines written (both quad rows) */
-      minleft = block(MIN2(xleft0, xleft1));
-      maxright = block(MAX2(xright0, xright1));
-      for (x = minleft; x <= maxright; x += 2) {
-         /* determine which of the four pixels is inside the span bounds */
-         uint mask = 0x0;
-         if (x >= xleft0 && x < xright0)
-            mask |= MASK_TOP_LEFT;
-         if (x >= xleft1 && x < xright1)
-            mask |= MASK_BOTTOM_LEFT;
-         if (x+1 >= xleft0 && x+1 < xright0)
-            mask |= MASK_TOP_RIGHT;
-         if (x+1 >= xleft1 && x+1 < xright1)
-            mask |= MASK_BOTTOM_RIGHT;
-         if (mask)
-            EMIT_QUAD( setup, x, setup->span.y, mask );
-      }
-      break;
-
-   case 0x1:
-      /* only even line written (quad top row) */
-      minleft = block(xleft0);
-      maxright = block(xright0);
-      for (x = minleft; x <= maxright; x += 2) {
-         uint mask = 0x0;
-         if (x >= xleft0 && x < xright0)
-            mask |= MASK_TOP_LEFT;
-         if (x+1 >= xleft0 && x+1 < xright0)
-            mask |= MASK_TOP_RIGHT;
-         if (mask)
-            EMIT_QUAD( setup, x, setup->span.y, mask );
-      }
-      break;
-
-   case 0x2:
-      /* only odd line written (quad bottom row) */
-      minleft = block(xleft1);
-      maxright = block(xright1);
-      for (x = minleft; x <= maxright; x += 2) {
-         uint mask = 0x0;
-         if (x >= xleft1 && x < xright1)
-            mask |= MASK_BOTTOM_LEFT;
-         if (x+1 >= xleft1 && x+1 < xright1)
-            mask |= MASK_BOTTOM_RIGHT;
-         if (mask)
-            EMIT_QUAD( setup, x, setup->span.y, mask );
-      }
-      break;
+   for (x = minleft; x < maxright; x += step) {
+      unsigned skip_left0 = CLAMP(xleft0 - x, 0, step);
+      unsigned skip_left1 = CLAMP(xleft1 - x, 0, step);
+      unsigned skip_right0 = CLAMP(x + step - xright0, 0, step);
+      unsigned skip_right1 = CLAMP(x + step - xright1, 0, step);
+      unsigned lx = x;
+      unsigned q = 0;
 
-   default:
-      return;
+      unsigned skipmask_left0 = (1U << skip_left0) - 1U;
+      unsigned skipmask_left1 = (1U << skip_left1) - 1U;
+
+      /* These calculations fail when step == 32 and skip_right == 0.
+       */
+      unsigned skipmask_right0 = ~0U << (unsigned)(step - skip_right0);
+      unsigned skipmask_right1 = ~0U << (unsigned)(step - skip_right1);
+
+      unsigned mask0 = ~skipmask_left0 & ~skipmask_right0;
+      unsigned mask1 = ~skipmask_left1 & ~skipmask_right1;
+
+      if (mask0 | mask1) {
+         do {
+            unsigned quadmask = (mask0 & 3) | ((mask1 & 3) << 2);
+            if (quadmask) {
+               setup->quad[q].input.x0 = lx;
+               setup->quad[q].input.y0 = setup->span.y;
+               setup->quad[q].input.facing = setup->facing;
+               setup->quad[q].inout.mask = quadmask;
+               setup->quad_ptrs[q] = &setup->quad[q];
+               q++;
+            }
+            mask0 >>= 2;
+            mask1 >>= 2;
+            lx += 2;
+         } while (mask0 | mask1);
+
+         pipe->run( pipe, setup->quad_ptrs, q );
+      }
    }
 
+
    setup->span.y = 0;
-   setup->span.y_flags = 0;
    setup->span.right[0] = 0;
    setup->span.right[1] = 0;
+   setup->span.left[0] = 1000000;     /* greater than right[0] */
+   setup->span.left[1] = 1000000;     /* greater than right[1] */
 }
 
 
@@ -495,8 +269,8 @@ static void print_vertex(const struct setup_context *setup,
                          const float (*v)[4])
 {
    int i;
-   debug_printf("   Vertex: (%p)\n", v);
-   for (i = 0; i < setup->quad.nr_attrs; i++) {
+   debug_printf("   Vertex: (%p)\n", (void *) v);
+   for (i = 0; i < setup->nr_vertex_attrs; i++) {
       debug_printf("     %d: %f %f %f %f\n",  i,
               v[i][0], v[i][1], v[i][2], v[i][3]);
       if (util_is_inf_or_nan(v[i][0])) {
@@ -601,7 +375,9 @@ static boolean setup_sort_vertices( struct setup_context *setup,
     *  - the GLSL gl_FrontFacing fragment attribute (bool)
     *  - two-sided stencil test
     */
-   setup->quad.input.facing = (det > 0.0) ^ (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW);
+   setup->facing = 
+      ((det > 0.0) ^ 
+       (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW));
 
    return TRUE;
 }
@@ -788,7 +564,7 @@ static void setup_tri_coefficients( struct setup_context *setup )
       }
 
       if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) {
-         setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing;
+         setup->coef[fragSlot].a0[0] = 1.0f - setup->facing;
          setup->coef[fragSlot].dadx[0] = 0.0;
          setup->coef[fragSlot].dady[0] = 0.0;
       }
@@ -844,11 +620,10 @@ static void subtriangle( struct setup_context *setup,
 
    /* clip top/bottom */
    start_y = sy;
-   finish_y = sy + lines;
-
    if (start_y < miny)
       start_y = miny;
 
+   finish_y = sy + lines;
    if (finish_y > maxy)
       finish_y = maxy;
 
@@ -885,7 +660,6 @@ static void subtriangle( struct setup_context *setup,
 
          setup->span.left[_y&1] = left;
          setup->span.right[_y&1] = right;
-         setup->span.y_flags |= 1<<(_y&1);
       }
    }
 
@@ -923,7 +697,7 @@ calc_det( const float (*v0)[4],
 /**
  * Do setup for triangle rasterization, then render the triangle.
  */
-void setup_tri( struct setup_context *setup,
+void sp_setup_tri( struct setup_context *setup,
                 const float (*v0)[4],
                 const float (*v1)[4],
                 const float (*v2)[4] )
@@ -958,10 +732,9 @@ void setup_tri( struct setup_context *setup,
    setup_tri_coefficients( setup );
    setup_tri_edges( setup );
 
-   setup->quad.input.prim = QUAD_PRIM_TRI;
+   assert(setup->softpipe->reduced_prim == PIPE_PRIM_TRIANGLES);
 
    setup->span.y = 0;
-   setup->span.y_flags = 0;
    setup->span.right[0] = 0;
    setup->span.right[1] = 0;
    /*   setup->span.z_mode = tri_z_mode( setup->ctx ); */
@@ -983,8 +756,6 @@ void setup_tri( struct setup_context *setup,
 
    flush_spans( setup );
 
-   WAIT_FOR_COMPLETION(setup);
-
 #if DEBUG_FRAGS
    printf("Tri: %u frags emitted, %u written\n",
           setup->numFragsEmitted,
@@ -1101,7 +872,7 @@ setup_line_coefficients(struct setup_context *setup,
       }
 
       if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) {
-         setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing;
+         setup->coef[fragSlot].a0[0] = 1.0f - setup->facing;
          setup->coef[fragSlot].dadx[0] = 0.0;
          setup->coef[fragSlot].dady[0] = 0.0;
       }
@@ -1122,20 +893,20 @@ plot(struct setup_context *setup, int x, int y)
    const int quadY = y - iy;
    const int mask = (1 << ix) << (2 * iy);
 
-   if (quadX != setup->quad.input.x0 ||
-       quadY != setup->quad.input.y0)
+   if (quadX != setup->quad[0].input.x0 ||
+       quadY != setup->quad[0].input.y0)
    {
       /* flush prev quad, start new quad */
 
-      if (setup->quad.input.x0 != -1)
-         CLIP_EMIT_QUAD(setup);
+      if (setup->quad[0].input.x0 != -1)
+         clip_emit_quad( setup, &setup->quad[0] );
 
-      setup->quad.input.x0 = quadX;
-      setup->quad.input.y0 = quadY;
-      setup->quad.inout.mask = 0x0;
+      setup->quad[0].input.x0 = quadX;
+      setup->quad[0].input.y0 = quadY;
+      setup->quad[0].inout.mask = 0x0;
    }
 
-   setup->quad.inout.mask |= mask;
+   setup->quad[0].inout.mask |= mask;
 }
 
 
@@ -1145,7 +916,7 @@ plot(struct setup_context *setup, int x, int y)
  * to handle stippling and wide lines.
  */
 void
-setup_line(struct setup_context *setup,
+sp_setup_line(struct setup_context *setup,
            const float (*v0)[4],
            const float (*v1)[4])
 {
@@ -1195,17 +966,18 @@ setup_line(struct setup_context *setup,
 
    assert(dx >= 0);
    assert(dy >= 0);
+   assert(setup->softpipe->reduced_prim == PIPE_PRIM_LINES);
+
+   setup->quad[0].input.x0 = setup->quad[0].input.y0 = -1;
+   setup->quad[0].inout.mask = 0x0;
 
-   setup->quad.input.x0 = setup->quad.input.y0 = -1;
-   setup->quad.inout.mask = 0x0;
-   setup->quad.input.prim = QUAD_PRIM_LINE;
    /* XXX temporary: set coverage to 1.0 so the line appears
     * if AA mode happens to be enabled.
     */
-   setup->quad.input.coverage[0] =
-   setup->quad.input.coverage[1] =
-   setup->quad.input.coverage[2] =
-   setup->quad.input.coverage[3] = 1.0;
+   setup->quad[0].input.coverage[0] =
+   setup->quad[0].input.coverage[1] =
+   setup->quad[0].input.coverage[2] =
+   setup->quad[0].input.coverage[3] = 1.0;
 
    if (dx > dy) {
       /*** X-major line ***/
@@ -1249,11 +1021,9 @@ setup_line(struct setup_context *setup,
    }
 
    /* draw final quad */
-   if (setup->quad.inout.mask) {
-      CLIP_EMIT_QUAD(setup);
+   if (setup->quad[0].inout.mask) {
+      clip_emit_quad( setup, &setup->quad[0] );
    }
-
-   WAIT_FOR_COMPLETION(setup);
 }
 
 
@@ -1276,7 +1046,7 @@ point_persp_coeff(const struct setup_context *setup,
  * XXX could optimize a lot for 1-pixel points.
  */
 void
-setup_point( struct setup_context *setup,
+sp_setup_point( struct setup_context *setup,
              const float (*v0)[4] )
 {
    struct softpipe_context *softpipe = setup->softpipe;
@@ -1300,6 +1070,8 @@ setup_point( struct setup_context *setup,
    if (softpipe->no_rast)
       return;
 
+   assert(setup->softpipe->reduced_prim == PIPE_PRIM_POINTS);
+
    /* For points, all interpolants are constant-valued.
     * However, for point sprites, we'll need to setup texcoords appropriately.
     * XXX: which coefficients are the texcoords???
@@ -1346,22 +1118,21 @@ setup_point( struct setup_context *setup,
       }
 
       if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) {
-         setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing;
+         setup->coef[fragSlot].a0[0] = 1.0f - setup->facing;
          setup->coef[fragSlot].dadx[0] = 0.0;
          setup->coef[fragSlot].dady[0] = 0.0;
       }
    }
 
-   setup->quad.input.prim = QUAD_PRIM_POINT;
 
    if (halfSize <= 0.5 && !round) {
       /* special case for 1-pixel points */
       const int ix = ((int) x) & 1;
       const int iy = ((int) y) & 1;
-      setup->quad.input.x0 = (int) x - ix;
-      setup->quad.input.y0 = (int) y - iy;
-      setup->quad.inout.mask = (1 << ix) << (2 * iy);
-      CLIP_EMIT_QUAD(setup);
+      setup->quad[0].input.x0 = (int) x - ix;
+      setup->quad[0].input.y0 = (int) y - iy;
+      setup->quad[0].inout.mask = (1 << ix) << (2 * iy);
+      clip_emit_quad( setup, &setup->quad[0] );
    }
    else {
       if (round) {
@@ -1381,15 +1152,15 @@ setup_point( struct setup_context *setup,
             for (ix = ixmin; ix <= ixmax; ix += 2) {
                float dx, dy, dist2, cover;
 
-               setup->quad.inout.mask = 0x0;
+               setup->quad[0].inout.mask = 0x0;
 
                dx = (ix + 0.5f) - x;
                dy = (iy + 0.5f) - y;
                dist2 = dx * dx + dy * dy;
                if (dist2 <= rmax2) {
                   cover = 1.0F - (dist2 - rmin2) * cscale;
-                  setup->quad.input.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f);
-                  setup->quad.inout.mask |= MASK_TOP_LEFT;
+                  setup->quad[0].input.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f);
+                  setup->quad[0].inout.mask |= MASK_TOP_LEFT;
                }
 
                dx = (ix + 1.5f) - x;
@@ -1397,8 +1168,8 @@ setup_point( struct setup_context *setup,
                dist2 = dx * dx + dy * dy;
                if (dist2 <= rmax2) {
                   cover = 1.0F - (dist2 - rmin2) * cscale;
-                  setup->quad.input.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f);
-                  setup->quad.inout.mask |= MASK_TOP_RIGHT;
+                  setup->quad[0].input.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f);
+                  setup->quad[0].inout.mask |= MASK_TOP_RIGHT;
                }
 
                dx = (ix + 0.5f) - x;
@@ -1406,8 +1177,8 @@ setup_point( struct setup_context *setup,
                dist2 = dx * dx + dy * dy;
                if (dist2 <= rmax2) {
                   cover = 1.0F - (dist2 - rmin2) * cscale;
-                  setup->quad.input.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f);
-                  setup->quad.inout.mask |= MASK_BOTTOM_LEFT;
+                  setup->quad[0].input.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f);
+                  setup->quad[0].inout.mask |= MASK_BOTTOM_LEFT;
                }
 
                dx = (ix + 1.5f) - x;
@@ -1415,14 +1186,14 @@ setup_point( struct setup_context *setup,
                dist2 = dx * dx + dy * dy;
                if (dist2 <= rmax2) {
                   cover = 1.0F - (dist2 - rmin2) * cscale;
-                  setup->quad.input.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f);
-                  setup->quad.inout.mask |= MASK_BOTTOM_RIGHT;
+                  setup->quad[0].input.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f);
+                  setup->quad[0].inout.mask |= MASK_BOTTOM_RIGHT;
                }
 
-               if (setup->quad.inout.mask) {
-                  setup->quad.input.x0 = ix;
-                  setup->quad.input.y0 = iy;
-                  CLIP_EMIT_QUAD(setup);
+               if (setup->quad[0].inout.mask) {
+                  setup->quad[0].input.x0 = ix;
+                  setup->quad[0].input.y0 = iy;
+                  clip_emit_quad( setup, &setup->quad[0] );
                }
             }
          }
@@ -1466,33 +1237,28 @@ setup_point( struct setup_context *setup,
                   mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT);
                }
 
-               setup->quad.inout.mask = mask;
-               setup->quad.input.x0 = ix;
-               setup->quad.input.y0 = iy;
-               CLIP_EMIT_QUAD(setup);
+               setup->quad[0].inout.mask = mask;
+               setup->quad[0].input.x0 = ix;
+               setup->quad[0].input.y0 = iy;
+               clip_emit_quad( setup, &setup->quad[0] );
             }
          }
       }
    }
-
-   WAIT_FOR_COMPLETION(setup);
 }
 
-void setup_prepare( struct setup_context *setup )
+void sp_setup_prepare( struct setup_context *setup )
 {
    struct softpipe_context *sp = setup->softpipe;
-   unsigned i;
 
    if (sp->dirty) {
       softpipe_update_derived(sp);
    }
 
    /* Note: nr_attrs is only used for debugging (vertex printing) */
-   setup->quad.nr_attrs = draw_num_vs_outputs(sp->draw);
+   setup->nr_vertex_attrs = draw_num_vs_outputs(sp->draw);
 
-   for (i = 0; i < SP_NUM_QUAD_THREADS; i++) {
-      sp->quad[i].first->begin( sp->quad[i].first );
-   }
+   sp->quad.first->begin( sp->quad.first );
 
    if (sp->reduced_api_prim == PIPE_PRIM_TRIANGLES &&
        sp->rasterizer->fill_cw == PIPE_POLYGON_MODE_FILL &&
@@ -1508,7 +1274,7 @@ void setup_prepare( struct setup_context *setup )
 
 
 
-void setup_destroy_context( struct setup_context *setup )
+void sp_setup_destroy_context( struct setup_context *setup )
 {
    FREE( setup );
 }
@@ -1517,33 +1283,20 @@ void setup_destroy_context( struct setup_context *setup )
 /**
  * Create a new primitive setup/render stage.
  */
-struct setup_context *setup_create_context( struct softpipe_context *softpipe )
+struct setup_context *sp_setup_create_context( struct softpipe_context *softpipe )
 {
    struct setup_context *setup = CALLOC_STRUCT(setup_context);
-#if SP_NUM_QUAD_THREADS > 1
-   uint i;
-#endif
+   unsigned i;
 
    setup->softpipe = softpipe;
 
-   setup->quad.coef = setup->coef;
-   setup->quad.posCoef = &setup->posCoef;
-
-#if SP_NUM_QUAD_THREADS > 1
-   setup->que.first = 0;
-   setup->que.last = 0;
-   pipe_mutex_init( setup->que.que_mutex );
-   pipe_condvar_init( setup->que.que_notfull_condvar );
-   pipe_condvar_init( setup->que.que_notempty_condvar );
-   setup->que.jobs_added = 0;
-   setup->que.jobs_done = 0;
-   pipe_condvar_init( setup->que.que_done_condvar );
-   for (i = 0; i < SP_NUM_QUAD_THREADS; i++) {
-      setup->threads[i].setup = setup;
-      setup->threads[i].id = i;
-      setup->threads[i].handle = pipe_thread_create( quad_thread, &setup->threads[i] );
+   for (i = 0; i < MAX_QUADS; i++) {
+      setup->quad[i].coef = setup->coef;
+      setup->quad[i].posCoef = &setup->posCoef;
    }
-#endif
+
+   setup->span.left[0] = 1000000;     /* greater than right[0] */
+   setup->span.left[1] = 1000000;     /* greater than right[1] */
 
    return setup;
 }
diff --git a/src/gallium/drivers/softpipe/sp_setup.h b/src/gallium/drivers/softpipe/sp_setup.h
index d54f334428..9c8844d2e8 100644
--- a/src/gallium/drivers/softpipe/sp_setup.h
+++ b/src/gallium/drivers/softpipe/sp_setup.h
@@ -31,23 +31,23 @@ struct setup_context;
 struct softpipe_context;
 
 void 
-setup_tri( struct setup_context *setup,
+sp_setup_tri( struct setup_context *setup,
 	   const float (*v0)[4],
 	   const float (*v1)[4],
 	   const float (*v2)[4] );
 
 void
-setup_line(struct setup_context *setup,
+sp_setup_line(struct setup_context *setup,
            const float (*v0)[4],
            const float (*v1)[4]);
 
 void
-setup_point( struct setup_context *setup,
+sp_setup_point( struct setup_context *setup,
              const float (*v0)[4] );
 
 
-struct setup_context *setup_create_context( struct softpipe_context *softpipe );
-void setup_prepare( struct setup_context *setup );
-void setup_destroy_context( struct setup_context *setup );
+struct setup_context *sp_setup_create_context( struct softpipe_context *softpipe );
+void sp_setup_prepare( struct setup_context *setup );
+void sp_setup_destroy_context( struct setup_context *setup );
 
 #endif
diff --git a/src/gallium/drivers/softpipe/sp_state.h b/src/gallium/drivers/softpipe/sp_state.h
index 9776e978e3..77ee3c1136 100644
--- a/src/gallium/drivers/softpipe/sp_state.h
+++ b/src/gallium/drivers/softpipe/sp_state.h
@@ -87,6 +87,7 @@ struct sp_fragment_shader {
 struct sp_vertex_shader {
    struct pipe_shader_state shader;
    struct draw_vertex_shader *draw_data;
+   int max_sampler;             /* -1 if no samplers */
 };
 
 
diff --git a/src/gallium/drivers/softpipe/sp_state_blend.c b/src/gallium/drivers/softpipe/sp_state_blend.c
index 384fe559af..efed082f82 100644
--- a/src/gallium/drivers/softpipe/sp_state_blend.c
+++ b/src/gallium/drivers/softpipe/sp_state_blend.c
@@ -45,7 +45,7 @@ void softpipe_bind_blend_state( struct pipe_context *pipe,
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
 
-   softpipe->blend = (const struct pipe_blend_state *)blend;
+   softpipe->blend = (struct pipe_blend_state *)blend;
 
    softpipe->dirty |= SP_NEW_BLEND;
 }
@@ -86,7 +86,7 @@ softpipe_bind_depth_stencil_state(struct pipe_context *pipe,
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
 
-   softpipe->depth_stencil = (const struct pipe_depth_stencil_alpha_state *)depth_stencil;
+   softpipe->depth_stencil = (struct pipe_depth_stencil_alpha_state *)depth_stencil;
 
    softpipe->dirty |= SP_NEW_DEPTH_STENCIL_ALPHA;
 }
diff --git a/src/gallium/drivers/softpipe/sp_state_derived.c b/src/gallium/drivers/softpipe/sp_state_derived.c
index 75551000c9..1faeca1c2a 100644
--- a/src/gallium/drivers/softpipe/sp_state_derived.c
+++ b/src/gallium/drivers/softpipe/sp_state_derived.c
@@ -32,7 +32,10 @@
 #include "draw/draw_vertex.h"
 #include "draw/draw_private.h"
 #include "sp_context.h"
+#include "sp_screen.h"
 #include "sp_state.h"
+#include "sp_texture.h"
+#include "sp_tex_tile_cache.h"
 
 
 /**
@@ -65,24 +68,19 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe)
       const struct sp_fragment_shader *spfs = softpipe->fs;
       const enum interp_mode colorInterp
          = softpipe->rasterizer->flatshade ? INTERP_CONSTANT : INTERP_LINEAR;
+      struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf;
+      const uint num = draw_num_vs_outputs(softpipe->draw);
       uint i;
 
-      if (softpipe->vbuf) {
-         /* if using the post-transform vertex buffer, tell draw_vbuf to
-          * simply emit the whole post-xform vertex as-is:
-          */
-         struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf;
-         const uint num = draw_num_vs_outputs(softpipe->draw);
-         uint i;
-
-         /* No longer any need to try and emit draw vertex_header info.
-          */
-         vinfo_vbuf->num_attribs = 0;
-         for (i = 0; i < num; i++) {
-            draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i);
-         }
-         draw_compute_vertex_size(vinfo_vbuf);
+      /* Tell draw_vbuf to simply emit the whole post-xform vertex
+       * as-is.  No longer any need to try and emit draw vertex_header
+       * info.
+       */
+      vinfo_vbuf->num_attribs = 0;
+      for (i = 0; i < num; i++) {
+	 draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i);
       }
+      draw_compute_vertex_size(vinfo_vbuf);
 
       /*
        * Loop over fragment shader inputs, searching for the matching output
@@ -91,6 +89,23 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe)
       vinfo->num_attribs = 0;
       for (i = 0; i < spfs->info.num_inputs; i++) {
          int src;
+         enum interp_mode interp;
+
+         switch (spfs->info.input_interpolate[i]) {
+         case TGSI_INTERPOLATE_CONSTANT:
+            interp = INTERP_CONSTANT;
+            break;
+         case TGSI_INTERPOLATE_LINEAR:
+            interp = INTERP_LINEAR;
+            break;
+         case TGSI_INTERPOLATE_PERSPECTIVE:
+            interp = INTERP_PERSPECTIVE;
+            break;
+         default:
+            assert(0);
+            interp = INTERP_LINEAR;
+         }
+
          switch (spfs->info.input_semantic_name[i]) {
          case TGSI_SEMANTIC_POSITION:
             src = draw_find_vs_output(softpipe->draw,
@@ -106,7 +121,7 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe)
 
          case TGSI_SEMANTIC_FOG:
             src = draw_find_vs_output(softpipe->draw, TGSI_SEMANTIC_FOG, 0);
-            draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
+            draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src);
             break;
 
          case TGSI_SEMANTIC_GENERIC:
@@ -114,7 +129,7 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe)
             /* this includes texcoords and varying vars */
             src = draw_find_vs_output(softpipe->draw, TGSI_SEMANTIC_GENERIC,
                                       spfs->info.input_semantic_index[i]);
-            draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
+            draw_emit_vertex_attr(vinfo, EMIT_4F, interp, src);
             break;
 
          default:
@@ -164,11 +179,19 @@ softpipe_get_vbuf_vertex_info(struct softpipe_context *softpipe)
 static void
 compute_cliprect(struct softpipe_context *sp)
 {
+   /* SP_NEW_FRAMEBUFFER
+    */
    uint surfWidth = sp->framebuffer.width;
    uint surfHeight = sp->framebuffer.height;
 
+   /* SP_NEW_RASTERIZER
+    */
    if (sp->rasterizer->scissor) {
-      /* clip to scissor rect */
+
+      /* SP_NEW_SCISSOR
+       *
+       * clip to scissor rect:
+       */
       sp->cliprect.minx = MAX2(sp->scissor.minx, 0);
       sp->cliprect.miny = MAX2(sp->scissor.miny, 0);
       sp->cliprect.maxx = MIN2(sp->scissor.maxx, surfWidth);
@@ -184,27 +207,63 @@ compute_cliprect(struct softpipe_context *sp)
 }
 
 
+static void
+update_tgsi_samplers( struct softpipe_context *softpipe )
+{
+   unsigned i;
+
+   softpipe_reset_sampler_varients( softpipe );
+
+   for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
+      struct softpipe_tex_tile_cache *tc = softpipe->tex_cache[i];
+      if (tc->texture) {
+         struct softpipe_texture *spt = softpipe_texture(tc->texture);
+         if (spt->timestamp != tc->timestamp) {
+	    sp_tex_tile_cache_validate_texture( tc );
+            /*
+            _debug_printf("INV %d %d\n", tc->timestamp, spt->timestamp);
+            */
+            tc->timestamp = spt->timestamp;
+         }
+      }
+   }
+}
+
+
 /* Hopefully this will remain quite simple, otherwise need to pull in
  * something like the state tracker mechanism.
  */
 void softpipe_update_derived( struct softpipe_context *softpipe )
 {
+   struct softpipe_screen *sp_screen = softpipe_screen(softpipe->pipe.screen);
+
+   /* Check for updated textures.
+    */
+   if (softpipe->tex_timestamp != sp_screen->timestamp) {
+      softpipe->tex_timestamp = sp_screen->timestamp;
+      softpipe->dirty |= SP_NEW_TEXTURE;
+   }
+      
+   if (softpipe->dirty & (SP_NEW_SAMPLER |
+                          SP_NEW_TEXTURE |
+                          SP_NEW_FS | 
+                          SP_NEW_VS))
+      update_tgsi_samplers( softpipe );
+
    if (softpipe->dirty & (SP_NEW_RASTERIZER |
                           SP_NEW_FS |
                           SP_NEW_VS))
       invalidate_vertex_layout( softpipe );
 
    if (softpipe->dirty & (SP_NEW_SCISSOR |
-                          SP_NEW_DEPTH_STENCIL_ALPHA |
+                          SP_NEW_RASTERIZER |
                           SP_NEW_FRAMEBUFFER))
       compute_cliprect(softpipe);
 
    if (softpipe->dirty & (SP_NEW_BLEND |
                           SP_NEW_DEPTH_STENCIL_ALPHA |
                           SP_NEW_FRAMEBUFFER |
-                          SP_NEW_RASTERIZER |
-                          SP_NEW_FS | 
-			  SP_NEW_QUERY))
+                          SP_NEW_FS))
       sp_build_quad_pipeline(softpipe);
 
    softpipe->dirty = 0;
diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c
index 4330c20393..256faa94b8 100644
--- a/src/gallium/drivers/softpipe/sp_state_fs.c
+++ b/src/gallium/drivers/softpipe/sp_state_fs.c
@@ -31,9 +31,8 @@
 
 #include "pipe/p_defines.h"
 #include "util/u_memory.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "pipe/p_shader_tokens.h"
 #include "draw/draw_context.h"
+#include "draw/draw_vs.h"
 #include "tgsi/tgsi_dump.h"
 #include "tgsi/tgsi_scan.h"
 #include "tgsi/tgsi_parse.h"
@@ -51,12 +50,9 @@ softpipe_create_fs_state(struct pipe_context *pipe,
       tgsi_dump(templ->tokens, 0);
 
    /* codegen */
-   state = softpipe_create_fs_llvm( softpipe, templ );
+   state = softpipe_create_fs_sse( softpipe, templ );
    if (!state) {
-      state = softpipe_create_fs_sse( softpipe, templ );
-      if (!state) {
-         state = softpipe_create_fs_exec( softpipe, templ );
-      }
+      state = softpipe_create_fs_exec( softpipe, templ );
    }
 
    assert(state);
@@ -111,6 +107,8 @@ softpipe_create_vs_state(struct pipe_context *pipe,
    if (state->draw_data == NULL) 
       goto fail;
 
+   state->max_sampler = state->draw_data->info.file_max[TGSI_FILE_SAMPLER];
+
    return state;
 
 fail:
@@ -128,7 +126,7 @@ softpipe_bind_vs_state(struct pipe_context *pipe, void *vs)
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
 
-   softpipe->vs = (const struct sp_vertex_shader *)vs;
+   softpipe->vs = (struct sp_vertex_shader *) vs;
 
    draw_bind_vertex_shader(softpipe->draw,
                            (softpipe->vs ? softpipe->vs->draw_data : NULL));
@@ -142,8 +140,7 @@ softpipe_delete_vs_state(struct pipe_context *pipe, void *vs)
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
 
-   struct sp_vertex_shader *state =
-      (struct sp_vertex_shader *)vs;
+   struct sp_vertex_shader *state = (struct sp_vertex_shader *) vs;
 
    draw_delete_vertex_shader(softpipe->draw, state->draw_data);
    FREE( state );
diff --git a/src/gallium/drivers/softpipe/sp_state_sampler.c b/src/gallium/drivers/softpipe/sp_state_sampler.c
index cb517b02e4..db0b8ab76b 100644
--- a/src/gallium/drivers/softpipe/sp_state_sampler.c
+++ b/src/gallium/drivers/softpipe/sp_state_sampler.c
@@ -32,21 +32,37 @@
 #include "util/u_memory.h"
 
 #include "draw/draw_context.h"
+#include "draw/draw_context.h"
 
 #include "sp_context.h"
-#include "sp_context.h"
 #include "sp_state.h"
 #include "sp_texture.h"
-#include "sp_tile_cache.h"
-#include "draw/draw_context.h"
+#include "sp_tex_sample.h"
+#include "sp_tex_tile_cache.h"
 
 
+struct sp_sampler {
+   struct pipe_sampler_state base;
+   struct sp_sampler_varient *varients;
+   struct sp_sampler_varient *current;
+};
+
+static struct sp_sampler *sp_sampler( struct pipe_sampler_state *sampler )
+{
+   return (struct sp_sampler *)sampler;
+}
+
 
 void *
 softpipe_create_sampler_state(struct pipe_context *pipe,
                               const struct pipe_sampler_state *sampler)
 {
-   return mem_dup(sampler, sizeof(*sampler));
+   struct sp_sampler *sp_sampler = CALLOC_STRUCT(sp_sampler);
+
+   sp_sampler->base = *sampler;
+   sp_sampler->varients = NULL;
+
+   return (void *)sp_sampler;
 }
 
 
@@ -97,7 +113,7 @@ softpipe_set_sampler_textures(struct pipe_context *pipe,
       struct pipe_texture *tex = i < num ? texture[i] : NULL;
 
       pipe_texture_reference(&softpipe->texture[i], tex);
-      sp_tile_cache_set_texture(pipe, softpipe->tex_cache[i], tex);
+      sp_tex_tile_cache_set_texture(softpipe->tex_cache[i], tex);
    }
 
    softpipe->num_textures = num;
@@ -106,10 +122,111 @@ softpipe_set_sampler_textures(struct pipe_context *pipe,
 }
 
 
+/**
+ * Find/create an sp_sampler_varient object for sampling the given texture,
+ * sampler and tex unit.
+ *
+ * Note that the tex unit is significant.  We can't re-use a sampler
+ * varient for multiple texture units because the sampler varient contains
+ * the texture object pointer.  If the texture object pointer were stored
+ * somewhere outside the sampler varient, we could re-use samplers for
+ * multiple texture units.
+ */
+static struct sp_sampler_varient *
+get_sampler_varient( unsigned unit,
+                     struct sp_sampler *sampler,
+                     struct pipe_texture *texture,
+                     unsigned processor )
+{
+   struct softpipe_texture *sp_texture = softpipe_texture(texture);
+   struct sp_sampler_varient *v = NULL;
+   union sp_sampler_key key;
+
+   /* if this fails, widen the key.unit field and update this assertion */
+   assert(PIPE_MAX_SAMPLERS <= 16);
+
+   key.bits.target = sp_texture->base.target;
+   key.bits.is_pot = sp_texture->pot;
+   key.bits.processor = processor;
+   key.bits.unit = unit;
+   key.bits.pad = 0;
+
+   if (sampler->current && 
+       key.value == sampler->current->key.value) {
+      v = sampler->current;
+   }
+
+   if (v == NULL) {
+      for (v = sampler->varients; v; v = v->next)
+         if (v->key.value == key.value)
+            break;
+
+      if (v == NULL) {
+         v = sp_create_sampler_varient( &sampler->base, key );
+         v->next = sampler->varients;
+         sampler->varients = v;
+      }
+   }
+   
+   sampler->current = v;
+   return v;
+}
+
+
+
+
+void
+softpipe_reset_sampler_varients(struct softpipe_context *softpipe)
+{
+   int i;
+
+   /* It's a bit hard to build these samplers ahead of time -- don't
+    * really know which samplers are going to be used for vertex and
+    * fragment programs.
+    */
+   for (i = 0; i <= softpipe->vs->max_sampler; i++) {
+      if (softpipe->sampler[i]) {
+         softpipe->tgsi.vert_samplers_list[i] = 
+            get_sampler_varient( i,
+                                 sp_sampler(softpipe->sampler[i]),
+                                 softpipe->texture[i],
+                                 TGSI_PROCESSOR_VERTEX );
+
+         sp_sampler_varient_bind_texture( softpipe->tgsi.vert_samplers_list[i], 
+                                          softpipe->tex_cache[i],
+                                          softpipe->texture[i] );
+      }
+   }
+
+   for (i = 0; i <= softpipe->fs->info.file_max[TGSI_FILE_SAMPLER]; i++) {
+      if (softpipe->sampler[i]) {
+         softpipe->tgsi.frag_samplers_list[i] =
+            get_sampler_varient( i,
+                                 sp_sampler(softpipe->sampler[i]),
+                                 softpipe->texture[i],
+                                 TGSI_PROCESSOR_FRAGMENT );
+
+         sp_sampler_varient_bind_texture( softpipe->tgsi.frag_samplers_list[i], 
+                                          softpipe->tex_cache[i],
+                                          softpipe->texture[i] );
+      }
+   }
+}
+
+
+
 void
 softpipe_delete_sampler_state(struct pipe_context *pipe,
                               void *sampler)
 {
+   struct sp_sampler *sp_sampler = (struct sp_sampler *)sampler;
+   struct sp_sampler_varient *v, *tmp;
+
+   for (v = sp_sampler->varients; v; v = tmp) {
+      tmp = v->next;
+      sp_sampler_varient_destroy(v);
+   }
+
    FREE( sampler );
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_state_surface.c b/src/gallium/drivers/softpipe/sp_state_surface.c
index 181bff8f75..bc0e201130 100644
--- a/src/gallium/drivers/softpipe/sp_state_surface.c
+++ b/src/gallium/drivers/softpipe/sp_state_surface.c
@@ -53,7 +53,7 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe,
       /* check if changing cbuf */
       if (sp->framebuffer.cbufs[i] != fb->cbufs[i]) {
          /* flush old */
-         sp_flush_tile_cache(sp, sp->cbuf_cache[i]);
+         sp_flush_tile_cache(sp->cbuf_cache[i]);
 
          /* assign new */
          pipe_surface_reference(&sp->framebuffer.cbufs[i], fb->cbufs[i]);
@@ -68,58 +68,28 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe,
    /* zbuf changing? */
    if (sp->framebuffer.zsbuf != fb->zsbuf) {
       /* flush old */
-      sp_flush_tile_cache(sp, sp->zsbuf_cache);
+      sp_flush_tile_cache(sp->zsbuf_cache);
 
       /* assign new */
       pipe_surface_reference(&sp->framebuffer.zsbuf, fb->zsbuf);
 
       /* update cache */
       sp_tile_cache_set_surface(sp->zsbuf_cache, fb->zsbuf);
-   }
-
-#if 0
-   /* XXX combined depth/stencil here */
-
-   /* sbuf changing? */
-   if (sp->framebuffer.sbuf != fb->sbuf) {
-      /* flush old */
-      sp_flush_tile_cache(sp, sp->sbuf_cache_sep);
-
-      /* assign new */
-      sp->framebuffer.sbuf = fb->sbuf;
-
-      /* update cache */
-      if (fb->sbuf != fb->zbuf) {
-         /* separate stencil buf */
-         sp->sbuf_cache = sp->sbuf_cache_sep;
-         sp_tile_cache_set_surface(sp->sbuf_cache, fb->sbuf);
-      }
-      else {
-         /* combined depth/stencil */
-         sp->sbuf_cache = sp->zbuf_cache;
-         sp_tile_cache_set_surface(sp->sbuf_cache, fb->sbuf);
-      }
-   }
-#endif
 
-   /* Tell draw module how deep the Z/depth buffer is */
-   {
-      int depth_bits;
-      double mrd;
+      /* Tell draw module how deep the Z/depth buffer is */
       if (sp->framebuffer.zsbuf) {
+         int depth_bits;
+         double mrd;
          depth_bits = pf_get_component_bits(sp->framebuffer.zsbuf->format,
                                             PIPE_FORMAT_COMP_Z);
+         if (depth_bits > 16) {
+            mrd = 0.0000001;
+         }
+         else {
+            mrd = 0.00002;
+         }
+         draw_set_mrd(sp->draw, mrd);
       }
-      else {
-         depth_bits = 0;
-      }
-      if (depth_bits > 16) {
-         mrd = 0.0000001;
-      }
-      else {
-         mrd = 0.00002;
-      }
-      draw_set_mrd(sp->draw, mrd);
    }
 
    sp->framebuffer.width = fb->width;
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c
index f99a30277d..c22ee86b66 100644
--- a/src/gallium/drivers/softpipe/sp_tex_sample.c
+++ b/src/gallium/drivers/softpipe/sp_tex_sample.c
@@ -31,29 +31,33 @@
  *
  * Authors:
  *   Brian Paul
+ *   Keith Whitwell
  */
 
-#include "sp_context.h"
-#include "sp_quad.h"
-#include "sp_surface.h"
-#include "sp_texture.h"
-#include "sp_tex_sample.h"
-#include "sp_tile_cache.h"
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
+#include "sp_quad.h"   /* only for #define QUAD_* tokens */
+#include "sp_tex_sample.h"
+#include "sp_tex_tile_cache.h"
 
 
 
 /*
- * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes
- * see 1-pixel bands of improperly weighted linear-filtered textures.
+ * Return fractional part of 'f'.  Used for computing interpolation weights.
+ * Need to be careful with negative values.
+ * Note, if this function isn't perfect you'll sometimes see 1-pixel bands
+ * of improperly weighted linear-filtered textures.
  * The tests/texwrap.c demo is a good test.
- * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.
- * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).
  */
-#define FRAC(f)  ((f) - util_ifloor(f))
+static INLINE float
+frac(float f)
+{
+   return f - util_ifloor(f);
+}
+
 
 
 /**
@@ -100,10 +104,16 @@ lerp_3d(float a, float b, float c,
 
 
 /**
- * If A is a signed integer, A % B doesn't give the right value for A < 0
- * (in terms of texture repeat).  Just casting to unsigned fixes that.
+ * Compute coord % size for repeat wrap modes.
+ * Note that if coord is a signed integer, coord % size doesn't give
+ * the right value for coord < 0 (in terms of texture repeat).  Just
+ * casting to unsigned fixes that.
  */
-#define REMAINDER(A, B) ((unsigned) (A) % (unsigned) (B))
+static INLINE int
+repeat(int coord, unsigned size)
+{
+   return (int) ((unsigned) coord % size);
+}
 
 
 /**
@@ -115,133 +125,153 @@ lerp_3d(float a, float b, float c,
  * \param icoord  returns the integer texcoords
  * \return  integer texture index
  */
-static INLINE void
-nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
-                   int icoord[4])
+static void
+wrap_nearest_repeat(const float s[4], unsigned size, int icoord[4])
 {
    uint ch;
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_REPEAT:
-      /* s limited to [0,1) */
-      /* i limited to [0,size-1] */
-      for (ch = 0; ch < 4; ch++) {
-         int i = util_ifloor(s[ch] * size);
-         icoord[ch] = REMAINDER(i, size);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP:
+   /* s limited to [0,1) */
+   /* i limited to [0,size-1] */
+   for (ch = 0; ch < 4; ch++) {
+      int i = util_ifloor(s[ch] * size);
+      icoord[ch] = repeat(i, size);
+   }
+}
+
+
+static void
+wrap_nearest_clamp(const float s[4], unsigned size, int icoord[4])
+{
+   uint ch;
+   /* s limited to [0,1] */
+   /* i limited to [0,size-1] */
+   for (ch = 0; ch < 4; ch++) {
+      if (s[ch] <= 0.0F)
+         icoord[ch] = 0;
+      else if (s[ch] >= 1.0F)
+         icoord[ch] = size - 1;
+      else
+         icoord[ch] = util_ifloor(s[ch] * size);
+   }
+}
+
+
+static void
+wrap_nearest_clamp_to_edge(const float s[4], unsigned size, int icoord[4])
+{
+   uint ch;
+   /* s limited to [min,max] */
+   /* i limited to [0, size-1] */
+   const float min = 1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   for (ch = 0; ch < 4; ch++) {
+      if (s[ch] < min)
+         icoord[ch] = 0;
+      else if (s[ch] > max)
+         icoord[ch] = size - 1;
+      else
+         icoord[ch] = util_ifloor(s[ch] * size);
+   }
+}
+
+
+static void
+wrap_nearest_clamp_to_border(const float s[4], unsigned size, int icoord[4])
+{
+   uint ch;
+   /* s limited to [min,max] */
+   /* i limited to [-1, size] */
+   const float min = -1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   for (ch = 0; ch < 4; ch++) {
+      if (s[ch] <= min)
+         icoord[ch] = -1;
+      else if (s[ch] >= max)
+         icoord[ch] = size;
+      else
+         icoord[ch] = util_ifloor(s[ch] * size);
+   }
+}
+
+
+static void
+wrap_nearest_mirror_repeat(const float s[4], unsigned size, int icoord[4])
+{
+   uint ch;
+   const float min = 1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   for (ch = 0; ch < 4; ch++) {
+      const int flr = util_ifloor(s[ch]);
+      float u;
+      if (flr & 1)
+         u = 1.0F - (s[ch] - (float) flr);
+      else
+         u = s[ch] - (float) flr;
+      if (u < min)
+         icoord[ch] = 0;
+      else if (u > max)
+         icoord[ch] = size - 1;
+      else
+         icoord[ch] = util_ifloor(u * size);
+   }
+}
+
+
+static void
+wrap_nearest_mirror_clamp(const float s[4], unsigned size, int icoord[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
       /* s limited to [0,1] */
       /* i limited to [0,size-1] */
-      for (ch = 0; ch < 4; ch++) {
-         if (s[ch] <= 0.0F)
-            icoord[ch] = 0;
-         else if (s[ch] >= 1.0F)
-            icoord[ch] = size - 1;
-         else
-            icoord[ch] = util_ifloor(s[ch] * size);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const float min = 1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            if (s[ch] < min)
-               icoord[ch] = 0;
-            else if (s[ch] > max)
-               icoord[ch] = size - 1;
-            else
-               icoord[ch] = util_ifloor(s[ch] * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [-1, size] */
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            if (s[ch] <= min)
-               icoord[ch] = -1;
-            else if (s[ch] >= max)
-               icoord[ch] = size;
-            else
-               icoord[ch] = util_ifloor(s[ch] * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_REPEAT:
-      {
-         const float min = 1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            const int flr = util_ifloor(s[ch]);
-            float u;
-            if (flr & 1)
-               u = 1.0F - (s[ch] - (float) flr);
-            else
-               u = s[ch] - (float) flr;
-            if (u < min)
-               icoord[ch] = 0;
-            else if (u > max)
-               icoord[ch] = size - 1;
-            else
-               icoord[ch] = util_ifloor(u * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         /* s limited to [0,1] */
-         /* i limited to [0,size-1] */
-         const float u = fabsf(s[ch]);
-         if (u <= 0.0F)
-            icoord[ch] = 0;
-         else if (u >= 1.0F)
-            icoord[ch] = size - 1;
-         else
-            icoord[ch] = util_ifloor(u * size);
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const float min = 1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            const float u = fabsf(s[ch]);
-            if (u < min)
-               icoord[ch] = 0;
-            else if (u > max)
-               icoord[ch] = size - 1;
-            else
-               icoord[ch] = util_ifloor(u * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            const float u = fabsf(s[ch]);
-            if (u < min)
-               icoord[ch] = -1;
-            else if (u > max)
-               icoord[ch] = size;
-            else
-               icoord[ch] = util_ifloor(u * size);
-         }
-      }
-      return;
-   default:
-      assert(0);
+      const float u = fabsf(s[ch]);
+      if (u <= 0.0F)
+         icoord[ch] = 0;
+      else if (u >= 1.0F)
+         icoord[ch] = size - 1;
+      else
+         icoord[ch] = util_ifloor(u * size);
+   }
+}
+
+
+static void
+wrap_nearest_mirror_clamp_to_edge(const float s[4], unsigned size,
+                                  int icoord[4])
+{
+   uint ch;
+   /* s limited to [min,max] */
+   /* i limited to [0, size-1] */
+   const float min = 1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   for (ch = 0; ch < 4; ch++) {
+      const float u = fabsf(s[ch]);
+      if (u < min)
+         icoord[ch] = 0;
+      else if (u > max)
+         icoord[ch] = size - 1;
+      else
+         icoord[ch] = util_ifloor(u * size);
+   }
+}
+
+
+static void
+wrap_nearest_mirror_clamp_to_border(const float s[4], unsigned size,
+                                    int icoord[4])
+{
+   uint ch;
+   /* s limited to [min,max] */
+   /* i limited to [0, size-1] */
+   const float min = -1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   for (ch = 0; ch < 4; ch++) {
+      const float u = fabsf(s[ch]);
+      if (u < min)
+         icoord[ch] = -1;
+      else if (u > max)
+         icoord[ch] = size;
+      else
+         icoord[ch] = util_ifloor(u * size);
    }
 }
 
@@ -256,125 +286,156 @@ nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
  * \param w  returns blend factor/weight between texture indexes
  * \param icoord  returns the computed integer texture coords
  */
-static INLINE void
-linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
+static void
+wrap_linear_repeat(const float s[4], unsigned size,
+                   int icoord0[4], int icoord1[4], float w[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = s[ch] * size - 0.5F;
+      icoord0[ch] = repeat(util_ifloor(u), size);
+      icoord1[ch] = repeat(icoord0[ch] + 1, size);
+      w[ch] = frac(u);
+   }
+}
+
+
+static void
+wrap_linear_clamp(const float s[4], unsigned size,
                   int icoord0[4], int icoord1[4], float w[4])
 {
    uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = CLAMP(s[ch], 0.0F, 1.0F);
+      u = u * size - 0.5f;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      w[ch] = frac(u);
+   }
+}
 
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_REPEAT:
-      for (ch = 0; ch < 4; ch++) {
-         float u = s[ch] * size - 0.5F;
-         icoord0[ch] = REMAINDER(util_ifloor(u), size);
-         icoord1[ch] = REMAINDER(icoord0[ch] + 1, size);
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         float u = CLAMP(s[ch], 0.0F, 1.0F);
-         u = u * size - 0.5f;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      for (ch = 0; ch < 4; ch++) {
-         float u = CLAMP(s[ch], 0.0F, 1.0F);
-         u = u * size - 0.5f;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord0[ch] < 0)
-            icoord0[ch] = 0;
-         if (icoord1[ch] >= (int) size)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      {
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            float u = CLAMP(s[ch], min, max);
-            u = u * size - 0.5f;
-            icoord0[ch] = util_ifloor(u);
-            icoord1[ch] = icoord0[ch] + 1;
-            w[ch] = FRAC(u);
-         }
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_REPEAT:
-      for (ch = 0; ch < 4; ch++) {
-         const int flr = util_ifloor(s[ch]);
-         float u;
-         if (flr & 1)
-            u = 1.0F - (s[ch] - (float) flr);
-         else
-            u = s[ch] - (float) flr;
-         u = u * size - 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord0[ch] < 0)
-            icoord0[ch] = 0;
-         if (icoord1[ch] >= (int) size)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         float u = fabsf(s[ch]);
-         if (u >= 1.0F)
-            u = (float) size;
-         else
-            u *= size;
-         u -= 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
-      for (ch = 0; ch < 4; ch++) {
-         float u = fabsf(s[ch]);
-         if (u >= 1.0F)
-            u = (float) size;
-         else
-            u *= size;
-         u -= 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord0[ch] < 0)
-            icoord0[ch] = 0;
-         if (icoord1[ch] >= (int) size)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
-      {
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            float u = fabsf(s[ch]);
-            if (u <= min)
-               u = min * size;
-            else if (u >= max)
-               u = max * size;
-            else
-               u *= size;
-            u -= 0.5F;
-            icoord0[ch] = util_ifloor(u);
-            icoord1[ch] = icoord0[ch] + 1;
-            w[ch] = FRAC(u);
-         }
-      }
-      break;;
-   default:
-      assert(0);
+
+static void
+wrap_linear_clamp_to_edge(const float s[4], unsigned size,
+                          int icoord0[4], int icoord1[4], float w[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = CLAMP(s[ch], 0.0F, 1.0F);
+      u = u * size - 0.5f;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      if (icoord0[ch] < 0)
+         icoord0[ch] = 0;
+      if (icoord1[ch] >= (int) size)
+         icoord1[ch] = size - 1;
+      w[ch] = frac(u);
+   }
+}
+
+
+static void
+wrap_linear_clamp_to_border(const float s[4], unsigned size,
+                            int icoord0[4], int icoord1[4], float w[4])
+{
+   const float min = -1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = CLAMP(s[ch], min, max);
+      u = u * size - 0.5f;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      w[ch] = frac(u);
+   }
+}
+
+
+static void
+wrap_linear_mirror_repeat(const float s[4], unsigned size,
+                          int icoord0[4], int icoord1[4], float w[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      const int flr = util_ifloor(s[ch]);
+      float u;
+      if (flr & 1)
+         u = 1.0F - (s[ch] - (float) flr);
+      else
+         u = s[ch] - (float) flr;
+      u = u * size - 0.5F;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      if (icoord0[ch] < 0)
+         icoord0[ch] = 0;
+      if (icoord1[ch] >= (int) size)
+         icoord1[ch] = size - 1;
+      w[ch] = frac(u);
+   }
+}
+
+
+static void
+wrap_linear_mirror_clamp(const float s[4], unsigned size,
+                         int icoord0[4], int icoord1[4], float w[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = fabsf(s[ch]);
+      if (u >= 1.0F)
+         u = (float) size;
+      else
+         u *= size;
+      u -= 0.5F;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      w[ch] = frac(u);
+   }
+}
+
+
+static void
+wrap_linear_mirror_clamp_to_edge(const float s[4], unsigned size,
+                                 int icoord0[4], int icoord1[4], float w[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = fabsf(s[ch]);
+      if (u >= 1.0F)
+         u = (float) size;
+      else
+         u *= size;
+      u -= 0.5F;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      if (icoord0[ch] < 0)
+         icoord0[ch] = 0;
+      if (icoord1[ch] >= (int) size)
+         icoord1[ch] = size - 1;
+      w[ch] = frac(u);
+   }
+}
+
+
+static void
+wrap_linear_mirror_clamp_to_border(const float s[4], unsigned size,
+                                   int icoord0[4], int icoord1[4], float w[4])
+{
+   const float min = -1.0F / (2.0F * size);
+   const float max = 1.0F - min;
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = fabsf(s[ch]);
+      if (u <= min)
+         u = min * size;
+      else if (u >= max)
+         u = max * size;
+      else
+         u *= size;
+      u -= 0.5F;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      w[ch] = frac(u);
    }
 }
 
@@ -383,27 +444,27 @@ linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
  * For RECT textures / unnormalized texcoords
  * Only a subset of wrap modes supported.
  */
-static INLINE void
-nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
-                          int icoord[4])
+static void
+wrap_nearest_unorm_clamp(const float s[4], unsigned size, int icoord[4])
 {
    uint ch;
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         int i = util_ifloor(s[ch]);
-         icoord[ch]= CLAMP(i, 0, (int) size-1);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      /* fall-through */
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      for (ch = 0; ch < 4; ch++) {
-         icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) );
-      }
-      return;
-   default:
-      assert(0);
+   for (ch = 0; ch < 4; ch++) {
+      int i = util_ifloor(s[ch]);
+      icoord[ch]= CLAMP(i, 0, (int) size-1);
+   }
+}
+
+
+/**
+ * Handles clamp_to_edge and clamp_to_border:
+ */
+static void
+wrap_nearest_unorm_clamp_to_border(const float s[4], unsigned size,
+                                   int icoord[4])
+{
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) );
    }
 }
 
@@ -412,358 +473,971 @@ nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
  * For RECT textures / unnormalized texcoords.
  * Only a subset of wrap modes supported.
  */
-static INLINE void
-linear_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
-                         int icoord0[4], int icoord1[4], float w[4])
+static void
+wrap_linear_unorm_clamp(const float s[4], unsigned size,
+                        int icoord0[4], int icoord1[4], float w[4])
 {
    uint ch;
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         /* Not exactly what the spec says, but it matches NVIDIA output */
-         float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f);
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         w[ch] = FRAC(u);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      /* fall-through */
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      for (ch = 0; ch < 4; ch++) {
-         float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F);
-         u -= 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord1[ch] > (int) size - 1)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;
-   default:
-      assert(0);
+   for (ch = 0; ch < 4; ch++) {
+      /* Not exactly what the spec says, but it matches NVIDIA output */
+      float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f);
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      w[ch] = frac(u);
    }
 }
 
 
-static unsigned
-choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
+static void
+wrap_linear_unorm_clamp_to_border(const float s[4], unsigned size,
+                                  int icoord0[4], int icoord1[4], float w[4])
 {
-   /*
-      major axis
-      direction     target                             sc     tc    ma
-      ----------    -------------------------------    ---    ---   ---
-       +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
-       -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
-       +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
-       -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
-       +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
-       -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
-   */
-   const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz);
-   unsigned face;
-   float sc, tc, ma;
-
-   if (arx >= ary && arx >= arz) {
-      if (rx >= 0.0F) {
-         face = PIPE_TEX_FACE_POS_X;
-         sc = -rz;
-         tc = -ry;
-         ma = arx;
+   uint ch;
+   for (ch = 0; ch < 4; ch++) {
+      float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F);
+      u -= 0.5F;
+      icoord0[ch] = util_ifloor(u);
+      icoord1[ch] = icoord0[ch] + 1;
+      if (icoord1[ch] > (int) size - 1)
+         icoord1[ch] = size - 1;
+      w[ch] = frac(u);
+   }
+}
+
+
+
+/**
+ * Examine the quad's texture coordinates to compute the partial
+ * derivatives w.r.t X and Y, then compute lambda (level of detail).
+ */
+static float
+compute_lambda_1d(const struct sp_sampler_varient *samp,
+                  const float s[QUAD_SIZE],
+                  const float t[QUAD_SIZE],
+                  const float p[QUAD_SIZE],
+                  float lodbias)
+{
+   const struct pipe_texture *texture = samp->texture;
+   const struct pipe_sampler_state *sampler = samp->sampler;
+   float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
+   float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
+   float rho = MAX2(dsdx, dsdy) * texture->width[0];
+   float lambda;
+
+   lambda = util_fast_log2(rho);
+   lambda += lodbias + sampler->lod_bias;
+   lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod);
+
+   return lambda;
+}
+
+
+static float
+compute_lambda_2d(const struct sp_sampler_varient *samp,
+                  const float s[QUAD_SIZE],
+                  const float t[QUAD_SIZE],
+                  const float p[QUAD_SIZE],
+                  float lodbias)
+{
+   const struct pipe_texture *texture = samp->texture;
+   const struct pipe_sampler_state *sampler = samp->sampler;
+   float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
+   float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
+   float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
+   float dtdy = fabsf(t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]);
+   float maxx = MAX2(dsdx, dsdy) * texture->width[0];
+   float maxy = MAX2(dtdx, dtdy) * texture->height[0];
+   float rho  = MAX2(maxx, maxy);
+   float lambda;
+
+   lambda = util_fast_log2(rho);
+   lambda += lodbias + sampler->lod_bias;
+   lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod);
+
+   return lambda;
+}
+
+
+static float
+compute_lambda_3d(const struct sp_sampler_varient *samp,
+                  const float s[QUAD_SIZE],
+                  const float t[QUAD_SIZE],
+                  const float p[QUAD_SIZE],
+                  float lodbias)
+{
+   const struct pipe_texture *texture = samp->texture;
+   const struct pipe_sampler_state *sampler = samp->sampler;
+   float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
+   float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
+   float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
+   float dtdy = fabsf(t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]);
+   float dpdx = fabsf(p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]);
+   float dpdy = fabsf(p[QUAD_TOP_LEFT]     - p[QUAD_BOTTOM_LEFT]);
+   float maxx = MAX2(dsdx, dsdy) * texture->width[0];
+   float maxy = MAX2(dtdx, dtdy) * texture->height[0];
+   float maxz = MAX2(dpdx, dpdy) * texture->depth[0];
+   float rho, lambda;
+
+   rho = MAX2(maxx, maxy);
+   rho = MAX2(rho, maxz);
+
+   lambda = util_fast_log2(rho);
+   lambda += lodbias + sampler->lod_bias;
+   lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod);
+
+   return lambda;
+}
+
+
+/**
+ * Compute lambda for a vertex texture sampler.
+ * Since there aren't derivatives to use, just return the LOD bias.
+ */
+static float
+compute_lambda_vert(const struct sp_sampler_varient *samp,
+                    const float s[QUAD_SIZE],
+                    const float t[QUAD_SIZE],
+                    const float p[QUAD_SIZE],
+                    float lodbias)
+{
+   return lodbias;
+}
+
+
+
+/**
+ * Get a texel from a texture, using the texture tile cache.
+ *
+ * \param addr  the template tex address containing cube, z, face info.
+ * \param x  the x coord of texel within 2D image
+ * \param y  the y coord of texel within 2D image
+ * \param rgba  the quad to put the texel/color into
+ *
+ * XXX maybe move this into sp_tex_tile_cache.c and merge with the
+ * sp_get_cached_tile_tex() function.  Also, get 4 texels instead of 1...
+ */
+
+
+
+
+static INLINE const float *
+get_texel_2d_no_border(const struct sp_sampler_varient *samp,
+		       union tex_tile_address addr, int x, int y)
+{
+   const struct softpipe_tex_cached_tile *tile;
+
+   addr.bits.x = x / TILE_SIZE;
+   addr.bits.y = y / TILE_SIZE;
+   y %= TILE_SIZE;
+   x %= TILE_SIZE;
+
+   tile = sp_get_cached_tile_tex(samp->cache, addr);
+
+   return &tile->data.color[y][x][0];
+}
+
+
+static INLINE const float *
+get_texel_2d(const struct sp_sampler_varient *samp,
+	     union tex_tile_address addr, int x, int y)
+{
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level = addr.bits.level;
+
+   if (x < 0 || x >= (int) texture->width[level] ||
+       y < 0 || y >= (int) texture->height[level]) {
+      return samp->sampler->border_color;
+   }
+   else {
+      return get_texel_2d_no_border( samp, addr, x, y );
+   }
+}
+
+
+/* Gather a quad of adjacent texels within a tile:
+ */
+static INLINE void
+get_texel_quad_2d_no_border_single_tile(const struct sp_sampler_varient *samp,
+					union tex_tile_address addr, 
+					unsigned x, unsigned y, 
+					const float *out[4])
+{
+   const struct softpipe_tex_cached_tile *tile;
+
+   addr.bits.x = x / TILE_SIZE;
+   addr.bits.y = y / TILE_SIZE;
+   y %= TILE_SIZE;
+   x %= TILE_SIZE;
+
+   tile = sp_get_cached_tile_tex(samp->cache, addr);
+      
+   out[0] = &tile->data.color[y  ][x  ][0];
+   out[1] = &tile->data.color[y  ][x+1][0];
+   out[2] = &tile->data.color[y+1][x  ][0];
+   out[3] = &tile->data.color[y+1][x+1][0];
+}
+
+
+/* Gather a quad of potentially non-adjacent texels:
+ */
+static INLINE void
+get_texel_quad_2d_no_border(const struct sp_sampler_varient *samp,
+			    union tex_tile_address addr,
+			    int x0, int y0, 
+			    int x1, int y1,
+			    const float *out[4])
+{
+   out[0] = get_texel_2d_no_border( samp, addr, x0, y0 );
+   out[1] = get_texel_2d_no_border( samp, addr, x1, y0 );
+   out[2] = get_texel_2d_no_border( samp, addr, x0, y1 );
+   out[3] = get_texel_2d_no_border( samp, addr, x1, y1 );
+}
+
+/* Can involve a lot of unnecessary checks for border color:
+ */
+static INLINE void
+get_texel_quad_2d(const struct sp_sampler_varient *samp,
+		  union tex_tile_address addr,
+		  int x0, int y0, 
+		  int x1, int y1,
+		  const float *out[4])
+{
+   out[0] = get_texel_2d( samp, addr, x0, y0 );
+   out[1] = get_texel_2d( samp, addr, x1, y0 );
+   out[3] = get_texel_2d( samp, addr, x1, y1 );
+   out[2] = get_texel_2d( samp, addr, x0, y1 );
+}
+
+
+
+/* 3d varients:
+ */
+static INLINE const float *
+get_texel_3d_no_border(const struct sp_sampler_varient *samp,
+                       union tex_tile_address addr, int x, int y, int z)
+{
+   const struct softpipe_tex_cached_tile *tile;
+
+   addr.bits.x = x / TILE_SIZE;
+   addr.bits.y = y / TILE_SIZE;
+   addr.bits.z = z;
+   y %= TILE_SIZE;
+   x %= TILE_SIZE;
+
+   tile = sp_get_cached_tile_tex(samp->cache, addr);
+
+   return &tile->data.color[y][x][0];
+}
+
+
+static INLINE const float *
+get_texel_3d(const struct sp_sampler_varient *samp,
+	     union tex_tile_address addr, int x, int y, int z)
+{
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level = addr.bits.level;
+
+   if (x < 0 || x >= (int) texture->width[level] ||
+       y < 0 || y >= (int) texture->height[level] ||
+       z < 0 || z >= (int) texture->depth[level]) {
+      return samp->sampler->border_color;
+   }
+   else {
+      return get_texel_3d_no_border( samp, addr, x, y, z );
+   }
+}
+
+
+/**
+ * Given the logbase2 of a mipmap's base level size and a mipmap level,
+ * return the size (in texels) of that mipmap level.
+ * For example, if level[0].width = 256 then base_pot will be 8.
+ * If level = 2, then we'll return 64 (the width at level=2).
+ * Return 1 if level > base_pot.
+ */
+static INLINE unsigned
+pot_level_size(unsigned base_pot, unsigned level)
+{
+   return (base_pot >= level) ? (1 << (base_pot - level)) : 1;
+}
+
+
+/* Some image-filter fastpaths:
+ */
+static INLINE void
+img_filter_2d_linear_repeat_POT(struct tgsi_sampler *tgsi_sampler,
+                                const float s[QUAD_SIZE],
+                                const float t[QUAD_SIZE],
+                                const float p[QUAD_SIZE],
+                                float lodbias,
+                                float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   unsigned  j;
+   unsigned level = samp->level;
+   unsigned xpot = pot_level_size(samp->xpot, level);
+   unsigned ypot = pot_level_size(samp->ypot, level);
+   unsigned xmax = (xpot - 1) & (TILE_SIZE - 1); /* MIN2(TILE_SIZE, xpot) - 1; */
+   unsigned ymax = (ypot - 1) & (TILE_SIZE - 1); /* MIN2(TILE_SIZE, ypot) - 1; */
+   union tex_tile_address addr;
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      int c;
+
+      float u = s[j] * xpot - 0.5F;
+      float v = t[j] * ypot - 0.5F;
+
+      int uflr = util_ifloor(u);
+      int vflr = util_ifloor(v);
+
+      float xw = u - (float)uflr;
+      float yw = v - (float)vflr;
+
+      int x0 = uflr & (xpot - 1);
+      int y0 = vflr & (ypot - 1);
+
+      const float *tx[4];      
+      
+      /* Can we fetch all four at once:
+       */
+      if (x0 < xmax && y0 < ymax) {
+         get_texel_quad_2d_no_border_single_tile(samp, addr, x0, y0, tx);
       }
       else {
-         face = PIPE_TEX_FACE_NEG_X;
-         sc = rz;
-         tc = -ry;
-         ma = arx;
+         unsigned x1 = (x0 + 1) & (xpot - 1);
+         unsigned y1 = (y0 + 1) & (ypot - 1);
+         get_texel_quad_2d_no_border(samp, addr, x0, y0, x1, y1, tx);
+      }
+
+      /* interpolate R, G, B, A */
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = lerp_2d(xw, yw, 
+                              tx[0][c], tx[1][c], 
+                              tx[2][c], tx[3][c]);
       }
    }
-   else if (ary >= arx && ary >= arz) {
-      if (ry >= 0.0F) {
-         face = PIPE_TEX_FACE_POS_Y;
-         sc = rx;
-         tc = rz;
-         ma = ary;
+}
+
+
+static INLINE void
+img_filter_2d_nearest_repeat_POT(struct tgsi_sampler *tgsi_sampler,
+                                 const float s[QUAD_SIZE],
+                                 const float t[QUAD_SIZE],
+                                 const float p[QUAD_SIZE],
+                                 float lodbias,
+                                 float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   unsigned  j;
+   unsigned level = samp->level;
+   unsigned xpot = pot_level_size(samp->xpot, level);
+   unsigned ypot = pot_level_size(samp->ypot, level);
+   union tex_tile_address addr;
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      int c;
+
+      float u = s[j] * xpot;
+      float v = t[j] * ypot;
+
+      int uflr = util_ifloor(u);
+      int vflr = util_ifloor(v);
+
+      int x0 = uflr & (xpot - 1);
+      int y0 = vflr & (ypot - 1);
+
+      const float *out = get_texel_2d_no_border(samp, addr, x0, y0);
+
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = out[c];
       }
-      else {
-         face = PIPE_TEX_FACE_NEG_Y;
-         sc = rx;
-         tc = -rz;
-         ma = ary;
+   }
+}
+
+
+static INLINE void
+img_filter_2d_nearest_clamp_POT(struct tgsi_sampler *tgsi_sampler,
+                                const float s[QUAD_SIZE],
+                                const float t[QUAD_SIZE],
+                                const float p[QUAD_SIZE],
+                                float lodbias,
+                                float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   unsigned  j;
+   unsigned level = samp->level;
+   unsigned xpot = pot_level_size(samp->xpot, level);
+   unsigned ypot = pot_level_size(samp->ypot, level);
+   union tex_tile_address addr;
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      int c;
+
+      float u = s[j] * xpot;
+      float v = t[j] * ypot;
+
+      int x0, y0;
+      const float *out;
+
+      x0 = util_ifloor(u);
+      if (x0 < 0) 
+         x0 = 0;
+      else if (x0 > xpot - 1)
+         x0 = xpot - 1;
+
+      y0 = util_ifloor(v);
+      if (y0 < 0) 
+         y0 = 0;
+      else if (y0 > ypot - 1)
+         y0 = ypot - 1;
+      
+      out = get_texel_2d_no_border(samp, addr, x0, y0);
+
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = out[c];
       }
    }
-   else {
-      if (rz > 0.0F) {
-         face = PIPE_TEX_FACE_POS_Z;
-         sc = rx;
-         tc = -ry;
-         ma = arz;
+}
+
+
+static void
+img_filter_1d_nearest(struct tgsi_sampler *tgsi_sampler,
+                        const float s[QUAD_SIZE],
+                        const float t[QUAD_SIZE],
+                        const float p[QUAD_SIZE],
+                        float lodbias,
+                        float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level0, j;
+   int width;
+   int x[4];
+   union tex_tile_address addr;
+
+   level0 = samp->level;
+   width = texture->width[level0];
+
+   assert(width > 0);
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   samp->nearest_texcoord_s(s, width, x);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      const float *out = get_texel_2d(samp, addr, x[j], 0);
+      int c;
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = out[c];
       }
-      else {
-         face = PIPE_TEX_FACE_NEG_Z;
-         sc = -rx;
-         tc = -ry;
-         ma = arz;
+   }
+}
+
+
+static void
+img_filter_2d_nearest(struct tgsi_sampler *tgsi_sampler,
+                      const float s[QUAD_SIZE],
+                      const float t[QUAD_SIZE],
+                      const float p[QUAD_SIZE],
+                      float lodbias,
+                      float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level0, j;
+   int width, height;
+   int x[4], y[4];
+   union tex_tile_address addr;
+
+
+   level0 = samp->level;
+   width = texture->width[level0];
+   height = texture->height[level0];
+
+   assert(width > 0);
+   assert(height > 0);
+ 
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   samp->nearest_texcoord_s(s, width, x);
+   samp->nearest_texcoord_t(t, height, y);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      const float *out = get_texel_2d(samp, addr, x[j], y[j]);
+      int c;
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = out[c];
       }
    }
+}
 
-   *newS = ( sc / ma + 1.0F ) * 0.5F;
-   *newT = ( tc / ma + 1.0F ) * 0.5F;
 
-   return face;
+static INLINE union tex_tile_address
+face(union tex_tile_address addr, unsigned face )
+{
+   addr.bits.face = face;
+   return addr;
 }
 
 
-/**
- * Examine the quad's texture coordinates to compute the partial
- * derivatives w.r.t X and Y, then compute lambda (level of detail).
- *
- * This is only done for fragment shaders, not vertex shaders.
- */
-static float
-compute_lambda(const struct pipe_texture *tex,
-               const struct pipe_sampler_state *sampler,
-               const float s[QUAD_SIZE],
-               const float t[QUAD_SIZE],
-               const float p[QUAD_SIZE],
-               float lodbias)
+static void
+img_filter_cube_nearest(struct tgsi_sampler *tgsi_sampler,
+                        const float s[QUAD_SIZE],
+                        const float t[QUAD_SIZE],
+                        const float p[QUAD_SIZE],
+                        float lodbias,
+                        float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
-   float rho, lambda;
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   const unsigned *faces = samp->faces; /* zero when not cube-mapping */
+   unsigned level0, j;
+   int width, height;
+   int x[4], y[4];
+   union tex_tile_address addr;
 
-   assert(sampler->normalized_coords);
+   level0 = samp->level;
+   width = texture->width[level0];
+   height = texture->height[level0];
 
-   assert(s);
-   {
-      float dsdx = s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT];
-      float dsdy = s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT];
-      dsdx = fabsf(dsdx);
-      dsdy = fabsf(dsdy);
-      rho = MAX2(dsdx, dsdy) * tex->width[0];
-   }
-   if (t) {
-      float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT];
-      float dtdy = t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT];
-      float max;
-      dtdx = fabsf(dtdx);
-      dtdy = fabsf(dtdy);
-      max = MAX2(dtdx, dtdy) * tex->height[0];
-      rho = MAX2(rho, max);
+   assert(width > 0);
+   assert(height > 0);
+ 
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   samp->nearest_texcoord_s(s, width, x);
+   samp->nearest_texcoord_t(t, height, y);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      const float *out = get_texel_2d(samp, face(addr, faces[j]), x[j], y[j]);
+      int c;
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = out[c];
+      }
    }
-   if (p) {
-      float dpdx = p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT];
-      float dpdy = p[QUAD_TOP_LEFT]     - p[QUAD_BOTTOM_LEFT];
-      float max;
-      dpdx = fabsf(dpdx);
-      dpdy = fabsf(dpdy);
-      max = MAX2(dpdx, dpdy) * tex->depth[0];
-      rho = MAX2(rho, max);
+}
+
+
+static void
+img_filter_3d_nearest(struct tgsi_sampler *tgsi_sampler,
+                      const float s[QUAD_SIZE],
+                      const float t[QUAD_SIZE],
+                      const float p[QUAD_SIZE],
+                      float lodbias,
+                      float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level0, j;
+   int width, height, depth;
+   int x[4], y[4], z[4];
+   union tex_tile_address addr;
+
+   level0 = samp->level;
+   width = texture->width[level0];
+   height = texture->height[level0];
+   depth = texture->depth[level0];
+
+   assert(width > 0);
+   assert(height > 0);
+   assert(depth > 0);
+
+   samp->nearest_texcoord_s(s, width,  x);
+   samp->nearest_texcoord_t(t, height, y);
+   samp->nearest_texcoord_p(p, depth,  z);
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      const float *out = get_texel_3d(samp, addr, x[j], y[j], z[j]);
+      int c;
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = out[c];
+      }      
    }
+}
 
-   lambda = util_fast_log2(rho);
-   lambda += lodbias + sampler->lod_bias;
-   lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod);
 
-   return lambda;
+static void
+img_filter_1d_linear(struct tgsi_sampler *tgsi_sampler,
+                     const float s[QUAD_SIZE],
+                     const float t[QUAD_SIZE],
+                     const float p[QUAD_SIZE],
+                     float lodbias,
+                     float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level0, j;
+   int width;
+   int x0[4], x1[4];
+   float xw[4]; /* weights */
+   union tex_tile_address addr;
+
+   level0 = samp->level;
+   width = texture->width[level0];
+
+   assert(width > 0);
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   samp->linear_texcoord_s(s, width, x0, x1, xw);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      const float *tx0 = get_texel_2d(samp, addr, x0[j], 0);
+      const float *tx1 = get_texel_2d(samp, addr, x1[j], 0);
+      int c;
+
+      /* interpolate R, G, B, A */
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = lerp(xw[j], tx0[c], tx1[c]);
+      }
+   }
 }
 
 
-/**
- * Do several things here:
- * 1. Compute lambda from the texcoords, if needed
- * 2. Determine if we're minifying or magnifying
- * 3. If minifying, choose mipmap levels
- * 4. Return image filter to use within mipmap images
- * \param level0  Returns first mipmap level to sample from
- * \param level1  Returns second mipmap level to sample from
- * \param levelBlend  Returns blend factor between levels, in [0,1]
- * \param imgFilter  Returns either the min or mag filter, depending on lambda
- */
 static void
-choose_mipmap_levels(const struct pipe_texture *texture,
-                     const struct pipe_sampler_state *sampler,
+img_filter_2d_linear(struct tgsi_sampler *tgsi_sampler,
                      const float s[QUAD_SIZE],
                      const float t[QUAD_SIZE],
                      const float p[QUAD_SIZE],
-                     boolean computeLambda,
                      float lodbias,
-                     unsigned *level0, unsigned *level1, float *levelBlend,
-                     unsigned *imgFilter)
-{
-   if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) {
-      /* no mipmap selection needed */
-      *level0 = *level1 = CLAMP((int) sampler->min_lod,
-                                0, (int) texture->last_level);
-
-      if (sampler->min_img_filter != sampler->mag_img_filter) {
-         /* non-mipmapped texture, but still need to determine if doing
-          * minification or magnification.
-          */
-         float lambda = compute_lambda(texture, sampler, s, t, p, lodbias);
-         if (lambda <= 0.0) {
-            *imgFilter = sampler->mag_img_filter;
-         }
-         else {
-            *imgFilter = sampler->min_img_filter;
-         }
+                     float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level0, j;
+   int width, height;
+   int x0[4], y0[4], x1[4], y1[4];
+   float xw[4], yw[4]; /* weights */
+   union tex_tile_address addr;
+
+   level0 = samp->level;
+   width = texture->width[level0];
+   height = texture->height[level0];
+
+   assert(width > 0);
+   assert(height > 0);
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   samp->linear_texcoord_s(s, width,  x0, x1, xw);
+   samp->linear_texcoord_t(t, height, y0, y1, yw);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      const float *tx0 = get_texel_2d(samp, addr, x0[j], y0[j]);
+      const float *tx1 = get_texel_2d(samp, addr, x1[j], y0[j]);
+      const float *tx2 = get_texel_2d(samp, addr, x0[j], y1[j]);
+      const float *tx3 = get_texel_2d(samp, addr, x1[j], y1[j]);
+      int c;
+
+      /* interpolate R, G, B, A */
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = lerp_2d(xw[j], yw[j],
+                              tx0[c], tx1[c],
+                              tx2[c], tx3[c]);
       }
-      else {
-         *imgFilter = sampler->mag_img_filter;
+   }
+}
+
+
+static void
+img_filter_cube_linear(struct tgsi_sampler *tgsi_sampler,
+                       const float s[QUAD_SIZE],
+                       const float t[QUAD_SIZE],
+                       const float p[QUAD_SIZE],
+                       float lodbias,
+                       float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   const unsigned *faces = samp->faces; /* zero when not cube-mapping */
+   unsigned level0, j;
+   int width, height;
+   int x0[4], y0[4], x1[4], y1[4];
+   float xw[4], yw[4]; /* weights */
+   union tex_tile_address addr;
+
+   level0 = samp->level;
+   width = texture->width[level0];
+   height = texture->height[level0];
+
+   assert(width > 0);
+   assert(height > 0);
+
+   addr.value = 0;
+   addr.bits.level = samp->level;
+
+   samp->linear_texcoord_s(s, width,  x0, x1, xw);
+   samp->linear_texcoord_t(t, height, y0, y1, yw);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      union tex_tile_address addrj = face(addr, faces[j]);
+      const float *tx0 = get_texel_2d(samp, addrj, x0[j], y0[j]);
+      const float *tx1 = get_texel_2d(samp, addrj, x1[j], y0[j]);
+      const float *tx2 = get_texel_2d(samp, addrj, x0[j], y1[j]);
+      const float *tx3 = get_texel_2d(samp, addrj, x1[j], y1[j]);
+      int c;
+
+      /* interpolate R, G, B, A */
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = lerp_2d(xw[j], yw[j],
+                              tx0[c], tx1[c],
+                              tx2[c], tx3[c]);
       }
    }
-   else {
-      float lambda;
+}
 
-      if (computeLambda)
-         /* fragment shader */
-         lambda = compute_lambda(texture, sampler, s, t, p, lodbias);
-      else
-         /* vertex shader */
-         lambda = lodbias; /* not really a bias, but absolute LOD */
 
-      if (lambda <= 0.0) { /* XXX threshold depends on the filter */
-         /* magnifying */
-         *imgFilter = sampler->mag_img_filter;
-         *level0 = *level1 = 0;
+static void
+img_filter_3d_linear(struct tgsi_sampler *tgsi_sampler,
+                     const float s[QUAD_SIZE],
+                     const float t[QUAD_SIZE],
+                     const float p[QUAD_SIZE],
+                     float lodbias,
+                     float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   unsigned level0, j;
+   int width, height, depth;
+   int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4];
+   float xw[4], yw[4], zw[4]; /* interpolation weights */
+   union tex_tile_address addr;
+
+   level0 = samp->level;
+   width = texture->width[level0];
+   height = texture->height[level0];
+   depth = texture->depth[level0];
+
+   addr.value = 0;
+   addr.bits.level = level0;
+
+   assert(width > 0);
+   assert(height > 0);
+   assert(depth > 0);
+
+   samp->linear_texcoord_s(s, width,  x0, x1, xw);
+   samp->linear_texcoord_t(t, height, y0, y1, yw);
+   samp->linear_texcoord_p(p, depth,  z0, z1, zw);
+
+   for (j = 0; j < QUAD_SIZE; j++) {
+      int c;
+
+      const float *tx00 = get_texel_3d(samp, addr, x0[j], y0[j], z0[j]);
+      const float *tx01 = get_texel_3d(samp, addr, x1[j], y0[j], z0[j]);
+      const float *tx02 = get_texel_3d(samp, addr, x0[j], y1[j], z0[j]);
+      const float *tx03 = get_texel_3d(samp, addr, x1[j], y1[j], z0[j]);
+      
+      const float *tx10 = get_texel_3d(samp, addr, x0[j], y0[j], z1[j]);
+      const float *tx11 = get_texel_3d(samp, addr, x1[j], y0[j], z1[j]);
+      const float *tx12 = get_texel_3d(samp, addr, x0[j], y1[j], z1[j]);
+      const float *tx13 = get_texel_3d(samp, addr, x1[j], y1[j], z1[j]);
+      
+      /* interpolate R, G, B, A */
+      for (c = 0; c < 4; c++) {
+         rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j],
+                              tx00[c], tx01[c],
+                              tx02[c], tx03[c],
+                              tx10[c], tx11[c],
+                              tx12[c], tx13[c]);
       }
-      else {
-         /* minifying */
-         *imgFilter = sampler->min_img_filter;
-
-         /* choose mipmap level(s) and compute the blend factor between them */
-         if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) {
-            /* Nearest mipmap level */
-            const int lvl = (int) (lambda + 0.5);
-            *level0 =
-            *level1 = CLAMP(lvl, 0, (int) texture->last_level);
-         }
-         else {
-            /* Linear interpolation between mipmap levels */
-            const int lvl = (int) lambda;
-            *level0 = CLAMP(lvl,     0, (int) texture->last_level);
-            *level1 = CLAMP(lvl + 1, 0, (int) texture->last_level);
-            *levelBlend = FRAC(lambda);  /* blending weight between levels */
+   }
+}
+
+
+static void
+mip_filter_linear(struct tgsi_sampler *tgsi_sampler,
+                  const float s[QUAD_SIZE],
+                  const float t[QUAD_SIZE],
+                  const float p[QUAD_SIZE],
+                  float lodbias,
+                  float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   int level0;
+   float lambda;
+
+   lambda = samp->compute_lambda(samp, s, t, p, lodbias);
+   level0 = (int)lambda;
+
+   if (lambda < 0.0) { 
+      samp->level = 0;
+      samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba );
+   }
+   else if (level0 >= texture->last_level) {
+      samp->level = texture->last_level;
+      samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba );
+   }
+   else {
+      float levelBlend = lambda - level0;
+      float rgba0[4][4];
+      float rgba1[4][4];
+      int c,j;
+
+      samp->level = level0;
+      samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba0 );
+
+      samp->level = level0+1;
+      samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba1 );
+
+      for (j = 0; j < QUAD_SIZE; j++) {
+         for (c = 0; c < 4; c++) {
+            rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]);
          }
       }
    }
 }
 
 
-/**
- * Get a texel from a texture, using the texture tile cache.
- *
- * \param face  the cube face in 0..5
- * \param level  the mipmap level
- * \param x  the x coord of texel within 2D image
- * \param y  the y coord of texel within 2D image
- * \param z  which slice of a 3D texture
- * \param rgba  the quad to put the texel/color into
- * \param j  which element of the rgba quad to write to
- *
- * XXX maybe move this into sp_tile_cache.c and merge with the
- * sp_get_cached_tile_tex() function.  Also, get 4 texels instead of 1...
- */
 static void
-get_texel(const struct tgsi_sampler *tgsi_sampler,
-          unsigned face, unsigned level, int x, int y, int z,
-          float rgba[NUM_CHANNELS][QUAD_SIZE], unsigned j)
+mip_filter_nearest(struct tgsi_sampler *tgsi_sampler,
+                   const float s[QUAD_SIZE],
+                   const float t[QUAD_SIZE],
+                   const float p[QUAD_SIZE],
+                   float lodbias,
+                   float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
-   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
-   struct softpipe_context *sp = samp->sp;
-   const uint unit = samp->unit;
-   const struct pipe_texture *texture = sp->texture[unit];
-   const struct pipe_sampler_state *sampler = sp->sampler[unit];
+   struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   float lambda;
 
-   if (x < 0 || x >= (int) texture->width[level] ||
-       y < 0 || y >= (int) texture->height[level] ||
-       z < 0 || z >= (int) texture->depth[level]) {
-      rgba[0][j] = sampler->border_color[0];
-      rgba[1][j] = sampler->border_color[1];
-      rgba[2][j] = sampler->border_color[2];
-      rgba[3][j] = sampler->border_color[3];
+   lambda = samp->compute_lambda(samp, s, t, p, lodbias);
+
+   if (lambda < 0.0) { 
+      samp->level = 0;
+      samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba );
    }
    else {
-      const int tx = x % TILE_SIZE;
-      const int ty = y % TILE_SIZE;
-      const struct softpipe_cached_tile *tile
-         = sp_get_cached_tile_tex(sp, samp->cache,
-                                  x, y, z, face, level);
-      rgba[0][j] = tile->data.color[ty][tx][0];
-      rgba[1][j] = tile->data.color[ty][tx][1];
-      rgba[2][j] = tile->data.color[ty][tx][2];
-      rgba[3][j] = tile->data.color[ty][tx][3];
-      if (0)
-      {
-         debug_printf("Get texel %f %f %f %f from %s\n",
-                      rgba[0][j], rgba[1][j], rgba[2][j], rgba[3][j],
-                      pf_name(texture->format));
-      }
+      samp->level = (int)(lambda + 0.5) ;
+      samp->level = MIN2(samp->level, (int)texture->last_level);
+      samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba );
+   }
+
+#if 0
+   printf("RGBA %g %g %g %g, %g %g %g %g, %g %g %g %g, %g %g %g %g\n",
+          rgba[0][0], rgba[1][0], rgba[2][0], rgba[3][0],
+          rgba[0][1], rgba[1][1], rgba[2][1], rgba[3][1],
+          rgba[0][2], rgba[1][2], rgba[2][2], rgba[3][2],
+          rgba[0][3], rgba[1][3], rgba[2][3], rgba[3][3]);
+#endif
+}
+
+
+static void
+mip_filter_none(struct tgsi_sampler *tgsi_sampler,
+                const float s[QUAD_SIZE],
+                const float t[QUAD_SIZE],
+                const float p[QUAD_SIZE],
+                float lodbias,
+                float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   float lambda = samp->compute_lambda(samp, s, t, p, lodbias);
+
+   if (lambda < 0.0) { 
+      samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba );
+   }
+   else {
+      samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba );
    }
 }
 
 
+
 /**
- * Compare texcoord 'p' (aka R) against texture value 'rgba[0]'
- * When we sampled the depth texture, the depth value was put into all
- * RGBA channels.  We look at the red channel here.
- * \param rgba  quad of (depth) texel values
- * \param p  texture 'P' components for four pixels in quad
- * \param j  which pixel in the quad to test [0..3]
+ * Specialized version of mip_filter_linear with hard-wired calls to
+ * 2d lambda calculation and 2d_linear_repeat_POT img filters.
  */
-static INLINE void
-shadow_compare(const struct pipe_sampler_state *sampler,
-               float rgba[NUM_CHANNELS][QUAD_SIZE],
-               const float p[QUAD_SIZE],
-               uint j)
+static void
+mip_filter_linear_2d_linear_repeat_POT(
+   struct tgsi_sampler *tgsi_sampler,
+   const float s[QUAD_SIZE],
+   const float t[QUAD_SIZE],
+   const float p[QUAD_SIZE],
+   float lodbias,
+   float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
-   int k;
-   switch (sampler->compare_func) {
-   case PIPE_FUNC_LESS:
-      k = p[j] < rgba[0][j];
-      break;
-   case PIPE_FUNC_LEQUAL:
-      k = p[j] <= rgba[0][j];
-      break;
-   case PIPE_FUNC_GREATER:
-      k = p[j] > rgba[0][j];
-      break;
-   case PIPE_FUNC_GEQUAL:
-      k = p[j] >= rgba[0][j];
-      break;
-   case PIPE_FUNC_EQUAL:
-      k = p[j] == rgba[0][j];
-      break;
-   case PIPE_FUNC_NOTEQUAL:
-      k = p[j] != rgba[0][j];
-      break;
-   case PIPE_FUNC_ALWAYS:
-      k = 1;
-      break;
-   case PIPE_FUNC_NEVER:
-      k = 0;
-      break;
-   default:
-      k = 0;
-      assert(0);
-      break;
+   struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_texture *texture = samp->texture;
+   int level0;
+   float lambda;
+
+   lambda = compute_lambda_2d(samp, s, t, p, lodbias);
+   level0 = (int)lambda;
+
+   /* Catches both negative and large values of level0:
+    */
+   if ((unsigned)level0 >= texture->last_level) { 
+      if (level0 < 0)
+         samp->level = 0;
+      else
+         samp->level = texture->last_level;
+
+      img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba );
    }
+   else {
+      float levelBlend = lambda - level0;
+      float rgba0[4][4];
+      float rgba1[4][4];
+      int c,j;
 
-   /* XXX returning result for default GL_DEPTH_TEXTURE_MODE = GL_LUMINANCE */
-   rgba[0][j] = rgba[1][j] = rgba[2][j] = (float) k;
-   rgba[3][j] = 1.0F;
+      samp->level = level0;
+      img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba0 );
+
+      samp->level = level0+1;
+      img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba1 );
+
+      for (j = 0; j < QUAD_SIZE; j++) {
+         for (c = 0; c < 4; c++) {
+            rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]);
+         }
+      }
+   }
 }
 
 
+
 /**
- * As above, but do four z/texture comparisons.
+ * Do shadow/depth comparisons.
  */
-static INLINE void
-shadow_compare4(const struct pipe_sampler_state *sampler,
-                float rgba[NUM_CHANNELS][QUAD_SIZE],
-                const float p[QUAD_SIZE])
+static void
+sample_compare(struct tgsi_sampler *tgsi_sampler,
+               const float s[QUAD_SIZE],
+               const float t[QUAD_SIZE],
+               const float p[QUAD_SIZE],
+               float lodbias,
+               float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
+   struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   const struct pipe_sampler_state *sampler = samp->sampler;
    int j, k0, k1, k2, k3;
    float val;
 
+   samp->mip_filter( tgsi_sampler, s, t, p, lodbias, rgba );
+
+   /**
+    * Compare texcoord 'p' (aka R) against texture value 'rgba[0]'
+    * When we sampled the depth texture, the depth value was put into all
+    * RGBA channels.  We look at the red channel here.
+    */
+
    /* compare four texcoords vs. four texture samples */
    switch (sampler->compare_func) {
    case PIPE_FUNC_LESS:
@@ -826,470 +1500,392 @@ shadow_compare4(const struct pipe_sampler_state *sampler,
 
 
 /**
- * Common code for sampling 1D/2D/cube textures.
- * Could probably extend for 3D...
+ * Compute which cube face is referenced by each texcoord and put that
+ * info into the sampler faces[] array.  Then sample the cube faces
  */
 static void
-sp_get_samples_2d_common(const struct tgsi_sampler *tgsi_sampler,
-                         const float s[QUAD_SIZE],
-                         const float t[QUAD_SIZE],
-                         const float p[QUAD_SIZE],
-                         boolean computeLambda,
-                         float lodbias,
-                         float rgba[NUM_CHANNELS][QUAD_SIZE],
-                         const unsigned faces[4])
-{
-   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
-   const struct softpipe_context *sp = samp->sp;
-   const uint unit = samp->unit;
-   const struct pipe_texture *texture = sp->texture[unit];
-   const struct pipe_sampler_state *sampler = sp->sampler[unit];
-   unsigned level0, level1, j, imgFilter;
-   int width, height;
-   float levelBlend;
-
-   choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias,
-                        &level0, &level1, &levelBlend, &imgFilter);
-
-   assert(sampler->normalized_coords);
-
-   width = texture->width[level0];
-   height = texture->height[level0];
-
-   assert(width > 0);
-
-   switch (imgFilter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      {
-         int x[4], y[4];
-         nearest_texcoord_4(sampler->wrap_s, s, width, x);
-         nearest_texcoord_4(sampler->wrap_t, t, height, y);
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            get_texel(tgsi_sampler, faces[j], level0, x[j], y[j], 0, rgba, j);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare(sampler, rgba, p, j);
-            }
+sample_cube(struct tgsi_sampler *tgsi_sampler,
+            const float s[QUAD_SIZE],
+            const float t[QUAD_SIZE],
+            const float p[QUAD_SIZE],
+            float lodbias,
+            float rgba[NUM_CHANNELS][QUAD_SIZE])
+{
+   struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler);
+   unsigned j;
+   float ssss[4], tttt[4];
 
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               unsigned c;
-               x[j] /= 2;
-               y[j] /= 2;
-               get_texel(tgsi_sampler, faces[j], level1, x[j], y[j], 0,
-                         rgba2, j);
-               if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
-                  shadow_compare(sampler, rgba2, p, j);
-               }
-
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
-               }
-            }
+   /*
+     major axis
+     direction     target                             sc     tc    ma
+     ----------    -------------------------------    ---    ---   ---
+     +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
+     -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
+     +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
+     -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
+     +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
+     -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
+   */
+   for (j = 0; j < QUAD_SIZE; j++) {
+      float rx = s[j];
+      float ry = t[j];
+      float rz = p[j];
+      const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz);
+      unsigned face;
+      float sc, tc, ma;
+
+      if (arx >= ary && arx >= arz) {
+         if (rx >= 0.0F) {
+            face = PIPE_TEX_FACE_POS_X;
+            sc = -rz;
+            tc = -ry;
+            ma = arx;
+         }
+         else {
+            face = PIPE_TEX_FACE_NEG_X;
+            sc = rz;
+            tc = -ry;
+            ma = arx;
          }
       }
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-   case PIPE_TEX_FILTER_ANISO:
+      else if (ary >= arx && ary >= arz) {
+         if (ry >= 0.0F) {
+            face = PIPE_TEX_FACE_POS_Y;
+            sc = rx;
+            tc = rz;
+            ma = ary;
+         }
+         else {
+            face = PIPE_TEX_FACE_NEG_Y;
+            sc = rx;
+            tc = -rz;
+            ma = ary;
+         }
+      }
+      else {
+         if (rz > 0.0F) {
+            face = PIPE_TEX_FACE_POS_Z;
+            sc = rx;
+            tc = -ry;
+            ma = arz;
+         }
+         else {
+            face = PIPE_TEX_FACE_NEG_Z;
+            sc = -rx;
+            tc = -ry;
+            ma = arz;
+         }
+      }
+
       {
-         int x0[4], y0[4], x1[4], y1[4];
-         float xw[4], yw[4]; /* weights */
-
-         linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw);
-         linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw);
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            float tx[4][4]; /* texels */
-            int c;
-            get_texel(tgsi_sampler, faces[j], level0, x0[j], y0[j], 0, tx, 0);
-            get_texel(tgsi_sampler, faces[j], level0, x1[j], y0[j], 0, tx, 1);
-            get_texel(tgsi_sampler, faces[j], level0, x0[j], y1[j], 0, tx, 2);
-            get_texel(tgsi_sampler, faces[j], level0, x1[j], y1[j], 0, tx, 3);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare4(sampler, tx, p);
-            }
+	 const float ima = 1.0 / ma;
+	 ssss[j] = ( sc * ima + 1.0F ) * 0.5F;
+	 tttt[j] = ( tc * ima + 1.0F ) * 0.5F;
+	 samp->faces[j] = face;
+      }
+   }
 
-            /* interpolate R, G, B, A */
-            for (c = 0; c < 4; c++) {
-               rgba[c][j] = lerp_2d(xw[j], yw[j],
-                                    tx[c][0], tx[c][1],
-                                    tx[c][2], tx[c][3]);
-            }
+   /* In our little pipeline, the compare stage is next.  If compare
+    * is not active, this will point somewhere deeper into the
+    * pipeline, eg. to mip_filter or even img_filter.
+    */
+   samp->compare(tgsi_sampler, ssss, tttt, NULL, lodbias, rgba);
+}
 
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               x0[j] /= 2;
-               y0[j] /= 2;
-               x1[j] /= 2;
-               y1[j] /= 2;
-               get_texel(tgsi_sampler, faces[j], level1, x0[j], y0[j], 0, tx, 0);
-               get_texel(tgsi_sampler, faces[j], level1, x1[j], y0[j], 0, tx, 1);
-               get_texel(tgsi_sampler, faces[j], level1, x0[j], y1[j], 0, tx, 2);
-               get_texel(tgsi_sampler, faces[j], level1, x1[j], y1[j], 0, tx, 3);
-               if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
-                  shadow_compare4(sampler, tx, p);
-               }
-
-               /* interpolate R, G, B, A */
-               for (c = 0; c < 4; c++) {
-                  rgba2[c][j] = lerp_2d(xw[j], yw[j],
-                                        tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
-               }
-
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
-               }
-            }
-         }
-      }
-      break;
+
+
+static wrap_nearest_func
+get_nearest_unorm_wrap(unsigned mode)
+{
+   switch (mode) {
+   case PIPE_TEX_WRAP_CLAMP:
+      return wrap_nearest_unorm_clamp;
+   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+      return wrap_nearest_unorm_clamp_to_border;
    default:
       assert(0);
+      return wrap_nearest_unorm_clamp;
    }
 }
 
 
-static INLINE void
-sp_get_samples_1d(const struct tgsi_sampler *sampler,
-                  const float s[QUAD_SIZE],
-                  const float t[QUAD_SIZE],
-                  const float p[QUAD_SIZE],
-                  boolean computeLambda,
-                  float lodbias,
-                  float rgba[NUM_CHANNELS][QUAD_SIZE])
+static wrap_nearest_func
+get_nearest_wrap(unsigned mode)
 {
-   static const unsigned faces[4] = {0, 0, 0, 0};
-   static const float tzero[4] = {0, 0, 0, 0};
-   sp_get_samples_2d_common(sampler, s, tzero, NULL,
-                            computeLambda, lodbias, rgba, faces);
+   switch (mode) {
+   case PIPE_TEX_WRAP_REPEAT:
+      return wrap_nearest_repeat;
+   case PIPE_TEX_WRAP_CLAMP:
+      return wrap_nearest_clamp;
+   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+      return wrap_nearest_clamp_to_edge;
+   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+      return wrap_nearest_clamp_to_border;
+   case PIPE_TEX_WRAP_MIRROR_REPEAT:
+      return wrap_nearest_mirror_repeat;
+   case PIPE_TEX_WRAP_MIRROR_CLAMP:
+      return wrap_nearest_mirror_clamp;
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+      return wrap_nearest_mirror_clamp_to_edge;
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+      return wrap_nearest_mirror_clamp_to_border;
+   default:
+      assert(0);
+      return wrap_nearest_repeat;
+   }
 }
 
 
-static INLINE void
-sp_get_samples_2d(const struct tgsi_sampler *sampler,
-                  const float s[QUAD_SIZE],
-                  const float t[QUAD_SIZE],
-                  const float p[QUAD_SIZE],
-                  boolean computeLambda,
-                  float lodbias,
-                  float rgba[NUM_CHANNELS][QUAD_SIZE])
+static wrap_linear_func
+get_linear_unorm_wrap(unsigned mode)
 {
-   static const unsigned faces[4] = {0, 0, 0, 0};
-   sp_get_samples_2d_common(sampler, s, t, p,
-                            computeLambda, lodbias, rgba, faces);
+   switch (mode) {
+   case PIPE_TEX_WRAP_CLAMP:
+      return wrap_linear_unorm_clamp;
+   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+      return wrap_linear_unorm_clamp_to_border;
+   default:
+      assert(0);
+      return wrap_linear_unorm_clamp;
+   }
 }
 
 
-static INLINE void
-sp_get_samples_3d(const struct tgsi_sampler *tgsi_sampler,
-                  const float s[QUAD_SIZE],
-                  const float t[QUAD_SIZE],
-                  const float p[QUAD_SIZE],
-                  boolean computeLambda,
-                  float lodbias,
-                  float rgba[NUM_CHANNELS][QUAD_SIZE])
+static wrap_linear_func
+get_linear_wrap(unsigned mode)
 {
-   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
-   const struct softpipe_context *sp = samp->sp;
-   const uint unit = samp->unit;
-   const struct pipe_texture *texture = sp->texture[unit];
-   const struct pipe_sampler_state *sampler = sp->sampler[unit];
-   /* get/map pipe_surfaces corresponding to 3D tex slices */
-   unsigned level0, level1, j, imgFilter;
-   int width, height, depth;
-   float levelBlend;
-   const uint face = 0;
-
-   choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias,
-                        &level0, &level1, &levelBlend, &imgFilter);
-
-   assert(sampler->normalized_coords);
-
-   width = texture->width[level0];
-   height = texture->height[level0];
-   depth = texture->depth[level0];
-
-   assert(width > 0);
-   assert(height > 0);
-   assert(depth > 0);
-
-   switch (imgFilter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      {
-         int x[4], y[4], z[4];
-         nearest_texcoord_4(sampler->wrap_s, s, width, x);
-         nearest_texcoord_4(sampler->wrap_t, t, height, y);
-         nearest_texcoord_4(sampler->wrap_r, p, depth, z);
-         for (j = 0; j < QUAD_SIZE; j++) {
-            get_texel(tgsi_sampler, face, level0, x[j], y[j], z[j], rgba, j);
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               unsigned c;
-               x[j] /= 2;
-               y[j] /= 2;
-               z[j] /= 2;
-               get_texel(tgsi_sampler, face, level1, x[j], y[j], z[j], rgba2, j);
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba2[c][j], rgba[c][j]);
-               }
-            }
-         }
-      }
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-   case PIPE_TEX_FILTER_ANISO:
-      {
-         int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4];
-         float xw[4], yw[4], zw[4]; /* interpolation weights */
-         linear_texcoord_4(sampler->wrap_s, s, width,  x0, x1, xw);
-         linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw);
-         linear_texcoord_4(sampler->wrap_r, p, depth,  z0, z1, zw);
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int c;
-            float tx0[4][4], tx1[4][4];
-            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z0[j], tx0, 0);
-            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z0[j], tx0, 1);
-            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z0[j], tx0, 2);
-            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z0[j], tx0, 3);
-            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z1[j], tx1, 0);
-            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z1[j], tx1, 1);
-            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z1[j], tx1, 2);
-            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z1[j], tx1, 3);
-
-            /* interpolate R, G, B, A */
-            for (c = 0; c < 4; c++) {
-               rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j],
-                                    tx0[c][0], tx0[c][1],
-                                    tx0[c][2], tx0[c][3],
-                                    tx1[c][0], tx1[c][1],
-                                    tx1[c][2], tx1[c][3]);
-            }
-
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               x0[j] /= 2;
-               y0[j] /= 2;
-               z0[j] /= 2;
-               x1[j] /= 2;
-               y1[j] /= 2;
-               z1[j] /= 2;
-               get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z0[j], tx0, 0);
-               get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z0[j], tx0, 1);
-               get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z0[j], tx0, 2);
-               get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z0[j], tx0, 3);
-               get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z1[j], tx1, 0);
-               get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z1[j], tx1, 1);
-               get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z1[j], tx1, 2);
-               get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z1[j], tx1, 3);
-
-               /* interpolate R, G, B, A */
-               for (c = 0; c < 4; c++) {
-                  rgba2[c][j] = lerp_3d(xw[j], yw[j], zw[j],
-                                        tx0[c][0], tx0[c][1],
-                                        tx0[c][2], tx0[c][3],
-                                        tx1[c][0], tx1[c][1],
-                                        tx1[c][2], tx1[c][3]);
-               }
-
-               /* blend mipmap levels */
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
-               }
-            }
-         }
-      }
-      break;
+   switch (mode) {
+   case PIPE_TEX_WRAP_REPEAT:
+      return wrap_linear_repeat;
+   case PIPE_TEX_WRAP_CLAMP:
+      return wrap_linear_clamp;
+   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+      return wrap_linear_clamp_to_edge;
+   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+      return wrap_linear_clamp_to_border;
+   case PIPE_TEX_WRAP_MIRROR_REPEAT:
+      return wrap_linear_mirror_repeat;
+   case PIPE_TEX_WRAP_MIRROR_CLAMP:
+      return wrap_linear_mirror_clamp;
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+      return wrap_linear_mirror_clamp_to_edge;
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+      return wrap_linear_mirror_clamp_to_border;
    default:
       assert(0);
+      return wrap_linear_repeat;
    }
 }
 
 
-static void
-sp_get_samples_cube(const struct tgsi_sampler *sampler,
-                    const float s[QUAD_SIZE],
-                    const float t[QUAD_SIZE],
-                    const float p[QUAD_SIZE],
-                    boolean computeLambda,
-                    float lodbias,
-                    float rgba[NUM_CHANNELS][QUAD_SIZE])
+static compute_lambda_func
+get_lambda_func(const union sp_sampler_key key)
 {
-   unsigned faces[QUAD_SIZE], j;
-   float ssss[4], tttt[4];
-   for (j = 0; j < QUAD_SIZE; j++) {
-      faces[j] = choose_cube_face(s[j], t[j], p[j], ssss + j, tttt + j);
+   if (key.bits.processor == TGSI_PROCESSOR_VERTEX)
+      return compute_lambda_vert;
+   
+   switch (key.bits.target) {
+   case PIPE_TEXTURE_1D:
+      return compute_lambda_1d;
+   case PIPE_TEXTURE_2D:
+   case PIPE_TEXTURE_CUBE:
+      return compute_lambda_2d;
+   case PIPE_TEXTURE_3D:
+      return compute_lambda_3d;
+   default:
+      assert(0);
+      return compute_lambda_1d;
    }
-   sp_get_samples_2d_common(sampler, ssss, tttt, NULL,
-                            computeLambda, lodbias, rgba, faces);
 }
 
 
-static void
-sp_get_samples_rect(const struct tgsi_sampler *tgsi_sampler,
-                    const float s[QUAD_SIZE],
-                    const float t[QUAD_SIZE],
-                    const float p[QUAD_SIZE],
-                    boolean computeLambda,
-                    float lodbias,
-                    float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
-   const struct softpipe_context *sp = samp->sp;
-   const uint unit = samp->unit;
-   const struct pipe_texture *texture = sp->texture[unit];
-   const struct pipe_sampler_state *sampler = sp->sampler[unit];
-   const uint face = 0;
-   unsigned level0, level1, j, imgFilter;
-   int width, height;
-   float levelBlend;
-
-   choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias,
-                        &level0, &level1, &levelBlend, &imgFilter);
-
-   /* texture RECTS cannot be mipmapped */
-   assert(level0 == level1);
-
-   width = texture->width[level0];
-   height = texture->height[level0];
-
-   assert(width > 0);
-
-   switch (imgFilter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      {
-         int x[4], y[4];
-         nearest_texcoord_unnorm_4(sampler->wrap_s, s, width, x);
-         nearest_texcoord_unnorm_4(sampler->wrap_t, t, height, y);
-         for (j = 0; j < QUAD_SIZE; j++) {
-            get_texel(tgsi_sampler, face, level0, x[j], y[j], 0, rgba, j);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare(sampler, rgba, p, j);
-            }
-         }
-      }
+static filter_func
+get_img_filter(const union sp_sampler_key key,
+               unsigned filter,
+               const struct pipe_sampler_state *sampler)
+{
+   switch (key.bits.target) {
+   case PIPE_TEXTURE_1D:
+      if (filter == PIPE_TEX_FILTER_NEAREST) 
+         return img_filter_1d_nearest;
+      else
+         return img_filter_1d_linear;
       break;
-   case PIPE_TEX_FILTER_LINEAR:
-   case PIPE_TEX_FILTER_ANISO:
+   case PIPE_TEXTURE_2D:
+      /* Try for fast path:
+       */
+      if (key.bits.is_pot &&
+          sampler->wrap_s == sampler->wrap_t &&
+          sampler->normalized_coords) 
       {
-         int x0[4], y0[4], x1[4], y1[4];
-         float xw[4], yw[4]; /* weights */
-         linear_texcoord_unnorm_4(sampler->wrap_s, s, width,  x0, x1, xw);
-         linear_texcoord_unnorm_4(sampler->wrap_t, t, height, y0, y1, yw);
-         for (j = 0; j < QUAD_SIZE; j++) {
-            float tx[4][4]; /* texels */
-            int c;
-            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], 0, tx, 0);
-            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], 0, tx, 1);
-            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], 0, tx, 2);
-            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], 0, tx, 3);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare4(sampler, tx, p);
+         switch (sampler->wrap_s) {
+         case PIPE_TEX_WRAP_REPEAT:
+            switch (filter) {
+            case PIPE_TEX_FILTER_NEAREST:
+               return img_filter_2d_nearest_repeat_POT;
+            case PIPE_TEX_FILTER_LINEAR:
+               return img_filter_2d_linear_repeat_POT;
+            default:
+               break;
             }
-            for (c = 0; c < 4; c++) {
-               rgba[c][j] = lerp_2d(xw[j], yw[j],
-                                    tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
+            break;
+         case PIPE_TEX_WRAP_CLAMP:
+            switch (filter) {
+            case PIPE_TEX_FILTER_NEAREST:
+               return img_filter_2d_nearest_clamp_POT;
+            default:
+               break;
             }
          }
       }
+      /* Otherwise use default versions:
+       */
+      if (filter == PIPE_TEX_FILTER_NEAREST) 
+         return img_filter_2d_nearest;
+      else
+         return img_filter_2d_linear;
+      break;
+   case PIPE_TEXTURE_CUBE:
+      if (filter == PIPE_TEX_FILTER_NEAREST) 
+         return img_filter_cube_nearest;
+      else
+         return img_filter_cube_linear;
+      break;
+   case PIPE_TEXTURE_3D:
+      if (filter == PIPE_TEX_FILTER_NEAREST) 
+         return img_filter_3d_nearest;
+      else
+         return img_filter_3d_linear;
       break;
    default:
       assert(0);
+      return img_filter_1d_nearest;
    }
 }
 
 
 /**
- * Common code for vertex/fragment program texture sampling.
+ * Bind the given texture object and texture cache to the sampler varient.
  */
-static INLINE void
-sp_get_samples(struct tgsi_sampler *tgsi_sampler,
-               const float s[QUAD_SIZE],
-               const float t[QUAD_SIZE],
-               const float p[QUAD_SIZE],
-               boolean computeLambda,
-               float lodbias,
-               float rgba[NUM_CHANNELS][QUAD_SIZE])
+void
+sp_sampler_varient_bind_texture( struct sp_sampler_varient *samp,
+                                 struct softpipe_tex_tile_cache *tex_cache,
+                                 const struct pipe_texture *texture )
 {
-   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
-   const struct softpipe_context *sp = samp->sp;
-   const uint unit = samp->unit;
-   const struct pipe_texture *texture = sp->texture[unit];
-   const struct pipe_sampler_state *sampler = sp->sampler[unit];
-
-   if (!texture)
-      return;
+   const struct pipe_sampler_state *sampler = samp->sampler;
 
-   switch (texture->target) {
-   case PIPE_TEXTURE_1D:
-      assert(sampler->normalized_coords);
-      sp_get_samples_1d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba);
-      break;
-   case PIPE_TEXTURE_2D:
-      if (sampler->normalized_coords)
-         sp_get_samples_2d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba);
-      else
-         sp_get_samples_rect(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba);
-      break;
-   case PIPE_TEXTURE_3D:
-      assert(sampler->normalized_coords);
-      sp_get_samples_3d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba);
-      break;
-   case PIPE_TEXTURE_CUBE:
-      assert(sampler->normalized_coords);
-      sp_get_samples_cube(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba);
-      break;
-   default:
-      assert(0);
-   }
-
-#if 0 /* DEBUG */
-   {
-      int i;
-      printf("Sampled at %f, %f, %f:\n", s[0], t[0], p[0]);
-      for (i = 0; i < 4; i++) {
-         printf("Frag %d: %f %f %f %f\n", i,
-                rgba[0][i],
-                rgba[1][i],
-                rgba[2][i],
-                rgba[3][i]);
-      }
-   }
-#endif
+   samp->texture = texture;
+   samp->cache = tex_cache;
+   samp->xpot = util_unsigned_logbase2( texture->width[0] );
+   samp->ypot = util_unsigned_logbase2( texture->height[0] );
+   samp->level = CLAMP((int) sampler->min_lod, 0, (int) texture->last_level);
 }
 
 
-/**
- * Called via tgsi_sampler::get_samples() when running a fragment shader.
- * Get four filtered RGBA values from the sampler's texture.
- */
 void
-sp_get_samples_fragment(struct tgsi_sampler *tgsi_sampler,
-                        const float s[QUAD_SIZE],
-                        const float t[QUAD_SIZE],
-                        const float p[QUAD_SIZE],
-                        float lodbias,
-                        float rgba[NUM_CHANNELS][QUAD_SIZE])
+sp_sampler_varient_destroy( struct sp_sampler_varient *samp )
 {
-   sp_get_samples(tgsi_sampler, s, t, p, TRUE, lodbias, rgba);
+   FREE(samp);
 }
 
 
 /**
- * Called via tgsi_sampler::get_samples() when running a vertex shader.
- * Get four filtered RGBA values from the sampler's texture.
+ * Create a sampler varient for a given set of non-orthogonal state.
  */
-void
-sp_get_samples_vertex(struct tgsi_sampler *tgsi_sampler,
-                      const float s[QUAD_SIZE],
-                      const float t[QUAD_SIZE],
-                      const float p[QUAD_SIZE],
-                      float lodbias,
-                      float rgba[NUM_CHANNELS][QUAD_SIZE])
+struct sp_sampler_varient *
+sp_create_sampler_varient( const struct pipe_sampler_state *sampler,
+                           const union sp_sampler_key key )
 {
-   sp_get_samples(tgsi_sampler, s, t, p, FALSE, lodbias, rgba);
+   struct sp_sampler_varient *samp = CALLOC_STRUCT(sp_sampler_varient);
+   if (!samp)
+      return NULL;
+
+   samp->sampler = sampler;
+   samp->key = key;
+
+   /* Note that (for instance) linear_texcoord_s and
+    * nearest_texcoord_s may be active at the same time, if the
+    * sampler min_img_filter differs from its mag_img_filter.
+    */
+   if (sampler->normalized_coords) {
+      samp->linear_texcoord_s = get_linear_wrap( sampler->wrap_s );
+      samp->linear_texcoord_t = get_linear_wrap( sampler->wrap_t );
+      samp->linear_texcoord_p = get_linear_wrap( sampler->wrap_r );
+      
+      samp->nearest_texcoord_s = get_nearest_wrap( sampler->wrap_s );
+      samp->nearest_texcoord_t = get_nearest_wrap( sampler->wrap_t );
+      samp->nearest_texcoord_p = get_nearest_wrap( sampler->wrap_r );
+   }
+   else {
+      samp->linear_texcoord_s = get_linear_unorm_wrap( sampler->wrap_s );
+      samp->linear_texcoord_t = get_linear_unorm_wrap( sampler->wrap_t );
+      samp->linear_texcoord_p = get_linear_unorm_wrap( sampler->wrap_r );
+      
+      samp->nearest_texcoord_s = get_nearest_unorm_wrap( sampler->wrap_s );
+      samp->nearest_texcoord_t = get_nearest_unorm_wrap( sampler->wrap_t );
+      samp->nearest_texcoord_p = get_nearest_unorm_wrap( sampler->wrap_r );
+   }
+   
+   samp->compute_lambda = get_lambda_func( key );
+
+   samp->min_img_filter = get_img_filter(key, sampler->min_img_filter, sampler);
+   samp->mag_img_filter = get_img_filter(key, sampler->mag_img_filter, sampler);
+
+   switch (sampler->min_mip_filter) {
+   case PIPE_TEX_MIPFILTER_NONE:
+      if (sampler->min_img_filter == sampler->mag_img_filter) 
+         samp->mip_filter = samp->min_img_filter;         
+      else
+         samp->mip_filter = mip_filter_none;
+      break;
+
+   case PIPE_TEX_MIPFILTER_NEAREST:
+      samp->mip_filter = mip_filter_nearest;
+      break;
+
+   case PIPE_TEX_MIPFILTER_LINEAR:
+      if (key.bits.is_pot &&
+          sampler->min_img_filter == sampler->mag_img_filter &&
+          sampler->normalized_coords &&
+          sampler->wrap_s == PIPE_TEX_WRAP_REPEAT &&
+          sampler->wrap_t == PIPE_TEX_WRAP_REPEAT &&
+          sampler->min_img_filter == PIPE_TEX_FILTER_LINEAR)
+      {
+         samp->mip_filter = mip_filter_linear_2d_linear_repeat_POT;
+      }
+      else 
+      {
+         samp->mip_filter = mip_filter_linear;
+      }
+      break;
+   }
+
+   if (sampler->compare_mode != FALSE) {
+      samp->compare = sample_compare;
+   }
+   else {
+      /* Skip compare operation by promoting the mip_filter function
+       * pointer:
+       */
+      samp->compare = samp->mip_filter;
+   }
+   
+   if (key.bits.target == PIPE_TEXTURE_CUBE) {
+      samp->base.get_samples = sample_cube;
+   }
+   else {
+      samp->faces[0] = 0;
+      samp->faces[1] = 0;
+      samp->faces[2] = 0;
+      samp->faces[3] = 0;
+
+      /* Skip cube face determination by promoting the compare
+       * function pointer:
+       */
+      samp->base.get_samples = samp->compare;
+   }
+
+   return samp;
 }
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.h b/src/gallium/drivers/softpipe/sp_tex_sample.h
index 40d8eb2c2a..b0797711d3 100644
--- a/src/gallium/drivers/softpipe/sp_tex_sample.h
+++ b/src/gallium/drivers/softpipe/sp_tex_sample.h
@@ -31,43 +31,122 @@
 
 #include "tgsi/tgsi_exec.h"
 
+struct sp_sampler_varient;
+
+typedef void (*wrap_nearest_func)(const float s[4],
+                                  unsigned size,
+                                  int icoord[4]);
+
+typedef void (*wrap_linear_func)(const float s[4], 
+                                 unsigned size,
+                                 int icoord0[4],
+                                 int icoord1[4],
+                                 float w[4]);
+
+typedef float (*compute_lambda_func)(const struct sp_sampler_varient *sampler,
+                                     const float s[QUAD_SIZE],
+                                     const float t[QUAD_SIZE],
+                                     const float p[QUAD_SIZE],
+                                     float lodbias);
+
+typedef void (*filter_func)(struct tgsi_sampler *tgsi_sampler,
+                            const float s[QUAD_SIZE],
+                            const float t[QUAD_SIZE],
+                            const float p[QUAD_SIZE],
+                            float lodbias,
+                            float rgba[NUM_CHANNELS][QUAD_SIZE]);
+
+
+union sp_sampler_key {
+   struct {
+      unsigned target:3;
+      unsigned is_pot:1;
+      unsigned processor:2;
+      unsigned unit:4;
+      unsigned pad:22;
+   } bits;
+   unsigned value;
+};
 
 /**
  * Subclass of tgsi_sampler
  */
-struct sp_shader_sampler
+struct sp_sampler_varient
 {
    struct tgsi_sampler base;  /**< base class */
 
-   uint unit;
-   struct softpipe_context *sp;
-   struct softpipe_tile_cache *cache;
+   union sp_sampler_key key;
+
+   /* The owner of this struct:
+    */
+   const struct pipe_sampler_state *sampler;
+
+
+   /* Currently bound texture:
+    */
+   const struct pipe_texture *texture;
+   struct softpipe_tex_tile_cache *cache;
+
+   unsigned processor;
+
+   /* For sp_get_samples_2d_linear_POT:
+    */
+   unsigned xpot;
+   unsigned ypot;
+   unsigned level;
+
+   unsigned faces[4];
+   
+   wrap_nearest_func nearest_texcoord_s;
+   wrap_nearest_func nearest_texcoord_t;
+   wrap_nearest_func nearest_texcoord_p;
+
+   wrap_linear_func linear_texcoord_s;
+   wrap_linear_func linear_texcoord_t;
+   wrap_linear_func linear_texcoord_p;
+
+   filter_func min_img_filter;
+   filter_func mag_img_filter;
+
+   compute_lambda_func compute_lambda;
+
+   filter_func mip_filter;
+   filter_func compare;
+   
+   /* Linked list:
+    */
+   struct sp_sampler_varient *next;
 };
 
+struct sp_sampler;
 
+/* Create a sampler varient for a given set of non-orthogonal state.  Currently the 
+ */
+struct sp_sampler_varient *
+sp_create_sampler_varient( const struct pipe_sampler_state *sampler,
+                           const union sp_sampler_key key );
 
-static INLINE const struct sp_shader_sampler *
-sp_shader_sampler(const struct tgsi_sampler *sampler)
-{
-   return (const struct sp_shader_sampler *) sampler;
-}
+void sp_sampler_varient_bind_texture( struct sp_sampler_varient *varient,
+                                      struct softpipe_tex_tile_cache *tex_cache,
+                                      const struct pipe_texture *tex );
 
+void sp_sampler_varient_destroy( struct sp_sampler_varient * );
 
-extern void
-sp_get_samples_fragment(struct tgsi_sampler *tgsi_sampler,
-                        const float s[QUAD_SIZE],
-                        const float t[QUAD_SIZE],
-                        const float p[QUAD_SIZE],
-                        float lodbias,
-                        float rgba[NUM_CHANNELS][QUAD_SIZE]);
+
+
+static INLINE struct sp_sampler_varient *
+sp_sampler_varient(const struct tgsi_sampler *sampler)
+{
+   return (struct sp_sampler_varient *) sampler;
+}
 
 extern void
-sp_get_samples_vertex(struct tgsi_sampler *tgsi_sampler,
-                      const float s[QUAD_SIZE],
-                      const float t[QUAD_SIZE],
-                      const float p[QUAD_SIZE],
-                      float lodbias,
-                      float rgba[NUM_CHANNELS][QUAD_SIZE]);
+sp_get_samples(struct tgsi_sampler *tgsi_sampler,
+               const float s[QUAD_SIZE],
+               const float t[QUAD_SIZE],
+               const float p[QUAD_SIZE],
+               float lodbias,
+               float rgba[NUM_CHANNELS][QUAD_SIZE]);
 
 
 #endif /* SP_TEX_SAMPLE_H */
diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
new file mode 100644
index 0000000000..407a22a9f4
--- /dev/null
+++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
@@ -0,0 +1,273 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * Texture tile caching.
+ *
+ * Author:
+ *    Brian Paul
+ */
+
+#include "pipe/p_inlines.h"
+#include "util/u_memory.h"
+#include "util/u_tile.h"
+#include "sp_context.h"
+#include "sp_surface.h"
+#include "sp_texture.h"
+#include "sp_tex_tile_cache.h"
+
+   
+
+struct softpipe_tex_tile_cache *
+sp_create_tex_tile_cache( struct pipe_screen *screen )
+{
+   struct softpipe_tex_tile_cache *tc;
+   uint pos;
+
+   tc = CALLOC_STRUCT( softpipe_tex_tile_cache );
+   if (tc) {
+      tc->screen = screen;
+      for (pos = 0; pos < NUM_ENTRIES; pos++) {
+         tc->entries[pos].addr.bits.invalid = 1;
+      }
+      tc->last_tile = &tc->entries[0]; /* any tile */
+   }
+   return tc;
+}
+
+
+void
+sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc)
+{
+   struct pipe_screen *screen;
+   uint pos;
+
+   for (pos = 0; pos < NUM_ENTRIES; pos++) {
+      /*assert(tc->entries[pos].x < 0);*/
+   }
+   if (tc->transfer) {
+      screen = tc->transfer->texture->screen;
+      screen->tex_transfer_destroy(tc->transfer);
+   }
+   if (tc->tex_trans) {
+      screen = tc->tex_trans->texture->screen;
+      screen->tex_transfer_destroy(tc->tex_trans);
+   }
+
+   FREE( tc );
+}
+
+
+
+
+void
+sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc)
+{
+   if (tc->tex_trans && !tc->tex_trans_map)
+      tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans);
+}
+
+
+void
+sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc)
+{
+   if (tc->tex_trans_map) {
+      tc->screen->transfer_unmap(tc->screen, tc->tex_trans);
+      tc->tex_trans_map = NULL;
+   }
+}
+
+/**
+ * Invalidate all cached tiles for the cached texture.
+ * Should be called when the texture is modified.
+ */
+void
+sp_tex_tile_cache_validate_texture(struct softpipe_tex_tile_cache *tc)
+{
+   unsigned i;
+
+   assert(tc);
+   assert(tc->texture);
+
+   for (i = 0; i < NUM_ENTRIES; i++) {
+      tc->entries[i].addr.bits.invalid = 1;
+   }
+}
+
+/**
+ * Specify the texture to cache.
+ */
+void
+sp_tex_tile_cache_set_texture(struct softpipe_tex_tile_cache *tc,
+                          struct pipe_texture *texture)
+{
+   uint i;
+
+   assert(!tc->transfer);
+
+   if (tc->texture != texture) {
+      pipe_texture_reference(&tc->texture, texture);
+
+      if (tc->tex_trans) {
+         struct pipe_screen *screen = tc->tex_trans->texture->screen;
+         
+         if (tc->tex_trans_map) {
+            screen->transfer_unmap(screen, tc->tex_trans);
+            tc->tex_trans_map = NULL;
+         }
+
+         screen->tex_transfer_destroy(tc->tex_trans);
+         tc->tex_trans = NULL;
+      }
+
+      /* mark as entries as invalid/empty */
+      /* XXX we should try to avoid this when the teximage hasn't changed */
+      for (i = 0; i < NUM_ENTRIES; i++) {
+         tc->entries[i].addr.bits.invalid = 1;
+      }
+
+      tc->tex_face = -1; /* any invalid value here */
+   }
+}
+
+
+
+
+/**
+ * Flush the tile cache: write all dirty tiles back to the transfer.
+ * any tiles "flagged" as cleared will be "really" cleared.
+ */
+void
+sp_flush_tex_tile_cache(struct softpipe_tex_tile_cache *tc)
+{
+   int pos;
+
+   if (tc->texture) {
+      /* caching a texture, mark all entries as empty */
+      for (pos = 0; pos < NUM_ENTRIES; pos++) {
+         tc->entries[pos].addr.bits.invalid = 1;
+      }
+      tc->tex_face = -1;
+   }
+
+}
+
+
+/**
+ * Given the texture face, level, zslice, x and y values, compute
+ * the cache entry position/index where we'd hope to find the
+ * cached texture tile.
+ * This is basically a direct-map cache.
+ * XXX There's probably lots of ways in which we can improve this.
+ */
+static INLINE uint
+tex_cache_pos( union tex_tile_address addr )
+{
+   uint entry = (addr.bits.x + 
+                 addr.bits.y * 9 + 
+                 addr.bits.z * 3 + 
+                 addr.bits.face + 
+                 addr.bits.level * 7);
+
+   return entry % NUM_ENTRIES;
+}
+
+/**
+ * Similar to sp_get_cached_tile() but for textures.
+ * Tiles are read-only and indexed with more params.
+ */
+const struct softpipe_tex_cached_tile *
+sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, 
+                        union tex_tile_address addr )
+{
+   struct pipe_screen *screen = tc->screen;
+   struct softpipe_tex_cached_tile *tile;
+   
+   tile = tc->entries + tex_cache_pos( addr );
+
+   if (addr.value != tile->addr.value) {
+
+      /* cache miss.  Most misses are because we've invaldiated the
+       * texture cache previously -- most commonly on binding a new
+       * texture.  Currently we effectively flush the cache on texture
+       * bind.
+       */
+#if 0
+      _debug_printf("miss at %u:  x=%d y=%d z=%d face=%d level=%d\n"
+                    "   tile %u:  x=%d y=%d z=%d face=%d level=%d\n",
+                    pos, x/TILE_SIZE, y/TILE_SIZE, z, face, level,
+                    pos, tile->addr.bits.x, tile->addr.bits.y, tile->z, tile->face, tile->level);
+#endif
+
+      /* check if we need to get a new transfer */
+      if (!tc->tex_trans ||
+          tc->tex_face != addr.bits.face ||
+          tc->tex_level != addr.bits.level ||
+          tc->tex_z != addr.bits.z) {
+         /* get new transfer (view into texture) */
+
+         if (tc->tex_trans) {
+            if (tc->tex_trans_map) {
+               tc->screen->transfer_unmap(tc->screen, tc->tex_trans);
+               tc->tex_trans_map = NULL;
+            }
+
+            screen->tex_transfer_destroy(tc->tex_trans);
+            tc->tex_trans = NULL;
+         }
+
+         tc->tex_trans = 
+            screen->get_tex_transfer(screen, tc->texture, 
+                                     addr.bits.face, 
+                                     addr.bits.level, 
+                                     addr.bits.z, 
+                                     PIPE_TRANSFER_READ, 0, 0,
+                                     tc->texture->width[addr.bits.level],
+                                     tc->texture->height[addr.bits.level]);
+
+         tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans);
+
+         tc->tex_face = addr.bits.face;
+         tc->tex_level = addr.bits.level;
+         tc->tex_z = addr.bits.z;
+      }
+
+      /* get tile from the transfer (view into texture) */
+      pipe_get_tile_rgba(tc->tex_trans,
+                         addr.bits.x * TILE_SIZE, 
+                         addr.bits.y * TILE_SIZE,
+                         TILE_SIZE, TILE_SIZE,
+                         (float *) tile->data.color);
+      tile->addr = addr;
+   }
+
+   tc->last_tile = tile;
+   return tile;
+}
+
+
+
diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h
new file mode 100644
index 0000000000..ac6886a3df
--- /dev/null
+++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h
@@ -0,0 +1,155 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef SP_TEX_TILE_CACHE_H
+#define SP_TEX_TILE_CACHE_H
+
+
+#include "pipe/p_compiler.h"
+
+
+struct softpipe_context;
+struct softpipe_tex_tile_cache;
+
+
+/**
+ * Cache tile size (width and height). This needs to be a power of two.
+ */
+#define TILE_SIZE 64
+
+
+/* If we need to support > 4096, just expand this to be a 64 bit
+ * union, or consider tiling in Z as well.
+ */
+union tex_tile_address {
+   struct {
+      unsigned x:6;             /* 4096 / TILE_SIZE */
+      unsigned y:6;             /* 4096 / TILE_SIZE */
+      unsigned z:12;            /* 4096 -- z not tiled */
+      unsigned face:3;
+      unsigned level:4;
+      unsigned invalid:1;
+   } bits;
+   unsigned value;
+};
+
+
+struct softpipe_tex_cached_tile
+{
+   union tex_tile_address addr;
+   union {
+      float color[TILE_SIZE][TILE_SIZE][4];
+   } data;
+};
+
+#define NUM_ENTRIES 50
+
+struct softpipe_tex_tile_cache
+{
+   struct pipe_screen *screen;
+   struct pipe_transfer *transfer;
+   void *transfer_map;
+
+   struct pipe_texture *texture;  /**< if caching a texture */
+   unsigned timestamp;
+
+   struct softpipe_tex_cached_tile entries[NUM_ENTRIES];
+
+   struct pipe_transfer *tex_trans;
+   void *tex_trans_map;
+   int tex_face, tex_level, tex_z;
+
+   struct softpipe_tex_cached_tile *last_tile;  /**< most recently retrieved tile */
+};
+
+
+extern struct softpipe_tex_tile_cache *
+sp_create_tex_tile_cache( struct pipe_screen *screen );
+
+extern void
+sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc);
+
+
+extern void
+sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc);
+
+extern void
+sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc);
+
+extern void
+sp_tex_tile_cache_set_texture(struct softpipe_tex_tile_cache *tc,
+                          struct pipe_texture *texture);
+
+void
+sp_tex_tile_cache_validate_texture(struct softpipe_tex_tile_cache *tc);
+
+extern void
+sp_flush_tex_tile_cache(struct softpipe_tex_tile_cache *tc);
+
+
+
+extern const struct softpipe_tex_cached_tile *
+sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, 
+                         union tex_tile_address addr );
+
+static INLINE union tex_tile_address
+tex_tile_address( unsigned x,
+		  unsigned y,
+		  unsigned z,
+		  unsigned face,
+		  unsigned level )
+{
+   union tex_tile_address addr;
+
+   addr.value = 0;
+   addr.bits.x = x / TILE_SIZE;
+   addr.bits.y = y / TILE_SIZE;
+   addr.bits.z = z;
+   addr.bits.face = face;
+   addr.bits.level = level;
+      
+   return addr;
+}
+
+/* Quickly retrieve tile if it matches last lookup.
+ */
+static INLINE const struct softpipe_tex_cached_tile *
+sp_get_cached_tile_tex(struct softpipe_tex_tile_cache *tc, 
+                         union tex_tile_address addr )
+{
+   if (tc->last_tile->addr.value == addr.value)
+      return tc->last_tile;
+
+   return sp_find_cached_tile_tex( tc, addr );
+}
+
+
+
+
+
+#endif /* SP_TEX_TILE_CACHE_H */
+
diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index 70f0932431..7caf2928b4 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -30,26 +30,21 @@
   *   Michel Dänzer <michel@tungstengraphics.com>
   */
 
-#include "pipe/p_context.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 
 #include "sp_context.h"
 #include "sp_state.h"
 #include "sp_texture.h"
-#include "sp_tile_cache.h"
 #include "sp_screen.h"
 #include "sp_winsys.h"
 
 
-/* Simple, maximally packed layout.
- */
-
-
-/* Conventional allocation path for non-display textures:
+/**
+ * Conventional allocation path for non-display textures:
+ * Use a simple, maximally packed layout.
  */
 static boolean
 softpipe_texture_layout(struct pipe_screen *screen,
@@ -89,6 +84,10 @@ softpipe_texture_layout(struct pipe_screen *screen,
    return spt->buffer != NULL;
 }
 
+
+/**
+ * Texture layout for simple color buffers.
+ */
 static boolean
 softpipe_displaytarget_layout(struct pipe_screen *screen,
                               struct softpipe_texture * spt)
@@ -112,21 +111,25 @@ softpipe_displaytarget_layout(struct pipe_screen *screen,
 }
 
 
-
-
-
+/**
+ * Create new pipe_texture given the template information.
+ */
 static struct pipe_texture *
 softpipe_texture_create(struct pipe_screen *screen,
-                        const struct pipe_texture *templat)
+                        const struct pipe_texture *template)
 {
    struct softpipe_texture *spt = CALLOC_STRUCT(softpipe_texture);
    if (!spt)
       return NULL;
 
-   spt->base = *templat;
+   spt->base = *template;
    pipe_reference_init(&spt->base.reference, 1);
    spt->base.screen = screen;
 
+   spt->pot = (util_is_power_of_two(template->width[0]) &&
+               util_is_power_of_two(template->height[0]) &&
+               util_is_power_of_two(template->depth[0]));
+
    if (spt->base.tex_usage & (PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
                               PIPE_TEXTURE_USAGE_PRIMARY)) {
       if (!softpipe_displaytarget_layout(screen, spt))
@@ -145,6 +148,9 @@ softpipe_texture_create(struct pipe_screen *screen,
 }
 
 
+/**
+ * Create a new pipe_texture which wraps an existing buffer.
+ */
 static struct pipe_texture *
 softpipe_texture_blanket(struct pipe_screen * screen,
                          const struct pipe_texture *base,
@@ -188,6 +194,9 @@ softpipe_texture_destroy(struct pipe_texture *pt)
 }
 
 
+/**
+ * Get a pipe_surface "view" into a texture.
+ */
 static struct pipe_surface *
 softpipe_get_tex_surface(struct pipe_screen *screen,
                          struct pipe_texture *pt,
@@ -222,6 +231,13 @@ softpipe_get_tex_surface(struct pipe_screen *screen,
       if (ps->usage & PIPE_BUFFER_USAGE_GPU_READ)
          ps->usage |= PIPE_BUFFER_USAGE_CPU_READ;
 
+      if (ps->usage & (PIPE_BUFFER_USAGE_CPU_WRITE |
+                       PIPE_BUFFER_USAGE_GPU_WRITE)) {
+         /* Mark the surface as dirty.  The tile cache will look for this. */
+         spt->timestamp++;
+         softpipe_screen(screen)->timestamp++;
+      }
+
       ps->face = face;
       ps->level = level;
       ps->zslice = zslice;
@@ -241,6 +257,9 @@ softpipe_get_tex_surface(struct pipe_screen *screen,
 }
 
 
+/**
+ * Free a pipe_surface which was created with softpipe_get_tex_surface().
+ */
 static void 
 softpipe_tex_surface_destroy(struct pipe_surface *surf)
 {
@@ -254,6 +273,18 @@ softpipe_tex_surface_destroy(struct pipe_surface *surf)
 }
 
 
+/**
+ * Geta pipe_transfer object which is used for moving data in/out of
+ * a texture object.
+ * \param face  one of PIPE_TEX_FACE_x or 0
+ * \param level  texture mipmap level
+ * \param zslice  2D slice of a 3D texture
+ * \param usage  one of PIPE_TRANSFER_READ/WRITE/READ_WRITE
+ * \param x  X position of region to read/write
+ * \param y  Y position of region to read/write
+ * \param width  width of region to read/write
+ * \param height  height of region to read/write
+ */
 static struct pipe_transfer *
 softpipe_get_tex_transfer(struct pipe_screen *screen,
                           struct pipe_texture *texture,
@@ -303,6 +334,10 @@ softpipe_get_tex_transfer(struct pipe_screen *screen,
 }
 
 
+/**
+ * Free a pipe_transfer object which was created with
+ * softpipe_get_tex_transfer().
+ */
 static void 
 softpipe_tex_transfer_destroy(struct pipe_transfer *transfer)
 {
@@ -316,40 +351,33 @@ softpipe_tex_transfer_destroy(struct pipe_transfer *transfer)
 }
 
 
+/**
+ * Create memory mapping for given pipe_transfer object.
+ */
 static void *
 softpipe_transfer_map( struct pipe_screen *screen,
                        struct pipe_transfer *transfer )
 {
    ubyte *map, *xfer_map;
    struct softpipe_texture *spt;
-   unsigned flags = 0;
 
    assert(transfer->texture);
    spt = softpipe_texture(transfer->texture);
 
-   if (transfer->usage != PIPE_TRANSFER_READ) {
-      flags |= PIPE_BUFFER_USAGE_CPU_WRITE;
-   }
-
-   if (transfer->usage != PIPE_TRANSFER_WRITE) {
-      flags |= PIPE_BUFFER_USAGE_CPU_READ;
-   }
-
-   map = pipe_buffer_map(screen, spt->buffer, flags);
+   map = pipe_buffer_map(screen, spt->buffer, pipe_transfer_buffer_flags(transfer));
    if (map == NULL)
       return NULL;
 
    /* May want to different things here depending on read/write nature
     * of the map:
     */
-   if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) 
-   {
+   if (transfer->texture && (transfer->usage & PIPE_TRANSFER_WRITE)) {
       /* Do something to notify sharing contexts of a texture change.
        * In softpipe, that would mean flushing the texture cache.
        */
       softpipe_screen(screen)->timestamp++;
    }
-   
+
    xfer_map = map + softpipe_transfer(transfer)->offset +
       transfer->y / transfer->block.height * transfer->stride +
       transfer->x / transfer->block.width * transfer->block.size;
@@ -358,9 +386,12 @@ softpipe_transfer_map( struct pipe_screen *screen,
 }
 
 
+/**
+ * Unmap memory mapping for given pipe_transfer object.
+ */
 static void
 softpipe_transfer_unmap(struct pipe_screen *screen,
-                       struct pipe_transfer *transfer)
+                        struct pipe_transfer *transfer)
 {
    struct softpipe_texture *spt;
 
@@ -369,16 +400,63 @@ softpipe_transfer_unmap(struct pipe_screen *screen,
 
    pipe_buffer_unmap( screen, spt->buffer );
 
-   if (transfer->usage != PIPE_TRANSFER_READ) {
+   if (transfer->usage & PIPE_TRANSFER_WRITE) {
       /* Mark the texture as dirty to expire the tile caches. */
-      spt->modified = TRUE;
+      spt->timestamp++;
    }
 }
 
 
-void
-softpipe_init_texture_funcs(struct softpipe_context *sp)
+static struct pipe_video_surface*
+softpipe_video_surface_create(struct pipe_screen *screen,
+                              enum pipe_video_chroma_format chroma_format,
+                              unsigned width, unsigned height)
+{
+   struct softpipe_video_surface *sp_vsfc;
+   struct pipe_texture template;
+
+   assert(screen);
+   assert(width && height);
+
+   sp_vsfc = CALLOC_STRUCT(softpipe_video_surface);
+   if (!sp_vsfc)
+      return NULL;
+
+   pipe_reference_init(&sp_vsfc->base.reference, 1);
+   sp_vsfc->base.screen = screen;
+   sp_vsfc->base.chroma_format = chroma_format;
+   /*sp_vsfc->base.surface_format = PIPE_VIDEO_SURFACE_FORMAT_VUYA;*/
+   sp_vsfc->base.width = width;
+   sp_vsfc->base.height = height;
+
+   memset(&template, 0, sizeof(struct pipe_texture));
+   template.target = PIPE_TEXTURE_2D;
+   template.format = PIPE_FORMAT_X8R8G8B8_UNORM;
+   template.last_level = 0;
+   /* vl_mpeg12_mc_renderer expects this when it's initialized with pot_buffers=true */
+   template.width[0] = util_next_power_of_two(width);
+   template.height[0] = util_next_power_of_two(height);
+   template.depth[0] = 1;
+   pf_get_block(template.format, &template.block);
+   template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET;
+
+   sp_vsfc->tex = screen->texture_create(screen, &template);
+   if (!sp_vsfc->tex) {
+      FREE(sp_vsfc);
+      return NULL;
+   }
+
+   return &sp_vsfc->base;
+}
+
+
+static void
+softpipe_video_surface_destroy(struct pipe_video_surface *vsfc)
 {
+   struct softpipe_video_surface *sp_vsfc = softpipe_video_surface(vsfc);
+
+   pipe_texture_reference(&sp_vsfc->tex, NULL);
+   FREE(sp_vsfc);
 }
 
 
@@ -396,15 +474,22 @@ softpipe_init_screen_texture_funcs(struct pipe_screen *screen)
    screen->tex_transfer_destroy = softpipe_tex_transfer_destroy;
    screen->transfer_map = softpipe_transfer_map;
    screen->transfer_unmap = softpipe_transfer_unmap;
+
+   screen->video_surface_create = softpipe_video_surface_create;
+   screen->video_surface_destroy = softpipe_video_surface_destroy;
 }
 
 
+/**
+ * Return pipe_buffer handle and stride for given texture object.
+ * XXX used for???
+ */
 boolean
 softpipe_get_texture_buffer( struct pipe_texture *texture,
                              struct pipe_buffer **buf,
                              unsigned *stride )
 {
-   struct softpipe_texture *tex = (struct softpipe_texture *)texture;
+   struct softpipe_texture *tex = (struct softpipe_texture *) texture;
 
    if (!tex)
       return FALSE;
diff --git a/src/gallium/drivers/softpipe/sp_texture.h b/src/gallium/drivers/softpipe/sp_texture.h
index 893aa7d11d..2ef64e1e7c 100644
--- a/src/gallium/drivers/softpipe/sp_texture.h
+++ b/src/gallium/drivers/softpipe/sp_texture.h
@@ -30,6 +30,7 @@
 
 
 #include "pipe/p_state.h"
+#include "pipe/p_video_state.h"
 
 
 struct pipe_context;
@@ -48,7 +49,11 @@ struct softpipe_texture
     */
    struct pipe_buffer *buffer;
 
-   boolean modified;
+   /* True if texture images are power-of-two in all dimensions:
+    */
+   boolean pot;
+
+   unsigned timestamp;
 };
 
 struct softpipe_transfer
@@ -58,6 +63,15 @@ struct softpipe_transfer
    unsigned long offset;
 };
 
+struct softpipe_video_surface
+{
+   struct pipe_video_surface base;
+
+   /* The data is held here:
+    */
+   struct pipe_texture *tex;
+};
+
 
 /** cast wrappers */
 static INLINE struct softpipe_texture *
@@ -72,9 +86,12 @@ softpipe_transfer(struct pipe_transfer *pt)
    return (struct softpipe_transfer *) pt;
 }
 
+static INLINE struct softpipe_video_surface *
+softpipe_video_surface(struct pipe_video_surface *pvs)
+{
+   return (struct softpipe_video_surface *) pvs;
+}
 
-extern void
-softpipe_init_texture_funcs( struct softpipe_context *softpipe );
 
 extern void
 softpipe_init_screen_texture_funcs(struct pipe_screen *screen);
diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c
index b2195ec6b5..65872cecc4 100644
--- a/src/gallium/drivers/softpipe/sp_tile_cache.c
+++ b/src/gallium/drivers/softpipe/sp_tile_cache.c
@@ -26,7 +26,7 @@
  **************************************************************************/
 
 /**
- * Texture tile caching.
+ * Render target tile caching.
  *
  * Author:
  *    Brian Paul
@@ -35,38 +35,8 @@
 #include "pipe/p_inlines.h"
 #include "util/u_memory.h"
 #include "util/u_tile.h"
-#include "sp_context.h"
-#include "sp_surface.h"
-#include "sp_texture.h"
 #include "sp_tile_cache.h"
 
-#define NUM_ENTRIES 50
-
-
-/** XXX move these */
-#define MAX_WIDTH 4096
-#define MAX_HEIGHT 4096
-
-
-struct softpipe_tile_cache
-{
-   struct pipe_screen *screen;
-   struct pipe_surface *surface;  /**< the surface we're caching */
-   struct pipe_transfer *transfer;
-   void *transfer_map;
-   struct pipe_texture *texture;  /**< if caching a texture */
-   struct softpipe_cached_tile entries[NUM_ENTRIES];
-   uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32];
-   float clear_color[4];  /**< for color bufs */
-   uint clear_val;        /**< for z+stencil, or packed color clear value */
-   boolean depth_stencil; /**< Is the surface a depth/stencil format? */
-
-   struct pipe_transfer *tex_trans;
-   void *tex_trans_map;
-   int tex_face, tex_level, tex_z;
-
-   struct softpipe_cached_tile tile;  /**< scratch tile for clears */
-};
 
 
 /**
@@ -76,7 +46,7 @@ struct softpipe_tile_cache
  * a LRU replacement policy.
  */
 #define CACHE_POS(x, y) \
-   (((x) / TILE_SIZE + ((y) / TILE_SIZE) * 5) % NUM_ENTRIES)
+   (((x) + (y) * 5) % NUM_ENTRIES)
 
 
 
@@ -84,12 +54,10 @@ struct softpipe_tile_cache
  * Is the tile at (x,y) in cleared state?
  */
 static INLINE uint
-is_clear_flag_set(const uint *bitvec, int x, int y)
+is_clear_flag_set(const uint *bitvec, union tile_address addr)
 {
    int pos, bit;
-   x /= TILE_SIZE;
-   y /= TILE_SIZE;
-   pos = y * (MAX_WIDTH / TILE_SIZE) + x;
+   pos = addr.bits.y * (MAX_WIDTH / TILE_SIZE) + addr.bits.x;
    assert(pos / 32 < (MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32);
    bit = bitvec[pos / 32] & (1 << (pos & 31));
    return bit;
@@ -100,12 +68,10 @@ is_clear_flag_set(const uint *bitvec, int x, int y)
  * Mark the tile at (x,y) as not cleared.
  */
 static INLINE void
-clear_clear_flag(uint *bitvec, int x, int y)
+clear_clear_flag(uint *bitvec, union tile_address addr)
 {
    int pos;
-   x /= TILE_SIZE;
-   y /= TILE_SIZE;
-   pos = y * (MAX_WIDTH / TILE_SIZE) + x;
+   pos = addr.bits.y * (MAX_WIDTH / TILE_SIZE) + addr.bits.x;
    assert(pos / 32 < (MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32);
    bitvec[pos / 32] &= ~(1 << (pos & 31));
 }
@@ -127,9 +93,9 @@ sp_create_tile_cache( struct pipe_screen *screen )
    if (tc) {
       tc->screen = screen;
       for (pos = 0; pos < NUM_ENTRIES; pos++) {
-         tc->entries[pos].x =
-         tc->entries[pos].y = -1;
+         tc->entries[pos].addr.bits.invalid = 1;
       }
+      tc->last_tile = &tc->entries[0]; /* any tile */
 
       /* XXX this code prevents valgrind warnings about use of uninitialized
        * memory in programs that don't clear the surface before rendering.
@@ -158,10 +124,6 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc)
       screen = tc->transfer->texture->screen;
       screen->tex_transfer_destroy(tc->transfer);
    }
-   if (tc->tex_trans) {
-      screen = tc->tex_trans->texture->screen;
-      screen->tex_transfer_destroy(tc->tex_trans);
-   }
 
    FREE( tc );
 }
@@ -174,8 +136,6 @@ void
 sp_tile_cache_set_surface(struct softpipe_tile_cache *tc,
                           struct pipe_surface *ps)
 {
-   assert(!tc->texture);
-
    if (tc->transfer) {
       struct pipe_screen *screen = tc->transfer->texture->screen;
 
@@ -227,9 +187,6 @@ sp_tile_cache_map_transfers(struct softpipe_tile_cache *tc)
 {
    if (tc->transfer && !tc->transfer_map)
       tc->transfer_map = tc->screen->transfer_map(tc->screen, tc->transfer);
-
-   if (tc->tex_trans && !tc->tex_trans_map)
-      tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans);
 }
 
 
@@ -240,47 +197,6 @@ sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc)
       tc->screen->transfer_unmap(tc->screen, tc->transfer);
       tc->transfer_map = NULL;
    }
-
-   if (tc->tex_trans_map) {
-      tc->screen->transfer_unmap(tc->screen, tc->tex_trans);
-      tc->tex_trans_map = NULL;
-   }
-}
-
-
-/**
- * Specify the texture to cache.
- */
-void
-sp_tile_cache_set_texture(struct pipe_context *pipe,
-                          struct softpipe_tile_cache *tc,
-                          struct pipe_texture *texture)
-{
-   uint i;
-
-   assert(!tc->transfer);
-
-   pipe_texture_reference(&tc->texture, texture);
-
-   if (tc->tex_trans) {
-      struct pipe_screen *screen = tc->tex_trans->texture->screen;
-
-      if (tc->tex_trans_map) {
-         screen->transfer_unmap(screen, tc->tex_trans);
-         tc->tex_trans_map = NULL;
-      }
-
-      screen->tex_transfer_destroy(tc->tex_trans);
-      tc->tex_trans = NULL;
-   }
-
-   /* mark as entries as invalid/empty */
-   /* XXX we should try to avoid this when the teximage hasn't changed */
-   for (i = 0; i < NUM_ENTRIES; i++) {
-      tc->entries[i].x = -1;
-   }
-
-   tc->tex_face = -1; /* any invalid value here */
 }
 
 
@@ -324,7 +240,7 @@ clear_tile(struct softpipe_cached_tile *tile,
 
    switch (pf_get_size(format)) {
    case 1:
-      memset(tile->data.any, 0, TILE_SIZE * TILE_SIZE);
+      memset(tile->data.any, clear_value, TILE_SIZE * TILE_SIZE);
       break;
    case 2:
       if (clear_value == 0) {
@@ -360,8 +276,7 @@ clear_tile(struct softpipe_cached_tile *tile,
  * Actually clear the tiles which were flagged as being in a clear state.
  */
 static void
-sp_tile_cache_flush_clear(struct pipe_context *pipe,
-                          struct softpipe_tile_cache *tc)
+sp_tile_cache_flush_clear(struct softpipe_tile_cache *tc)
 {
    struct pipe_transfer *pt = tc->transfer;
    const uint w = tc->transfer->width;
@@ -375,13 +290,15 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe,
    /* push the tile to all positions marked as clear */
    for (y = 0; y < h; y += TILE_SIZE) {
       for (x = 0; x < w; x += TILE_SIZE) {
-         if (is_clear_flag_set(tc->clear_flags, x, y)) {
+         union tile_address addr = tile_address(x, y);
+
+         if (is_clear_flag_set(tc->clear_flags, addr)) {
             pipe_put_tile_raw(pt,
                               x, y, TILE_SIZE, TILE_SIZE,
                               tc->tile.data.color32, 0/*STRIDE*/);
 
             /* do this? */
-            clear_clear_flag(tc->clear_flags, x, y);
+            clear_clear_flag(tc->clear_flags, addr);
 
             numCleared++;
          }
@@ -398,8 +315,7 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe,
  * any tiles "flagged" as cleared will be "really" cleared.
  */
 void
-sp_flush_tile_cache(struct softpipe_context *softpipe,
-                    struct softpipe_tile_cache *tc)
+sp_flush_tile_cache(struct softpipe_tile_cache *tc)
 {
    struct pipe_transfer *pt = tc->transfer;
    int inuse = 0, pos;
@@ -408,33 +324,30 @@ sp_flush_tile_cache(struct softpipe_context *softpipe,
       /* caching a drawing transfer */
       for (pos = 0; pos < NUM_ENTRIES; pos++) {
          struct softpipe_cached_tile *tile = tc->entries + pos;
-         if (tile->x >= 0) {
+         if (!tile->addr.bits.invalid) {
             if (tc->depth_stencil) {
                pipe_put_tile_raw(pt,
-                                 tile->x, tile->y, TILE_SIZE, TILE_SIZE,
+                                 tile->addr.bits.x * TILE_SIZE, 
+                                 tile->addr.bits.y * TILE_SIZE, 
+                                 TILE_SIZE, TILE_SIZE,
                                  tile->data.depth32, 0/*STRIDE*/);
             }
             else {
                pipe_put_tile_rgba(pt,
-                                  tile->x, tile->y, TILE_SIZE, TILE_SIZE,
+                                  tile->addr.bits.x * TILE_SIZE, 
+                                  tile->addr.bits.y * TILE_SIZE, 
+                                  TILE_SIZE, TILE_SIZE,
                                   (float *) tile->data.color);
             }
-            tile->x = tile->y = -1;  /* mark as empty */
+            tile->addr.bits.invalid = 1;  /* mark as empty */
             inuse++;
          }
       }
 
 #if TILE_CLEAR_OPTIMIZATION
-      sp_tile_cache_flush_clear(&softpipe->pipe, tc);
+      sp_tile_cache_flush_clear(tc);
 #endif
    }
-   else if (tc->texture) {
-      /* caching a texture, mark all entries as empty */
-      for (pos = 0; pos < NUM_ENTRIES; pos++) {
-         tc->entries[pos].x = -1;
-      }
-      tc->tex_face = -1;
-   }
 
 #if 0
    debug_printf("flushed tiles in use: %d\n", inuse);
@@ -447,40 +360,39 @@ sp_flush_tile_cache(struct softpipe_context *softpipe,
  * \param x, y  position of tile, in pixels
  */
 struct softpipe_cached_tile *
-sp_get_cached_tile(struct softpipe_context *softpipe,
-                   struct softpipe_tile_cache *tc, int x, int y)
+sp_find_cached_tile(struct softpipe_tile_cache *tc, 
+                    union tile_address addr )
 {
    struct pipe_transfer *pt = tc->transfer;
-
-   /* tile pos in framebuffer: */
-   const int tile_x = x & ~(TILE_SIZE - 1);
-   const int tile_y = y & ~(TILE_SIZE - 1);
-
+   
    /* cache pos/entry: */
-   const int pos = CACHE_POS(x, y);
+   const int pos = CACHE_POS(addr.bits.x,
+                             addr.bits.y);
    struct softpipe_cached_tile *tile = tc->entries + pos;
 
-   if (tile_x != tile->x ||
-       tile_y != tile->y) {
+   if (addr.value != tile->addr.value) {
 
-      if (tile->x != -1) {
+      if (tile->addr.bits.invalid == 0) {
          /* put dirty tile back in framebuffer */
          if (tc->depth_stencil) {
             pipe_put_tile_raw(pt,
-                              tile->x, tile->y, TILE_SIZE, TILE_SIZE,
+                              tile->addr.bits.x * TILE_SIZE,
+                              tile->addr.bits.y * TILE_SIZE,
+                              TILE_SIZE, TILE_SIZE,
                               tile->data.depth32, 0/*STRIDE*/);
          }
          else {
             pipe_put_tile_rgba(pt,
-                               tile->x, tile->y, TILE_SIZE, TILE_SIZE,
+                               tile->addr.bits.x * TILE_SIZE,
+                               tile->addr.bits.y * TILE_SIZE,
+                               TILE_SIZE, TILE_SIZE,
                                (float *) tile->data.color);
          }
       }
 
-      tile->x = tile_x;
-      tile->y = tile_y;
+      tile->addr = addr;
 
-      if (is_clear_flag_set(tc->clear_flags, x, y)) {
+      if (is_clear_flag_set(tc->clear_flags, addr)) {
          /* don't get tile from framebuffer, just clear it */
          if (tc->depth_stencil) {
             clear_tile(tile, pt->format, tc->clear_val);
@@ -488,125 +400,33 @@ sp_get_cached_tile(struct softpipe_context *softpipe,
          else {
             clear_tile_rgba(tile, pt->format, tc->clear_color);
          }
-         clear_clear_flag(tc->clear_flags, x, y);
+         clear_clear_flag(tc->clear_flags, addr);
       }
       else {
          /* get new tile data from transfer */
          if (tc->depth_stencil) {
             pipe_get_tile_raw(pt,
-                              tile->x, tile->y, TILE_SIZE, TILE_SIZE,
+                              tile->addr.bits.x * TILE_SIZE, 
+                              tile->addr.bits.y * TILE_SIZE, 
+                              TILE_SIZE, TILE_SIZE,
                               tile->data.depth32, 0/*STRIDE*/);
          }
          else {
             pipe_get_tile_rgba(pt,
-                               tile->x, tile->y, TILE_SIZE, TILE_SIZE,
+                               tile->addr.bits.x * TILE_SIZE, 
+                               tile->addr.bits.y * TILE_SIZE,
+                               TILE_SIZE, TILE_SIZE,
                                (float *) tile->data.color);
          }
       }
    }
 
+   tc->last_tile = tile;
    return tile;
 }
 
 
-/**
- * Given the texture face, level, zslice, x and y values, compute
- * the cache entry position/index where we'd hope to find the
- * cached texture tile.
- * This is basically a direct-map cache.
- * XXX There's probably lots of ways in which we can improve this.
- */
-static INLINE uint
-tex_cache_pos(int x, int y, int z, int face, int level)
-{
-   uint entry = x + y * 9 + z * 3 + face + level * 7;
-   return entry % NUM_ENTRIES;
-}
-
 
-/**
- * Similar to sp_get_cached_tile() but for textures.
- * Tiles are read-only and indexed with more params.
- */
-const struct softpipe_cached_tile *
-sp_get_cached_tile_tex(struct softpipe_context *sp,
-                       struct softpipe_tile_cache *tc, int x, int y, int z,
-                       int face, int level)
-{
-   struct pipe_screen *screen = sp->pipe.screen;
-   /* tile pos in framebuffer: */
-   const int tile_x = x & ~(TILE_SIZE - 1);
-   const int tile_y = y & ~(TILE_SIZE - 1);
-   /* cache pos/entry: */
-   const uint pos = tex_cache_pos(x / TILE_SIZE, y / TILE_SIZE, z,
-                                  face, level);
-   struct softpipe_cached_tile *tile = tc->entries + pos;
-
-   if (tc->texture) {
-      struct softpipe_texture *spt = softpipe_texture(tc->texture);
-      if (spt->modified) {
-         /* texture was modified, invalidate all cached tiles */
-         uint p;
-         for (p = 0; p < NUM_ENTRIES; p++) {
-            tile = tc->entries + p;
-            tile->x = -1;
-         }
-         spt->modified = FALSE;
-      }
-   }
-
-   if (tile_x != tile->x ||
-       tile_y != tile->y ||
-       z != tile->z ||
-       face != tile->face ||
-       level != tile->level) {
-      /* cache miss */
-
-#if 0
-      printf("miss at %u  x=%d y=%d z=%d face=%d level=%d\n", pos,
-             x/TILE_SIZE, y/TILE_SIZE, z, face, level);
-#endif
-      /* check if we need to get a new transfer */
-      if (!tc->tex_trans ||
-          tc->tex_face != face ||
-          tc->tex_level != level ||
-          tc->tex_z != z) {
-         /* get new transfer (view into texture) */
-
-         if (tc->tex_trans) {
-            if (tc->tex_trans_map) {
-               tc->screen->transfer_unmap(tc->screen, tc->tex_trans);
-               tc->tex_trans_map = NULL;
-            }
-
-            screen->tex_transfer_destroy(tc->tex_trans);
-            tc->tex_trans = NULL;
-         }
-
-         tc->tex_trans = screen->get_tex_transfer(screen, tc->texture, face, level, z, 
-                                                  PIPE_TRANSFER_READ, 0, 0,
-                                                  tc->texture->width[level],
-                                                  tc->texture->height[level]);
-         tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans);
-
-         tc->tex_face = face;
-         tc->tex_level = level;
-         tc->tex_z = z;
-      }
-
-      /* get tile from the transfer (view into texture) */
-      pipe_get_tile_rgba(tc->tex_trans,
-                         tile_x, tile_y, TILE_SIZE, TILE_SIZE,
-                         (float *) tile->data.color);
-      tile->x = tile_x;
-      tile->y = tile_y;
-      tile->z = z;
-      tile->face = face;
-      tile->level = level;
-   }
-
-   return tile;
-}
 
 
 /**
@@ -637,6 +457,6 @@ sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float *rgba,
 
    for (pos = 0; pos < NUM_ENTRIES; pos++) {
       struct softpipe_cached_tile *tile = tc->entries + pos;
-      tile->x = tile->y = -1;
+      tile->addr.bits.invalid = 1;
    }
 }
diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.h b/src/gallium/drivers/softpipe/sp_tile_cache.h
index 8f247d0e58..a12092702a 100644
--- a/src/gallium/drivers/softpipe/sp_tile_cache.h
+++ b/src/gallium/drivers/softpipe/sp_tile_cache.h
@@ -34,7 +34,6 @@
 #include "pipe/p_compiler.h"
 
 
-struct softpipe_context;
 struct softpipe_tile_cache;
 
 
@@ -44,11 +43,23 @@ struct softpipe_tile_cache;
 #define TILE_SIZE 64
 
 
+/* If we need to support > 4096, just expand this to be a 64 bit
+ * union, or consider tiling in Z as well.
+ */
+union tile_address {
+   struct {
+      unsigned x:6;             /* 4096 / TILE_SIZE */
+      unsigned y:6;             /* 4096 / TILE_SIZE */
+      unsigned invalid:1;
+      unsigned pad:19;
+   } bits;
+   unsigned value;
+};
+
 
 struct softpipe_cached_tile
 {
-   int x, y;           /**< pos of tile in window coords */
-   int z, face, level; /**< Extra texture indexes */
+   union tile_address addr;
    union {
       float color[TILE_SIZE][TILE_SIZE][4];
       uint color32[TILE_SIZE][TILE_SIZE];
@@ -59,6 +70,32 @@ struct softpipe_cached_tile
    } data;
 };
 
+#define NUM_ENTRIES 50
+
+
+/** XXX move these */
+#define MAX_WIDTH 4096
+#define MAX_HEIGHT 4096
+
+
+struct softpipe_tile_cache
+{
+   struct pipe_screen *screen;
+   struct pipe_surface *surface;  /**< the surface we're caching */
+   struct pipe_transfer *transfer;
+   void *transfer_map;
+
+   struct softpipe_cached_tile entries[NUM_ENTRIES];
+   uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32];
+   float clear_color[4];  /**< for color bufs */
+   uint clear_val;        /**< for z+stencil, or packed color clear value */
+   boolean depth_stencil; /**< Is the surface a depth/stencil format? */
+
+   struct softpipe_cached_tile tile;  /**< scratch tile for clears */
+
+   struct softpipe_cached_tile *last_tile;  /**< most recently retrieved tile */
+};
+
 
 extern struct softpipe_tile_cache *
 sp_create_tile_cache( struct pipe_screen *screen );
@@ -80,26 +117,45 @@ extern void
 sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc);
 
 extern void
-sp_tile_cache_set_texture(struct pipe_context *pipe,
-                          struct softpipe_tile_cache *tc,
-                          struct pipe_texture *texture);
-
-extern void
-sp_flush_tile_cache(struct softpipe_context *softpipe,
-                    struct softpipe_tile_cache *tc);
+sp_flush_tile_cache(struct softpipe_tile_cache *tc);
 
 extern void
 sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float *rgba,
                     uint clearValue);
 
 extern struct softpipe_cached_tile *
-sp_get_cached_tile(struct softpipe_context *softpipe,
-                   struct softpipe_tile_cache *tc, int x, int y);
+sp_find_cached_tile(struct softpipe_tile_cache *tc, 
+                    union tile_address addr );
+
+
+static INLINE union tile_address
+tile_address( unsigned x,
+              unsigned y )
+{
+   union tile_address addr;
+
+   addr.value = 0;
+   addr.bits.x = x / TILE_SIZE;
+   addr.bits.y = y / TILE_SIZE;
+      
+   return addr;
+}
+
+/* Quickly retrieve tile if it matches last lookup.
+ */
+static INLINE struct softpipe_cached_tile *
+sp_get_cached_tile(struct softpipe_tile_cache *tc, 
+                   int x, int y )
+{
+   union tile_address addr = tile_address( x, y );
+
+   if (tc->last_tile->addr.value == addr.value)
+      return tc->last_tile;
+
+   return sp_find_cached_tile( tc, addr );
+}
+
 
-extern const struct softpipe_cached_tile *
-sp_get_cached_tile_tex(struct softpipe_context *softpipe,
-                       struct softpipe_tile_cache *tc, int x, int y, int z,
-                       int face, int level);
 
 
 #endif /* SP_TILE_CACHE_H */
diff --git a/src/gallium/drivers/softpipe/sp_video_context.c b/src/gallium/drivers/softpipe/sp_video_context.c
new file mode 100644
index 0000000000..cae2d3efc5
--- /dev/null
+++ b/src/gallium/drivers/softpipe/sp_video_context.c
@@ -0,0 +1,304 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "sp_video_context.h"
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include "softpipe/sp_winsys.h"
+#include "softpipe/sp_texture.h"
+
+static void
+sp_mpeg12_destroy(struct pipe_video_context *vpipe)
+{
+   struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe;
+
+   assert(vpipe);
+	
+   /* Asserted in softpipe_delete_fs_state() for some reason */
+   ctx->pipe->bind_vs_state(ctx->pipe, NULL);
+   ctx->pipe->bind_fs_state(ctx->pipe, NULL);
+
+   ctx->pipe->delete_blend_state(ctx->pipe, ctx->blend);
+   ctx->pipe->delete_rasterizer_state(ctx->pipe, ctx->rast);
+   ctx->pipe->delete_depth_stencil_alpha_state(ctx->pipe, ctx->dsa);
+
+   pipe_video_surface_reference(&ctx->decode_target, NULL);
+   vl_compositor_cleanup(&ctx->compositor);
+   vl_mpeg12_mc_renderer_cleanup(&ctx->mc_renderer);
+   ctx->pipe->destroy(ctx->pipe);
+
+   FREE(ctx);
+}
+
+static void
+sp_mpeg12_decode_macroblocks(struct pipe_video_context *vpipe,
+                             struct pipe_video_surface *past,
+                             struct pipe_video_surface *future,
+                             unsigned num_macroblocks,
+                             struct pipe_macroblock *macroblocks,
+                             struct pipe_fence_handle **fence)
+{
+   struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe;
+   struct pipe_mpeg12_macroblock *mpeg12_macroblocks = (struct pipe_mpeg12_macroblock*)macroblocks;
+
+   assert(vpipe);
+   assert(num_macroblocks);
+   assert(macroblocks);
+   assert(macroblocks->codec == PIPE_VIDEO_CODEC_MPEG12);
+   assert(ctx->decode_target);
+
+   vl_mpeg12_mc_renderer_render_macroblocks(&ctx->mc_renderer,
+                                            softpipe_video_surface(ctx->decode_target)->tex,
+                                            past ? softpipe_video_surface(past)->tex : NULL,
+                                            future ? softpipe_video_surface(future)->tex : NULL,
+                                            num_macroblocks, mpeg12_macroblocks, fence);
+}
+
+static void
+sp_mpeg12_clear_surface(struct pipe_video_context *vpipe,
+                        unsigned x, unsigned y,
+                        unsigned width, unsigned height,
+                        unsigned value,
+                        struct pipe_surface *surface)
+{
+   struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe;
+
+   assert(vpipe);
+   assert(surface);
+
+   ctx->pipe->surface_fill(ctx->pipe, surface, x, y, width, height, value);
+}
+
+static void
+sp_mpeg12_render_picture(struct pipe_video_context     *vpipe,
+                         /*struct pipe_surface         *backround,
+                         struct pipe_video_rect        *backround_area,*/
+                         struct pipe_video_surface     *src_surface,
+                         enum pipe_mpeg12_picture_type picture_type,
+                         /*unsigned                    num_past_surfaces,
+                         struct pipe_video_surface     *past_surfaces,
+                         unsigned                      num_future_surfaces,
+                         struct pipe_video_surface     *future_surfaces,*/
+                         struct pipe_video_rect        *src_area,
+                         struct pipe_surface           *dst_surface,
+                         struct pipe_video_rect        *dst_area,
+                         /*unsigned                      num_layers,
+                         struct pipe_surface           *layers,
+                         struct pipe_video_rect        *layer_src_areas,
+                         struct pipe_video_rect        *layer_dst_areas*/
+                         struct pipe_fence_handle      **fence)
+{
+   struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe;
+	
+   assert(vpipe);
+   assert(src_surface);
+   assert(src_area);
+   assert(dst_surface);
+   assert(dst_area);
+	
+   vl_compositor_render(&ctx->compositor, softpipe_video_surface(src_surface)->tex,
+                        picture_type, src_area, dst_surface->texture, dst_area, fence);
+}
+
+static void
+sp_mpeg12_set_decode_target(struct pipe_video_context *vpipe,
+                            struct pipe_video_surface *dt)
+{
+   struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe;
+
+   assert(vpipe);
+   assert(dt);
+
+   pipe_video_surface_reference(&ctx->decode_target, dt);
+}
+
+static void sp_mpeg12_set_csc_matrix(struct pipe_video_context *vpipe, const float *mat)
+{
+   struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe;
+
+   assert(vpipe);
+
+   vl_compositor_set_csc_matrix(&ctx->compositor, mat);
+}
+
+static bool
+init_pipe_state(struct sp_mpeg12_context *ctx)
+{
+   struct pipe_rasterizer_state rast;
+   struct pipe_blend_state blend;
+   struct pipe_depth_stencil_alpha_state dsa;
+   unsigned i;
+
+   assert(ctx);
+	
+   rast.flatshade = 1;
+   rast.flatshade_first = 0;
+   rast.light_twoside = 0;
+   rast.front_winding = PIPE_WINDING_CCW;
+   rast.cull_mode = PIPE_WINDING_CW;
+   rast.fill_cw = PIPE_POLYGON_MODE_FILL;
+   rast.fill_ccw = PIPE_POLYGON_MODE_FILL;
+   rast.offset_cw = 0;
+   rast.offset_ccw = 0;
+   rast.scissor = 0;
+   rast.poly_smooth = 0;
+   rast.poly_stipple_enable = 0;
+   rast.point_sprite = 0;
+   rast.point_size_per_vertex = 0;
+   rast.multisample = 0;
+   rast.line_smooth = 0;
+   rast.line_stipple_enable = 0;
+   rast.line_stipple_factor = 0;
+   rast.line_stipple_pattern = 0;
+   rast.line_last_pixel = 0;
+   rast.bypass_vs_clip_and_viewport = 0;
+   rast.line_width = 1;
+   rast.point_smooth = 0;
+   rast.point_size = 1;
+   rast.offset_units = 1;
+   rast.offset_scale = 1;
+   /*rast.sprite_coord_mode[i] = ;*/
+   ctx->rast = ctx->pipe->create_rasterizer_state(ctx->pipe, &rast);
+   ctx->pipe->bind_rasterizer_state(ctx->pipe, ctx->rast);
+
+   blend.blend_enable = 0;
+   blend.rgb_func = PIPE_BLEND_ADD;
+   blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE;
+   blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
+   blend.alpha_func = PIPE_BLEND_ADD;
+   blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE;
+   blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
+   blend.logicop_enable = 0;
+   blend.logicop_func = PIPE_LOGICOP_CLEAR;
+   /* Needed to allow color writes to FB, even if blending disabled */
+   blend.colormask = PIPE_MASK_RGBA;
+   blend.dither = 0;
+   ctx->blend = ctx->pipe->create_blend_state(ctx->pipe, &blend);
+   ctx->pipe->bind_blend_state(ctx->pipe, ctx->blend);
+
+   dsa.depth.enabled = 0;
+   dsa.depth.writemask = 0;
+   dsa.depth.func = PIPE_FUNC_ALWAYS;
+   for (i = 0; i < 2; ++i) {
+      dsa.stencil[i].enabled = 0;
+      dsa.stencil[i].func = PIPE_FUNC_ALWAYS;
+      dsa.stencil[i].fail_op = PIPE_STENCIL_OP_KEEP;
+      dsa.stencil[i].zpass_op = PIPE_STENCIL_OP_KEEP;
+      dsa.stencil[i].zfail_op = PIPE_STENCIL_OP_KEEP;
+      dsa.stencil[i].ref_value = 0;
+      dsa.stencil[i].valuemask = 0;
+      dsa.stencil[i].writemask = 0;
+   }
+   dsa.alpha.enabled = 0;
+   dsa.alpha.func = PIPE_FUNC_ALWAYS;
+   dsa.alpha.ref_value = 0;
+   ctx->dsa = ctx->pipe->create_depth_stencil_alpha_state(ctx->pipe, &dsa);
+   ctx->pipe->bind_depth_stencil_alpha_state(ctx->pipe, ctx->dsa);
+	
+   return true;
+}
+
+static struct pipe_video_context *
+sp_mpeg12_create(struct pipe_screen *screen, enum pipe_video_profile profile,
+                 enum pipe_video_chroma_format chroma_format,
+                 unsigned width, unsigned height)
+{
+   struct sp_mpeg12_context *ctx;
+
+   assert(u_reduce_video_profile(profile) == PIPE_VIDEO_CODEC_MPEG12);
+
+   ctx = CALLOC_STRUCT(sp_mpeg12_context);
+
+   if (!ctx)
+      return NULL;
+
+   ctx->base.profile = profile;
+   ctx->base.chroma_format = chroma_format;
+   ctx->base.width = width;
+   ctx->base.height = height;
+
+   ctx->base.screen = screen;
+   ctx->base.destroy = sp_mpeg12_destroy;
+   ctx->base.decode_macroblocks = sp_mpeg12_decode_macroblocks;
+   ctx->base.clear_surface = sp_mpeg12_clear_surface;
+   ctx->base.render_picture = sp_mpeg12_render_picture;
+   ctx->base.set_decode_target = sp_mpeg12_set_decode_target;
+   ctx->base.set_csc_matrix = sp_mpeg12_set_csc_matrix;
+
+   ctx->pipe = softpipe_create(screen);
+   if (!ctx->pipe) {
+      FREE(ctx);
+      return NULL;
+   }
+
+   /* TODO: Use slice buffering for softpipe when implemented, no advantage to buffering an entire picture */
+   if (!vl_mpeg12_mc_renderer_init(&ctx->mc_renderer, ctx->pipe,
+                                   width, height, chroma_format,
+                                   VL_MPEG12_MC_RENDERER_BUFFER_PICTURE,
+                                   /* TODO: Use XFER_NONE when implemented */
+                                   VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE,
+                                   true)) {
+      ctx->pipe->destroy(ctx->pipe);
+      FREE(ctx);
+      return NULL;
+   }
+	
+   if (!vl_compositor_init(&ctx->compositor, ctx->pipe)) {
+      vl_mpeg12_mc_renderer_cleanup(&ctx->mc_renderer);
+      ctx->pipe->destroy(ctx->pipe);
+      FREE(ctx);
+      return NULL;
+   }
+	
+   if (!init_pipe_state(ctx)) {
+      vl_compositor_cleanup(&ctx->compositor);
+      vl_mpeg12_mc_renderer_cleanup(&ctx->mc_renderer);
+      ctx->pipe->destroy(ctx->pipe);
+      FREE(ctx);
+      return NULL;
+   }
+
+   return &ctx->base;
+}
+
+struct pipe_video_context *
+sp_video_create(struct pipe_screen *screen, enum pipe_video_profile profile,
+                enum pipe_video_chroma_format chroma_format,
+                unsigned width, unsigned height)
+{
+   assert(screen);
+   assert(width && height);
+
+   switch (u_reduce_video_profile(profile)) {
+      case PIPE_VIDEO_CODEC_MPEG12:
+         return sp_mpeg12_create(screen, profile,
+                                 chroma_format,
+                                 width, height);
+      default:
+         return NULL;
+   }
+}
diff --git a/src/gallium/drivers/i965simple/brw_batch.h b/src/gallium/drivers/softpipe/sp_video_context.h
index 5f5932a488..ccbd1ffe4c 100644
--- a/src/gallium/drivers/i965simple/brw_batch.h
+++ b/src/gallium/drivers/softpipe/sp_video_context.h
@@ -1,8 +1,8 @@
 /**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * 
+ * Copyright 2009 Younes Manton.
  * All Rights Reserved.
- *
+ * 
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -10,11 +10,11 @@
  * distribute, sub license, and/or sell copies of the Software, and to
  * permit persons to whom the Software is furnished to do so, subject to
  * the following conditions:
- *
+ * 
  * The above copyright notice and this permission notice (including the
  * next paragraph) shall be included in all copies or substantial portions
  * of the Software.
- *
+ * 
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
@@ -22,38 +22,36 @@
  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
+ * 
  **************************************************************************/
 
-#ifndef BRW_BATCH_H
-#define BRW_BATCH_H
+#ifndef SP_VIDEO_CONTEXT_H
+#define SP_VIDEO_CONTEXT_H
 
-#include "brw_winsys.h"
+#include <pipe/p_video_context.h>
+#include <vl/vl_mpeg12_mc_renderer.h>
+#include <vl/vl_compositor.h>
 
-#define BATCH_LOCALS
+struct pipe_screen;
+struct pipe_context;
+struct pipe_video_surface;
 
-#define INTEL_BATCH_NO_CLIPRECTS 0x1
-#define INTEL_BATCH_CLIPRECTS    0x2
+struct sp_mpeg12_context
+{
+   struct pipe_video_context base;
+   struct pipe_context *pipe;
+   struct pipe_video_surface *decode_target;
+   struct vl_mpeg12_mc_renderer mc_renderer;
+   struct vl_compositor compositor;
 
-#define BEGIN_BATCH( dwords, relocs ) \
-   brw->winsys->batch_start(brw->winsys, dwords, relocs)
+   void *rast;
+   void *dsa;
+   void *blend;
+};
 
-#define OUT_BATCH( dword ) \
-   brw->winsys->batch_dword(brw->winsys, dword)
+struct pipe_video_context *
+sp_video_create(struct pipe_screen *screen, enum pipe_video_profile profile,
+                enum pipe_video_chroma_format chroma_format,
+                unsigned width, unsigned height);
 
-#define OUT_RELOC( buf, flags, delta ) \
-   brw->winsys->batch_reloc(brw->winsys, buf, flags, delta)
-
-#define ADVANCE_BATCH() \
-   brw->winsys->batch_end( brw->winsys )
-
-/* XXX: this is bogus - need proper handling for out-of-memory in batchbuffer.
- */
-#define FLUSH_BATCH(fence) do {				\
-   brw->winsys->batch_flush(brw->winsys, fence);	\
-   brw->hardware_dirty = ~0;				\
-} while (0)
-
-#define BRW_BATCH_STRUCT(brw, s) brw_batchbuffer_data( brw->winsys, (s), sizeof(*(s)))
-
-#endif
+#endif /* SP_VIDEO_CONTEXT_H */
diff --git a/src/gallium/drivers/trace/tr_context.c b/src/gallium/drivers/trace/tr_context.c
index ae0af4d055..bf470b46ae 100644
--- a/src/gallium/drivers/trace/tr_context.c
+++ b/src/gallium/drivers/trace/tr_context.c
@@ -125,11 +125,11 @@ trace_context_draw_block(struct trace_context *tr_ctx, int flag)
    } else if ((tr_ctx->draw_rule.blocker & flag) &&
               (tr_ctx->draw_blocker & 4)) {
       boolean block = FALSE;
-      debug_printf("%s (%lu %lu) (%lu %lu) (%lu %u) (%lu %u)\n", __FUNCTION__,
-					tr_ctx->draw_rule.fs, tr_ctx->curr.fs,
-					tr_ctx->draw_rule.vs, tr_ctx->curr.vs,
-					tr_ctx->draw_rule.surf, 0,
-					tr_ctx->draw_rule.tex, 0);
+      debug_printf("%s (%p %p) (%p %p) (%p %u) (%p %u)\n", __FUNCTION__,
+                   (void *) tr_ctx->draw_rule.fs, (void *) tr_ctx->curr.fs,
+                   (void *) tr_ctx->draw_rule.vs, (void *) tr_ctx->curr.vs,
+                   (void *) tr_ctx->draw_rule.surf, 0,
+                   (void *) tr_ctx->draw_rule.tex, 0);
       if (tr_ctx->draw_rule.fs &&
           tr_ctx->draw_rule.fs == tr_ctx->curr.fs)
          block = TRUE;
diff --git a/src/gallium/drivers/trace/tr_drm.c b/src/gallium/drivers/trace/tr_drm.c
index 781ca5d3bc..48d1c4051c 100644
--- a/src/gallium/drivers/trace/tr_drm.c
+++ b/src/gallium/drivers/trace/tr_drm.c
@@ -150,7 +150,9 @@ trace_drm_destroy(struct drm_api *_api)
 {
    struct trace_drm_api *tr_api = trace_drm_api(_api);
    struct drm_api *api = tr_api->api;
-   api->destroy(api);
+
+   if (api->destroy)
+      api->destroy(api);
 
    free(tr_api);
 }
diff --git a/src/gallium/drivers/trace/tr_rbug.c b/src/gallium/drivers/trace/tr_rbug.c
index e85ac15edc..81e0a6f3b0 100644
--- a/src/gallium/drivers/trace/tr_rbug.c
+++ b/src/gallium/drivers/trace/tr_rbug.c
@@ -44,7 +44,7 @@
 
 #if defined(PIPE_SUBSYSTEM_WINDOWS_USER)
 #  define sleep Sleep
-#elif defined(PIPE_OS_LINUX)
+#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 void usleep(int);
 #  define sleep usleep
 #else
diff --git a/src/gallium/drivers/trace/tr_screen.c b/src/gallium/drivers/trace/tr_screen.c
index 26f1c04594..7da9bd3866 100644
--- a/src/gallium/drivers/trace/tr_screen.c
+++ b/src/gallium/drivers/trace/tr_screen.c
@@ -366,7 +366,8 @@ trace_screen_get_tex_transfer(struct pipe_screen *_screen,
 
    trace_dump_call_end();
 
-   result = trace_transfer_create(tr_tex, result);
+   if (result)
+      result = trace_transfer_create(tr_tex, result);
 
    return result;
 }
@@ -403,7 +404,7 @@ trace_screen_transfer_map(struct pipe_screen *_screen,
 
    map = screen->transfer_map(screen, transfer);
    if(map) {
-      if(transfer->usage != PIPE_TRANSFER_READ) {
+      if(transfer->usage & PIPE_TRANSFER_WRITE) {
          assert(!tr_trans->map);
          tr_trans->map = map;
       }
diff --git a/src/gallium/include/pipe/p_compiler.h b/src/gallium/include/pipe/p_compiler.h
index c13cffceb0..c36286f9be 100644
--- a/src/gallium/include/pipe/p_compiler.h
+++ b/src/gallium/include/pipe/p_compiler.h
@@ -167,11 +167,17 @@ typedef unsigned char boolean;
 #define ALIGN16_ASSIGN(NAME) NAME##___aligned
 #define ALIGN16_ATTRIB  __attribute__(( aligned( 16 ) ))
 #define ALIGN8_ATTRIB  __attribute__(( aligned( 8 ) ))
+#if __GNUC__ > 4 || (__GNUC__ == 4 &&__GNUC_MINOR__>1)
+#define ALIGN_STACK __attribute__((force_align_arg_pointer))
+#else
+#define ALIGN_STACK
+#endif
 #else
 #define ALIGN16_DECL(TYPE, NAME, SIZE)  TYPE NAME##___unaligned[SIZE + 1]
 #define ALIGN16_ASSIGN(NAME) align16(NAME##___unaligned)
 #define ALIGN16_ATTRIB
 #define ALIGN8_ATTRIB
+#define ALIGN_STACK
 #endif
 
 
diff --git a/src/gallium/include/pipe/p_config.h b/src/gallium/include/pipe/p_config.h
index de99957d9d..f6feea5f74 100644
--- a/src/gallium/include/pipe/p_config.h
+++ b/src/gallium/include/pipe/p_config.h
@@ -122,18 +122,35 @@
 
 #if defined(__linux__)
 #define PIPE_OS_LINUX
+#define PIPE_OS_UNIX
 #endif
 
 #if defined(__FreeBSD__)
+#define PIPE_OS_FREEBSD
 #define PIPE_OS_BSD
+#define PIPE_OS_UNIX
+#endif
+
+#if defined(__OpenBSD__)
+#define PIPE_OS_OPENBSD
+#define PIPE_OS_BSD
+#define PIPE_OS_UNIX
+#endif
+
+#if defined(__NetBSD__)
+#define PIPE_OS_NETBSD
+#define PIPE_OS_BSD
+#define PIPE_OS_UNIX
 #endif
 
 #if defined(__sun)
 #define PIPE_OS_SOLARIS
+#define PIPE_OS_UNIX
 #endif
 
 #if defined(__APPLE__)
 #define PIPE_OS_APPLE
+#define PIPE_OS_UNIX
 #endif
 
 #if defined(_WIN32) || defined(WIN32)
@@ -142,6 +159,7 @@
 
 #if defined(__HAIKU__)
 #define PIPE_OS_HAIKU
+#define PIPE_OS_UNIX
 #endif
 
 /*
diff --git a/src/gallium/include/pipe/p_context.h b/src/gallium/include/pipe/p_context.h
index 39620a7198..5569001e60 100644
--- a/src/gallium/include/pipe/p_context.h
+++ b/src/gallium/include/pipe/p_context.h
@@ -189,6 +189,9 @@ struct pipe_context {
 
    /**
     * Surface functions
+    *
+    * The pipe driver is allowed to set these functions to NULL, and in that
+    * case, they will not be available.
     */
    /*@{*/
 
diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h
index b01ab6d137..f8fa1e3f49 100644
--- a/src/gallium/include/pipe/p_defines.h
+++ b/src/gallium/include/pipe/p_defines.h
@@ -193,7 +193,18 @@ enum pipe_texture_target {
 enum pipe_transfer_usage {
    PIPE_TRANSFER_READ = (1 << 0),
    PIPE_TRANSFER_WRITE = (1 << 1),
-   PIPE_TRANSFER_READ_WRITE = PIPE_TRANSFER_READ | PIPE_TRANSFER_WRITE /**< Read/modify/write */
+   /** Read/modify/write */
+   PIPE_TRANSFER_READ_WRITE = PIPE_TRANSFER_READ | PIPE_TRANSFER_WRITE,
+   /** 
+    * The transfer should map the texture storage directly. The driver may
+    * return NULL if that isn't possible, and the state tracker needs to cope
+    * with that and use an alternative path without this flag.
+    *
+    * E.g. the state tracker could have a simpler path which maps textures and
+    * does read/modify/write cycles on them directly, and a more complicated
+    * path which uses minimal read and write transfers.
+    */
+   PIPE_TRANSFER_MAP_DIRECTLY = (1 << 2)
 };
 
 
@@ -281,7 +292,7 @@ enum pipe_transfer_usage {
 #define PIPE_CAP_NPOT_TEXTURES           2
 #define PIPE_CAP_TWO_SIDED_STENCIL       3
 #define PIPE_CAP_GLSL                    4  /* XXX need something better */
-#define PIPE_CAP_S3TC                    5
+#define PIPE_CAP_S3TC                    5  /* XXX: deprecated; cap determined via supported sampler formats */
 #define PIPE_CAP_ANISOTROPIC_FILTER      6
 #define PIPE_CAP_POINT_SPRITE            7
 #define PIPE_CAP_MAX_RENDER_TARGETS      8
@@ -315,6 +326,31 @@ enum pipe_transfer_usage {
 #define PIPE_REFERENCED_FOR_READ  (1 << 0)
 #define PIPE_REFERENCED_FOR_WRITE (1 << 1)
 
+
+enum pipe_video_codec
+{
+   PIPE_VIDEO_CODEC_UNKNOWN = 0,
+   PIPE_VIDEO_CODEC_MPEG12,   /**< MPEG1, MPEG2 */
+   PIPE_VIDEO_CODEC_MPEG4,    /**< DIVX, XVID */
+   PIPE_VIDEO_CODEC_VC1,      /**< WMV */
+   PIPE_VIDEO_CODEC_MPEG4_AVC /**< H.264 */
+};
+
+enum pipe_video_profile
+{
+   PIPE_VIDEO_PROFILE_MPEG1,
+   PIPE_VIDEO_PROFILE_MPEG2_SIMPLE,
+   PIPE_VIDEO_PROFILE_MPEG2_MAIN,
+   PIPE_VIDEO_PROFILE_MPEG4_SIMPLE,
+   PIPE_VIDEO_PROFILE_MPEG4_ADVANCED_SIMPLE,
+   PIPE_VIDEO_PROFILE_VC1_SIMPLE,
+   PIPE_VIDEO_PROFILE_VC1_MAIN,
+   PIPE_VIDEO_PROFILE_VC1_ADVANCED,
+   PIPE_VIDEO_PROFILE_MPEG4_AVC_BASELINE,
+   PIPE_VIDEO_PROFILE_MPEG4_AVC_MAIN,
+   PIPE_VIDEO_PROFILE_MPEG4_AVC_HIGH
+};
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h
index c4469d4a9e..af23080920 100644
--- a/src/gallium/include/pipe/p_format.h
+++ b/src/gallium/include/pipe/p_format.h
@@ -613,6 +613,24 @@ pf_has_alpha( enum pipe_format format )
    }
 }
 
+enum pipe_video_chroma_format
+{
+   PIPE_VIDEO_CHROMA_FORMAT_420,
+   PIPE_VIDEO_CHROMA_FORMAT_422,
+   PIPE_VIDEO_CHROMA_FORMAT_444
+};
+
+#if 0
+enum pipe_video_surface_format
+{
+   PIPE_VIDEO_SURFACE_FORMAT_NV12,  /**< Planar; Y plane, UV plane */
+   PIPE_VIDEO_SURFACE_FORMAT_YV12,  /**< Planar; Y plane, U plane, V plane */
+   PIPE_VIDEO_SURFACE_FORMAT_YUYV,  /**< Interleaved; Y,U,Y,V,Y,U,Y,V */
+   PIPE_VIDEO_SURFACE_FORMAT_UYVY,  /**< Interleaved; U,Y,V,Y,U,Y,V,Y */
+   PIPE_VIDEO_SURFACE_FORMAT_VUYA   /**< Packed; A31-24|Y23-16|U15-8|V7-0 */
+};
+#endif
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/gallium/include/pipe/p_inlines.h b/src/gallium/include/pipe/p_inlines.h
index a5c1e8270a..5fbd62a03d 100644
--- a/src/gallium/include/pipe/p_inlines.h
+++ b/src/gallium/include/pipe/p_inlines.h
@@ -118,7 +118,7 @@ pipe_buffer_write(struct pipe_screen *screen,
                   unsigned offset, unsigned size,
                   const void *data)
 {
-   uint8_t *map;
+   void *map;
    
    assert(offset < buf->size);
    assert(offset + size <= buf->size);
@@ -129,7 +129,7 @@ pipe_buffer_write(struct pipe_screen *screen,
                                PIPE_BUFFER_USAGE_FLUSH_EXPLICIT);
    assert(map);
    if(map) {
-      memcpy(map + offset, data, size);
+      memcpy((uint8_t *)map + offset, data, size);
       pipe_buffer_flush_mapped_range(screen, buf, offset, size);
       pipe_buffer_unmap(screen, buf);
    }
@@ -141,7 +141,7 @@ pipe_buffer_read(struct pipe_screen *screen,
                  unsigned offset, unsigned size,
                  void *data)
 {
-   uint8_t *map;
+   void *map;
    
    assert(offset < buf->size);
    assert(offset + size <= buf->size);
@@ -150,11 +150,47 @@ pipe_buffer_read(struct pipe_screen *screen,
    map = pipe_buffer_map_range(screen, buf, offset, size, PIPE_BUFFER_USAGE_CPU_READ);
    assert(map);
    if(map) {
-      memcpy(data, map + offset, size);
+      memcpy(data, (const uint8_t *)map + offset, size);
       pipe_buffer_unmap(screen, buf);
    }
 }
 
+static INLINE void *
+pipe_transfer_map( struct pipe_transfer *transf )
+{
+   struct pipe_screen *screen = transf->texture->screen;
+   return screen->transfer_map(screen, transf);
+}
+
+static INLINE void
+pipe_transfer_unmap( struct pipe_transfer *transf )
+{
+   struct pipe_screen *screen = transf->texture->screen;
+   screen->transfer_unmap(screen, transf);
+}
+
+static INLINE void
+pipe_transfer_destroy( struct pipe_transfer *transf )
+{
+   struct pipe_screen *screen = transf->texture->screen;
+   screen->tex_transfer_destroy(transf);
+}
+
+static INLINE unsigned
+pipe_transfer_buffer_flags( struct pipe_transfer *transf )
+{
+   switch (transf->usage & PIPE_TRANSFER_READ_WRITE) {
+   case PIPE_TRANSFER_READ_WRITE:
+      return PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE;
+   case PIPE_TRANSFER_READ:
+      return PIPE_BUFFER_USAGE_CPU_READ;
+   case PIPE_TRANSFER_WRITE:
+      return PIPE_BUFFER_USAGE_CPU_WRITE;
+   default:
+      debug_assert(0);
+      return 0;
+   }
+}
 
 #ifdef __cplusplus
 }
diff --git a/src/gallium/include/pipe/p_screen.h b/src/gallium/include/pipe/p_screen.h
index 3f30c52a16..f0a4de5df3 100644
--- a/src/gallium/include/pipe/p_screen.h
+++ b/src/gallium/include/pipe/p_screen.h
@@ -53,7 +53,10 @@ extern "C" {
 struct pipe_fence_handle;
 struct pipe_winsys;
 struct pipe_buffer;
-
+struct pipe_texture;
+struct pipe_surface;
+struct pipe_video_surface;
+struct pipe_transfer;
 
 
 /**
@@ -252,6 +255,17 @@ struct pipe_screen {
 
    void (*buffer_destroy)( struct pipe_buffer *buf );
 
+   /**
+    * Create a video surface suitable for use as a decoding target by the
+    * driver's pipe_video_context.
+    */
+   struct pipe_video_surface*
+   (*video_surface_create)( struct pipe_screen *screen,
+                            enum pipe_video_chroma_format chroma_format,
+                            unsigned width, unsigned height );
+
+   void (*video_surface_destroy)( struct pipe_video_surface *vsfc );
+
 
    /**
     * Do any special operations to ensure frontbuffer contents are
diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
index 626bedb35a..287b424e4a 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -114,11 +114,29 @@ struct pipe_rasterizer_state
     * the vertex shader, clipping and viewport processing.  Note that
     * a vertex shader is still needed though, to indicate the mapping
     * from vertex elements to fragment shader input semantics.
+    *
+    * XXX: considered for removal.
     */
    unsigned bypass_vs_clip_and_viewport:1;
 
-   unsigned flatshade_first:1;   /**< take color attribute from the first vertex of a primitive */
-   unsigned gl_rasterization_rules:1; /**< enable tweaks for GL rasterization?  */
+   /** 
+    * Use the first vertex of a primitive as the provoking vertex for
+    * flat shading.
+    */
+   unsigned flatshade_first:1;   
+
+   /** 
+    * When true, triangle rasterization uses (0.5, 0.5) pixel centers
+    * for determining pixel ownership.
+    *
+    * When false, triangle rasterization uses (0,0) pixel centers for
+    * determining pixel ownership.
+    *
+    * Triangle rasterization always uses a 'top,left' rule for pixel
+    * ownership, this just alters which point we consider the pixel
+    * center for that test.
+    */
+   unsigned gl_rasterization_rules:1;
 
    float line_width;
    float point_size;           /**< used when no per-vertex size */
@@ -179,7 +197,6 @@ struct pipe_depth_state
    unsigned enabled:1;         /**< depth test enabled? */
    unsigned writemask:1;       /**< allow depth buffer writes? */
    unsigned func:3;            /**< depth test func (PIPE_FUNC_x) */
-   unsigned occlusion_count:1; /**< do occlusion counting? */
 };
 
 
@@ -307,7 +324,7 @@ struct pipe_transfer
    unsigned nblocksx;            /**< allocated width in blocks */
    unsigned nblocksy;            /**< allocated height in blocks */
    unsigned stride;              /**< stride in bytes between rows of blocks */
-   unsigned usage;               /**< PIPE_TRANSFER_*  */
+   enum pipe_transfer_usage usage; /**< PIPE_TRANSFER_*  */
 
    struct pipe_texture *texture; /**< texture to transfer to/from  */
    unsigned face;
diff --git a/src/gallium/include/pipe/p_video_context.h b/src/gallium/include/pipe/p_video_context.h
new file mode 100644
index 0000000000..6ae31418fa
--- /dev/null
+++ b/src/gallium/include/pipe/p_video_context.h
@@ -0,0 +1,121 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef PIPE_VIDEO_CONTEXT_H
+#define PIPE_VIDEO_CONTEXT_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <pipe/p_video_state.h>
+
+struct pipe_screen;
+struct pipe_buffer;
+struct pipe_surface;
+struct pipe_video_surface;
+struct pipe_macroblock;
+struct pipe_picture_desc;
+struct pipe_fence_handle;
+
+/**
+ * Gallium video rendering context
+ */
+struct pipe_video_context
+{
+   struct pipe_screen *screen;
+   enum pipe_video_profile profile;
+   enum pipe_video_chroma_format chroma_format;
+   unsigned width;
+   unsigned height;
+
+   void *priv; /**< context private data (for DRI for example) */
+
+   void (*destroy)(struct pipe_video_context *vpipe);
+
+   /**
+    * Picture decoding and displaying
+    */
+   /*@{*/
+   void (*decode_bitstream)(struct pipe_video_context *vpipe,
+                            unsigned num_bufs,
+                            struct pipe_buffer **bitstream_buf);
+
+   void (*decode_macroblocks)(struct pipe_video_context *vpipe,
+                              struct pipe_video_surface *past,
+                              struct pipe_video_surface *future,
+                              unsigned num_macroblocks,
+                              struct pipe_macroblock *macroblocks,
+                              struct pipe_fence_handle **fence);
+
+   void (*clear_surface)(struct pipe_video_context *vpipe,
+                         unsigned x, unsigned y,
+                         unsigned width, unsigned height,
+                         unsigned value,
+                         struct pipe_surface *surface);
+
+   void (*render_picture)(struct pipe_video_context     *vpipe,
+                          /*struct pipe_surface         *backround,
+                          struct pipe_video_rect        *backround_area,*/
+                          struct pipe_video_surface     *src_surface,
+                          enum pipe_mpeg12_picture_type picture_type,
+                          /*unsigned                    num_past_surfaces,
+                          struct pipe_video_surface     *past_surfaces,
+                          unsigned                      num_future_surfaces,
+                          struct pipe_video_surface     *future_surfaces,*/
+                          struct pipe_video_rect        *src_area,
+                          struct pipe_surface           *dst_surface,
+                          struct pipe_video_rect        *dst_area,
+                          /*unsigned                      num_layers,
+                          struct pipe_texture           *layers,
+                          struct pipe_video_rect        *layer_src_areas,
+                          struct pipe_video_rect        *layer_dst_areas,*/
+                          struct pipe_fence_handle      **fence);
+   /*@}*/
+
+   /**
+    * Parameter-like states (or properties)
+    */
+   /*@{*/
+   void (*set_picture_desc)(struct pipe_video_context *vpipe,
+                            const struct pipe_picture_desc *desc);
+
+   void (*set_decode_target)(struct pipe_video_context *vpipe,
+                             struct pipe_video_surface *dt);
+
+   void (*set_csc_matrix)(struct pipe_video_context *vpipe, const float *mat);
+
+   /* TODO: Interface for scaling modes, post-processing, etc. */
+   /*@}*/
+};
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* PIPE_VIDEO_CONTEXT_H */
diff --git a/src/gallium/include/pipe/p_video_state.h b/src/gallium/include/pipe/p_video_state.h
new file mode 100644
index 0000000000..4da26d608c
--- /dev/null
+++ b/src/gallium/include/pipe/p_video_state.h
@@ -0,0 +1,184 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef PIPE_VIDEO_STATE_H
+#define PIPE_VIDEO_STATE_H
+
+/* u_reduce_video_profile() needs these */
+#include <pipe/p_compiler.h>
+#include <util/u_debug.h>
+
+#include <pipe/p_defines.h>
+#include <pipe/p_format.h>
+#include <pipe/p_refcnt.h>
+#include <pipe/p_screen.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct pipe_video_surface
+{
+   struct pipe_reference reference;
+   struct pipe_screen *screen;
+   enum pipe_video_chroma_format chroma_format;
+   /*enum pipe_video_surface_format surface_format;*/
+   unsigned width;
+   unsigned height;
+};
+
+static INLINE void
+pipe_video_surface_reference(struct pipe_video_surface **ptr, struct pipe_video_surface *surf)
+{
+   struct pipe_video_surface *old_surf = *ptr;
+
+   if (pipe_reference((struct pipe_reference **)ptr, &surf->reference))
+      old_surf->screen->video_surface_destroy(old_surf);
+}
+
+struct pipe_video_rect
+{
+   unsigned x, y, w, h;
+};
+
+static INLINE enum pipe_video_codec
+u_reduce_video_profile(enum pipe_video_profile profile)
+{
+   switch (profile)
+   {
+      case PIPE_VIDEO_PROFILE_MPEG1:
+      case PIPE_VIDEO_PROFILE_MPEG2_SIMPLE:
+      case PIPE_VIDEO_PROFILE_MPEG2_MAIN:
+         return PIPE_VIDEO_CODEC_MPEG12;
+
+      case PIPE_VIDEO_PROFILE_MPEG4_SIMPLE:
+      case PIPE_VIDEO_PROFILE_MPEG4_ADVANCED_SIMPLE:
+         return PIPE_VIDEO_CODEC_MPEG4;
+
+      case PIPE_VIDEO_PROFILE_VC1_SIMPLE:
+      case PIPE_VIDEO_PROFILE_VC1_MAIN:
+      case PIPE_VIDEO_PROFILE_VC1_ADVANCED:
+         return PIPE_VIDEO_CODEC_VC1;
+
+      case PIPE_VIDEO_PROFILE_MPEG4_AVC_BASELINE:
+      case PIPE_VIDEO_PROFILE_MPEG4_AVC_MAIN:
+      case PIPE_VIDEO_PROFILE_MPEG4_AVC_HIGH:
+         return PIPE_VIDEO_CODEC_MPEG4_AVC;
+
+      default:
+         assert(0);
+         return PIPE_VIDEO_CODEC_UNKNOWN;
+   }
+}
+
+enum pipe_mpeg12_picture_type
+{
+   PIPE_MPEG12_PICTURE_TYPE_FIELD_TOP,
+   PIPE_MPEG12_PICTURE_TYPE_FIELD_BOTTOM,
+   PIPE_MPEG12_PICTURE_TYPE_FRAME
+};
+
+enum pipe_mpeg12_macroblock_type
+{
+   PIPE_MPEG12_MACROBLOCK_TYPE_INTRA,
+   PIPE_MPEG12_MACROBLOCK_TYPE_FWD,
+   PIPE_MPEG12_MACROBLOCK_TYPE_BKWD,
+   PIPE_MPEG12_MACROBLOCK_TYPE_BI,
+	
+   PIPE_MPEG12_MACROBLOCK_NUM_TYPES
+};
+
+enum pipe_mpeg12_motion_type
+{
+   PIPE_MPEG12_MOTION_TYPE_FIELD,
+   PIPE_MPEG12_MOTION_TYPE_FRAME,
+   PIPE_MPEG12_MOTION_TYPE_DUALPRIME,
+   PIPE_MPEG12_MOTION_TYPE_16x8
+};
+
+enum pipe_mpeg12_dct_type
+{
+   PIPE_MPEG12_DCT_TYPE_FIELD,
+   PIPE_MPEG12_DCT_TYPE_FRAME
+};
+
+struct pipe_macroblock
+{
+   enum pipe_video_codec codec;
+};
+
+struct pipe_mpeg12_macroblock
+{
+   struct pipe_macroblock base;
+
+   unsigned mbx;
+   unsigned mby;
+   enum pipe_mpeg12_macroblock_type mb_type;
+   enum pipe_mpeg12_motion_type mo_type;
+   enum pipe_mpeg12_dct_type dct_type;
+   signed pmv[2][2][2];
+   unsigned cbp;
+   void *blocks;
+};
+
+#if 0
+struct pipe_picture_desc
+{
+   enum pipe_video_format format;
+};
+
+struct pipe_mpeg12_picture_desc
+{
+   struct pipe_picture_desc base;
+
+   /* TODO: Use bitfields where possible? */
+   struct pipe_surface *forward_reference;
+   struct pipe_surface *backward_reference;
+   unsigned picture_coding_type;
+   unsigned fcode;
+   unsigned intra_dc_precision;
+   unsigned picture_structure;
+   unsigned top_field_first;
+   unsigned frame_pred_frame_dct;
+   unsigned concealment_motion_vectors;
+   unsigned q_scale_type;
+   unsigned intra_vlc_format;
+   unsigned alternate_scan;
+   unsigned full_pel_forward_vector;
+   unsigned full_pel_backward_vector;
+   struct pipe_buffer *intra_quantizer_matrix;
+   struct pipe_buffer *non_intra_quantizer_matrix;
+   struct pipe_buffer *chroma_intra_quantizer_matrix;
+   struct pipe_buffer *chroma_non_intra_quantizer_matrix;
+};
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* PIPE_VIDEO_STATE_H */
diff --git a/src/gallium/state_trackers/dri/dri_drawable.c b/src/gallium/state_trackers/dri/dri_drawable.c
index 5cec9e329d..5625ff53cf 100644
--- a/src/gallium/state_trackers/dri/dri_drawable.c
+++ b/src/gallium/state_trackers/dri/dri_drawable.c
@@ -45,6 +45,7 @@
 #include "state_tracker/st_cb_fbo.h"
 
 #include "util/u_memory.h"
+#include "util/u_rect.h"
 
 static struct pipe_surface *
 dri_surface_from_handle(struct drm_api *api,
@@ -179,7 +180,6 @@ dri_get_buffers(__DRIdrawablePrivate * dPriv)
 
       switch (buffers[i].attachment) {
       case __DRI_BUFFER_FRONT_LEFT:
-	 continue;
       case __DRI_BUFFER_FAKE_FRONT_LEFT:
 	 index = ST_SURFACE_FRONT_LEFT;
 	 format = drawable->color_format;
@@ -214,6 +214,7 @@ dri_get_buffers(__DRIdrawablePrivate * dPriv)
 					dri_drawable->h, buffers[i].pitch);
 
       switch (buffers[i].attachment) {
+      case __DRI_BUFFER_FRONT_LEFT:
       case __DRI_BUFFER_FAKE_FRONT_LEFT:
       case __DRI_BUFFER_BACK_LEFT:
 	 drawable->color_format = surface->format;
@@ -223,6 +224,9 @@ dri_get_buffers(__DRIdrawablePrivate * dPriv)
       case __DRI_BUFFER_STENCIL:
 	 drawable->depth_stencil_format = surface->format;
 	 break;
+      case __DRI_BUFFER_ACCUM:
+      default:
+	 assert(0);
       }
 
       st_set_framebuffer_surface(drawable->stfb, index, surface);
@@ -250,6 +254,9 @@ void dri2_set_tex_buffer2(__DRIcontext *pDRICtx, GLint target,
    dri_get_buffers(drawable->dPriv);
    st_get_framebuffer_surface(drawable->stfb, ST_SURFACE_FRONT_LEFT, &ps);
 
+   if (!ps)
+      return;
+
    st_bind_texture_surface(ps, target == GL_TEXTURE_2D ? ST_TEXTURE_2D :
                            ST_TEXTURE_RECT, 0, drawable->color_format);
 }
@@ -360,8 +367,6 @@ dri_create_buffer(__DRIscreenPrivate * sPriv,
 
    if (visual->doubleBufferMode)
       drawable->attachments[i++] = __DRI_BUFFER_BACK_LEFT;
-   else
-      drawable->attachments[i++] = __DRI_BUFFER_FAKE_FRONT_LEFT;
    if (visual->depthBits && visual->stencilBits)
       drawable->attachments[i++] = __DRI_BUFFER_DEPTH_STENCIL;
    else if (visual->depthBits)
@@ -537,12 +542,21 @@ dri1_swap_copy(struct dri_context *ctx,
    cur = dPriv->pClipRects;
 
    for (i = 0; i < dPriv->numClipRects; ++i) {
-      if (dri1_intersect_src_bbox(&clip, dPriv->x, dPriv->y, cur++, bbox))
-	 pipe->surface_copy(pipe, dst, clip.x1, clip.y1,
-			    src,
-			    (int)clip.x1 - dPriv->x,
-			    (int)clip.y1 - dPriv->y,
-			    clip.x2 - clip.x1, clip.y2 - clip.y1);
+      if (dri1_intersect_src_bbox(&clip, dPriv->x, dPriv->y, cur++, bbox)) {
+         if (pipe->surface_copy) {
+            pipe->surface_copy(pipe, dst, clip.x1, clip.y1,
+                               src,
+                               (int)clip.x1 - dPriv->x,
+                               (int)clip.y1 - dPriv->y,
+                               clip.x2 - clip.x1, clip.y2 - clip.y1);
+         } else {
+            util_surface_copy(pipe, FALSE, dst, clip.x1, clip.y1,
+                              src,
+                              (int)clip.x1 - dPriv->x,
+                              (int)clip.y1 - dPriv->y,
+                              clip.x2 - clip.x1, clip.y2 - clip.y1);
+         }
+      }
    }
 }
 
diff --git a/src/gallium/state_trackers/dri/dri_extensions.c b/src/gallium/state_trackers/dri/dri_extensions.c
index 4349a4d1d2..800d462e32 100644
--- a/src/gallium/state_trackers/dri/dri_extensions.c
+++ b/src/gallium/state_trackers/dri/dri_extensions.c
@@ -37,8 +37,10 @@
 #define need_GL_ARB_multisample
 #define need_GL_ARB_occlusion_query
 #define need_GL_ARB_point_parameters
+#define need_GL_ARB_provoking_vertex
 #define need_GL_ARB_shader_objects
 #define need_GL_ARB_texture_compression
+#define need_GL_ARB_vertex_array_object
 #define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_vertex_program
 #define need_GL_ARB_vertex_shader
@@ -51,7 +53,9 @@
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_framebuffer_object
 #define need_GL_EXT_multi_draw_arrays
+#define need_GL_EXT_provoking_vertex
 #define need_GL_EXT_secondary_color
+#define need_GL_APPLE_vertex_array_object
 #define need_GL_NV_vertex_program
 #define need_GL_VERSION_2_0
 #define need_GL_VERSION_2_1
@@ -67,6 +71,7 @@ const struct dri_extension card_extensions[] = {
    {"GL_ARB_multitexture", NULL},
    {"GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions},
    {"GL_ARB_pixel_buffer_object", NULL},
+   {"GL_ARB_provoking_vertex", GL_ARB_provoking_vertex_functions},
    {"GL_ARB_point_parameters", GL_ARB_point_parameters_functions},
    {"GL_ARB_shading_language_100", GL_VERSION_2_0_functions },
    {"GL_ARB_shading_language_120", GL_VERSION_2_1_functions },
@@ -79,6 +84,7 @@ const struct dri_extension card_extensions[] = {
    {"GL_ARB_texture_env_dot3", NULL},
    {"GL_ARB_texture_mirrored_repeat", NULL},
    {"GL_ARB_texture_rectangle", NULL},
+   {"GL_ARB_vertex_array_object", GL_ARB_vertex_array_object_functions},
    {"GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions},
    {"GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions},
    {"GL_ARB_vertex_program", GL_ARB_vertex_program_functions},
@@ -94,6 +100,7 @@ const struct dri_extension card_extensions[] = {
    {"GL_EXT_multi_draw_arrays", GL_EXT_multi_draw_arrays_functions},
    {"GL_EXT_packed_depth_stencil", NULL},
    {"GL_EXT_pixel_buffer_object", NULL},
+   {"GL_EXT_provoking_vertex", GL_EXT_provoking_vertex_functions},
    {"GL_EXT_secondary_color", GL_EXT_secondary_color_functions},
    {"GL_EXT_stencil_wrap", NULL},
    {"GL_EXT_texture_edge_clamp", NULL},
@@ -103,6 +110,7 @@ const struct dri_extension card_extensions[] = {
    {"GL_EXT_texture_lod_bias", NULL},
    {"GL_3DFX_texture_compression_FXT1", NULL},
    {"GL_APPLE_client_storage", NULL},
+   {"GL_APPLE_vertex_array_object", GL_APPLE_vertex_array_object_functions},
    {"GL_MESA_pack_invert", NULL},
    {"GL_MESA_ycbcr_texture", NULL},
    {"GL_NV_blend_square", NULL},
diff --git a/src/gallium/state_trackers/egl/egl_context.c b/src/gallium/state_trackers/egl/egl_context.c
index c4f7361ca0..e21a4a1095 100644
--- a/src/gallium/state_trackers/egl/egl_context.c
+++ b/src/gallium/state_trackers/egl/egl_context.c
@@ -138,7 +138,6 @@ drm_destroy_context(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *context)
 	struct drm_context *c = lookup_drm_context(context);
 	if (!_eglIsContextBound(&c->base)) {
 		st_destroy_context(c->st);
-		c->pipe->destroy(c->pipe);
 		free(c);
 	}
 	return EGL_TRUE;
@@ -160,18 +159,12 @@ drm_make_current(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *draw, _EGLSurfa
 		if (!drawSurf || !readSurf)
 			return EGL_FALSE;
 
-		drawSurf->user = ctx;
-		readSurf->user = ctx;
-
 		st_make_current(ctx->st, drawSurf->stfb, readSurf->stfb);
 
 		/* st_resize_framebuffer needs a bound context to work */
 		st_resize_framebuffer(drawSurf->stfb, drawSurf->w, drawSurf->h);
 		st_resize_framebuffer(readSurf->stfb, readSurf->w, readSurf->h);
 	} else {
-		drawSurf->user = NULL;
-		readSurf->user = NULL;
-
 		st_make_current(NULL, NULL, NULL);
 	}
 
diff --git a/src/gallium/state_trackers/egl/egl_surface.c b/src/gallium/state_trackers/egl/egl_surface.c
index 69e2d6b708..71c013756d 100644
--- a/src/gallium/state_trackers/egl/egl_surface.c
+++ b/src/gallium/state_trackers/egl/egl_surface.c
@@ -12,6 +12,8 @@
 
 #include "state_tracker/drm_api.h"
 
+#include "util/u_rect.h"
+
 /*
  * Util functions
  */
@@ -152,7 +154,6 @@ drm_takedown_shown_screen(_EGLDisplay *dpy, struct drm_screen *screen)
 
 	pipe_surface_reference(&screen->surface, NULL);
 	pipe_texture_reference(&screen->tex, NULL);
-	pipe_buffer_reference(&screen->buffer, NULL);
 
 	screen->shown = 0;
 }
@@ -250,8 +251,8 @@ drm_show_screen_surface_mesa(_EGLDriver *drv, _EGLDisplay *dpy,
 
 
 	drm_create_texture(dpy, scrn, mode->Width, mode->Height);
-	if (!scrn->buffer)
-		return EGL_FALSE;
+	if (!scrn->tex)
+		goto err_tex;
 
 	ret = drmModeAddFB(dev->drmFD,
 	                   scrn->front.width, scrn->front.height,
@@ -325,8 +326,8 @@ err_fb:
 err_bo:
 	pipe_surface_reference(&scrn->surface, NULL);
 	pipe_texture_reference(&scrn->tex, NULL);
-	pipe_buffer_reference(&scrn->buffer, NULL);
 
+err_tex:
 	return EGL_FALSE;
 }
 
@@ -353,24 +354,30 @@ drm_swap_buffers(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *draw)
 	if (!surf)
 		return EGL_FALSE;
 
-	/* error checking */
-	if (!_eglSwapBuffers(drv, dpy, draw))
-		return EGL_FALSE;
-
 	st_get_framebuffer_surface(surf->stfb, ST_SURFACE_BACK_LEFT, &back_surf);
 
 	if (back_surf) {
+		struct drm_context *ctx = lookup_drm_context(draw->Binding);
 
 		st_notify_swapbuffers(surf->stfb);
 
-		if (surf->screen) {
-			surf->user->pipe->surface_copy(surf->user->pipe,
-				surf->screen->surface,
-				0, 0,
-				back_surf,
-				0, 0,
-				surf->w, surf->h);
-			surf->user->pipe->flush(surf->user->pipe, PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_TEXTURE_CACHE, NULL);
+		if (ctx && surf->screen) {
+            if (ctx->pipe->surface_copy) {
+                ctx->pipe->surface_copy(ctx->pipe,
+                    surf->screen->surface,
+                    0, 0,
+                    back_surf,
+                    0, 0,
+                    surf->w, surf->h);
+            } else {
+                util_surface_copy(ctx->pipe, FALSE,
+                    surf->screen->surface,
+                    0, 0,
+                    back_surf,
+                    0, 0,
+                    surf->w, surf->h);
+            }
+			ctx->pipe->flush(ctx->pipe, PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_TEXTURE_CACHE, NULL);
 
 #ifdef DRM_MODE_FEATURE_DIRTYFB
 			/* TODO query connector property to see if this is needed */
diff --git a/src/gallium/state_trackers/egl/egl_tracker.h b/src/gallium/state_trackers/egl/egl_tracker.h
index dd4730f957..73eb1a1226 100644
--- a/src/gallium/state_trackers/egl/egl_tracker.h
+++ b/src/gallium/state_trackers/egl/egl_tracker.h
@@ -69,7 +69,6 @@ struct drm_surface
 	 * drm
 	 */
 
-	struct drm_context *user;
 	struct drm_screen *screen;
 
 	int w;
@@ -94,7 +93,6 @@ struct drm_screen
 	 * pipe
 	 */
 
-	struct pipe_buffer *buffer;
 	struct pipe_texture *tex;
 	struct pipe_surface *surface;
 
diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile
deleted file mode 100644
index f9f4d6be3c..0000000000
--- a/src/gallium/state_trackers/g3dvl/Makefile
+++ /dev/null
@@ -1,21 +0,0 @@
-TARGET		= libg3dvl.a
-OBJECTS		= vl_display.o vl_screen.o vl_context.o vl_surface.o vl_shader_build.o vl_util.o vl_basic_csc.o	\
-		  vl_r16snorm_mc_buf.o
-GALLIUMDIR	= ../..
-
-CFLAGS		+= -g -Wall -Werror-implicit-function-declaration -fPIC	\
-		   -I${GALLIUMDIR}/include				\
-		   -I${GALLIUMDIR}/auxiliary				\
-		   -I${GALLIUMDIR}/winsys/g3dvl				\
-
-#############################################
-
-.PHONY	= all clean
-
-all: ${TARGET}
-
-${TARGET}: ${OBJECTS}
-	ar rcs $@ $^
-
-clean:
-	rm -rf ${OBJECTS} ${TARGET}
diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c
deleted file mode 100644
index 20d682de3f..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c
+++ /dev/null
@@ -1,714 +0,0 @@
-#define VL_INTERNAL
-#include "vl_basic_csc.h"
-#include <assert.h>
-#include <pipe/p_context.h>
-#include <pipe/p_state.h>
-#include <pipe/p_inlines.h>
-#include <tgsi/tgsi_parse.h>
-#include <tgsi/tgsi_build.h>
-#include <util/u_memory.h>
-#include "vl_csc.h"
-#include "vl_surface.h"
-#include "vl_shader_build.h"
-#include "vl_types.h"
-
-struct vlVertexShaderConsts
-{
-	struct vlVertex4f	dst_scale;
-	struct vlVertex4f	dst_trans;
-	struct vlVertex4f	src_scale;
-	struct vlVertex4f	src_trans;
-};
-
-struct vlFragmentShaderConsts
-{
-	struct vlVertex4f	bias;
-	float			matrix[16];
-};
-
-struct vlBasicCSC
-{
-	struct vlCSC				base;
-
-	struct pipe_context			*pipe;
-	struct pipe_viewport_state		viewport;
-	struct pipe_framebuffer_state		framebuffer;
-	struct pipe_texture			*framebuffer_tex;
-	void					*sampler;
-	void					*vertex_shader, *fragment_shader;
-	struct pipe_vertex_buffer 		vertex_bufs[2];
-	struct pipe_vertex_element		vertex_elems[2];
-	struct pipe_constant_buffer		vs_const_buf, fs_const_buf;
-};
-
-static int vlResizeFrameBuffer
-(
-	struct vlCSC *csc,
-	unsigned int width,
-	unsigned int height
-)
-{
-	struct vlBasicCSC	*basic_csc;
-	struct pipe_context	*pipe;
-	struct pipe_texture	template;
-
-	assert(csc);
-
-	basic_csc = (struct vlBasicCSC*)csc;
-	pipe = basic_csc->pipe;
-
-	if (basic_csc->framebuffer.width == width && basic_csc->framebuffer.height == height)
-		return 0;
-
-	basic_csc->viewport.scale[0] = width;
-	basic_csc->viewport.scale[1] = height;
-	basic_csc->viewport.scale[2] = 1;
-	basic_csc->viewport.scale[3] = 1;
-	basic_csc->viewport.translate[0] = 0;
-	basic_csc->viewport.translate[1] = 0;
-	basic_csc->viewport.translate[2] = 0;
-	basic_csc->viewport.translate[3] = 0;
-	
-	if (basic_csc->framebuffer_tex)
-	{
-		pipe_surface_reference(&basic_csc->framebuffer.cbufs[0], NULL);
-		pipe_texture_reference(&basic_csc->framebuffer_tex, NULL);
-	}
-	
-	memset(&template, 0, sizeof(struct pipe_texture));
-	template.target = PIPE_TEXTURE_2D;
-	template.format = PIPE_FORMAT_A8R8G8B8_UNORM;
-	template.last_level = 0;
-	template.width[0] = width;
-	template.height[0] = height;
-	template.depth[0] = 1;
-	pf_get_block(template.format, &template.block);
-	template.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
-
-	basic_csc->framebuffer_tex = pipe->screen->texture_create(pipe->screen, &template);
-
-	basic_csc->framebuffer.width = width;
-	basic_csc->framebuffer.height = height;
-	basic_csc->framebuffer.cbufs[0] = pipe->screen->get_tex_surface
-	(
-		pipe->screen,
-		basic_csc->framebuffer_tex,
-		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
-	);
-
-	/* Clear to black, in case video doesn't fill the entire window */
-	pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer);
-	pipe->clear(pipe, PIPE_CLEAR_COLOR, 0, 0.0f, 0);
-
-	return 0;
-}
-
-static int vlBegin
-(
-	struct vlCSC *csc
-)
-{
-	struct vlBasicCSC	*basic_csc;
-	struct pipe_context	*pipe;
-
-	assert(csc);
-
-	basic_csc = (struct vlBasicCSC*)csc;
-	pipe = basic_csc->pipe;
-
-	pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer);
-	pipe->set_viewport_state(pipe, &basic_csc->viewport);
-	pipe->bind_sampler_states(pipe, 1, (void**)&basic_csc->sampler);
-	/* Source texture set in vlPutPictureCSC() */
-	pipe->bind_vs_state(pipe, basic_csc->vertex_shader);
-	pipe->bind_fs_state(pipe, basic_csc->fragment_shader);
-	pipe->set_vertex_buffers(pipe, 2, basic_csc->vertex_bufs);
-	pipe->set_vertex_elements(pipe, 2, basic_csc->vertex_elems);
-	pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &basic_csc->vs_const_buf);
-	pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &basic_csc->fs_const_buf);
-
-	return 0;
-}
-
-static int vlPutPictureCSC
-(
-	struct vlCSC *csc,
-	struct vlSurface *surface,
-	int srcx,
-	int srcy,
-	int srcw,
-	int srch,
-	int destx,
-	int desty,
-	int destw,
-	int desth,
-	enum vlPictureType picture_type
-)
-{
-	struct vlBasicCSC		*basic_csc;
-	struct pipe_context		*pipe;
-	struct vlVertexShaderConsts	*vs_consts;
-
-	assert(csc);
-	assert(surface);
-
-	basic_csc = (struct vlBasicCSC*)csc;
-	pipe = basic_csc->pipe;
-
-	vs_consts = pipe_buffer_map
-	(
-		pipe->screen,
-		basic_csc->vs_const_buf.buffer,
-		PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
-	);
-
-	vs_consts->dst_scale.x = destw / (float)basic_csc->framebuffer.cbufs[0]->width;
-	vs_consts->dst_scale.y = desth / (float)basic_csc->framebuffer.cbufs[0]->height;
-	vs_consts->dst_scale.z = 1;
-	vs_consts->dst_scale.w = 1;
-	vs_consts->dst_trans.x = destx / (float)basic_csc->framebuffer.cbufs[0]->width;
-	vs_consts->dst_trans.y = desty / (float)basic_csc->framebuffer.cbufs[0]->height;
-	vs_consts->dst_trans.z = 0;
-	vs_consts->dst_trans.w = 0;
-
-	vs_consts->src_scale.x = srcw / (float)surface->texture->width[0];
-	vs_consts->src_scale.y = srch / (float)surface->texture->height[0];
-	vs_consts->src_scale.z = 1;
-	vs_consts->src_scale.w = 1;
-	vs_consts->src_trans.x = srcx / (float)surface->texture->width[0];
-	vs_consts->src_trans.y = srcy / (float)surface->texture->height[0];
-	vs_consts->src_trans.z = 0;
-	vs_consts->src_trans.w = 0;
-
-	pipe_buffer_unmap(pipe->screen, basic_csc->vs_const_buf.buffer);
-
-	pipe->set_sampler_textures(pipe, 1, &surface->texture);
-	pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4);
-
-	return 0;
-}
-
-static int vlEnd
-(
-	struct vlCSC *csc
-)
-{
-	assert(csc);
-
-	return 0;
-}
-
-static struct pipe_surface* vlGetFrameBuffer
-(
-	struct vlCSC *csc
-)
-{
-	struct vlBasicCSC	*basic_csc;
-
-	assert(csc);
-
-	basic_csc = (struct vlBasicCSC*)csc;
-
-	return basic_csc->framebuffer.cbufs[0];
-}
-
-static int vlDestroy
-(
-	struct vlCSC *csc
-)
-{
-	struct vlBasicCSC	*basic_csc;
-	struct pipe_context	*pipe;
-	unsigned int		i;
-
-	assert(csc);
-
-	basic_csc = (struct vlBasicCSC*)csc;
-	pipe = basic_csc->pipe;
-
-	if (basic_csc->framebuffer_tex)
-	{
-		pipe_surface_reference(&basic_csc->framebuffer.cbufs[0], NULL);
-		pipe_texture_reference(&basic_csc->framebuffer_tex, NULL);
-	}
-
-	pipe->delete_sampler_state(pipe, basic_csc->sampler);
-	pipe->delete_vs_state(pipe, basic_csc->vertex_shader);
-	pipe->delete_fs_state(pipe, basic_csc->fragment_shader);
-
-	for (i = 0; i < 2; ++i)
-		pipe_buffer_reference(&basic_csc->vertex_bufs[i].buffer, NULL);
-
-	pipe_buffer_reference(&basic_csc->vs_const_buf.buffer, NULL);
-	pipe_buffer_reference(&basic_csc->fs_const_buf.buffer, NULL);
-
-	FREE(basic_csc);
-
-	return 0;
-}
-
-/*
- * Represents 2 triangles in a strip in normalized coords.
- * Used to render the surface onto the frame buffer.
- */
-static const struct vlVertex2f surface_verts[4] =
-{
-	{0.0f, 0.0f},
-	{0.0f, 1.0f},
-	{1.0f, 0.0f},
-	{1.0f, 1.0f}
-};
-
-/*
- * Represents texcoords for the above. We can use the position values directly.
- * TODO: Duplicate these in the shader, no need to create a buffer.
- */
-static const struct vlVertex2f *surface_texcoords = surface_verts;
-
-/*
- * Identity color conversion constants, for debugging
- */
-static const struct vlFragmentShaderConsts identity =
-{
-	{
-		0.0f, 0.0f, 0.0f, 0.0f
-	},
-	{
-		1.0f, 0.0f, 0.0f, 0.0f,
-		0.0f, 1.0f, 0.0f, 0.0f,
-		0.0f, 0.0f, 1.0f, 0.0f,
-		0.0f, 0.0f, 0.0f, 1.0f
-	}
-};
-
-/*
- * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where:
- * Y is in [16,235], Cb and Cr are in [16,240]
- * R, G, and B are in [16,235]
- */
-static const struct vlFragmentShaderConsts bt_601 =
-{
-	{
-		0.0f,		0.501960784f,	0.501960784f,	0.0f
-	},
-	{
-		1.0f,		0.0f,		1.371f,		0.0f,
-		1.0f,		-0.336f,	-0.698f,	0.0f,
-		1.0f,		1.732f,		0.0f,		0.0f,
-		0.0f,		0.0f,		0.0f,		1.0f
-	}
-};
-
-/*
- * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where:
- * Y is in [16,235], Cb and Cr are in [16,240]
- * R, G, and B are in [0,255]
- */
-static const struct vlFragmentShaderConsts bt_601_full =
-{
-	{
-		0.062745098f,	0.501960784f,	0.501960784f,	0.0f
-	},
-	{
-		1.164f,		0.0f,		1.596f,		0.0f,
-		1.164f,		-0.391f,	-0.813f,	0.0f,
-		1.164f,		2.018f,		0.0f,		0.0f,
-		0.0f,		0.0f,		0.0f,		1.0f
-	}
-};
-
-/*
- * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where:
- * Y is in [16,235], Cb and Cr are in [16,240]
- * R, G, and B are in [16,235]
- */
-static const struct vlFragmentShaderConsts bt_709 =
-{
-	{
-		0.0f,		0.501960784f,	0.501960784f,	0.0f
-	},
-	{
-		1.0f,		0.0f,		1.540f,		0.0f,
-		1.0f,		-0.183f,	-0.459f,	0.0f,
-		1.0f,		1.816f,		0.0f,		0.0f,
-		0.0f,		0.0f,		0.0f,		1.0f
-	}
-};
-
-/*
- * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where:
- * Y is in [16,235], Cb and Cr are in [16,240]
- * R, G, and B are in [0,255]
- */
-const struct vlFragmentShaderConsts bt_709_full =
-{
-	{
-		0.062745098f,	0.501960784f,	0.501960784f,	0.0f
-	},
-	{
-		1.164f,		0.0f,		1.793f,		0.0f,
-		1.164f,		-0.213f,	-0.534f,	0.0f,
-		1.164f,		2.115f,		0.0f,		0.0f,
-		0.0f,		0.0f,		0.0f,		1.0f
-	}
-};
-
-static int vlCreateVertexShader
-(
-	struct vlBasicCSC *csc
-)
-{
-	const unsigned int		max_tokens = 50;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	vs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(csc);
-
-	pipe = csc->pipe;
-	tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0		; Vertex pos
-	 * decl i1		; Vertex texcoords
-	 */
-	for (i = 0; i < 2; i++)
-	{
-		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl c0		; Scaling vector to scale vertex pos rect to destination size
-	 * decl c1		; Translation vector to move vertex pos rect into position
-	 * decl c2		; Scaling vector to scale texcoord rect to source size
-	 * decl c3		; Translation vector to move texcoord rect into position
-	 */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl o0		; Vertex pos
-	 * decl o1		; Vertex texcoords
-	 */
-	for (i = 0; i < 2; i++)
-	{
-		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* decl t0, t1 */
-	decl = vl_decl_temps(0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * madd o0, i0, c0, c1	; Scale and translate unit output rect to destination size and pos
-	 * madd o1, i1, c2, c3	; Scale and translate unit texcoord rect to source size and pos
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_inst4(TGSI_OPCODE_MADD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	vs.tokens = tokens;
-	csc->vertex_shader = pipe->create_vs_state(pipe, &vs);
-	FREE(tokens);
-
-	return 0;
-}
-
-static int vlCreateFragmentShader
-(
-	struct vlBasicCSC *csc
-)
-{
-	const unsigned int		max_tokens = 50;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	fs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(csc);
-
-	pipe = csc->pipe;
-	tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
-
-	ti = 3;
-
-	/* decl i0		; Texcoords for s0 */
-	decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl c0		; Bias vector for CSC
-	 * decl c1-c4		; CSC matrix c1-c4
-	 */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl o0		; Fragment color */
-	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl t0 */
-	decl = vl_decl_temps(0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl s0		; Sampler for tex containing picture to display */
-	decl = vl_decl_samplers(0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* tex2d t0, i0, s0	; Read src pixel */
-	inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* sub t0, t0, c0	; Subtract bias vector from pixel */
-	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * dp4 o0.x, t0, c1	; Multiply pixel by the color conversion matrix
-	 * dp4 o0.y, t0, c2
-	 * dp4 o0.z, t0, c3
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1);
-		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	fs.tokens = tokens;
-	csc->fragment_shader = pipe->create_fs_state(pipe, &fs);
-	FREE(tokens);
-
-	return 0;
-}
-
-static int vlCreateDataBufs
-(
-	struct vlBasicCSC *csc
-)
-{
-	struct pipe_context *pipe;
-
-	assert(csc);
-
-	pipe = csc->pipe;
-
-	/*
-	 * Create our vertex buffer and vertex buffer element
-	 * VB contains 4 vertices that render a quad covering the entire window
-	 * to display a rendered surface
-	 * Quad is rendered as a tri strip
-	 */
-	csc->vertex_bufs[0].stride = sizeof(struct vlVertex2f);
-	csc->vertex_bufs[0].max_index = 3;
-	csc->vertex_bufs[0].buffer_offset = 0;
-	csc->vertex_bufs[0].buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		1,
-		PIPE_BUFFER_USAGE_VERTEX,
-		sizeof(struct vlVertex2f) * 4
-	);
-
-	memcpy
-	(
-		pipe_buffer_map(pipe->screen, csc->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
-		surface_verts,
-		sizeof(struct vlVertex2f) * 4
-	);
-
-	pipe_buffer_unmap(pipe->screen, csc->vertex_bufs[0].buffer);
-
-	csc->vertex_elems[0].src_offset = 0;
-	csc->vertex_elems[0].vertex_buffer_index = 0;
-	csc->vertex_elems[0].nr_components = 2;
-	csc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/*
-	 * Create our texcoord buffer and texcoord buffer element
-	 * Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices
-	 */
-	csc->vertex_bufs[1].stride = sizeof(struct vlVertex2f);
-	csc->vertex_bufs[1].max_index = 3;
-	csc->vertex_bufs[1].buffer_offset = 0;
-	csc->vertex_bufs[1].buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		1,
-		PIPE_BUFFER_USAGE_VERTEX,
-		sizeof(struct vlVertex2f) * 4
-	);
-
-	memcpy
-	(
-		pipe_buffer_map(pipe->screen, csc->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
-		surface_texcoords,
-		sizeof(struct vlVertex2f) * 4
-	);
-
-	pipe_buffer_unmap(pipe->screen, csc->vertex_bufs[1].buffer);
-
-	csc->vertex_elems[1].src_offset = 0;
-	csc->vertex_elems[1].vertex_buffer_index = 1;
-	csc->vertex_elems[1].nr_components = 2;
-	csc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/*
-	 * Create our vertex shader's constant buffer
-	 * Const buffer contains scaling and translation vectors
-	 */
-	csc->vs_const_buf.buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		1,
-		PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD,
-		sizeof(struct vlVertexShaderConsts)
-	);
-
-	/*
-	 * Create our fragment shader's constant buffer
-	 * Const buffer contains the color conversion matrix and bias vectors
-	 */
-	csc->fs_const_buf.buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		1,
-		PIPE_BUFFER_USAGE_CONSTANT,
-		sizeof(struct vlFragmentShaderConsts)
-	);
-
-	/*
-	 * TODO: Refactor this into a seperate function,
-	 * allow changing the CSC matrix at runtime to switch between regular & full versions
-	 */
-	memcpy
-	(
-		pipe_buffer_map(pipe->screen, csc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
-		&bt_601_full,
-		sizeof(struct vlFragmentShaderConsts)
-	);
-
-	pipe_buffer_unmap(pipe->screen, csc->fs_const_buf.buffer);
-
-	return 0;
-}
-
-static int vlInit
-(
-	struct vlBasicCSC *csc
-)
-{
-	struct pipe_context		*pipe;
-	struct pipe_sampler_state	sampler;
-
-	assert(csc);
-
-	pipe = csc->pipe;
-
-	/* Delay creating the FB until vlPutPictureCSC() so we know window size */
-	csc->framebuffer_tex = NULL;
-	csc->framebuffer.width = 0;
-	csc->framebuffer.height = 0;
-	csc->framebuffer.nr_cbufs = 1;
-	csc->framebuffer.cbufs[0] = NULL;
-	csc->framebuffer.zsbuf = NULL;
-
-	sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
-	sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
-	sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
-	sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR;
-	sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
-	sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR;
-	sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
-	sampler.compare_func = PIPE_FUNC_ALWAYS;
-	sampler.normalized_coords = 1;
-	/*sampler.prefilter = ;*/
-	/*sampler.lod_bias = ;*/
-	/*sampler.min_lod = ;*/
-	/*sampler.max_lod = ;*/
-	/*sampler.border_color[i] = ;*/
-	/*sampler.max_anisotropy = ;*/
-	csc->sampler = pipe->create_sampler_state(pipe, &sampler);
-
-	vlCreateVertexShader(csc);
-	vlCreateFragmentShader(csc);
-	vlCreateDataBufs(csc);
-
-	return 0;
-}
-
-int vlCreateBasicCSC
-(
-	struct pipe_context *pipe,
-	struct vlCSC **csc
-)
-{
-	struct vlBasicCSC *basic_csc;
-
-	assert(pipe);
-	assert(csc);
-
-	basic_csc = CALLOC_STRUCT(vlBasicCSC);
-
-	if (!basic_csc)
-		return 1;
-
-	basic_csc->base.vlResizeFrameBuffer = &vlResizeFrameBuffer;
-	basic_csc->base.vlBegin = &vlBegin;
-	basic_csc->base.vlPutPicture = &vlPutPictureCSC;
-	basic_csc->base.vlEnd = &vlEnd;
-	basic_csc->base.vlGetFrameBuffer = &vlGetFrameBuffer;
-	basic_csc->base.vlDestroy = &vlDestroy;
-	basic_csc->pipe = pipe;
-
-	vlInit(basic_csc);
-
-	*csc = &basic_csc->base;
-
-	return 0;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.h b/src/gallium/state_trackers/g3dvl/vl_basic_csc.h
deleted file mode 100644
index 2e17f1d814..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_basic_csc.h
+++ /dev/null
@@ -1,13 +0,0 @@
-#ifndef vl_basic_csc_h
-#define vl_basic_csc_h
-
-struct pipe_context;
-struct vlCSC;
-
-int vlCreateBasicCSC
-(
-	struct pipe_context *pipe,
-	struct vlCSC **csc
-);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c
deleted file mode 100644
index 5cfd233c4c..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_context.c
+++ /dev/null
@@ -1,205 +0,0 @@
-#define VL_INTERNAL
-#include "vl_context.h"
-#include <assert.h>
-#include <pipe/p_context.h>
-#include <pipe/p_state.h>
-#include <util/u_memory.h>
-#include "vl_render.h"
-#include "vl_r16snorm_mc_buf.h"
-#include "vl_csc.h"
-#include "vl_basic_csc.h"
-
-static int vlInitCommon(struct vlContext *context)
-{
-	struct pipe_context			*pipe;
-	struct pipe_rasterizer_state		rast;
-	struct pipe_blend_state			blend;
-	struct pipe_depth_stencil_alpha_state	dsa;
-	unsigned int				i;
-
-	assert(context);
-
-	pipe = context->pipe;
-
-	rast.flatshade = 1;
-	rast.flatshade_first = 0;
-	rast.light_twoside = 0;
-	rast.front_winding = PIPE_WINDING_CCW;
-	rast.cull_mode = PIPE_WINDING_CW;
-	rast.fill_cw = PIPE_POLYGON_MODE_FILL;
-	rast.fill_ccw = PIPE_POLYGON_MODE_FILL;
-	rast.offset_cw = 0;
-	rast.offset_ccw = 0;
-	rast.scissor = 0;
-	rast.poly_smooth = 0;
-	rast.poly_stipple_enable = 0;
-	rast.point_sprite = 0;
-	rast.point_size_per_vertex = 0;
-	rast.multisample = 0;
-	rast.line_smooth = 0;
-	rast.line_stipple_enable = 0;
-	rast.line_stipple_factor = 0;
-	rast.line_stipple_pattern = 0;
-	rast.line_last_pixel = 0;
-	rast.bypass_vs_clip_and_viewport = 0;
-	rast.line_width = 1;
-	rast.point_smooth = 0;
-	rast.point_size = 1;
-	rast.offset_units = 1;
-	rast.offset_scale = 1;
-	/*rast.sprite_coord_mode[i] = ;*/
-	context->raster = pipe->create_rasterizer_state(pipe, &rast);
-	pipe->bind_rasterizer_state(pipe, context->raster);
-
-	blend.blend_enable = 0;
-	blend.rgb_func = PIPE_BLEND_ADD;
-	blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE;
-	blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
-	blend.alpha_func = PIPE_BLEND_ADD;
-	blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE;
-	blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
-	blend.logicop_enable = 0;
-	blend.logicop_func = PIPE_LOGICOP_CLEAR;
-	/* Needed to allow color writes to FB, even if blending disabled */
-	blend.colormask = PIPE_MASK_RGBA;
-	blend.dither = 0;
-	context->blend = pipe->create_blend_state(pipe, &blend);
-	pipe->bind_blend_state(pipe, context->blend);
-
-	dsa.depth.enabled = 0;
-	dsa.depth.writemask = 0;
-	dsa.depth.func = PIPE_FUNC_ALWAYS;
-	dsa.depth.occlusion_count = 0;
-	for (i = 0; i < 2; ++i)
-	{
-		dsa.stencil[i].enabled = 0;
-		dsa.stencil[i].func = PIPE_FUNC_ALWAYS;
-		dsa.stencil[i].fail_op = PIPE_STENCIL_OP_KEEP;
-		dsa.stencil[i].zpass_op = PIPE_STENCIL_OP_KEEP;
-		dsa.stencil[i].zfail_op = PIPE_STENCIL_OP_KEEP;
-		dsa.stencil[i].ref_value = 0;
-		dsa.stencil[i].valuemask = 0;
-		dsa.stencil[i].writemask = 0;
-	}
-	dsa.alpha.enabled = 0;
-	dsa.alpha.func = PIPE_FUNC_ALWAYS;
-	dsa.alpha.ref_value = 0;
-	context->dsa = pipe->create_depth_stencil_alpha_state(pipe, &dsa);
-	pipe->bind_depth_stencil_alpha_state(pipe, context->dsa);
-
-	return 0;
-}
-
-int vlCreateContext
-(
-	struct vlScreen *screen,
-	struct pipe_context *pipe,
-	unsigned int picture_width,
-	unsigned int picture_height,
-	enum vlFormat picture_format,
-	enum vlProfile profile,
-	enum vlEntryPoint entry_point,
-	struct vlContext **context
-)
-{
-	struct vlContext *ctx;
-
-	assert(screen);
-	assert(context);
-	assert(pipe);
-
-	ctx = CALLOC_STRUCT(vlContext);
-
-	if (!ctx)
-		return 1;
-
-	ctx->screen = screen;
-	ctx->pipe = pipe;
-	ctx->picture_width = picture_width;
-	ctx->picture_height = picture_height;
-	ctx->picture_format = picture_format;
-	ctx->profile = profile;
-	ctx->entry_point = entry_point;
-
-	vlInitCommon(ctx);
-
-	vlCreateR16SNormBufferedMC(pipe, picture_width, picture_height, picture_format, &ctx->render);
-	vlCreateBasicCSC(pipe, &ctx->csc);
-
-	*context = ctx;
-
-	return 0;
-}
-
-int vlDestroyContext
-(
-	struct vlContext *context
-)
-{
-	assert(context);
-
-	/* XXX: Must unbind shaders before we can delete them for some reason */
-	context->pipe->bind_vs_state(context->pipe, NULL);
-	context->pipe->bind_fs_state(context->pipe, NULL);
-
-	context->render->vlDestroy(context->render);
-	context->csc->vlDestroy(context->csc);
-
-	context->pipe->delete_blend_state(context->pipe, context->blend);
-	context->pipe->delete_rasterizer_state(context->pipe, context->raster);
-	context->pipe->delete_depth_stencil_alpha_state(context->pipe, context->dsa);
-
-	FREE(context);
-
-	return 0;
-}
-
-struct vlScreen* vlContextGetScreen
-(
-	struct vlContext *context
-)
-{
-	assert(context);
-
-	return context->screen;
-}
-
-struct pipe_context* vlGetPipeContext
-(
-	struct vlContext *context
-)
-{
-	assert(context);
-
-	return context->pipe;
-}
-
-unsigned int vlGetPictureWidth
-(
-	struct vlContext *context
-)
-{
-	assert(context);
-
-	return context->picture_width;
-}
-
-unsigned int vlGetPictureHeight
-(
-	struct vlContext *context
-)
-{
-	assert(context);
-
-	return context->picture_height;
-}
-
-enum vlFormat vlGetPictureFormat
-(
-	struct vlContext *context
-)
-{
-	assert(context);
-
-	return context->picture_format;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h
deleted file mode 100644
index 3d14634c44..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_context.h
+++ /dev/null
@@ -1,73 +0,0 @@
-#ifndef vl_context_h
-#define vl_context_h
-
-#include "vl_types.h"
-
-struct pipe_context;
-
-#ifdef VL_INTERNAL
-struct vlRender;
-struct vlCSC;
-
-struct vlContext
-{
-	struct vlScreen		*screen;
-	struct pipe_context	*pipe;
-	unsigned int		picture_width;
-	unsigned int		picture_height;
-	enum vlFormat		picture_format;
-	enum vlProfile		profile;
-	enum vlEntryPoint	entry_point;
-
-	void			*raster;
-	void			*dsa;
-	void			*blend;
-
-	struct vlRender		*render;
-	struct vlCSC		*csc;
-};
-#endif
-
-int vlCreateContext
-(
-	struct vlScreen *screen,
-	struct pipe_context *pipe,
-	unsigned int picture_width,
-	unsigned int picture_height,
-	enum vlFormat picture_format,
-	enum vlProfile profile,
-	enum vlEntryPoint entry_point,
-	struct vlContext **context
-);
-
-int vlDestroyContext
-(
-	struct vlContext *context
-);
-
-struct vlScreen* vlContextGetScreen
-(
-	struct vlContext *context
-);
-
-struct pipe_context* vlGetPipeContext
-(
-	struct vlContext *context
-);
-
-unsigned int vlGetPictureWidth
-(
-	struct vlContext *context
-);
-
-unsigned int vlGetPictureHeight
-(
-	struct vlContext *context
-);
-
-enum vlFormat vlGetPictureFormat
-(
-	struct vlContext *context
-);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_csc.h b/src/gallium/state_trackers/g3dvl/vl_csc.h
deleted file mode 100644
index 36417a2792..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_csc.h
+++ /dev/null
@@ -1,53 +0,0 @@
-#ifndef vl_csc_h
-#define vl_csc_h
-
-#include "vl_types.h"
-
-struct pipe_surface;
-
-struct vlCSC
-{
-	int (*vlResizeFrameBuffer)
-	(
-		struct vlCSC *csc,
-		unsigned int width,
-		unsigned int height
-	);
-
-	int (*vlBegin)
-	(
-		struct vlCSC *csc
-	);
-
-	int (*vlPutPicture)
-	(
-		struct vlCSC *csc,
-		struct vlSurface *surface,
-		int srcx,
-		int srcy,
-		int srcw,
-		int srch,
-		int destx,
-		int desty,
-		int destw,
-		int desth,
-		enum vlPictureType picture_type
-	);
-
-	int (*vlEnd)
-	(
-		struct vlCSC *csc
-	);
-
-	struct pipe_surface* (*vlGetFrameBuffer)
-	(
-		struct vlCSC *csc
-	);
-
-	int (*vlDestroy)
-	(
-		struct vlCSC *csc
-	);
-};
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_defs.h b/src/gallium/state_trackers/g3dvl/vl_defs.h
deleted file mode 100644
index d612d02502..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_defs.h
+++ /dev/null
@@ -1,11 +0,0 @@
-#ifndef vl_defs_h
-#define vl_defs_h
-
-#define VL_BLOCK_WIDTH		8
-#define VL_BLOCK_HEIGHT		8
-#define VL_BLOCK_SIZE		(VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT)
-#define VL_MACROBLOCK_WIDTH	16
-#define VL_MACROBLOCK_HEIGHT	16
-#define VL_MACROBLOCK_SIZE	(VL_MACROBLOCK_WIDTH * VL_MACROBLOCK_HEIGHT)
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_display.c b/src/gallium/state_trackers/g3dvl/vl_display.c
deleted file mode 100644
index dce06de758..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_display.c
+++ /dev/null
@@ -1,48 +0,0 @@
-#define VL_INTERNAL
-#include "vl_display.h"
-#include <assert.h>
-#include <util/u_memory.h>
-
-int vlCreateDisplay
-(
-	vlNativeDisplay native_display,
-	struct vlDisplay **display
-)
-{
-	struct vlDisplay *dpy;
-
-	assert(native_display);
-	assert(display);
-
-	dpy = CALLOC_STRUCT(vlDisplay);
-
-	if (!dpy)
-		return 1;
-
-	dpy->native = native_display;
-	*display = dpy;
-
-	return 0;
-}
-
-int vlDestroyDisplay
-(
-	struct vlDisplay *display
-)
-{
-	assert(display);
-
-	FREE(display);
-
-	return 0;
-}
-
-vlNativeDisplay vlGetNativeDisplay
-(
-	struct vlDisplay *display
-)
-{
-	assert(display);
-
-	return display->native;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_display.h b/src/gallium/state_trackers/g3dvl/vl_display.h
deleted file mode 100644
index e11fd40799..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_display.h
+++ /dev/null
@@ -1,29 +0,0 @@
-#ifndef vl_display_h
-#define vl_display_h
-
-#include "vl_types.h"
-
-#ifdef VL_INTERNAL
-struct vlDisplay
-{
-	vlNativeDisplay native;
-};
-#endif
-
-int vlCreateDisplay
-(
-	vlNativeDisplay native_display,
-	struct vlDisplay **display
-);
-
-int vlDestroyDisplay
-(
-	struct vlDisplay *display
-);
-
-vlNativeDisplay vlGetNativeDisplay
-(
-	struct vlDisplay *display
-);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c
deleted file mode 100644
index 23631adb69..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c
+++ /dev/null
@@ -1,1155 +0,0 @@
-#define VL_INTERNAL
-#include "vl_r16snorm_mc_buf.h"
-#include <assert.h>
-#include <pipe/p_context.h>
-#include <pipe/p_screen.h>
-#include <pipe/p_state.h>
-#include <pipe/p_inlines.h>
-#include <tgsi/tgsi_parse.h>
-#include <tgsi/tgsi_build.h>
-#include <util/u_math.h>
-#include <util/u_memory.h>
-#include "vl_render.h"
-#include "vl_shader_build.h"
-#include "vl_surface.h"
-#include "vl_util.h"
-#include "vl_types.h"
-#include "vl_defs.h"
-
-const unsigned int DEFAULT_BUF_ALIGNMENT = 1;
-
-enum vlMacroBlockTypeEx
-{
-	vlMacroBlockExTypeIntra,
-	vlMacroBlockExTypeFwdPredictedFrame,
-	vlMacroBlockExTypeFwdPredictedField,
-	vlMacroBlockExTypeBkwdPredictedFrame,
-	vlMacroBlockExTypeBkwdPredictedField,
-	vlMacroBlockExTypeBiPredictedFrame,
-	vlMacroBlockExTypeBiPredictedField,
-
-	vlNumMacroBlockExTypes
-};
-
-struct vlVertexShaderConsts
-{
-	struct vlVertex4f denorm;
-};
-
-struct vlFragmentShaderConsts
-{
-	struct vlVertex4f multiplier;
-	struct vlVertex4f div;
-};
-
-struct vlMacroBlockVertexStream0
-{
-	struct vlVertex2f pos;
-	struct vlVertex2f luma_tc;
-	struct vlVertex2f cb_tc;
-	struct vlVertex2f cr_tc;
-};
-
-struct vlR16SnormBufferedMC
-{
-	struct vlRender				base;
-
-	unsigned int				picture_width;
-	unsigned int				picture_height;
-	enum vlFormat				picture_format;
-	unsigned int				macroblocks_per_picture;
-
-	struct vlSurface			*buffered_surface;
-	struct vlSurface			*past_surface;
-	struct vlSurface			*future_surface;
-	struct vlVertex2f			surface_tex_inv_size;
-	struct vlVertex2f			zero_block[3];
-	unsigned int				num_macroblocks;
-	struct vlMpeg2MacroBlock		*macroblocks;
-	struct pipe_transfer			*tex_transfer[3];
-	short					*texels[3];
-
-	struct pipe_context			*pipe;
-	struct pipe_viewport_state		viewport;
-	struct pipe_framebuffer_state		render_target;
-
-	union
-	{
-		void					*all[5];
-		struct
-		{
-			void				*y;
-			void				*cb;
-			void				*cr;
-			void				*ref[2];
-		};
-	} samplers;
-
-	union
-	{
-		struct pipe_texture			*all[5];
-		struct
-		{
-			struct pipe_texture		*y;
-			struct pipe_texture		*cb;
-			struct pipe_texture		*cr;
-			struct pipe_texture		*ref[2];
-		};
-	} textures;
-
-	union
-	{
-		struct pipe_vertex_buffer 		all[3];
-		struct
-		{
-			struct pipe_vertex_buffer	ycbcr;
-			struct pipe_vertex_buffer	ref[2];
-		};
-	} vertex_bufs;
-
-	void					*i_vs, *p_vs[2], *b_vs[2];
-	void					*i_fs, *p_fs[2], *b_fs[2];
-	struct pipe_vertex_element		vertex_elems[8];
-	struct pipe_constant_buffer		vs_const_buf;
-	struct pipe_constant_buffer		fs_const_buf;
-};
-
-static inline int vlBegin
-(
-	struct vlRender *render
-)
-{
-	assert(render);
-
-	return 0;
-}
-
-static inline int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch)
-{
-	unsigned int y;
-
-	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
-		memcpy
-		(
-			dst + y * dst_pitch,
-			src + y * VL_BLOCK_WIDTH,
-			VL_BLOCK_WIDTH * 2
-		);
-
-	return 0;
-}
-
-static inline int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch)
-{
-	unsigned int y;
-
-	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
-		memcpy
-		(
-			dst + y * dst_pitch * 2,
-			src + y * VL_BLOCK_WIDTH,
-			VL_BLOCK_WIDTH * 2
-		);
-
-	return 0;
-}
-
-static inline int vlGrabNoBlock(short *dst, unsigned int dst_pitch)
-{
-	unsigned int y;
-
-	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
-		memset
-		(
-			dst + y * dst_pitch,
-			0,
-			VL_BLOCK_WIDTH * 2
-		);
-
-	return 0;
-}
-
-static inline int vlGrabBlocks
-(
-	struct vlR16SnormBufferedMC *mc,
-	unsigned int mbx,
-	unsigned int mby,
-	enum vlDCTType dct_type,
-	unsigned int coded_block_pattern,
-	short *blocks
-)
-{
-	short			*texels;
-	unsigned int		tex_pitch;
-	unsigned int		x, y, tb = 0, sb = 0;
-	unsigned int		mbpx = mbx * VL_MACROBLOCK_WIDTH, mbpy = mby * VL_MACROBLOCK_HEIGHT;
-
-	assert(mc);
-	assert(blocks);
-
-	tex_pitch = mc->tex_transfer[0]->stride / mc->tex_transfer[0]->block.size;
-	texels = mc->texels[0] + mbpy * tex_pitch + mbpx;
-
-	for (y = 0; y < 2; ++y)
-	{
-		for (x = 0; x < 2; ++x, ++tb)
-		{
-			if ((coded_block_pattern >> (5 - tb)) & 1)
-			{
-				short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT;
-
-				if (dct_type == vlDCTTypeFrameCoded)
-				{
-					vlGrabFrameCodedBlock
-					(
-						cur_block,
-						texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH,
-						tex_pitch
-					);
-				}
-				else
-				{
-					vlGrabFieldCodedBlock
-					(
-						cur_block,
-						texels + y * tex_pitch + x * VL_BLOCK_WIDTH,
-						tex_pitch
-					);
-				}
-
-				++sb;
-			}
-			else if (mc->zero_block[0].x < 0.0f)
-			{
-				vlGrabNoBlock(texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, tex_pitch);
-
-				mc->zero_block[0].x = (mbpx + x * 8) * mc->surface_tex_inv_size.x;
-				mc->zero_block[0].y = (mbpy + y * 8) * mc->surface_tex_inv_size.y;
-			}
-		}
-	}
-
-	/* TODO: Implement 422, 444 */
-	mbpx >>= 1;
-	mbpy >>= 1;
-
-	for (tb = 0; tb < 2; ++tb)
-	{
-		tex_pitch = mc->tex_transfer[tb + 1]->stride / mc->tex_transfer[tb + 1]->block.size;
-		texels = mc->texels[tb + 1] + mbpy * tex_pitch + mbpx;
-
-		if ((coded_block_pattern >> (1 - tb)) & 1)
-		{
-			short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT;
-
-			vlGrabFrameCodedBlock
-			(
-				cur_block,
-				texels,
-				tex_pitch
-			);
-
-			++sb;
-		}
-		else if (mc->zero_block[tb + 1].x < 0.0f)
-		{
-			vlGrabNoBlock(texels, tex_pitch);
-
-			mc->zero_block[tb + 1].x = (mbpx << 1) * mc->surface_tex_inv_size.x;
-			mc->zero_block[tb + 1].y = (mbpy << 1) * mc->surface_tex_inv_size.y;
-		}
-	}
-
-	return 0;
-}
-
-static inline enum vlMacroBlockTypeEx vlGetMacroBlockTypeEx(struct vlMpeg2MacroBlock *mb)
-{
-	assert(mb);
-
-	switch (mb->mb_type)
-	{
-		case vlMacroBlockTypeIntra:
-			return vlMacroBlockExTypeIntra;
-		case vlMacroBlockTypeFwdPredicted:
-			return mb->mo_type == vlMotionTypeFrame ?
-				vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField;
-		case vlMacroBlockTypeBkwdPredicted:
-			return mb->mo_type == vlMotionTypeFrame ?
-				vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField;
-		case vlMacroBlockTypeBiPredicted:
-			return mb->mo_type == vlMotionTypeFrame ?
-				vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField;
-		default:
-			assert(0);
-	}
-
-	/* Unreachable */
-	return -1;
-}
-
-static inline int vlGrabMacroBlock
-(
-	struct vlR16SnormBufferedMC *mc,
-	struct vlMpeg2MacroBlock *macroblock
-)
-{
-	assert(mc);
-	assert(macroblock);
-	assert(mc->num_macroblocks < mc->macroblocks_per_picture);
-
-	mc->macroblocks[mc->num_macroblocks].mbx = macroblock->mbx;
-	mc->macroblocks[mc->num_macroblocks].mby = macroblock->mby;
-	mc->macroblocks[mc->num_macroblocks].mb_type = macroblock->mb_type;
-	mc->macroblocks[mc->num_macroblocks].mo_type = macroblock->mo_type;
-	mc->macroblocks[mc->num_macroblocks].dct_type = macroblock->dct_type;
-	mc->macroblocks[mc->num_macroblocks].PMV[0][0][0] = macroblock->PMV[0][0][0];
-	mc->macroblocks[mc->num_macroblocks].PMV[0][0][1] = macroblock->PMV[0][0][1];
-	mc->macroblocks[mc->num_macroblocks].PMV[0][1][0] = macroblock->PMV[0][1][0];
-	mc->macroblocks[mc->num_macroblocks].PMV[0][1][1] = macroblock->PMV[0][1][1];
-	mc->macroblocks[mc->num_macroblocks].PMV[1][0][0] = macroblock->PMV[1][0][0];
-	mc->macroblocks[mc->num_macroblocks].PMV[1][0][1] = macroblock->PMV[1][0][1];
-	mc->macroblocks[mc->num_macroblocks].PMV[1][1][0] = macroblock->PMV[1][1][0];
-	mc->macroblocks[mc->num_macroblocks].PMV[1][1][1] = macroblock->PMV[1][1][1];
-	mc->macroblocks[mc->num_macroblocks].cbp = macroblock->cbp;
-	mc->macroblocks[mc->num_macroblocks].blocks = macroblock->blocks;
-
-	vlGrabBlocks
-	(
-		mc,
-		macroblock->mbx,
-		macroblock->mby,
-		macroblock->dct_type,
-		macroblock->cbp,
-		macroblock->blocks
-	);
-
-	mc->num_macroblocks++;
-
-	return 0;
-}
-
-#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, zb)					\
-	do {															\
-	(vb)[0].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[0].pos.y = (mby) * (unity) + (ofsy);			\
-	(vb)[1].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
-	(vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].pos.y = (mby) * (unity) + (ofsy);			\
-	(vb)[3].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].pos.y = (mby) * (unity) + (ofsy);			\
-	(vb)[4].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
-	(vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
-																\
-	if ((cbp) & (lm))													\
-	{															\
-		(vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].luma_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-		(vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].luma_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].luma_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-		(vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-	}															\
-	else															\
-	{															\
-		(vb)[0].luma_tc.x = (zb)[0].x;		(vb)[0].luma_tc.y = (zb)[0].y;						\
-		(vb)[1].luma_tc.x = (zb)[0].x;		(vb)[1].luma_tc.y = (zb)[0].y + (hy);					\
-		(vb)[2].luma_tc.x = (zb)[0].x + (hx);	(vb)[2].luma_tc.y = (zb)[0].y;						\
-		(vb)[3].luma_tc.x = (zb)[0].x + (hx);	(vb)[3].luma_tc.y = (zb)[0].y;						\
-		(vb)[4].luma_tc.x = (zb)[0].x;		(vb)[4].luma_tc.y = (zb)[0].y + (hy);					\
-		(vb)[5].luma_tc.x = (zb)[0].x + (hx);	(vb)[5].luma_tc.y = (zb)[0].y + (hy);					\
-	}															\
-																\
-	if ((cbp) & (cbm))													\
-	{															\
-		(vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].cb_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-		(vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].cb_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].cb_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-		(vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-	}															\
-	else															\
-	{															\
-		(vb)[0].cb_tc.x = (zb)[1].x;		(vb)[0].cb_tc.y = (zb)[1].y;						\
-		(vb)[1].cb_tc.x = (zb)[1].x;		(vb)[1].cb_tc.y = (zb)[1].y + (hy);					\
-		(vb)[2].cb_tc.x = (zb)[1].x + (hx);	(vb)[2].cb_tc.y = (zb)[1].y;						\
-		(vb)[3].cb_tc.x = (zb)[1].x + (hx);	(vb)[3].cb_tc.y = (zb)[1].y;						\
-		(vb)[4].cb_tc.x = (zb)[1].x;		(vb)[4].cb_tc.y = (zb)[1].y + (hy);					\
-		(vb)[5].cb_tc.x = (zb)[1].x + (hx);	(vb)[5].cb_tc.y = (zb)[1].y + (hy);					\
-	}															\
-																\
-	if ((cbp) & (crm))													\
-	{															\
-		(vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].cr_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-		(vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].cr_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].cr_tc.y = (mby) * (unity) + (ofsy);		\
-		(vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-		(vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
-	}															\
-	else															\
-	{															\
-		(vb)[0].cr_tc.x = (zb)[2].x;		(vb)[0].cr_tc.y = (zb)[2].y;						\
-		(vb)[1].cr_tc.x = (zb)[2].x;		(vb)[1].cr_tc.y = (zb)[2].y + (hy);					\
-		(vb)[2].cr_tc.x = (zb)[2].x + (hx);	(vb)[2].cr_tc.y = (zb)[2].y;						\
-		(vb)[3].cr_tc.x = (zb)[2].x + (hx);	(vb)[3].cr_tc.y = (zb)[2].y;						\
-		(vb)[4].cr_tc.x = (zb)[2].x;		(vb)[4].cr_tc.y = (zb)[2].y + (hy);					\
-		(vb)[5].cr_tc.x = (zb)[2].x + (hx);	(vb)[5].cr_tc.y = (zb)[2].y + (hy);					\
-	}															\
-	} while (0)
-
-static inline int vlGenMacroblockVerts
-(
-	struct vlR16SnormBufferedMC *mc,
-	struct vlMpeg2MacroBlock *macroblock,
-	unsigned int pos,
-	struct vlMacroBlockVertexStream0 *ycbcr_vb,
-	struct vlVertex2f **ref_vb
-)
-{
-	struct vlVertex2f	mo_vec[2];
-	unsigned int		i;
-
-	assert(mc);
-	assert(macroblock);
-	assert(ycbcr_vb);
-	assert(pos < mc->macroblocks_per_picture);
-
-	switch (macroblock->mb_type)
-	{
-		case vlMacroBlockTypeBiPredicted:
-		{
-			struct vlVertex2f *vb;
-
-			assert(ref_vb && ref_vb[1]);
-
-			vb = ref_vb[1] + pos * 2 * 24;
-
-			mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x;
-			mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y;
-
-			if (macroblock->mo_type == vlMotionTypeFrame)
-			{
-				for (i = 0; i < 24 * 2; i += 2)
-				{
-					vb[i].x = mo_vec[0].x;
-					vb[i].y = mo_vec[0].y;
-				}
-			}
-			else
-			{
-				mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x;
-				mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y;
-
-				for (i = 0; i < 24 * 2; i += 2)
-				{
-					vb[i].x = mo_vec[0].x;
-					vb[i].y = mo_vec[0].y;
-					vb[i + 1].x = mo_vec[1].x;
-					vb[i + 1].y = mo_vec[1].y;
-				}
-			}
-
-			/* fall-through */
-		}
-		case vlMacroBlockTypeFwdPredicted:
-		case vlMacroBlockTypeBkwdPredicted:
-		{
-			struct vlVertex2f *vb;
-
-			assert(ref_vb && ref_vb[0]);
-
-			vb = ref_vb[0] + pos * 2 * 24;
-
-			if (macroblock->mb_type == vlMacroBlockTypeBkwdPredicted)
-			{
-				mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x;
-				mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y;
-
-				if (macroblock->mo_type == vlMotionTypeField)
-				{
-					mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x;
-					mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y;
-				}
-			}
-			else
-			{
-				mo_vec[0].x = macroblock->PMV[0][0][0] * 0.5f * mc->surface_tex_inv_size.x;
-				mo_vec[0].y = macroblock->PMV[0][0][1] * 0.5f * mc->surface_tex_inv_size.y;
-
-				if (macroblock->mo_type == vlMotionTypeField)
-				{
-					mo_vec[1].x = macroblock->PMV[1][0][0] * 0.5f * mc->surface_tex_inv_size.x;
-					mo_vec[1].y = macroblock->PMV[1][0][1] * 0.5f * mc->surface_tex_inv_size.y;
-				}
-			}
-
-			if (macroblock->mo_type == vlMotionTypeFrame)
-			{
-				for (i = 0; i < 24 * 2; i += 2)
-				{
-					vb[i].x = mo_vec[0].x;
-					vb[i].y = mo_vec[0].y;
-				}
-			}
-			else
-			{
-				for (i = 0; i < 24 * 2; i += 2)
-				{
-					vb[i].x = mo_vec[0].x;
-					vb[i].y = mo_vec[0].y;
-					vb[i + 1].x = mo_vec[1].x;
-					vb[i + 1].y = mo_vec[1].y;
-				}
-			}
-
-			/* fall-through */
-		}
-		case vlMacroBlockTypeIntra:
-		{
-			const struct vlVertex2f	unit =
-			{
-				mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH,
-				mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT
-			};
-			const struct vlVertex2f half =
-			{
-				mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2),
-				mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2)
-			};
-
-			struct vlMacroBlockVertexStream0 *vb;
-
-			vb = ycbcr_vb + pos * 24;
-
-			SET_BLOCK
-			(
-				vb,
-				macroblock->cbp, macroblock->mbx, macroblock->mby,
-				unit.x, unit.y, 0, 0, half.x, half.y,
-				32, 2, 1, mc->zero_block
-			);
-
-			SET_BLOCK
-			(
-				vb + 6,
-				macroblock->cbp, macroblock->mbx, macroblock->mby,
-				unit.x, unit.y, half.x, 0, half.x, half.y,
-				16, 2, 1, mc->zero_block
-			);
-
-			SET_BLOCK
-			(
-				vb + 12,
-				macroblock->cbp, macroblock->mbx, macroblock->mby,
-				unit.x, unit.y, 0, half.y, half.x, half.y,
-				8, 2, 1, mc->zero_block
-			);
-
-			SET_BLOCK
-			(
-				vb + 18,
-				macroblock->cbp, macroblock->mbx, macroblock->mby,
-				unit.x, unit.y, half.x, half.y, half.x, half.y,
-				4, 2, 1, mc->zero_block
-			);
-
-			break;
-		}
-		default:
-			assert(0);
-	}
-
-	return 0;
-}
-
-static int vlFlush
-(
-	struct vlRender *render
-)
-{
-	struct vlR16SnormBufferedMC	*mc;
-	struct pipe_context		*pipe;
-	struct vlVertexShaderConsts	*vs_consts;
-	unsigned int			num_macroblocks[vlNumMacroBlockExTypes] = {0};
-	unsigned int			offset[vlNumMacroBlockExTypes];
-	unsigned int			vb_start = 0;
-	unsigned int			i;
-
-	assert(render);
-
-	mc = (struct vlR16SnormBufferedMC*)render;
-
-	if (!mc->buffered_surface)
-		return 0;
-
-	if (mc->num_macroblocks < mc->macroblocks_per_picture)
-		return 0;
-
-	assert(mc->num_macroblocks <= mc->macroblocks_per_picture);
-
-	pipe = mc->pipe;
-
-	for (i = 0; i < mc->num_macroblocks; ++i)
-	{
-		enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]);
-
-		num_macroblocks[mb_type_ex]++;
-	}
-
-	offset[0] = 0;
-
-	for (i = 1; i < vlNumMacroBlockExTypes; ++i)
-		offset[i] = offset[i - 1] + num_macroblocks[i - 1];
-
-	{
-		struct vlMacroBlockVertexStream0	*ycbcr_vb;
-		struct vlVertex2f			*ref_vb[2];
-
-		ycbcr_vb = (struct vlMacroBlockVertexStream0*)pipe_buffer_map
-		(
-			pipe->screen,
-			mc->vertex_bufs.ycbcr.buffer,
-			PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
-		);
-
-		for (i = 0; i < 2; ++i)
-			ref_vb[i] = (struct vlVertex2f*)pipe_buffer_map
-			(
-				pipe->screen,
-				mc->vertex_bufs.ref[i].buffer,
-				PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
-			);
-
-		for (i = 0; i < mc->num_macroblocks; ++i)
-		{
-			enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]);
-
-			vlGenMacroblockVerts(mc, &mc->macroblocks[i], offset[mb_type_ex], ycbcr_vb, ref_vb);
-
-			offset[mb_type_ex]++;
-		}
-
-		pipe_buffer_unmap(pipe->screen, mc->vertex_bufs.ycbcr.buffer);
-		for (i = 0; i < 2; ++i)
-			pipe_buffer_unmap(pipe->screen, mc->vertex_bufs.ref[i].buffer);
-	}
-
-	for (i = 0; i < 3; ++i)
-	{
-		pipe->screen->transfer_unmap(pipe->screen, mc->tex_transfer[i]);
-		pipe->screen->tex_transfer_destroy(mc->tex_transfer[i]);
-	}
-
-	mc->render_target.cbufs[0] = pipe->screen->get_tex_surface
-	(
-		pipe->screen,
-		mc->buffered_surface->texture,
-		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
-	);
-
-	pipe->set_framebuffer_state(pipe, &mc->render_target);
-	pipe->set_viewport_state(pipe, &mc->viewport);
-	vs_consts = pipe_buffer_map
-	(
-		pipe->screen,
-		mc->vs_const_buf.buffer,
-		PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
-	);
-
-	vs_consts->denorm.x = mc->buffered_surface->texture->width[0];
-	vs_consts->denorm.y = mc->buffered_surface->texture->height[0];
-
-	pipe_buffer_unmap(pipe->screen, mc->vs_const_buf.buffer);
-	pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &mc->vs_const_buf);
-	pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf);
-
-	if (num_macroblocks[vlMacroBlockExTypeIntra] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 4, mc->vertex_elems);
-		pipe->set_sampler_textures(pipe, 3, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 3, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->i_vs);
-		pipe->bind_fs_state(pipe, mc->i_fs);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeIntra] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeIntra] * 24;
-	}
-
-	if (num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
-		mc->textures.ref[0] = mc->past_surface->texture;
-		pipe->set_sampler_textures(pipe, 4, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 4, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->p_vs[0]);
-		pipe->bind_fs_state(pipe, mc->p_fs[0]);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24;
-	}
-
-	if (num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
-		mc->textures.ref[0] = mc->past_surface->texture;
-		pipe->set_sampler_textures(pipe, 4, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 4, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->p_vs[1]);
-		pipe->bind_fs_state(pipe, mc->p_fs[1]);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24;
-	}
-
-	if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
-		mc->textures.ref[0] = mc->future_surface->texture;
-		pipe->set_sampler_textures(pipe, 4, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 4, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->p_vs[0]);
-		pipe->bind_fs_state(pipe, mc->p_fs[0]);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24;
-	}
-
-	if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
-		mc->textures.ref[0] = mc->future_surface->texture;
-		pipe->set_sampler_textures(pipe, 4, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 4, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->p_vs[1]);
-		pipe->bind_fs_state(pipe, mc->p_fs[1]);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24;
-	}
-
-	if (num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 8, mc->vertex_elems);
-		mc->textures.ref[0] = mc->past_surface->texture;
-		mc->textures.ref[1] = mc->future_surface->texture;
-		pipe->set_sampler_textures(pipe, 5, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 5, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->b_vs[0]);
-		pipe->bind_fs_state(pipe, mc->b_fs[0]);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24;
-	}
-
-	if (num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0)
-	{
-		pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs.all);
-		pipe->set_vertex_elements(pipe, 8, mc->vertex_elems);
-		mc->textures.ref[0] = mc->past_surface->texture;
-		mc->textures.ref[1] = mc->future_surface->texture;
-		pipe->set_sampler_textures(pipe, 5, mc->textures.all);
-		pipe->bind_sampler_states(pipe, 5, mc->samplers.all);
-		pipe->bind_vs_state(pipe, mc->b_vs[1]);
-		pipe->bind_fs_state(pipe, mc->b_fs[1]);
-
-		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24);
-		vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24;
-	}
-
-	pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &mc->buffered_surface->render_fence);
-	pipe_surface_reference(&mc->render_target.cbufs[0], NULL);
-
-	for (i = 0; i < 3; ++i)
-		mc->zero_block[i].x = -1.0f;
-
-	mc->buffered_surface = NULL;
-	mc->num_macroblocks = 0;
-
-	return 0;
-}
-
-static int vlRenderMacroBlocksMpeg2R16SnormBuffered
-(
-	struct vlRender *render,
-	struct vlMpeg2MacroBlockBatch *batch,
-	struct vlSurface *surface
-)
-{
-	struct vlR16SnormBufferedMC	*mc;
-	bool				new_surface = false;
-	unsigned int			i;
-
-	assert(render);
-
-	mc = (struct vlR16SnormBufferedMC*)render;
-
-	if (mc->buffered_surface)
-	{
-		if (mc->buffered_surface != surface)
-		{
-			vlFlush(&mc->base);
-			new_surface = true;
-		}
-	}
-	else
-		new_surface = true;
-
-	if (new_surface)
-	{
-		mc->buffered_surface = surface;
-		mc->past_surface = batch->past_surface;
-		mc->future_surface = batch->future_surface;
-		mc->surface_tex_inv_size.x = 1.0f / surface->texture->width[0];
-		mc->surface_tex_inv_size.y = 1.0f / surface->texture->height[0];
-
-		for (i = 0; i < 3; ++i)
-		{
-			mc->tex_transfer[i] = mc->pipe->screen->get_tex_transfer
-			(
-				mc->pipe->screen,
-				mc->textures.all[i],
-				0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0,
-				surface->texture->width[0],
-				surface->texture->height[0]
-			);
-
-			mc->texels[i] = mc->pipe->screen->transfer_map(mc->pipe->screen, mc->tex_transfer[i]);
-		}
-	}
-
-	for (i = 0; i < batch->num_macroblocks; ++i)
-		vlGrabMacroBlock(mc, &batch->macroblocks[i]);
-
-	return 0;
-}
-
-static inline int vlEnd
-(
-	struct vlRender *render
-)
-{
-	assert(render);
-
-	return 0;
-}
-
-static int vlDestroy
-(
-	struct vlRender *render
-)
-{
-	struct vlR16SnormBufferedMC	*mc;
-	struct pipe_context		*pipe;
-	unsigned int			i;
-
-	assert(render);
-
-	mc = (struct vlR16SnormBufferedMC*)render;
-	pipe = mc->pipe;
-
-	for (i = 0; i < 5; ++i)
-		pipe->delete_sampler_state(pipe, mc->samplers.all[i]);
-
-	for (i = 0; i < 3; ++i)
-		pipe_buffer_reference(&mc->vertex_bufs.all[i].buffer, NULL);
-
-	/* Textures 3 & 4 are not created directly, no need to release them here */
-	for (i = 0; i < 3; ++i)
-		pipe_texture_reference(&mc->textures.all[i], NULL);
-
-	pipe->delete_vs_state(pipe, mc->i_vs);
-	pipe->delete_fs_state(pipe, mc->i_fs);
-
-	for (i = 0; i < 2; ++i)
-	{
-		pipe->delete_vs_state(pipe, mc->p_vs[i]);
-		pipe->delete_fs_state(pipe, mc->p_fs[i]);
-		pipe->delete_vs_state(pipe, mc->b_vs[i]);
-		pipe->delete_fs_state(pipe, mc->b_fs[i]);
-	}
-
-	pipe_buffer_reference(&mc->vs_const_buf.buffer, NULL);
-	pipe_buffer_reference(&mc->fs_const_buf.buffer, NULL);
-
-	FREE(mc->macroblocks);
-	FREE(mc);
-
-	return 0;
-}
-
-/*
- * Muliplier renormalizes block samples from 16 bits to 12 bits.
- * Divider is used when calculating Y % 2 for choosing top or bottom
- * field for P or B macroblocks.
- * TODO: Use immediates.
- */
-static const struct vlFragmentShaderConsts fs_consts =
-{
-	{32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f},
-	{0.5f, 2.0f, 0.0f, 0.0f}
-};
-
-#include "vl_r16snorm_mc_buf_shaders.inc"
-
-static int vlCreateDataBufs
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int	mbw = align(mc->picture_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH;
-	const unsigned int	mbh = align(mc->picture_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT;
-
-	struct pipe_context	*pipe;
-	unsigned int		i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	mc->macroblocks_per_picture = mbw * mbh;
-
-	/* Create our vertex buffers */
-	mc->vertex_bufs.ycbcr.stride = sizeof(struct vlVertex2f) * 4;
-	mc->vertex_bufs.ycbcr.max_index = 24 * mc->macroblocks_per_picture - 1;
-	mc->vertex_bufs.ycbcr.buffer_offset = 0;
-	mc->vertex_bufs.ycbcr.buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		DEFAULT_BUF_ALIGNMENT,
-		PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD,
-		sizeof(struct vlVertex2f) * 4 * 24 * mc->macroblocks_per_picture
-	);
-
-	for (i = 1; i < 3; ++i)
-	{
-		mc->vertex_bufs.all[i].stride = sizeof(struct vlVertex2f) * 2;
-		mc->vertex_bufs.all[i].max_index = 24 * mc->macroblocks_per_picture - 1;
-		mc->vertex_bufs.all[i].buffer_offset = 0;
-		mc->vertex_bufs.all[i].buffer = pipe_buffer_create
-		(
-			pipe->screen,
-			DEFAULT_BUF_ALIGNMENT,
-			PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD,
-			sizeof(struct vlVertex2f) * 2 * 24 * mc->macroblocks_per_picture
-		);
-	}
-
-	/* Position element */
-	mc->vertex_elems[0].src_offset = 0;
-	mc->vertex_elems[0].vertex_buffer_index = 0;
-	mc->vertex_elems[0].nr_components = 2;
-	mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* Luma, texcoord element */
-	mc->vertex_elems[1].src_offset = sizeof(struct vlVertex2f);
-	mc->vertex_elems[1].vertex_buffer_index = 0;
-	mc->vertex_elems[1].nr_components = 2;
-	mc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* Chroma Cr texcoord element */
-	mc->vertex_elems[2].src_offset = sizeof(struct vlVertex2f) * 2;
-	mc->vertex_elems[2].vertex_buffer_index = 0;
-	mc->vertex_elems[2].nr_components = 2;
-	mc->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* Chroma Cb texcoord element */
-	mc->vertex_elems[3].src_offset = sizeof(struct vlVertex2f) * 3;
-	mc->vertex_elems[3].vertex_buffer_index = 0;
-	mc->vertex_elems[3].nr_components = 2;
-	mc->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* First ref surface top field texcoord element */
-	mc->vertex_elems[4].src_offset = 0;
-	mc->vertex_elems[4].vertex_buffer_index = 1;
-	mc->vertex_elems[4].nr_components = 2;
-	mc->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* First ref surface bottom field texcoord element */
-	mc->vertex_elems[5].src_offset = sizeof(struct vlVertex2f);
-	mc->vertex_elems[5].vertex_buffer_index = 1;
-	mc->vertex_elems[5].nr_components = 2;
-	mc->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* Second ref surface top field texcoord element */
-	mc->vertex_elems[6].src_offset = 0;
-	mc->vertex_elems[6].vertex_buffer_index = 2;
-	mc->vertex_elems[6].nr_components = 2;
-	mc->vertex_elems[6].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* Second ref surface bottom field texcoord element */
-	mc->vertex_elems[7].src_offset = sizeof(struct vlVertex2f);
-	mc->vertex_elems[7].vertex_buffer_index = 2;
-	mc->vertex_elems[7].nr_components = 2;
-	mc->vertex_elems[7].src_format = PIPE_FORMAT_R32G32_FLOAT;
-
-	/* Create our constant buffer */
-	mc->vs_const_buf.buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		DEFAULT_BUF_ALIGNMENT,
-		PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD,
-		sizeof(struct vlVertexShaderConsts)
-	);
-
-	mc->fs_const_buf.buffer = pipe_buffer_create
-	(
-		pipe->screen,
-		DEFAULT_BUF_ALIGNMENT,
-		PIPE_BUFFER_USAGE_CONSTANT,
-		sizeof(struct vlFragmentShaderConsts)
-	);
-
-	memcpy
-	(
-		pipe_buffer_map(pipe->screen, mc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
-		&fs_consts,
-		sizeof(struct vlFragmentShaderConsts)
-	);
-
-	pipe_buffer_unmap(pipe->screen, mc->fs_const_buf.buffer);
-
-	mc->macroblocks = MALLOC(sizeof(struct vlMpeg2MacroBlock) * mc->macroblocks_per_picture);
-
-	return 0;
-}
-
-static int vlInit
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	struct pipe_context		*pipe;
-	struct pipe_sampler_state	sampler;
-	struct pipe_texture		template;
-	unsigned int			filters[5];
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-
-	mc->buffered_surface = NULL;
-	mc->past_surface = NULL;
-	mc->future_surface = NULL;
-	for (i = 0; i < 3; ++i)
-		mc->zero_block[i].x = -1.0f;
-	mc->num_macroblocks = 0;
-
-	/* For MC we render to textures, which are rounded up to nearest POT */
-	mc->viewport.scale[0] = vlRoundUpPOT(mc->picture_width);
-	mc->viewport.scale[1] = vlRoundUpPOT(mc->picture_height);
-	mc->viewport.scale[2] = 1;
-	mc->viewport.scale[3] = 1;
-	mc->viewport.translate[0] = 0;
-	mc->viewport.translate[1] = 0;
-	mc->viewport.translate[2] = 0;
-	mc->viewport.translate[3] = 0;
-
-	mc->render_target.width = vlRoundUpPOT(mc->picture_width);
-	mc->render_target.height = vlRoundUpPOT(mc->picture_height);
-	mc->render_target.nr_cbufs = 1;
-	/* FB for MC stage is a vlSurface created by the user, set at render time */
-	mc->render_target.zsbuf = NULL;
-
-	filters[0] = PIPE_TEX_FILTER_NEAREST;
-	/* FIXME: Linear causes discoloration around block edges */
-	filters[1] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/;
-	filters[2] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/;
-	filters[3] = PIPE_TEX_FILTER_LINEAR;
-	filters[4] = PIPE_TEX_FILTER_LINEAR;
-
-	for (i = 0; i < 5; ++i)
-	{
-		sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
-		sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
-		sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
-		sampler.min_img_filter = filters[i];
-		sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
-		sampler.mag_img_filter = filters[i];
-		sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
-		sampler.compare_func = PIPE_FUNC_ALWAYS;
-		sampler.normalized_coords = 1;
-		/*sampler.prefilter = ;*/
-		/*sampler.lod_bias = ;*/
-		sampler.min_lod = 0;
-		/*sampler.max_lod = ;*/
-		/*sampler.border_color[i] = ;*/
-		/*sampler.max_anisotropy = ;*/
-		mc->samplers.all[i] = pipe->create_sampler_state(pipe, &sampler);
-	}
-
-	memset(&template, 0, sizeof(struct pipe_texture));
-	template.target = PIPE_TEXTURE_2D;
-	template.format = PIPE_FORMAT_R16_SNORM;
-	template.last_level = 0;
-	template.width[0] = vlRoundUpPOT(mc->picture_width);
-	template.height[0] = vlRoundUpPOT(mc->picture_height);
-	template.depth[0] = 1;
-	pf_get_block(template.format, &template.block);
-	template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_DYNAMIC;
-
-	mc->textures.y = pipe->screen->texture_create(pipe->screen, &template);
-
-	if (mc->picture_format == vlFormatYCbCr420)
-	{
-		template.width[0] = vlRoundUpPOT(mc->picture_width / 2);
-		template.height[0] = vlRoundUpPOT(mc->picture_height / 2);
-	}
-	else if (mc->picture_format == vlFormatYCbCr422)
-		template.height[0] = vlRoundUpPOT(mc->picture_height / 2);
-
-	mc->textures.cb = pipe->screen->texture_create(pipe->screen, &template);
-	mc->textures.cr = pipe->screen->texture_create(pipe->screen, &template);
-
-	/* textures.all[3] & textures.all[4] are assigned from vlSurfaces for P and B macroblocks at render time */
-
-	vlCreateVertexShaderIMB(mc);
-	vlCreateFragmentShaderIMB(mc);
-	vlCreateVertexShaderFramePMB(mc);
-	vlCreateVertexShaderFieldPMB(mc);
-	vlCreateFragmentShaderFramePMB(mc);
-	vlCreateFragmentShaderFieldPMB(mc);
-	vlCreateVertexShaderFrameBMB(mc);
-	vlCreateVertexShaderFieldBMB(mc);
-	vlCreateFragmentShaderFrameBMB(mc);
-	vlCreateFragmentShaderFieldBMB(mc);
-	vlCreateDataBufs(mc);
-
-	return 0;
-}
-
-int vlCreateR16SNormBufferedMC
-(
-	struct pipe_context *pipe,
-	unsigned int picture_width,
-	unsigned int picture_height,
-	enum vlFormat picture_format,
-	struct vlRender **render
-)
-{
-	struct vlR16SnormBufferedMC *mc;
-
-	assert(pipe);
-	assert(render);
-
-	mc = CALLOC_STRUCT(vlR16SnormBufferedMC);
-
-	mc->base.vlBegin = &vlBegin;
-	mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16SnormBuffered;
-	mc->base.vlEnd = &vlEnd;
-	mc->base.vlFlush = &vlFlush;
-	mc->base.vlDestroy = &vlDestroy;
-	mc->pipe = pipe;
-	mc->picture_width = picture_width;
-	mc->picture_height = picture_height;
-
-	vlInit(mc);
-
-	*render = &mc->base;
-
-	return 0;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h
deleted file mode 100644
index 27177d64ca..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h
+++ /dev/null
@@ -1,18 +0,0 @@
-#ifndef vl_r16snorm_mc_buf_h
-#define vl_r16snorm_mc_buf_h
-
-#include "vl_types.h"
-
-struct pipe_context;
-struct vlRender;
-
-int vlCreateR16SNormBufferedMC
-(
-	struct pipe_context *pipe,
-	unsigned int picture_width,
-	unsigned int picture_height,
-	enum vlFormat picture_format,
-	struct vlRender **render
-);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc
deleted file mode 100644
index ef4a4b2add..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc
+++ /dev/null
@@ -1,1185 +0,0 @@
-static int vlCreateVertexShaderIMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 50;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	vs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0		; Vertex pos
-	 * decl i1		; Luma texcoords
-	 * decl i2		; Chroma Cb texcoords
-	 * decl i3		; Chroma Cr texcoords
-	 */
-	for (i = 0; i < 4; i++)
-	{
-		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl o0		; Vertex pos
-	 * decl o1		; Luma texcoords
-	 * decl o2		; Chroma Cb texcoords
-	 * decl o3		; Chroma Cr texcoords
-	 */
-	for (i = 0; i < 4; i++)
-	{
-		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * mov o0, i0		; Move input vertex pos to output
-	 * mov o1, i1		; Move input luma texcoords to output
-	 * mov o2, i2		; Move input chroma Cb texcoords to output
-	 * mov o3, i3		; Move input chroma Cr texcoords to output
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	vs.tokens = tokens;
-	mc->i_vs = pipe->create_vs_state(pipe, &vs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateFragmentShaderIMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	fs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0			; Luma texcoords
-	 * decl i1			; Chroma Cb texcoords
-	 * decl i2			; Chroma Cr texcoords
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl o0			; Fragment color */
-	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl t0, t1 */
-	decl = vl_decl_temps(0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl s0			; Sampler for luma texture
-	 * decl s1			; Sampler for chroma Cb texture
-	 * decl s2			; Sampler for chroma Cr texture
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		decl = vl_decl_samplers(i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti);
-	}
-
-	/*
-	 * tex2d t1, i0, s0		; Read texel from luma texture
-	 * mov t0.x, t1.x		; Move luma sample into .x component
-	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
-	 * mov t0.y, t1.x		; Move Cb sample into .y component
-	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
-	 * mov t0.z, t1.x		; Move Cr sample into .z component
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul o0, t0, c0		; Rescale texel to correct range */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	fs.tokens = tokens;
-	mc->i_fs = pipe->create_fs_state(pipe, &fs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateVertexShaderFramePMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	vs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0		; Vertex pos
-	 * decl i1		; Luma texcoords
-	 * decl i2		; Chroma Cb texcoords
-	 * decl i3		; Chroma Cr texcoords
-	 * decl i4		; Ref surface top field texcoords
-	 * decl i5		; Ref surface bottom field texcoords (unused, packed in the same stream)
-	 */
-	for (i = 0; i < 6; i++)
-	{
-		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl o0		; Vertex pos
-	 * decl o1		; Luma texcoords
-	 * decl o2		; Chroma Cb texcoords
-	 * decl o3		; Chroma Cr texcoords
-	 * decl o4		; Ref macroblock texcoords
-	 */
-	for (i = 0; i < 5; i++)
-	{
-		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * mov o0, i0		; Move input vertex pos to output
-	 * mov o1, i1		; Move input luma texcoords to output
-	 * mov o2, i2		; Move input chroma Cb texcoords to output
-	 * mov o3, i3		; Move input chroma Cr texcoords to output
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* add o4, i0, i4	; Translate vertex pos by motion vec to form ref macroblock texcoords */
-	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 4);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	vs.tokens = tokens;
-	mc->p_vs[0] = pipe->create_vs_state(pipe, &vs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateVertexShaderFieldPMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	vs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0		; Vertex pos
-	 * decl i1		; Luma texcoords
-	 * decl i2		; Chroma Cb texcoords
-	 * decl i3		; Chroma Cr texcoords
-	 * decl i4              ; Ref macroblock top field texcoords
-	 * decl i5              ; Ref macroblock bottom field texcoords
-	 */
-	for (i = 0; i < 6; i++)
-	{
-		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* decl c0		; Render target dimensions */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl o0		; Vertex pos
-	 * decl o1		; Luma texcoords
-	 * decl o2		; Chroma Cb texcoords
-	 * decl o3		; Chroma Cr texcoords
-	 * decl o4		; Ref macroblock top field texcoords
-	 * decl o5		; Ref macroblock bottom field texcoords
-	 * decl o6		; Denormalized vertex pos
-	 */
-	for (i = 0; i < 7; i++)
-	{
-		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * mov o0, i0		; Move input vertex pos to output
-	 * mov o1, i1		; Move input luma texcoords to output
-	 * mov o2, i2		; Move input chroma Cb texcoords to output
-	 * mov o3, i3		; Move input chroma Cr texcoords to output
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * add o4, i0, i4	; Translate vertex pos by motion vec to form top field macroblock texcoords
-	 * add o5, i0, i5	; Translate vertex pos by motion vec to form bottom field macroblock texcoords
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul o6, i0, c0	; Denorm vertex pos */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 6, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	vs.tokens = tokens;
-	mc->p_vs[1] = pipe->create_vs_state(pipe, &vs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateFragmentShaderFramePMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	fs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0			; Luma texcoords
-	 * decl i1			; Chroma Cb texcoords
-	 * decl i2			; Chroma Cr texcoords
-	 * decl i3			; Ref macroblock texcoords
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl o0			; Fragment color */
-	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl t0, t1 */
-	decl = vl_decl_temps(0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl s0			; Sampler for luma texture
-	 * decl s1			; Sampler for chroma Cb texture
-	 * decl s2			; Sampler for chroma Cr texture
-	 * decl s3			; Sampler for ref surface texture
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		decl = vl_decl_samplers(i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * tex2d t1, i0, s0		; Read texel from luma texture
-	 * mov t0.x, t1.x		; Move luma sample into .x component
-	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
-	 * mov t0.y, t1.x		; Move Cb sample into .y component
-	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
-	 * mov t0.z, t1.x		; Move Cr sample into .z component
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul t0, t0, c0		; Rescale texel to correct range */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* tex2d t1, i3, s3		; Read texel from ref macroblock */
-	inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 3, TGSI_FILE_SAMPLER, 3);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* add o0, t0, t1		; Add ref and differential to form final output */
-	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	fs.tokens = tokens;
-	mc->p_fs[0] = pipe->create_fs_state(pipe, &fs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateFragmentShaderFieldPMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 200;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	fs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0			; Luma texcoords
-	 * decl i1			; Chroma Cb texcoords
-	 * decl i2			; Chroma Cr texcoords
-	 * decl i3			; Ref macroblock top field texcoords
-	 * decl i4			; Ref macroblock bottom field texcoords
-	 * decl i5			; Denormalized vertex pos
-	 */
-	for (i = 0; i < 6; ++i)
-	{
-		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
-	 * decl c1			; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection
-	 */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl o0			; Fragment color */
-	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl t0-t4 */
-	decl = vl_decl_temps(0, 4);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl s0			; Sampler for luma texture
-	 * decl s1			; Sampler for chroma Cb texture
-	 * decl s2			; Sampler for chroma Cr texture
-	 * decl s3			; Sampler for ref surface texture
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		decl = vl_decl_samplers(i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * tex2d t1, i0, s0		; Read texel from luma texture
-	 * mov t0.x, t1.x		; Move luma sample into .x component
-	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
-	 * mov t0.y, t1.x		; Move Cb sample into .y component
-	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
-	 * mov t0.z, t1.x		; Move Cr sample into .z component
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul t0, t0, c0		; Rescale texel to correct range */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * tex2d t1, i3, s3		; Read texel from ref macroblock top field
-	 * tex2d t2, i4, s3		; Read texel from ref macroblock bottom field
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* XXX: Pos values off by 0.5? */
-	/* sub t4, i5.y, c1.x		; Sub 0.5 from denormalized pos */
-	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 5, TGSI_FILE_CONSTANT, 1);
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* mul t3, t4, c1.x		; Multiply pos Y-coord by 1/2 */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1);
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* floor t3, t3			; Get rid of fractional part */
-	inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* mul t3, t3, c1.y		; Multiply by 2 */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1);
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* sub t3, t4, t3		; Subtract from original Y to get Y % 2 */
-	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
-	/* lerp t1, t3, t1, t2		; Choose between top and bottom fields based on Y % 2 */
-	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* add o0, t0, t1		; Add ref and differential to form final output */
-	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	fs.tokens = tokens;
-	mc->p_fs[1] = pipe->create_fs_state(pipe, &fs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateVertexShaderFrameBMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	vs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0		; Vertex pos
-	 * decl i1		; Luma texcoords
-	 * decl i2		; Chroma Cb texcoords
-	 * decl i3		; Chroma Cr texcoords
-	 * decl i4              ; First ref macroblock top field texcoords
-	 * decl i5              ; First ref macroblock bottom field texcoords (unused, packed in the same stream)
-	 * decl i6		; Second ref macroblock top field texcoords
-	 * decl i7		; Second ref macroblock bottom field texcoords (unused, packed in the same stream)
-	 */
-	for (i = 0; i < 8; i++)
-	{
-		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl o0		; Vertex pos
-	 * decl o1		; Luma texcoords
-	 * decl o2		; Chroma Cb texcoords
-	 * decl o3		; Chroma Cr texcoords
-	 * decl o4		; First ref macroblock texcoords
-	 * decl o5		; Second ref macroblock texcoords
-	 */
-	for (i = 0; i < 6; i++)
-	{
-		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * mov o0, i0		; Move input vertex pos to output
-	 * mov o1, i1		; Move input luma texcoords to output
-	 * mov o2, i2		; Move input chroma Cb texcoords to output
-	 * mov o3, i3		; Move input chroma Cr texcoords to output
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * add o4, i0, i4	; Translate vertex pos by motion vec to form first ref macroblock texcoords
-	 * add o5, i0, i6	; Translate vertex pos by motion vec to form second ref macroblock texcoords
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 2) * 2);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	vs.tokens = tokens;
-	mc->b_vs[0] = pipe->create_vs_state(pipe, &vs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateVertexShaderFieldBMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	vs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0		; Vertex pos
-	 * decl i1		; Luma texcoords
-	 * decl i2		; Chroma Cb texcoords
-	 * decl i3		; Chroma Cr texcoords
-	 * decl i4              ; First ref macroblock top field texcoords
-	 * decl i5              ; First ref macroblock bottom field texcoords
-	 * decl i6              ; Second ref macroblock top field texcoords
-	 * decl i7              ; Second ref macroblock bottom field texcoords
-	 */
-	for (i = 0; i < 8; i++)
-	{
-		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* decl c0		; Render target dimensions */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl o0		; Vertex pos
-	 * decl o1		; Luma texcoords
-	 * decl o2		; Chroma Cb texcoords
-	 * decl o3		; Chroma Cr texcoords
-	 * decl o4		; First ref macroblock top field texcoords
-	 * decl o5		; First ref macroblock Bottom field texcoords
-	 * decl o6		; Second ref macroblock top field texcoords
-	 * decl o7		; Second ref macroblock Bottom field texcoords
-	 * decl o8		; Denormalized vertex pos
-	 */
-	for (i = 0; i < 9; i++)
-	{
-		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* decl t0, t1 */
-	decl = vl_decl_temps(0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * mov o0, i0		; Move input vertex pos to output
-	 * mov o1, i1		; Move input luma texcoords to output
-	 * mov o2, i2		; Move input chroma Cb texcoords to output
-	 * mov o3, i3		; Move input chroma Cr texcoords to output
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * add o4, i0, i4	; Translate vertex pos by motion vec to form first top field macroblock texcoords
-	 * add o5, i0, i5	; Translate vertex pos by motion vec to form first bottom field macroblock texcoords
-	 * add o6, i0, i6	; Translate vertex pos by motion vec to form second top field macroblock texcoords
-	 * add o7, i0, i7	; Translate vertex pos by motion vec to form second bottom field macroblock texcoords
-	 */
-	for (i = 0; i < 4; ++i)
-	{
-		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul o8, i0, c0	; Denorm vertex pos */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 8, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	vs.tokens = tokens;
-	mc->b_vs[1] = pipe->create_vs_state(pipe, &vs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateFragmentShaderFrameBMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 100;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	fs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0			; Luma texcoords
-	 * decl i1			; Chroma Cb texcoords
-	 * decl i2			; Chroma Cr texcoords
-	 * decl i3			; First ref macroblock texcoords
-	 * decl i4			; Second ref macroblock texcoords
-	 */
-	for (i = 0; i < 5; ++i)
-	{
-		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
-	 * decl c1			; Constant 1/2 in .x channel to use as weight to blend past and future texels
-	 */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl o0			; Fragment color */
-	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl t0-t2 */
-	decl = vl_decl_temps(0, 2);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl s0			; Sampler for luma texture
-	 * decl s1			; Sampler for chroma Cb texture
-	 * decl s2			; Sampler for chroma Cr texture
-	 * decl s3			; Sampler for first ref surface texture
-	 * decl s4			; Sampler for second ref surface texture
-	 */
-	for (i = 0; i < 5; ++i)
-	{
-		decl = vl_decl_samplers(i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * tex2d t1, i0, s0		; Read texel from luma texture
-	 * mov t0.x, t1.x		; Move luma sample into .x component
-	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
-	 * mov t0.y, t1.x		; Move Cb sample into .y component
-	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
-	 * mov t0.z, t1.x		; Move Cr sample into .z component
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul t0, t0, c0		; Rescale texel to correct range */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * tex2d t1, i3, s3		; Read texel from first ref macroblock
-	 * tex2d t2, i4, s4		; Read texel from second ref macroblock
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, i + 3);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* lerp t1, c1.x, t1, t2	; Blend past and future texels */
-	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* add o0, t0, t1		; Add past/future ref and differential to form final output */
-	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	fs.tokens = tokens;
-	mc->b_fs[0] = pipe->create_fs_state(pipe, &fs);
-	free(tokens);
-
-	return 0;
-}
-
-static int vlCreateFragmentShaderFieldBMB
-(
-	struct vlR16SnormBufferedMC *mc
-)
-{
-	const unsigned int		max_tokens = 200;
-
-	struct pipe_context		*pipe;
-	struct pipe_shader_state	fs;
-	struct tgsi_token		*tokens;
-	struct tgsi_header		*header;
-
-	struct tgsi_full_declaration	decl;
-	struct tgsi_full_instruction	inst;
-
-	unsigned int			ti;
-	unsigned int			i;
-
-	assert(mc);
-
-	pipe = mc->pipe;
-	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
-
-	/* Version */
-	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
-	/* Header */
-	header = (struct tgsi_header*)&tokens[1];
-	*header = tgsi_build_header();
-	/* Processor */
-	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
-
-	ti = 3;
-
-	/*
-	 * decl i0			; Luma texcoords
-	 * decl i1			; Chroma Cb texcoords
-	 * decl i2			; Chroma Cr texcoords
-	 * decl i3			; First ref macroblock top field texcoords
-	 * decl i4			; First ref macroblock bottom field texcoords
-	 * decl i5			; Second ref macroblock top field texcoords
-	 * decl i6			; Second ref macroblock bottom field texcoords
-	 * decl i7			; Denormalized vertex pos
-	 */
-	for (i = 0; i < 8; ++i)
-	{
-		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
-	 * decl c1			; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels
-	 *				; and for Y-mod-2 top/bottom field selection
-	 */
-	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl o0			; Fragment color */
-	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/* decl t0-t5 */
-	decl = vl_decl_temps(0, 5);
-	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * decl s0			; Sampler for luma texture
-	 * decl s1			; Sampler for chroma Cb texture
-	 * decl s2			; Sampler for chroma Cr texture
-	 * decl s3			; Sampler for first ref surface texture
-	 * decl s4			; Sampler for second ref surface texture
-	 */
-	for (i = 0; i < 5; ++i)
-	{
-		decl = vl_decl_samplers(i, i);
-		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/*
-	 * tex2d t1, i0, s0		; Read texel from luma texture
-	 * mov t0.x, t1.x		; Move luma sample into .x component
-	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
-	 * mov t0.y, t1.x		; Move Cb sample into .y component
-	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
-	 * mov t0.z, t1.x		; Move Cr sample into .z component
-	 */
-	for (i = 0; i < 3; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* mul t0, t0, c0		; Rescale texel to correct range */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* XXX: Pos values off by 0.5? */
-	/* sub t4, i7.y, c1.x		; Sub 0.5 from denormalized pos */
-	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 7, TGSI_FILE_CONSTANT, 1);
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* mul t3, t4, c1.x		; Multiply pos Y-coord by 1/2 */
-	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1);
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* floor t3, t3			; Get rid of fractional part */
-	inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* mul t3, t3, c1.y		; Multiply by 2 */
-	inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1);
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
-	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* sub t3, t4, t3		; Subtract from original Y to get Y % 2 */
-	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * tex2d t1, i3, s3		; Read texel from past ref macroblock top field
-	 * tex2d t2, i4, s3		; Read texel from past ref macroblock bottom field
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
-	/* lerp t1, t3, t1, t2		; Choose between top and bottom fields based on Y % 2 */
-	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/*
-	 * tex2d t4, i5, s4		; Read texel from future ref macroblock top field
-	 * tex2d t5, i6, s4		; Read texel from future ref macroblock bottom field
-	 */
-	for (i = 0; i < 2; ++i)
-	{
-		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 5, TGSI_FILE_SAMPLER, 4);
-		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-	}
-
-	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
-	/* lerp t2, t3, t4, t5		; Choose between top and bottom fields based on Y % 2 */
-	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* lerp t1, c1.x, t1, t2	; Blend past and future texels */
-	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
-	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* add o0, t0, t1		; Add past/future ref and differential to form final output */
-	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	/* end */
-	inst = vl_end();
-	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
-
-	fs.tokens = tokens;
-	mc->b_fs[1] = pipe->create_fs_state(pipe, &fs);
-	free(tokens);
-
-	return 0;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_render.h b/src/gallium/state_trackers/g3dvl/vl_render.h
deleted file mode 100644
index 166030b498..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_render.h
+++ /dev/null
@@ -1,38 +0,0 @@
-#ifndef vl_render_h
-#define vl_render_h
-
-#include "vl_types.h"
-
-struct pipe_surface;
-
-struct vlRender
-{
-	int (*vlBegin)
-	(
-		struct vlRender *render
-	);
-
-	int (*vlRenderMacroBlocksMpeg2)
-	(
-		struct vlRender *render,
-		struct vlMpeg2MacroBlockBatch *batch,
-		struct vlSurface *surface
-	);
-
-	int (*vlEnd)
-	(
-		struct vlRender *render
-	);
-
-	int (*vlFlush)
-	(
-		struct vlRender *render
-	);
-
-	int (*vlDestroy)
-	(
-		struct vlRender *render
-	);
-};
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.c b/src/gallium/state_trackers/g3dvl/vl_screen.c
deleted file mode 100644
index ade8643a66..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_screen.c
+++ /dev/null
@@ -1,115 +0,0 @@
-#define VL_INTERNAL
-#include "vl_screen.h"
-#include <assert.h>
-#include <util/u_memory.h>
-
-int vlCreateScreen
-(
-	struct vlDisplay *display,
-	int screen,
-	struct pipe_screen *pscreen,
-	struct vlScreen **vl_screen
-)
-{
-	struct vlScreen *scrn;
-
-	assert(display);
-	assert(pscreen);
-	assert(vl_screen);
-
-	scrn = CALLOC_STRUCT(vlScreen);
-
-	if (!scrn)
-		return 1;
-
-	scrn->display = display;
-	scrn->ordinal = screen;
-	scrn->pscreen = pscreen;
-	*vl_screen = scrn;
-
-	return 0;
-}
-
-int vlDestroyScreen
-(
-	struct vlScreen *screen
-)
-{
-	assert(screen);
-
-	FREE(screen);
-
-	return 0;
-}
-
-struct vlDisplay* vlGetDisplay
-(
-	struct vlScreen *screen
-)
-{
-	assert(screen);
-
-	return screen->display;
-}
-
-struct pipe_screen* vlGetPipeScreen
-(
-	struct vlScreen *screen
-)
-{
-	assert(screen);
-
-	return screen->pscreen;
-}
-
-unsigned int vlGetMaxProfiles
-(
-	struct vlScreen *screen
-)
-{
-	assert(screen);
-
-	return vlProfileCount;
-}
-
-int vlQueryProfiles
-(
-	struct vlScreen *screen,
-	enum vlProfile *profiles
-)
-{
-	assert(screen);
-	assert(profiles);
-
-	profiles[0] = vlProfileMpeg2Simple;
-	profiles[1] = vlProfileMpeg2Main;
-
-	return 0;
-}
-
-unsigned int vlGetMaxEntryPoints
-(
-	struct vlScreen *screen
-)
-{
-	assert(screen);
-
-	return vlEntryPointCount;
-}
-
-int vlQueryEntryPoints
-(
-	struct vlScreen *screen,
-	enum vlProfile profile,
-	enum vlEntryPoint *entry_points
-)
-{
-	assert(screen);
-	assert(entry_points);
-
-	entry_points[0] = vlEntryPointIDCT;
-	entry_points[1] = vlEntryPointMC;
-	entry_points[2] = vlEntryPointCSC;
-
-	return 0;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.h b/src/gallium/state_trackers/g3dvl/vl_screen.h
deleted file mode 100644
index 98f3d429b6..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_screen.h
+++ /dev/null
@@ -1,63 +0,0 @@
-#ifndef vl_screen_h
-#define vl_screen_h
-
-#include "vl_types.h"
-
-struct pipe_screen;
-
-#ifdef VL_INTERNAL
-struct vlScreen
-{
-	struct vlDisplay	*display;
-	unsigned int		ordinal;
-	struct pipe_screen	*pscreen;
-};
-#endif
-
-int vlCreateScreen
-(
-	struct vlDisplay *display,
-	int screen,
-	struct pipe_screen *pscreen,
-	struct vlScreen **vl_screen
-);
-
-int vlDestroyScreen
-(
-	struct vlScreen *screen
-);
-
-struct vlDisplay* vlGetDisplay
-(
-	struct vlScreen *screen
-);
-
-struct pipe_screen* vlGetPipeScreen
-(
-	struct vlScreen *screen
-);
-
-unsigned int vlGetMaxProfiles
-(
-	struct vlScreen *screen
-);
-
-int vlQueryProfiles
-(
-	struct vlScreen *screen,
-	enum vlProfile *profiles
-);
-
-unsigned int vlGetMaxEntryPoints
-(
-	struct vlScreen *screen
-);
-
-int vlQueryEntryPoints
-(
-	struct vlScreen *screen,
-	enum vlProfile profile,
-	enum vlEntryPoint *entry_points
-);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c
deleted file mode 100644
index 51f1721a33..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_shader_build.c
+++ /dev/null
@@ -1,215 +0,0 @@
-#include "vl_shader_build.h"
-#include <assert.h>
-#include <tgsi/tgsi_parse.h>
-#include <tgsi/tgsi_build.h>
-
-struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
-{
-	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
-
-	decl.Declaration.File = TGSI_FILE_INPUT;
-	decl.Declaration.Semantic = 1;
-	decl.Semantic.SemanticName = name;
-	decl.Semantic.SemanticIndex = index;
-	decl.DeclarationRange.First = first;
-	decl.DeclarationRange.Last = last;
-
-	return decl;
-}
-
-struct tgsi_full_declaration vl_decl_interpolated_input
-(
-	unsigned int name,
-	unsigned int index,
-	unsigned int first,
-	unsigned int last,
-	int interpolation
-)
-{
-	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
-
-	assert
-	(
-		interpolation == TGSI_INTERPOLATE_CONSTANT ||
-		interpolation == TGSI_INTERPOLATE_LINEAR ||
-		interpolation == TGSI_INTERPOLATE_PERSPECTIVE
-	);
-
-	decl.Declaration.File = TGSI_FILE_INPUT;
-	decl.Declaration.Semantic = 1;
-	decl.Semantic.SemanticName = name;
-	decl.Semantic.SemanticIndex = index;
-	decl.Declaration.Interpolate = interpolation;;
-	decl.DeclarationRange.First = first;
-	decl.DeclarationRange.Last = last;
-
-	return decl;
-}
-
-struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
-{
-	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
-
-	decl.Declaration.File = TGSI_FILE_CONSTANT;
-	decl.Declaration.Semantic = 1;
-	decl.Semantic.SemanticName = name;
-	decl.Semantic.SemanticIndex = index;
-	decl.DeclarationRange.First = first;
-	decl.DeclarationRange.Last = last;
-
-	return decl;
-}
-
-struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
-{
-	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
-
-	decl.Declaration.File = TGSI_FILE_OUTPUT;
-	decl.Declaration.Semantic = 1;
-	decl.Semantic.SemanticName = name;
-	decl.Semantic.SemanticIndex = index;
-	decl.DeclarationRange.First = first;
-	decl.DeclarationRange.Last = last;
-
-	return decl;
-}
-
-struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last)
-{
-	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
-
-	decl = tgsi_default_full_declaration();
-	decl.Declaration.File = TGSI_FILE_TEMPORARY;
-	decl.DeclarationRange.First = first;
-	decl.DeclarationRange.Last = last;
-
-	return decl;
-}
-
-struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last)
-{
-	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
-
-	decl = tgsi_default_full_declaration();
-	decl.Declaration.File = TGSI_FILE_SAMPLER;
-	decl.DeclarationRange.First = first;
-	decl.DeclarationRange.Last = last;
-
-	return decl;
-}
-
-struct tgsi_full_instruction vl_inst2
-(
-	int opcode,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src_file,
-	unsigned int src_index
-)
-{
-	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
-
-	inst.Instruction.Opcode = opcode;
-	inst.Instruction.NumDstRegs = 1;
-	inst.FullDstRegisters[0].DstRegister.File = dst_file;
-	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
-	inst.Instruction.NumSrcRegs = 1;
-	inst.FullSrcRegisters[0].SrcRegister.File = src_file;
-	inst.FullSrcRegisters[0].SrcRegister.Index = src_index;
-
-	return inst;
-}
-
-struct tgsi_full_instruction vl_inst3
-(
-	int opcode,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src1_file,
-	unsigned int src1_index,
-	enum tgsi_file_type src2_file,
-	unsigned int src2_index
-)
-{
-	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
-
-	inst.Instruction.Opcode = opcode;
-	inst.Instruction.NumDstRegs = 1;
-	inst.FullDstRegisters[0].DstRegister.File = dst_file;
-	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
-	inst.Instruction.NumSrcRegs = 2;
-	inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
-	inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
-	inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
-	inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
-
-	return inst;
-}
-
-struct tgsi_full_instruction vl_tex
-(
-	int tex,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src1_file,
-	unsigned int src1_index,
-	enum tgsi_file_type src2_file,
-	unsigned int src2_index
-)
-{
-	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
-
-	inst.Instruction.Opcode = TGSI_OPCODE_TEX;
-	inst.Instruction.NumDstRegs = 1;
-	inst.FullDstRegisters[0].DstRegister.File = dst_file;
-	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
-	inst.Instruction.NumSrcRegs = 2;
-	inst.InstructionExtTexture.Texture = tex;
-	inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
-	inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
-	inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
-	inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
-
-	return inst;
-}
-
-struct tgsi_full_instruction vl_inst4
-(
-	int opcode,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src1_file,
-	unsigned int src1_index,
-	enum tgsi_file_type src2_file,
-	unsigned int src2_index,
-	enum tgsi_file_type src3_file,
-	unsigned int src3_index
-)
-{
-	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
-
-	inst.Instruction.Opcode = opcode;
-	inst.Instruction.NumDstRegs = 1;
-	inst.FullDstRegisters[0].DstRegister.File = dst_file;
-	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
-	inst.Instruction.NumSrcRegs = 3;
-	inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
-	inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
-	inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
-	inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
-	inst.FullSrcRegisters[2].SrcRegister.File = src3_file;
-	inst.FullSrcRegisters[2].SrcRegister.Index = src3_index;
-
-	return inst;
-}
-
-struct tgsi_full_instruction vl_end(void)
-{
-	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
-
-	inst.Instruction.Opcode = TGSI_OPCODE_END;
-	inst.Instruction.NumDstRegs = 0;
-	inst.Instruction.NumSrcRegs = 0;
-
-	return inst;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.h b/src/gallium/state_trackers/g3dvl/vl_shader_build.h
deleted file mode 100644
index dc615cb156..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_shader_build.h
+++ /dev/null
@@ -1,61 +0,0 @@
-#ifndef vl_shader_build_h
-#define vl_shader_build_h
-
-#include <pipe/p_shader_tokens.h>
-
-struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
-struct tgsi_full_declaration vl_decl_interpolated_input
-(
-	unsigned int name,
-	unsigned int index,
-	unsigned int first,
-	unsigned int last,
-	int interpolation
-);
-struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
-struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
-struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last);
-struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last);
-struct tgsi_full_instruction vl_inst2
-(
-	int opcode,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src_file,
-	unsigned int src_index
-);
-struct tgsi_full_instruction vl_inst3
-(
-	int opcode,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src1_file,
-	unsigned int src1_index,
-	enum tgsi_file_type src2_file,
-	unsigned int src2_index
-);
-struct tgsi_full_instruction vl_tex
-(
-	int tex,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src1_file,
-	unsigned int src1_index,
-	enum tgsi_file_type src2_file,
-	unsigned int src2_index
-);
-struct tgsi_full_instruction vl_inst4
-(
-	int opcode,
-	enum tgsi_file_type dst_file,
-	unsigned int dst_index,
-	enum tgsi_file_type src1_file,
-	unsigned int src1_index,
-	enum tgsi_file_type src2_file,
-	unsigned int src2_index,
-	enum tgsi_file_type src3_file,
-	unsigned int src3_index
-);
-struct tgsi_full_instruction vl_end(void);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c
deleted file mode 100644
index 7f60852cae..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_surface.c
+++ /dev/null
@@ -1,242 +0,0 @@
-#define VL_INTERNAL
-#include "vl_surface.h"
-#include <assert.h>
-#include <string.h>
-#include <pipe/p_screen.h>
-#include <pipe/p_state.h>
-#include <pipe/p_inlines.h>
-#include <util/u_memory.h>
-#include <vl_winsys.h>
-#include "vl_screen.h"
-#include "vl_context.h"
-#include "vl_render.h"
-#include "vl_csc.h"
-#include "vl_util.h"
-
-int vlCreateSurface
-(
-	struct vlScreen *screen,
-	unsigned int width,
-	unsigned int height,
-	enum vlFormat format,
-	struct vlSurface **surface
-)
-{
-	struct vlSurface	*sfc;
-	struct pipe_texture	template;
-
-	assert(screen);
-	assert(surface);
-
-	sfc = CALLOC_STRUCT(vlSurface);
-
-	if (!sfc)
-		return 1;
-
-	sfc->screen = screen;
-	sfc->width = width;
-	sfc->height = height;
-	sfc->format = format;
-
-	memset(&template, 0, sizeof(struct pipe_texture));
-	template.target = PIPE_TEXTURE_2D;
-	template.format = PIPE_FORMAT_A8R8G8B8_UNORM;
-	template.last_level = 0;
-	template.width[0] = vlRoundUpPOT(sfc->width);
-	template.height[0] = vlRoundUpPOT(sfc->height);
-	template.depth[0] = 1;
-	pf_get_block(template.format, &template.block);
-	template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET;
-
-	sfc->texture = vlGetPipeScreen(screen)->texture_create(vlGetPipeScreen(screen), &template);
-
-	if (!sfc->texture)
-	{
-		FREE(sfc);
-		return 1;
-	}
-
-	*surface = sfc;
-
-	return 0;
-}
-
-int vlDestroySurface
-(
-	struct vlSurface *surface
-)
-{
-	assert(surface);
-
-	pipe_texture_reference(&surface->texture, NULL);
-	FREE(surface);
-
-	return 0;
-}
-
-int vlRenderMacroBlocksMpeg2
-(
-	struct vlMpeg2MacroBlockBatch *batch,
-	struct vlSurface *surface
-)
-{
-	assert(batch);
-	assert(surface);
-	assert(surface->context);
-
-	surface->context->render->vlBegin(surface->context->render);
-
-	surface->context->render->vlRenderMacroBlocksMpeg2
-	(
-		surface->context->render,
-		batch,
-		surface
-	);
-
-	surface->context->render->vlEnd(surface->context->render);
-
-	return 0;
-}
-
-int vlPutPicture
-(
-	struct vlSurface *surface,
-	vlNativeDrawable drawable,
-	int srcx,
-	int srcy,
-	int srcw,
-	int srch,
-	int destx,
-	int desty,
-	int destw,
-	int desth,
-	int drawable_w,
-	int drawable_h,
-	enum vlPictureType picture_type
-)
-{
-	struct vlCSC		*csc;
-	struct pipe_context	*pipe;
-
-	assert(surface);
-	assert(surface->context);
-
-	surface->context->render->vlFlush(surface->context->render);
-
-	csc = surface->context->csc;
-	pipe = surface->context->pipe;
-
-	csc->vlResizeFrameBuffer(csc, drawable_w, drawable_h);
-
-	csc->vlBegin(csc);
-
-	csc->vlPutPicture
-	(
-		csc,
-		surface,
-		srcx,
-		srcy,
-		srcw,
-		srch,
-		destx,
-		desty,
-		destw,
-		desth,
-		picture_type
-	);
-
-	csc->vlEnd(csc);
-
-	pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &surface->disp_fence);
-
-	bind_pipe_drawable(pipe, drawable);
-
-	pipe->screen->flush_frontbuffer
-	(
-		pipe->screen,
-		csc->vlGetFrameBuffer(csc),
-		pipe->priv
-	);
-
-	return 0;
-}
-
-int vlSurfaceGetStatus
-(
-	struct vlSurface *surface,
-	enum vlResourceStatus *status
-)
-{
-	assert(surface);
-	assert(surface->context);
-	assert(status);
-
-	if (surface->render_fence && !surface->context->pipe->screen->fence_signalled(surface->context->pipe->screen, surface->render_fence, 0))
-	{
-		*status = vlResourceStatusRendering;
-		return 0;
-	}
-
-	if (surface->disp_fence && !surface->context->pipe->screen->fence_signalled(surface->context->pipe->screen, surface->disp_fence, 0))
-	{
-		*status = vlResourceStatusDisplaying;
-		return 0;
-	}
-
-	*status = vlResourceStatusFree;
-
-	return 0;
-}
-
-int vlSurfaceFlush
-(
-	struct vlSurface *surface
-)
-{
-	assert(surface);
-	assert(surface->context);
-
-	surface->context->render->vlFlush(surface->context->render);
-
-	return 0;
-}
-
-int vlSurfaceSync
-(
-	struct vlSurface *surface
-)
-{
-	assert(surface);
-	assert(surface->context);
-	assert(surface->render_fence);
-
-	surface->context->pipe->screen->fence_finish(surface->context->pipe->screen, surface->render_fence, 0);
-
-	return 0;
-}
-
-struct vlScreen* vlSurfaceGetScreen
-(
-	struct vlSurface *surface
-)
-{
-	assert(surface);
-
-	return surface->screen;
-}
-
-struct vlContext* vlBindToContext
-(
-	struct vlSurface *surface,
-	struct vlContext *context
-)
-{
-	struct vlContext *old;
-
-	assert(surface);
-
-	old = surface->context;
-	surface->context = context;
-
-	return old;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h
deleted file mode 100644
index 133e1515ef..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_surface.h
+++ /dev/null
@@ -1,86 +0,0 @@
-#ifndef vl_surface_h
-#define vl_surface_h
-
-#include "vl_types.h"
-
-#ifdef VL_INTERNAL
-struct pipe_texture;
-
-struct vlSurface
-{
-	struct vlScreen			*screen;
-	struct vlContext		*context;
-	unsigned int			width;
-	unsigned int			height;
-	enum vlFormat			format;
-	struct pipe_texture		*texture;
-	struct pipe_fence_handle	*render_fence;
-	struct pipe_fence_handle	*disp_fence;
-};
-#endif
-
-int vlCreateSurface
-(
-	struct vlScreen *screen,
-	unsigned int width,
-	unsigned int height,
-	enum vlFormat format,
-	struct vlSurface **surface
-);
-
-int vlDestroySurface
-(
-	struct vlSurface *surface
-);
-
-int vlRenderMacroBlocksMpeg2
-(
-	struct vlMpeg2MacroBlockBatch *batch,
-	struct vlSurface *surface
-);
-
-int vlPutPicture
-(
-	struct vlSurface *surface,
-	vlNativeDrawable drawable,
-	int srcx,
-	int srcy,
-	int srcw,
-	int srch,
-	int destx,
-	int desty,
-	int destw,
-	int desth,
-	int drawable_w,
-	int drawable_h,
-	enum vlPictureType picture_type
-);
-
-int vlSurfaceGetStatus
-(
-	struct vlSurface *surface,
-	enum vlResourceStatus *status
-);
-
-int vlSurfaceFlush
-(
-	struct vlSurface *surface
-);
-
-int vlSurfaceSync
-(
-	struct vlSurface *surface
-);
-
-struct vlScreen* vlSurfaceGetScreen
-(
-	struct vlSurface *surface
-);
-
-struct vlContext* vlBindToContext
-(
-	struct vlSurface *surface,
-	struct vlContext *context
-);
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h
deleted file mode 100644
index 274e1f7437..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_types.h
+++ /dev/null
@@ -1,115 +0,0 @@
-#ifndef vl_types_h
-#define vl_types_h
-
-#if 1 /*#ifdef X11*/
-#include <X11/Xlib.h>
-
-typedef Display* vlNativeDisplay;
-typedef Drawable vlNativeDrawable;
-#endif
-
-struct vlDisplay;
-struct vlScreen;
-struct vlContext;
-struct vlSurface;
-
-enum vlResourceStatus
-{
-	vlResourceStatusFree,
-	vlResourceStatusRendering,
-	vlResourceStatusDisplaying
-};
-
-enum vlProfile
-{
-	vlProfileMpeg2Simple,
-	vlProfileMpeg2Main,
-
-	vlProfileCount
-};
-
-enum vlEntryPoint
-{
-	vlEntryPointIDCT,
-	vlEntryPointMC,
-	vlEntryPointCSC,
-
-	vlEntryPointCount
-};
-
-enum vlFormat
-{
-	vlFormatYCbCr420,
-	vlFormatYCbCr422,
-	vlFormatYCbCr444
-};
-
-enum vlPictureType
-{
-	vlPictureTypeTopField,
-	vlPictureTypeBottomField,
-	vlPictureTypeFrame
-};
-
-enum vlMotionType
-{
-	vlMotionTypeField,
-	vlMotionTypeFrame,
-	vlMotionTypeDualPrime,
-	vlMotionType16x8
-};
-
-enum vlFieldOrder
-{
-	vlFieldOrderFirst,
-	vlFieldOrderSecond
-};
-
-enum vlDCTType
-{
-	vlDCTTypeFrameCoded,
-	vlDCTTypeFieldCoded
-};
-
-struct vlVertex2f
-{
-	float x, y;
-};
-
-struct vlVertex4f
-{
-	float x, y, z, w;
-};
-
-enum vlMacroBlockType
-{
-	vlMacroBlockTypeIntra,
-	vlMacroBlockTypeFwdPredicted,
-	vlMacroBlockTypeBkwdPredicted,
-	vlMacroBlockTypeBiPredicted,
-
-	vlNumMacroBlockTypes
-};
-
-struct vlMpeg2MacroBlock
-{
-	unsigned int		mbx, mby;
-	enum vlMacroBlockType	mb_type;
-	enum vlMotionType	mo_type;
-	enum vlDCTType		dct_type;
-	int			PMV[2][2][2];
-	unsigned int		cbp;
-	short			*blocks;
-};
-
-struct vlMpeg2MacroBlockBatch
-{
-	struct vlSurface		*past_surface;
-	struct vlSurface		*future_surface;
-	enum vlPictureType		picture_type;
-	enum vlFieldOrder		field_order;
-	unsigned int			num_macroblocks;
-	struct vlMpeg2MacroBlock	*macroblocks;
-};
-
-#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_util.c b/src/gallium/state_trackers/g3dvl/vl_util.c
deleted file mode 100644
index 50aa9af66f..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_util.c
+++ /dev/null
@@ -1,16 +0,0 @@
-#include "vl_util.h"
-#include <assert.h>
-
-unsigned int vlRoundUpPOT(unsigned int x)
-{
-	unsigned int i;
-
-	assert(x > 0);
-
-	--x;
-
-	for (i = 1; i < sizeof(unsigned int) * 8; i <<= 1)
-		x |= x >> i;
-
-	return x + 1;
-}
diff --git a/src/gallium/state_trackers/g3dvl/vl_util.h b/src/gallium/state_trackers/g3dvl/vl_util.h
deleted file mode 100644
index bc98e79df4..0000000000
--- a/src/gallium/state_trackers/g3dvl/vl_util.h
+++ /dev/null
@@ -1,6 +0,0 @@
-#ifndef vl_util_h
-#define vl_util_h
-
-unsigned int vlRoundUpPOT(unsigned int x);
-
-#endif
diff --git a/src/gallium/state_trackers/python/retrace/interpreter.py b/src/gallium/state_trackers/python/retrace/interpreter.py
index 6f0bd6ae52..f4ed2fde4d 100755
--- a/src/gallium/state_trackers/python/retrace/interpreter.py
+++ b/src/gallium/state_trackers/python/retrace/interpreter.py
@@ -314,7 +314,7 @@ class Screen(Object):
         if texture is None:
             return None
         transfer = Transfer(texture.get_surface(face, level, zslice), x, y, w, h)
-        if transfer and usage != gallium.PIPE_TRANSFER_WRITE:
+        if transfer and usage & gallium.PIPE_TRANSFER_READ
             if self.interpreter.options.all:
                 self.interpreter.present(transfer.surface, 'transf_read', x, y, w, h)
         return transfer
diff --git a/src/gallium/state_trackers/python/st_softpipe_winsys.c b/src/gallium/state_trackers/python/st_softpipe_winsys.c
index f0a4826a00..f0abd12e3d 100644
--- a/src/gallium/state_trackers/python/st_softpipe_winsys.c
+++ b/src/gallium/state_trackers/python/st_softpipe_winsys.c
@@ -172,6 +172,7 @@ st_softpipe_surface_buffer_create(struct pipe_winsys *winsys,
                                   unsigned width, unsigned height,
                                   enum pipe_format format,
                                   unsigned usage,
+                                  unsigned tex_usage,
                                   unsigned *stride)
 {
    const unsigned alignment = 64;
diff --git a/src/gallium/state_trackers/vega/renderer.c b/src/gallium/state_trackers/vega/renderer.c
index f7c5f2f0cd..396c88aa3d 100644
--- a/src/gallium/state_trackers/vega/renderer.c
+++ b/src/gallium/state_trackers/vega/renderer.c
@@ -37,6 +37,7 @@
 #include "util/u_draw_quad.h"
 #include "util/u_simple_shaders.h"
 #include "util/u_memory.h"
+#include "util/u_rect.h"
 
 #include "cso_cache/cso_context.h"
 
@@ -457,10 +458,17 @@ void renderer_copy_surface(struct renderer *ctx,
                                      PIPE_BUFFER_USAGE_GPU_WRITE);
 
    /* load temp texture */
-   pipe->surface_copy(pipe,
-                      texSurf, 0, 0,   /* dest */
-                      src, srcLeft, srcTop, /* src */
-                      srcW, srcH);     /* size */
+   if (pipe->surface_copy) {
+      pipe->surface_copy(pipe,
+                         texSurf, 0, 0,   /* dest */
+                         src, srcLeft, srcTop, /* src */
+                         srcW, srcH);     /* size */
+   } else {
+      util_surface_copy(pipe, FALSE,
+                        texSurf, 0, 0,   /* dest */
+                        src, srcLeft, srcTop, /* src */
+                        srcW, srcH);     /* size */
+   }
 
    /* free the surface, update the texture if necessary.*/
    screen->tex_surface_destroy(texSurf);
diff --git a/src/gallium/state_trackers/vega/st_inlines.h b/src/gallium/state_trackers/vega/st_inlines.h
index 1f331dfcdb..610755e063 100644
--- a/src/gallium/state_trackers/vega/st_inlines.h
+++ b/src/gallium/state_trackers/vega/st_inlines.h
@@ -57,8 +57,7 @@ st_cond_flush_get_tex_transfer(struct vg_context *st,
       pipe->is_texture_referenced(pipe, pt, face, level);
 
    if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) ||
-		      usage == PIPE_TRANSFER_WRITE ||
-		      usage == PIPE_TRANSFER_READ_WRITE))
+		      (usage & PIPE_TRANSFER_WRITE)))
       vgFlush();
 
    return screen->get_tex_transfer(screen, pt, face, level, zslice, usage,
diff --git a/src/gallium/state_trackers/vega/vg_tracker.c b/src/gallium/state_trackers/vega/vg_tracker.c
index 56cc60aebe..c4da01e52c 100644
--- a/src/gallium/state_trackers/vega/vg_tracker.c
+++ b/src/gallium/state_trackers/vega/vg_tracker.c
@@ -235,13 +235,23 @@ static void setup_new_alpha_mask(struct vg_context *ctx,
          old_texture,
          0, 0, 0,
          PIPE_BUFFER_USAGE_GPU_READ);
-      pipe->surface_copy(pipe,
-                         surface,
-                         0, 0,
-                         old_surface,
-                         0, 0,
-                         MIN2(old_surface->width, width),
-                         MIN2(old_surface->height, height));
+      if (pipe->surface_copy) {
+         pipe->surface_copy(pipe,
+                            surface,
+                            0, 0,
+                            old_surface,
+                            0, 0,
+                            MIN2(old_surface->width, width),
+                            MIN2(old_surface->height, height));
+      } else {
+         util_surface_copy(pipe, FALSE,
+                           surface,
+                           0, 0,
+                           old_surface,
+                           0, 0,
+                           MIN2(old_surface->width, width),
+                           MIN2(old_surface->height, height));
+      }
       if (surface)
          pipe_surface_reference(&surface, NULL);
       if (old_surface)
diff --git a/src/gallium/state_trackers/wgl/SConscript b/src/gallium/state_trackers/wgl/SConscript
index 69b88618ec..b05944a33b 100644
--- a/src/gallium/state_trackers/wgl/SConscript
+++ b/src/gallium/state_trackers/wgl/SConscript
@@ -18,20 +18,17 @@ if env['platform'] in ['windows']:
     ])
      
     sources = [
-        'icd/stw_icd.c',
-
-        'wgl/stw_wgl.c',
-
-        'shared/stw_context.c',
-        'shared/stw_device.c',
-        'shared/stw_framebuffer.c',
-        'shared/stw_pixelformat.c',
-        'shared/stw_extensionsstring.c',
-        'shared/stw_extswapinterval.c',
-        'shared/stw_getprocaddress.c',
-        'shared/stw_extgallium.c',
-        'shared/stw_arbpixelformat.c',
-        'shared/stw_tls.c',
+        'stw_context.c',
+        'stw_device.c',
+        'stw_ext_extensionsstring.c',
+        'stw_ext_gallium.c',
+        'stw_ext_pixelformat.c',
+        'stw_ext_swapinterval.c',
+        'stw_framebuffer.c',
+        'stw_getprocaddress.c',
+        'stw_pixelformat.c',
+        'stw_tls.c',
+        'stw_wgl.c',
     ]
 
     wgl = env.ConvenienceLibrary(
diff --git a/src/gallium/state_trackers/wgl/opengl32.def b/src/gallium/state_trackers/wgl/opengl32.def
index 596417ed84..5daa6ddd41 100644
--- a/src/gallium/state_trackers/wgl/opengl32.def
+++ b/src/gallium/state_trackers/wgl/opengl32.def
@@ -376,6 +376,7 @@ EXPORTS
 	DrvDescribePixelFormat
 	DrvGetLayerPaletteEntries
 	DrvGetProcAddress
+	DrvPresentBuffers
 	DrvRealizeLayerPalette
 	DrvReleaseContext
 	DrvSetCallbackProcs
diff --git a/src/gallium/state_trackers/wgl/opengl32.mingw.def b/src/gallium/state_trackers/wgl/opengl32.mingw.def
index 1f03ea3b37..6ebb31a6f1 100644
--- a/src/gallium/state_trackers/wgl/opengl32.mingw.def
+++ b/src/gallium/state_trackers/wgl/opengl32.mingw.def
@@ -375,6 +375,7 @@ EXPORTS
 	DrvDescribePixelFormat = DrvDescribePixelFormat@16
 	DrvGetLayerPaletteEntries = DrvGetLayerPaletteEntries@20
 	DrvGetProcAddress = DrvGetProcAddress@4
+	DrvPresentBuffers = DrvPresentBuffers@8
 	DrvRealizeLayerPalette = DrvRealizeLayerPalette@12
 	DrvReleaseContext = DrvReleaseContext@4
 	DrvSetCallbackProcs = DrvSetCallbackProcs@8
diff --git a/src/gallium/state_trackers/wgl/shared/stw_context.c b/src/gallium/state_trackers/wgl/shared/stw_context.c
deleted file mode 100644
index 4968ecc692..0000000000
--- a/src/gallium/state_trackers/wgl/shared/stw_context.c
+++ /dev/null
@@ -1,382 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-#include <windows.h>
-
-#include "main/mtypes.h"
-#include "main/context.h"
-#include "pipe/p_compiler.h"
-#include "pipe/p_context.h"
-#include "state_tracker/st_context.h"
-#include "state_tracker/st_public.h"
-
-#ifdef DEBUG
-#include "trace/tr_screen.h"
-#include "trace/tr_context.h"
-#endif
-
-#include "shared/stw_device.h"
-#include "shared/stw_winsys.h"
-#include "shared/stw_framebuffer.h"
-#include "shared/stw_pixelformat.h"
-#include "stw_public.h"
-#include "stw_context.h"
-#include "stw_tls.h"
-
-
-static INLINE struct stw_context *
-stw_context(GLcontext *glctx)
-{
-   if(!glctx)
-      return NULL;
-   assert(glctx->DriverCtx);
-   return (struct stw_context *)glctx->DriverCtx;
-}
-
-static INLINE struct stw_context *
-stw_current_context(void)
-{
-   /* We must check if multiple threads are being used or GET_CURRENT_CONTEXT 
-    * might return the current context of the thread first seen. */
-   _glapi_check_multithread();
-
-   {
-      GET_CURRENT_CONTEXT( glctx );
-      return stw_context(glctx);
-   }
-}
-
-BOOL
-stw_copy_context(
-   UINT_PTR hglrcSrc,
-   UINT_PTR hglrcDst,
-   UINT mask )
-{
-   struct stw_context *src;
-   struct stw_context *dst;
-   BOOL ret = FALSE;
-
-   pipe_mutex_lock( stw_dev->ctx_mutex );
-   
-   src = stw_lookup_context_locked( hglrcSrc );
-   dst = stw_lookup_context_locked( hglrcDst );
-
-   if (src && dst) { 
-      /* FIXME */
-      assert(0);
-      (void) src;
-      (void) dst;
-      (void) mask;
-   }
-
-   pipe_mutex_unlock( stw_dev->ctx_mutex );
-   
-   return ret;
-}
-
-BOOL
-stw_share_lists(
-   UINT_PTR hglrc1, 
-   UINT_PTR hglrc2 )
-{
-   struct stw_context *ctx1;
-   struct stw_context *ctx2;
-   BOOL ret = FALSE;
-
-   pipe_mutex_lock( stw_dev->ctx_mutex );
-   
-   ctx1 = stw_lookup_context_locked( hglrc1 );
-   ctx2 = stw_lookup_context_locked( hglrc2 );
-
-   if (ctx1 && ctx2 &&
-       ctx1->iPixelFormat == ctx2->iPixelFormat) { 
-      ret = _mesa_share_state(ctx2->st->ctx, ctx1->st->ctx);
-   }
-
-   pipe_mutex_unlock( stw_dev->ctx_mutex );
-   
-   return ret;
-}
-
-static void
-stw_viewport(GLcontext * glctx, GLint x, GLint y,
-             GLsizei width, GLsizei height)
-{
-   struct stw_context *ctx = (struct stw_context *)glctx->DriverCtx;
-   struct stw_framebuffer *fb;
-   
-   fb = stw_framebuffer_from_hdc( ctx->hdc );
-   if(fb) {
-      stw_framebuffer_update(fb);
-      stw_framebuffer_release(fb);
-   }
-}
-
-UINT_PTR
-stw_create_layer_context(
-   HDC hdc,
-   int iLayerPlane )
-{
-   int iPixelFormat;
-   const struct stw_pixelformat_info *pfi;
-   GLvisual visual;
-   struct stw_context *ctx = NULL;
-   struct pipe_screen *screen = NULL;
-   struct pipe_context *pipe = NULL;
-   
-   if(!stw_dev)
-      return 0;
-   
-   if (iLayerPlane != 0)
-      return 0;
-
-   iPixelFormat = GetPixelFormat(hdc);
-   if(!iPixelFormat)
-      return 0;
-   
-   pfi = stw_pixelformat_get_info( iPixelFormat - 1 );
-   stw_pixelformat_visual(&visual, pfi);
-   
-   ctx = CALLOC_STRUCT( stw_context );
-   if (ctx == NULL)
-      goto no_ctx;
-
-   ctx->hdc = hdc;
-   ctx->iPixelFormat = iPixelFormat;
-
-   screen = stw_dev->screen;
-
-#ifdef DEBUG
-   /* Unwrap screen */
-   if(stw_dev->trace_running)
-      screen = trace_screen(screen)->screen;
-#endif
-
-   pipe = stw_dev->stw_winsys->create_context( screen );
-   if (pipe == NULL) 
-      goto no_pipe;
-
-#ifdef DEBUG
-   /* Wrap context */
-   if(stw_dev->trace_running)
-      pipe = trace_context_create(stw_dev->screen, pipe);
-#endif
-
-   /* pass to stw_flush_frontbuffer as context_private */
-   assert(!pipe->priv);
-   pipe->priv = hdc;
-
-   ctx->st = st_create_context( pipe, &visual, NULL );
-   if (ctx->st == NULL) 
-      goto no_st_ctx;
-
-   ctx->st->ctx->DriverCtx = ctx;
-   ctx->st->ctx->Driver.Viewport = stw_viewport;
-
-   pipe_mutex_lock( stw_dev->ctx_mutex );
-   ctx->hglrc = handle_table_add(stw_dev->ctx_table, ctx);
-   pipe_mutex_unlock( stw_dev->ctx_mutex );
-   if (!ctx->hglrc)
-      goto no_hglrc;
-
-   return ctx->hglrc;
-
-no_hglrc:
-   st_destroy_context(ctx->st);
-   goto no_pipe; /* st_context_destroy already destroys pipe */
-no_st_ctx:
-   pipe->destroy( pipe );
-no_pipe:
-   FREE(ctx);
-no_ctx:
-   return 0;
-}
-
-BOOL
-stw_delete_context(
-   UINT_PTR hglrc )
-{
-   struct stw_context *ctx ;
-   BOOL ret = FALSE;
-   
-   if (!stw_dev)
-      return FALSE;
-
-   pipe_mutex_lock( stw_dev->ctx_mutex );
-   ctx = stw_lookup_context_locked(hglrc);
-   handle_table_remove(stw_dev->ctx_table, hglrc);
-   pipe_mutex_unlock( stw_dev->ctx_mutex );
-
-   if (ctx) {
-      struct stw_context *curctx = stw_current_context();
-      
-      /* Unbind current if deleting current context. */
-      if (curctx == ctx)
-         st_make_current( NULL, NULL, NULL );
-
-      st_destroy_context(ctx->st);
-      FREE(ctx);
-
-      ret = TRUE;
-   }
-
-   return ret;
-}
-
-BOOL
-stw_release_context(
-   UINT_PTR hglrc )
-{
-   struct stw_context *ctx;
-
-   if (!stw_dev)
-      return FALSE;
-
-   pipe_mutex_lock( stw_dev->ctx_mutex );
-   ctx = stw_lookup_context_locked( hglrc );
-   pipe_mutex_unlock( stw_dev->ctx_mutex );
-
-   if (!ctx)
-      return FALSE;
-   
-   /* The expectation is that ctx is the same context which is
-    * current for this thread.  We should check that and return False
-    * if not the case.
-    */
-   if (ctx != stw_current_context())
-      return FALSE;
-
-   if (stw_make_current( NULL, 0 ) == FALSE)
-      return FALSE;
-
-   return TRUE;
-}
-
-
-UINT_PTR
-stw_get_current_context( void )
-{
-   struct stw_context *ctx;
-
-   ctx = stw_current_context();
-   if(!ctx)
-      return 0;
-   
-   return ctx->hglrc;
-}
-
-HDC
-stw_get_current_dc( void )
-{
-   struct stw_context *ctx;
-
-   ctx = stw_current_context();
-   if(!ctx)
-      return NULL;
-   
-   return ctx->hdc;
-}
-
-BOOL
-stw_make_current(
-   HDC hdc,
-   UINT_PTR hglrc )
-{
-   struct stw_context *curctx = NULL;
-   struct stw_context *ctx = NULL;
-   struct stw_framebuffer *fb = NULL;
-
-   if (!stw_dev)
-      goto fail;
-
-   curctx = stw_current_context();
-   if (curctx != NULL) {
-      if (curctx->hglrc != hglrc)
-	 st_flush(curctx->st, PIPE_FLUSH_RENDER_CACHE, NULL);
-      
-      /* Return if already current. */
-      if (curctx->hglrc == hglrc && curctx->hdc == hdc) {
-         ctx = curctx;
-         fb = stw_framebuffer_from_hdc( hdc );
-         goto success;
-      }
-   }
-
-   if (hdc == NULL || hglrc == 0) {
-      return st_make_current( NULL, NULL, NULL );
-   }
-
-   pipe_mutex_lock( stw_dev->ctx_mutex ); 
-   ctx = stw_lookup_context_locked( hglrc );
-   pipe_mutex_unlock( stw_dev->ctx_mutex ); 
-   if(!ctx)
-      goto fail;
-
-   fb = stw_framebuffer_from_hdc( hdc );
-   if(!fb) { 
-      /* Applications should call SetPixelFormat before creating a context,
-       * but not all do, and the opengl32 runtime seems to use a default pixel
-       * format in some cases, so we must create a framebuffer for those here
-       */
-      int iPixelFormat = GetPixelFormat(hdc);
-      if(iPixelFormat)
-         fb = stw_framebuffer_create( hdc, iPixelFormat );
-      if(!fb) 
-         goto fail;
-   }
-   
-   if(fb->iPixelFormat != ctx->iPixelFormat)
-      goto fail;
-
-   /* Lazy allocation of the frame buffer */
-   if(!stw_framebuffer_allocate(fb))
-      goto fail;
-
-   /* Bind the new framebuffer */
-   ctx->hdc = hdc;
-   
-   /* pass to stw_flush_frontbuffer as context_private */
-   ctx->st->pipe->priv = hdc;
-   
-   if(!st_make_current( ctx->st, fb->stfb, fb->stfb ))
-      goto fail;
-
-success:
-   assert(fb);
-   if(fb) {
-      stw_framebuffer_update(fb);
-      stw_framebuffer_release(fb);
-   }
-   
-   return TRUE;
-
-fail:
-   if(fb)
-      stw_framebuffer_release(fb);
-   st_make_current( NULL, NULL, NULL );
-   return FALSE;
-}
diff --git a/src/gallium/state_trackers/wgl/shared/stw_public.h b/src/gallium/state_trackers/wgl/shared/stw_public.h
deleted file mode 100644
index 7fe9cfb356..0000000000
--- a/src/gallium/state_trackers/wgl/shared/stw_public.h
+++ /dev/null
@@ -1,73 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-#ifndef STW_PUBLIC_H
-#define STW_PUBLIC_H
-
-#include <windows.h>
-
-BOOL stw_copy_context( UINT_PTR hglrcSrc,
-                       UINT_PTR hglrcDst,
-                       UINT mask );
-
-UINT_PTR stw_create_layer_context( HDC hdc, 
-                                   int iLayerPlane );
-
-BOOL stw_share_lists( UINT_PTR hglrc1, UINT_PTR hglrc2 );
-
-BOOL stw_delete_context( UINT_PTR hglrc );
-
-BOOL
-stw_release_context( UINT_PTR dhglrc );
-
-UINT_PTR stw_get_current_context( void );
-
-HDC stw_get_current_dc( void );
-
-BOOL stw_make_current( HDC hdc, UINT_PTR hglrc );
-
-BOOL stw_swap_buffers( HDC hdc );
-
-BOOL
-stw_swap_layer_buffers( HDC hdc, UINT fuPlanes );
-
-PROC stw_get_proc_address( LPCSTR lpszProc );
-
-int stw_pixelformat_describe( HDC hdc,
-                              int iPixelFormat,
-                              UINT nBytes,
-                              LPPIXELFORMATDESCRIPTOR ppfd );
-
-int stw_pixelformat_get( HDC hdc );
-
-BOOL stw_pixelformat_set( HDC hdc,
-                          int iPixelFormat );
-
-int stw_pixelformat_choose( HDC hdc,
-                            CONST PIXELFORMATDESCRIPTOR *ppfd );
-
-#endif
diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.c b/src/gallium/state_trackers/wgl/stw_context.c
index 347f40aa06..f2f0264844 100644
--- a/src/gallium/state_trackers/wgl/icd/stw_icd.c
+++ b/src/gallium/state_trackers/wgl/stw_context.c
@@ -26,18 +26,49 @@
  **************************************************************************/
 
 #include <windows.h>
-#include <stdio.h>
 
-#include "GL/gl.h"
-
-#include "util/u_debug.h"
-#include "pipe/p_thread.h"
-
-#include "shared/stw_public.h"
-#include "icd/stw_icd.h"
+#include "main/mtypes.h"
+#include "main/context.h"
+#include "pipe/p_compiler.h"
+#include "pipe/p_context.h"
+#include "state_tracker/st_context.h"
+#include "state_tracker/st_public.h"
+
+#ifdef DEBUG
+#include "trace/tr_screen.h"
+#include "trace/tr_context.h"
+#endif
+
+#include "stw_icd.h"
+#include "stw_device.h"
+#include "stw_winsys.h"
+#include "stw_framebuffer.h"
+#include "stw_pixelformat.h"
+#include "stw_context.h"
+#include "stw_tls.h"
+
+
+static INLINE struct stw_context *
+stw_context(GLcontext *glctx)
+{
+   if(!glctx)
+      return NULL;
+   assert(glctx->DriverCtx);
+   return (struct stw_context *)glctx->DriverCtx;
+}
 
-#define DBG 0
+static INLINE struct stw_context *
+stw_current_context(void)
+{
+   /* We must check if multiple threads are being used or GET_CURRENT_CONTEXT 
+    * might return the current context of the thread first seen. */
+   _glapi_check_multithread();
 
+   {
+      GET_CURRENT_CONTEXT( glctx );
+      return stw_context(glctx);
+   }
+}
 
 BOOL APIENTRY
 DrvCopyContext(
@@ -45,24 +76,64 @@ DrvCopyContext(
    DHGLRC dhrcDest,
    UINT fuMask )
 {
-   return stw_copy_context(dhrcSource, dhrcDest, fuMask);
-}
+   struct stw_context *src;
+   struct stw_context *dst;
+   BOOL ret = FALSE;
 
+   pipe_mutex_lock( stw_dev->ctx_mutex );
+   
+   src = stw_lookup_context_locked( dhrcSource );
+   dst = stw_lookup_context_locked( dhrcDest );
+
+   if (src && dst) { 
+      /* FIXME */
+      assert(0);
+      (void) src;
+      (void) dst;
+      (void) fuMask;
+   }
 
-DHGLRC APIENTRY
-DrvCreateLayerContext(
-   HDC hdc,
-   INT iLayerPlane )
+   pipe_mutex_unlock( stw_dev->ctx_mutex );
+   
+   return ret;
+}
+
+BOOL APIENTRY
+DrvShareLists(
+   DHGLRC dhglrc1,
+   DHGLRC dhglrc2 )
 {
-   DHGLRC r;
+   struct stw_context *ctx1;
+   struct stw_context *ctx2;
+   BOOL ret = FALSE;
+
+   pipe_mutex_lock( stw_dev->ctx_mutex );
    
-   r = stw_create_layer_context( hdc, iLayerPlane );
+   ctx1 = stw_lookup_context_locked( dhglrc1 );
+   ctx2 = stw_lookup_context_locked( dhglrc2 );
+
+   if (ctx1 && ctx2 &&
+       ctx1->iPixelFormat == ctx2->iPixelFormat) { 
+      ret = _mesa_share_state(ctx2->st->ctx, ctx1->st->ctx);
+   }
+
+   pipe_mutex_unlock( stw_dev->ctx_mutex );
    
-   if (DBG)
-      debug_printf( "%s( %p, %i ) = %u\n",
-                    __FUNCTION__, hdc, iLayerPlane, r );
+   return ret;
+}
+
+static void
+stw_viewport(GLcontext * glctx, GLint x, GLint y,
+             GLsizei width, GLsizei height)
+{
+   struct stw_context *ctx = (struct stw_context *)glctx->DriverCtx;
+   struct stw_framebuffer *fb;
    
-   return r;
+   fb = stw_framebuffer_from_hdc( ctx->hdc );
+   if(fb) {
+      stw_framebuffer_update(fb);
+      stw_framebuffer_release(fb);
+   }
 }
 
 DHGLRC APIENTRY
@@ -72,114 +143,253 @@ DrvCreateContext(
    return DrvCreateLayerContext( hdc, 0 );
 }
 
-BOOL APIENTRY
-DrvDeleteContext(
-   DHGLRC dhglrc )
+DHGLRC APIENTRY
+DrvCreateLayerContext(
+   HDC hdc,
+   INT iLayerPlane )
 {
-   BOOL r;
+   int iPixelFormat;
+   const struct stw_pixelformat_info *pfi;
+   GLvisual visual;
+   struct stw_context *ctx = NULL;
+   struct pipe_screen *screen = NULL;
+   struct pipe_context *pipe = NULL;
    
-   r = stw_delete_context( dhglrc );
+   if(!stw_dev)
+      return 0;
    
-   if (DBG)
-      debug_printf( "%s( %u ) = %u\n",
-                    __FUNCTION__, dhglrc, r );
+   if (iLayerPlane != 0)
+      return 0;
+
+   iPixelFormat = GetPixelFormat(hdc);
+   if(!iPixelFormat)
+      return 0;
    
-   return r;
+   pfi = stw_pixelformat_get_info( iPixelFormat - 1 );
+   stw_pixelformat_visual(&visual, pfi);
+   
+   ctx = CALLOC_STRUCT( stw_context );
+   if (ctx == NULL)
+      goto no_ctx;
+
+   ctx->hdc = hdc;
+   ctx->iPixelFormat = iPixelFormat;
+
+   screen = stw_dev->screen;
+
+#ifdef DEBUG
+   /* Unwrap screen */
+   if(stw_dev->trace_running)
+      screen = trace_screen(screen)->screen;
+#endif
+
+   pipe = stw_dev->stw_winsys->create_context( screen );
+   if (pipe == NULL) 
+      goto no_pipe;
+
+#ifdef DEBUG
+   /* Wrap context */
+   if(stw_dev->trace_running)
+      pipe = trace_context_create(stw_dev->screen, pipe);
+#endif
+
+   /* pass to stw_flush_frontbuffer as context_private */
+   assert(!pipe->priv);
+   pipe->priv = hdc;
+
+   ctx->st = st_create_context( pipe, &visual, NULL );
+   if (ctx->st == NULL) 
+      goto no_st_ctx;
+
+   ctx->st->ctx->DriverCtx = ctx;
+   ctx->st->ctx->Driver.Viewport = stw_viewport;
+
+   pipe_mutex_lock( stw_dev->ctx_mutex );
+   ctx->dhglrc = handle_table_add(stw_dev->ctx_table, ctx);
+   pipe_mutex_unlock( stw_dev->ctx_mutex );
+   if (!ctx->dhglrc)
+      goto no_hglrc;
+
+   return ctx->dhglrc;
+
+no_hglrc:
+   st_destroy_context(ctx->st);
+   goto no_pipe; /* st_context_destroy already destroys pipe */
+no_st_ctx:
+   pipe->destroy( pipe );
+no_pipe:
+   FREE(ctx);
+no_ctx:
+   return 0;
 }
 
 BOOL APIENTRY
-DrvDescribeLayerPlane(
-   HDC hdc,
-   INT iPixelFormat,
-   INT iLayerPlane,
-   UINT nBytes,
-   LPLAYERPLANEDESCRIPTOR plpd )
+DrvDeleteContext(
+   DHGLRC dhglrc )
 {
-   if (DBG) 
-      debug_printf( "%s\n", __FUNCTION__ );
+   struct stw_context *ctx ;
+   BOOL ret = FALSE;
+   
+   if (!stw_dev)
+      return FALSE;
 
-   return FALSE;
-}
+   pipe_mutex_lock( stw_dev->ctx_mutex );
+   ctx = stw_lookup_context_locked(dhglrc);
+   handle_table_remove(stw_dev->ctx_table, dhglrc);
+   pipe_mutex_unlock( stw_dev->ctx_mutex );
 
-LONG APIENTRY
-DrvDescribePixelFormat(
-   HDC hdc,
-   INT iPixelFormat,
-   ULONG cjpfd,
-   PIXELFORMATDESCRIPTOR *ppfd )
-{
-   LONG r;
+   if (ctx) {
+      struct stw_context *curctx = stw_current_context();
+      
+      /* Unbind current if deleting current context. */
+      if (curctx == ctx)
+         st_make_current( NULL, NULL, NULL );
 
-   r = stw_pixelformat_describe( hdc, iPixelFormat, cjpfd, ppfd );
+      st_destroy_context(ctx->st);
+      FREE(ctx);
 
-   if (DBG)
-      debug_printf( "%s( %p, %d, %u, %p ) = %d\n",
-                    __FUNCTION__, hdc, iPixelFormat, cjpfd, ppfd, r );
+      ret = TRUE;
+   }
 
-   return r;
+   return ret;
 }
 
-int APIENTRY
-DrvGetLayerPaletteEntries(
-   HDC hdc,
-   INT iLayerPlane,
-   INT iStart,
-   INT cEntries,
-   COLORREF *pcr )
+BOOL APIENTRY
+DrvReleaseContext(
+   DHGLRC dhglrc )
 {
-   if (DBG)
-      debug_printf( "%s\n", __FUNCTION__ );
+   struct stw_context *ctx;
 
-   return 0;
-}
+   if (!stw_dev)
+      return FALSE;
 
-PROC APIENTRY
-DrvGetProcAddress(
-   LPCSTR lpszProc )
-{
-   PROC r;
+   pipe_mutex_lock( stw_dev->ctx_mutex );
+   ctx = stw_lookup_context_locked( dhglrc );
+   pipe_mutex_unlock( stw_dev->ctx_mutex );
 
-   r = stw_get_proc_address( lpszProc );
+   if (!ctx)
+      return FALSE;
+   
+   /* The expectation is that ctx is the same context which is
+    * current for this thread.  We should check that and return False
+    * if not the case.
+    */
+   if (ctx != stw_current_context())
+      return FALSE;
 
-   if (DBG)
-      debug_printf( "%s( \"%s\" ) = %p\n", __FUNCTION__, lpszProc, r );
+   if (stw_make_current( NULL, 0 ) == FALSE)
+      return FALSE;
 
-   return r;
+   return TRUE;
 }
 
-BOOL APIENTRY
-DrvRealizeLayerPalette(
-   HDC hdc,
-   INT iLayerPlane,
-   BOOL bRealize )
+
+DHGLRC
+stw_get_current_context( void )
 {
-   if (DBG)
-      debug_printf( "%s\n", __FUNCTION__ );
+   struct stw_context *ctx;
 
-   return FALSE;
+   ctx = stw_current_context();
+   if(!ctx)
+      return 0;
+   
+   return ctx->dhglrc;
 }
 
-BOOL APIENTRY
-DrvReleaseContext(
-   DHGLRC dhglrc )
+HDC
+stw_get_current_dc( void )
 {
-   return stw_release_context(dhglrc);
+   struct stw_context *ctx;
+
+   ctx = stw_current_context();
+   if(!ctx)
+      return NULL;
+   
+   return ctx->hdc;
 }
 
-void APIENTRY
-DrvSetCallbackProcs(
-   INT nProcs,
-   PROC *pProcs )
+BOOL
+stw_make_current(
+   HDC hdc,
+   DHGLRC dhglrc )
 {
-   if (DBG)
-      debug_printf( "%s( %d, %p )\n", __FUNCTION__, nProcs, pProcs );
+   struct stw_context *curctx = NULL;
+   struct stw_context *ctx = NULL;
+   struct stw_framebuffer *fb = NULL;
 
-   return;
-}
+   if (!stw_dev)
+      goto fail;
+
+   curctx = stw_current_context();
+   if (curctx != NULL) {
+      if (curctx->dhglrc != dhglrc)
+	 st_flush(curctx->st, PIPE_FLUSH_RENDER_CACHE, NULL);
+      
+      /* Return if already current. */
+      if (curctx->dhglrc == dhglrc && curctx->hdc == hdc) {
+         ctx = curctx;
+         fb = stw_framebuffer_from_hdc( hdc );
+         goto success;
+      }
+   }
+
+   if (hdc == NULL || dhglrc == 0) {
+      return st_make_current( NULL, NULL, NULL );
+   }
+
+   pipe_mutex_lock( stw_dev->ctx_mutex ); 
+   ctx = stw_lookup_context_locked( dhglrc );
+   pipe_mutex_unlock( stw_dev->ctx_mutex ); 
+   if(!ctx)
+      goto fail;
+
+   fb = stw_framebuffer_from_hdc( hdc );
+   if(!fb) { 
+      /* Applications should call SetPixelFormat before creating a context,
+       * but not all do, and the opengl32 runtime seems to use a default pixel
+       * format in some cases, so we must create a framebuffer for those here
+       */
+      int iPixelFormat = GetPixelFormat(hdc);
+      if(iPixelFormat)
+         fb = stw_framebuffer_create( hdc, iPixelFormat );
+      if(!fb) 
+         goto fail;
+   }
+   
+   if(fb->iPixelFormat != ctx->iPixelFormat)
+      goto fail;
+
+   /* Lazy allocation of the frame buffer */
+   if(!stw_framebuffer_allocate(fb))
+      goto fail;
+
+   /* Bind the new framebuffer */
+   ctx->hdc = hdc;
+   
+   /* pass to stw_flush_frontbuffer as context_private */
+   ctx->st->pipe->priv = hdc;
+   
+   if(!st_make_current( ctx->st, fb->stfb, fb->stfb ))
+      goto fail;
+
+success:
+   assert(fb);
+   if(fb) {
+      stw_framebuffer_update(fb);
+      stw_framebuffer_release(fb);
+   }
+   
+   return TRUE;
 
+fail:
+   if(fb)
+      stw_framebuffer_release(fb);
+   st_make_current( NULL, NULL, NULL );
+   return FALSE;
+}
 
 /**
- * Although WGL allows different dispatch entrypoints per context 
+ * Although WGL allows different dispatch entrypoints per context
  */
 static const GLCLTPROCTABLE cpt =
 {
@@ -524,7 +734,6 @@ static const GLCLTPROCTABLE cpt =
    }
 };
 
-
 PGLCLTPROCTABLE APIENTRY
 DrvSetContext(
    HDC hdc,
@@ -532,86 +741,9 @@ DrvSetContext(
    PFN_SETPROCTABLE pfnSetProcTable )
 {
    PGLCLTPROCTABLE r = (PGLCLTPROCTABLE)&cpt;
-   
+
    if (!stw_make_current( hdc, dhglrc ))
       r = NULL;
-      
-   if (DBG)
-      debug_printf( "%s( 0x%p, %u, 0x%p ) = %p\n", 
-                    __FUNCTION__, hdc, dhglrc, pfnSetProcTable, r );
-
-   return r;
-}
-
-int APIENTRY
-DrvSetLayerPaletteEntries(
-   HDC hdc,
-   INT iLayerPlane,
-   INT iStart,
-   INT cEntries,
-   CONST COLORREF *pcr )
-{
-   if (DBG)
-      debug_printf( "%s\n", __FUNCTION__ );
-
-   return 0;
-}
-
-BOOL APIENTRY
-DrvSetPixelFormat(
-   HDC hdc,
-   LONG iPixelFormat )
-{
-   BOOL r;
-
-   r = stw_pixelformat_set( hdc, iPixelFormat );
-
-   if (DBG)
-      debug_printf( "%s( %p, %d ) = %s\n", __FUNCTION__, hdc, iPixelFormat, r ? "TRUE" : "FALSE" );
 
    return r;
 }
-
-BOOL APIENTRY
-DrvShareLists(
-   DHGLRC dhglrc1,
-   DHGLRC dhglrc2 )
-{
-   if (DBG)
-      debug_printf( "%s\n", __FUNCTION__ );
-
-   return stw_share_lists(dhglrc1, dhglrc2);
-}
-
-BOOL APIENTRY
-DrvSwapBuffers(
-   HDC hdc )
-{
-   if (DBG)
-      debug_printf( "%s( %p )\n", __FUNCTION__, hdc );
-
-   return stw_swap_buffers( hdc );
-}
-
-BOOL APIENTRY
-DrvSwapLayerBuffers(
-   HDC hdc,
-   UINT fuPlanes )
-{
-   if (DBG)
-      debug_printf( "%s\n", __FUNCTION__ );
-
-   return stw_swap_layer_buffers( hdc, fuPlanes );
-}
-
-BOOL APIENTRY
-DrvValidateVersion(
-   ULONG ulVersion )
-{
-   if (DBG)
-      debug_printf( "%s( %u )\n", __FUNCTION__, ulVersion );
-
-   /* TODO: get the expected version from the winsys */
-   
-   return ulVersion == 1;
-}
diff --git a/src/gallium/state_trackers/wgl/shared/stw_context.h b/src/gallium/state_trackers/wgl/stw_context.h
index 166471de5e..256c27e21e 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_context.h
+++ b/src/gallium/state_trackers/wgl/stw_context.h
@@ -35,9 +35,15 @@ struct st_context;
 struct stw_context
 {
    struct st_context *st;
-   UINT_PTR hglrc;
+   DHGLRC dhglrc;
    int iPixelFormat;
    HDC hdc;
 };
 
+DHGLRC stw_get_current_context( void );
+
+HDC stw_get_current_dc( void );
+
+BOOL stw_make_current( HDC hdc, DHGLRC dhglrc );
+
 #endif /* STW_CONTEXT_H */
diff --git a/src/gallium/state_trackers/wgl/shared/stw_device.c b/src/gallium/state_trackers/wgl/stw_device.c
index 0b6954915a..985b8f0456 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_device.c
+++ b/src/gallium/state_trackers/wgl/stw_device.c
@@ -29,6 +29,7 @@
 
 #include "glapi/glthread.h"
 #include "util/u_debug.h"
+#include "util/u_math.h"
 #include "pipe/p_screen.h"
 #include "state_tracker/st_public.h"
 
@@ -37,12 +38,12 @@
 #include "trace/tr_texture.h"
 #endif
 
-#include "shared/stw_device.h"
-#include "shared/stw_winsys.h"
-#include "shared/stw_pixelformat.h"
-#include "shared/stw_public.h"
-#include "shared/stw_tls.h"
-#include "shared/stw_framebuffer.h"
+#include "stw_device.h"
+#include "stw_winsys.h"
+#include "stw_pixelformat.h"
+#include "stw_icd.h"
+#include "stw_tls.h"
+#include "stw_framebuffer.h"
 
 #ifdef WIN32_THREADS
 extern _glthread_Mutex OneTimeLock;
@@ -62,38 +63,28 @@ stw_flush_frontbuffer(struct pipe_screen *screen,
                      struct pipe_surface *surface,
                      void *context_private )
 {
-   const struct stw_winsys *stw_winsys = stw_dev->stw_winsys;
    HDC hdc = (HDC)context_private;
    struct stw_framebuffer *fb;
    
    fb = stw_framebuffer_from_hdc( hdc );
-   /* fb can be NULL if window was destroyed already */
-   if (fb) {
+   if (!fb) {
+      /* fb can be NULL if window was destroyed already */
+      return;
+   }
+
 #if DEBUG
-      {
-         struct pipe_surface *surface2;
-   
-         if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_FRONT_LEFT, &surface2 ))
-            assert(0);
-         else
-            assert(surface2 == surface);
-      }
-#endif
+   {
+      /* ensure that a random surface was not passed to us */
+      struct pipe_surface *surface2;
 
-#ifdef DEBUG
-      if(stw_dev->trace_running) {
-         screen = trace_screen(screen)->screen;
-         surface = trace_surface(surface)->surface;
-      }
-#endif
-   }
-   
-   stw_winsys->flush_frontbuffer(screen, surface, hdc);
-   
-   if(fb) {
-      stw_framebuffer_update(fb);
-      stw_framebuffer_release(fb);
+      if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_FRONT_LEFT, &surface2 ))
+         assert(0);
+      else
+         assert(surface2 == surface);
    }
+#endif
+
+   stw_framebuffer_present_locked(hdc, fb, ST_SURFACE_FRONT_LEFT);
 }
 
 
@@ -126,6 +117,9 @@ stw_init(const struct stw_winsys *stw_winsys)
    if(!screen)
       goto error1;
 
+   if(stw_winsys->get_adapter_luid)
+      stw_winsys->get_adapter_luid(screen, &stw_dev->AdapterLuid);
+
 #ifdef DEBUG
    stw_dev->screen = trace_screen_create(screen);
    stw_dev->trace_running = stw_dev->screen != screen ? TRUE : FALSE;
@@ -182,7 +176,7 @@ stw_cleanup(void)
       /* Ensure all contexts are destroyed */
       i = handle_table_get_first_handle(stw_dev->ctx_table);
       while (i) {
-         stw_delete_context(i);
+         DrvDeleteContext(i);
          i = handle_table_get_next_handle(stw_dev->ctx_table, i);
       }
       handle_table_destroy(stw_dev->ctx_table);
@@ -212,7 +206,7 @@ stw_cleanup(void)
 
 
 struct stw_context *
-stw_lookup_context_locked( UINT_PTR dhglrc )
+stw_lookup_context_locked( DHGLRC dhglrc )
 {
    if (dhglrc == 0)
       return NULL;
@@ -223,3 +217,28 @@ stw_lookup_context_locked( UINT_PTR dhglrc )
    return (struct stw_context *) handle_table_get(stw_dev->ctx_table, dhglrc);
 }
 
+
+void APIENTRY
+DrvSetCallbackProcs(
+   INT nProcs,
+   PROC *pProcs )
+{
+   size_t size;
+
+   if (stw_dev == NULL)
+      return;
+
+   size = MIN2(nProcs * sizeof *pProcs, sizeof stw_dev->callbacks);
+   memcpy(&stw_dev->callbacks, pProcs, size);
+
+   return;
+}
+
+
+BOOL APIENTRY
+DrvValidateVersion(
+   ULONG ulVersion )
+{
+   /* TODO: get the expected version from the winsys */
+   return ulVersion == 1;
+}
diff --git a/src/gallium/state_trackers/wgl/shared/stw_device.h b/src/gallium/state_trackers/wgl/stw_device.h
index e1bb9518dd..0bf3b0da82 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_device.h
+++ b/src/gallium/state_trackers/wgl/stw_device.h
@@ -29,11 +29,10 @@
 #define STW_DEVICE_H_
 
 
-#include <windows.h>
-
 #include "pipe/p_compiler.h"
 #include "pipe/p_thread.h"
 #include "util/u_handle_table.h"
+#include "stw_icd.h"
 #include "stw_pixelformat.h"
 
 
@@ -53,10 +52,14 @@ struct stw_device
    boolean trace_running;
 #endif
 
+   LUID AdapterLuid;
+
    struct stw_pixelformat_info pixelformats[STW_MAX_PIXELFORMATS];
    unsigned pixelformat_count;
    unsigned pixelformat_extended_count;
 
+   GLCALLBACKTABLE callbacks;
+
    pipe_mutex ctx_mutex;
    struct handle_table *ctx_table;
    
@@ -69,7 +72,7 @@ struct stw_device
 };
 
 struct stw_context *
-stw_lookup_context_locked( UINT_PTR hglrc );
+stw_lookup_context_locked( DHGLRC hglrc );
 
 extern struct stw_device *stw_dev;
 
diff --git a/src/gallium/state_trackers/wgl/shared/stw_extensionsstring.c b/src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c
index 62c859e1f9..62c859e1f9 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_extensionsstring.c
+++ b/src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c
diff --git a/src/gallium/state_trackers/wgl/shared/stw_extgallium.c b/src/gallium/state_trackers/wgl/stw_ext_gallium.c
index fc22737d7e..fb30ec5dba 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_extgallium.c
+++ b/src/gallium/state_trackers/wgl/stw_ext_gallium.c
@@ -27,9 +27,9 @@
 
 
 #include "pipe/p_screen.h"
-#include "stw_public.h"
 #include "stw_device.h"
 #include "stw_winsys.h"
+#include "stw_ext_gallium.h"
 
 #ifdef DEBUG
 #include "trace/tr_screen.h"
diff --git a/src/gallium/state_trackers/wgl/shared/stw_extgallium.h b/src/gallium/state_trackers/wgl/stw_ext_gallium.h
index cc35f2bb7f..cc35f2bb7f 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_extgallium.h
+++ b/src/gallium/state_trackers/wgl/stw_ext_gallium.h
diff --git a/src/gallium/state_trackers/wgl/shared/stw_arbpixelformat.c b/src/gallium/state_trackers/wgl/stw_ext_pixelformat.c
index 0e2d407699..8a9995aba8 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_arbpixelformat.c
+++ b/src/gallium/state_trackers/wgl/stw_ext_pixelformat.c
@@ -43,7 +43,6 @@
 
 #include "pipe/p_compiler.h"
 #include "util/u_memory.h"
-#include "stw_public.h"
 #include "stw_pixelformat.h"
 
 
diff --git a/src/gallium/state_trackers/wgl/shared/stw_extswapinterval.c b/src/gallium/state_trackers/wgl/stw_ext_swapinterval.c
index 9eac6a1d09..9eac6a1d09 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_extswapinterval.c
+++ b/src/gallium/state_trackers/wgl/stw_ext_swapinterval.c
diff --git a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.c b/src/gallium/state_trackers/wgl/stw_framebuffer.c
index b8956bb550..8a3e11b6b4 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.c
+++ b/src/gallium/state_trackers/wgl/stw_framebuffer.c
@@ -1,8 +1,8 @@
 /**************************************************************************
- * 
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ *
+ * Copyright 2008-2009 Vmware, Inc.
  * All Rights Reserved.
- * 
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -10,19 +10,19 @@
  * distribute, sub license, and/or sell copies of the Software, and to
  * permit persons to whom the Software is furnished to do so, subject to
  * the following conditions:
- * 
+ *
  * The above copyright notice and this permission notice (including the
  * next paragraph) shall be included in all copies or substantial portions
  * of the Software.
- * 
+ *
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
+ *
  **************************************************************************/
 
 #include <windows.h>
@@ -38,9 +38,9 @@
 #include "trace/tr_texture.h"
 #endif
 
+#include "stw_icd.h"
 #include "stw_framebuffer.h"
 #include "stw_device.h"
-#include "stw_public.h"
 #include "stw_winsys.h"
 #include "stw_tls.h"
 
@@ -83,6 +83,9 @@ stw_framebuffer_destroy_locked(
    *link = fb->next;
    fb->next = NULL;
 
+   if(fb->shared_surface)
+      stw_dev->stw_winsys->shared_surface_close(stw_dev->screen, fb->shared_surface);
+
    st_unreference_framebuffer(fb->stfb);
    
    pipe_mutex_unlock( fb->mutex );
@@ -106,13 +109,18 @@ static INLINE void
 stw_framebuffer_get_size( struct stw_framebuffer *fb )
 {
    unsigned width, height;
-   RECT rect;
+   RECT client_rect;
+   RECT window_rect;
+   POINT client_pos;
 
    assert(fb->hWnd);
    
-   GetClientRect( fb->hWnd, &rect );
-   width = rect.right - rect.left;
-   height = rect.bottom - rect.top;
+   /* Get the client area size. */
+   GetClientRect( fb->hWnd, &client_rect );
+   assert(client_rect.left == 0);
+   assert(client_rect.top == 0);
+   width = client_rect.right - client_rect.left;
+   height = client_rect.bottom - client_rect.top;
 
    if(width < 1)
       width = 1;
@@ -124,6 +132,31 @@ stw_framebuffer_get_size( struct stw_framebuffer *fb )
       fb->width = width; 
       fb->height = height; 
    }
+
+   client_pos.x = 0;
+   client_pos.y = 0;
+   ClientToScreen(fb->hWnd, &client_pos);
+
+   GetWindowRect(fb->hWnd, &window_rect);
+
+   fb->client_rect.left = client_pos.x - window_rect.left;
+   fb->client_rect.top =  client_pos.y - window_rect.top;
+   fb->client_rect.right = fb->client_rect.left + fb->width;
+   fb->client_rect.bottom = fb->client_rect.top + fb->height;
+
+#if 0
+   debug_printf("\n");
+   debug_printf("%s: client_position = (%i, %i)\n",
+                __FUNCTION__, client_pos.x, client_pos.y);
+   debug_printf("%s: window_rect = (%i, %i) - (%i, %i)\n",
+                __FUNCTION__,
+                window_rect.left, window_rect.top,
+                window_rect.right, window_rect.bottom);
+   debug_printf("%s: client_rect = (%i, %i) - (%i, %i)\n",
+                __FUNCTION__,
+                fb->client_rect.left, fb->client_rect.top,
+                fb->client_rect.right, fb->client_rect.bottom);
+#endif
 }
 
 
@@ -155,6 +188,7 @@ stw_call_window_proc(
        * can be masked out by the application. */
       LPWINDOWPOS lpWindowPos = (LPWINDOWPOS)pParams->lParam;
       if((lpWindowPos->flags & SWP_SHOWWINDOW) || 
+         !(lpWindowPos->flags & SWP_NOMOVE) ||
          !(lpWindowPos->flags & SWP_NOSIZE)) {
          fb = stw_framebuffer_from_hwnd( pParams->hwnd );
          if(fb) {
@@ -379,10 +413,10 @@ stw_framebuffer_from_hwnd(
 }
 
 
-BOOL
-stw_pixelformat_set(
+BOOL APIENTRY
+DrvSetPixelFormat(
    HDC hdc,
-   int iPixelFormat )
+   LONG iPixelFormat )
 {
    uint count;
    uint index;
@@ -435,35 +469,24 @@ stw_pixelformat_get(
 }
 
 
-BOOL
-stw_swap_buffers(
-   HDC hdc )
+BOOL APIENTRY
+DrvPresentBuffers(HDC hdc, PGLPRESENTBUFFERSDATA data)
 {
    struct stw_framebuffer *fb;
    struct pipe_screen *screen;
    struct pipe_surface *surface;
+   unsigned surface_index;
+   BOOL ret = FALSE;
 
    fb = stw_framebuffer_from_hdc( hdc );
    if (fb == NULL)
       return FALSE;
 
-   if (!(fb->pfi->pfd.dwFlags & PFD_DOUBLEBUFFER)) {
-      stw_framebuffer_release(fb);
-      return TRUE;
-   }
-
-   /* If we're swapping the buffer associated with the current context
-    * we have to flush any pending rendering commands first.
-    */
-   st_notify_swapbuffers( fb->stfb );
-
    screen = stw_dev->screen;
-   
-   if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_BACK_LEFT, &surface )) {
-      /* FIXME: this shouldn't happen, but does on glean */
-      stw_framebuffer_release(fb);
-      return FALSE;
-   }
+
+   surface_index = (unsigned)(uintptr_t)data->pPrivateData;
+   if(!st_get_framebuffer_surface( fb->stfb, surface_index, &surface ))
+      goto fail;
 
 #ifdef DEBUG
    if(stw_dev->trace_running) {
@@ -472,22 +495,127 @@ stw_swap_buffers(
    }
 #endif
 
-   stw_dev->stw_winsys->flush_frontbuffer( screen, surface, hdc );
-   
+   if(data->hSharedSurface != fb->hSharedSurface) {
+      if(fb->shared_surface) {
+         stw_dev->stw_winsys->shared_surface_close(screen, fb->shared_surface);
+         fb->shared_surface = NULL;
+      }
+
+      fb->hSharedSurface = data->hSharedSurface;
+
+      if(data->hSharedSurface &&
+         stw_dev->stw_winsys->shared_surface_open) {
+         fb->shared_surface = stw_dev->stw_winsys->shared_surface_open(screen, fb->hSharedSurface);
+      }
+   }
+
+   if(fb->shared_surface) {
+      stw_dev->stw_winsys->compose(screen,
+                                   surface,
+                                   fb->shared_surface,
+                                   &fb->client_rect,
+                                   data->PresentHistoryToken);
+   }
+   else {
+      stw_dev->stw_winsys->present( screen, surface, hdc );
+   }
+
+   ret = TRUE;
+
+fail:
+
    stw_framebuffer_update(fb);
+
    stw_framebuffer_release(fb);
-   
-   return TRUE;
+
+   return ret;
 }
 
 
+/**
+ * Queue a composition.
+ *
+ * It will drop the lock on success.
+ */
 BOOL
-stw_swap_layer_buffers(
+stw_framebuffer_present_locked(HDC hdc,
+                               struct stw_framebuffer *fb,
+                               unsigned surface_index)
+{
+   if(stw_dev->callbacks.wglCbPresentBuffers &&
+      stw_dev->stw_winsys->compose) {
+      GLCBPRESENTBUFFERSDATA data;
+
+      memset(&data, 0, sizeof data);
+      data.magic1 = 2;
+      data.magic2 = 0;
+      data.AdapterLuid = stw_dev->AdapterLuid;
+      data.rect = fb->client_rect;
+      data.pPrivateData = (void *)(uintptr_t)surface_index;
+
+      stw_framebuffer_release(fb);
+
+      return stw_dev->callbacks.wglCbPresentBuffers(hdc, &data);
+   }
+   else {
+      struct pipe_screen *screen = stw_dev->screen;
+      struct pipe_surface *surface;
+
+      if(!st_get_framebuffer_surface( fb->stfb, surface_index, &surface )) {
+         /* FIXME: this shouldn't happen, but does on glean */
+         stw_framebuffer_release(fb);
+         return FALSE;
+      }
+
+#ifdef DEBUG
+      if(stw_dev->trace_running) {
+         screen = trace_screen(screen)->screen;
+         surface = trace_surface(surface)->surface;
+      }
+#endif
+
+      stw_dev->stw_winsys->present( screen, surface, hdc );
+
+      stw_framebuffer_update(fb);
+
+      stw_framebuffer_release(fb);
+
+      return TRUE;
+   }
+}
+
+
+BOOL APIENTRY
+DrvSwapBuffers(
+   HDC hdc )
+{
+   struct stw_framebuffer *fb;
+
+   fb = stw_framebuffer_from_hdc( hdc );
+   if (fb == NULL)
+      return FALSE;
+
+   if (!(fb->pfi->pfd.dwFlags & PFD_DOUBLEBUFFER)) {
+      stw_framebuffer_release(fb);
+      return TRUE;
+   }
+
+   /* If we're swapping the buffer associated with the current context
+    * we have to flush any pending rendering commands first.
+    */
+   st_notify_swapbuffers( fb->stfb );
+
+   return stw_framebuffer_present_locked(hdc, fb, ST_SURFACE_BACK_LEFT);
+}
+
+
+BOOL APIENTRY
+DrvSwapLayerBuffers(
    HDC hdc,
    UINT fuPlanes )
 {
    if(fuPlanes & WGL_SWAP_MAIN_PLANE)
-      return stw_swap_buffers(hdc);
+      return DrvSwapBuffers(hdc);
 
    return FALSE;
 }
diff --git a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.h b/src/gallium/state_trackers/wgl/stw_framebuffer.h
index 13d29f37e4..5afbe74908 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.h
+++ b/src/gallium/state_trackers/wgl/stw_framebuffer.h
@@ -73,9 +73,20 @@ struct stw_framebuffer
    
    /* FIXME: Make this work for multiple contexts bound to the same framebuffer */
    boolean must_resize;
+
    unsigned width;
    unsigned height;
    
+   /**
+    * Client area rectangle, relative to the window upper-left corner.
+    *
+    * @sa GLCBPRESENTBUFFERSDATA::rect.
+    */
+   RECT client_rect;
+
+   HANDLE hSharedSurface;
+   struct stw_shared_surface *shared_surface;
+
    /** 
     * This is protected by stw_device::fb_mutex, not the mutex above.
     * 
@@ -126,6 +137,11 @@ BOOL
 stw_framebuffer_allocate(
    struct stw_framebuffer *fb );
 
+BOOL
+stw_framebuffer_present_locked(HDC hdc,
+                               struct stw_framebuffer *fb,
+                               unsigned surface_index);
+
 void
 stw_framebuffer_update(
    struct stw_framebuffer *fb);
diff --git a/src/gallium/state_trackers/wgl/shared/stw_getprocaddress.c b/src/gallium/state_trackers/wgl/stw_getprocaddress.c
index 879ced925a..8875dc22f3 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_getprocaddress.c
+++ b/src/gallium/state_trackers/wgl/stw_getprocaddress.c
@@ -33,8 +33,7 @@
 #include <GL/wglext.h>
 
 #include "glapi/glapi.h"
-#include "stw_public.h"
-#include "stw_extgallium.h"
+#include "stw_ext_gallium.h"
 
 struct stw_extension_entry
 {
@@ -68,8 +67,8 @@ static const struct stw_extension_entry stw_extension_entries[] = {
    { NULL, NULL }
 };
 
-PROC
-stw_get_proc_address(
+PROC APIENTRY
+DrvGetProcAddress(
    LPCSTR lpszProc )
 {
    const struct stw_extension_entry *entry;
diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.h b/src/gallium/state_trackers/wgl/stw_icd.h
index cbc1a66548..02eb543fef 100644
--- a/src/gallium/state_trackers/wgl/icd/stw_icd.h
+++ b/src/gallium/state_trackers/wgl/stw_icd.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  *
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2008-2009 Vmware, Inc.
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -18,7 +18,7 @@
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
@@ -388,6 +388,113 @@ typedef struct _GLCLTPROCTABLE
 
 typedef VOID (APIENTRY * PFN_SETPROCTABLE)(PGLCLTPROCTABLE);
 
+/**
+ * Presentation data passed to opengl32!wglCbPresentBuffers.
+ *
+ * Pure software drivers don't need to worry about this -- if they stick to the
+ * GDI API then will integrate with the Desktop Window Manager (DWM) without
+ * problems. Hardware drivers, however, cannot present directly to the primary
+ * surface while the DWM is active, as DWM gets exclusive access to the primary
+ * surface.
+ *
+ * Proper DWM integration requires:
+ * - advertise the PFD_SUPPORT_COMPOSITION flag
+ * - redirect glFlush/glfinish/wglSwapBuffers into a surface shared with the
+ * DWM process.
+ *
+ * @sa http://www.opengl.org/pipeline/article/vol003_7/
+ * @sa http://blogs.msdn.com/greg_schechter/archive/2006/05/02/588934.aspx
+ */
+typedef struct _GLCBPRESENTBUFFERSDATA
+{
+   /**
+    * wglCbPresentBuffers enforces this to be 2.
+    */
+   DWORD magic1;
+
+   /**
+    * wglCbPresentBuffers enforces to be 0 or 1, but it is most commonly
+    * set to 0.
+    */
+   DWORD magic2;
+
+   /**
+    * Locally unique identifier (LUID) of the graphics adapter.
+    *
+    * This should contain the value returned by D3DKMTOpenAdapterFromHdc. It
+    * is passed to dwmapi!DwmpDxGetWindowSharedSurface in order to obtain
+    * the shared surface handle for the bound drawable (window).
+    *
+    * @sa http://msdn.microsoft.com/en-us/library/ms799177.aspx
+    */
+   LUID AdapterLuid;
+
+   /**
+    * This is passed unmodified to DrvPresentBuffers
+    */
+   LPVOID pPrivateData;
+
+   /**
+    * Client area rectangle to update, relative to the window upper-left corner.
+    */
+   RECT rect;
+} GLCBPRESENTBUFFERSDATA, *PGLCBPRESENTBUFFERSDATA;
+
+/**
+ * Callbacks supplied to DrvSetCallbackProcs by the OpenGL runtime.
+ *
+ * Pointers to several callback functions in opengl32.dll.
+ */
+typedef struct _GLCALLBACKTABLE
+{
+   /** Unused */
+   PROC wglCbSetCurrentValue;
+
+   /** Unused */
+   PROC wglCbGetCurrentValue;
+
+   /** Unused */
+   PROC wglCbGetDhglrc;
+
+   /** Unused */
+   PROC wglCbGetDdHandle;
+
+   /**
+    * Queue a present composition.
+    *
+    * Makes the runtime call DrvPresentBuffers with the composition information.
+    */
+   BOOL (APIENTRY *wglCbPresentBuffers)(HDC hdc, PGLCBPRESENTBUFFERSDATA data);
+
+} GLCALLBACKTABLE;
+
+typedef struct _GLPRESENTBUFFERSDATA
+{
+   /**
+    * The shared surface handle.
+    *
+    * Return by dwmapi!DwmpDxGetWindowSharedSurface.
+    *
+    * @sa http://channel9.msdn.com/forums/TechOff/251261-Help-Getting-the-shared-window-texture-out-of-DWM-/
+    */
+   HANDLE hSharedSurface;
+
+   LUID AdapterLuid;
+
+   /**
+    * Present history token.
+    *
+    * This is returned by dwmapi!DwmpDxGetWindowSharedSurface and
+    * should be passed to D3DKMTRender in D3DKMT_RENDER::PresentHistoryToken.
+    *
+    * @sa http://msdn.microsoft.com/en-us/library/ms799176.aspx
+    */
+   ULONGLONG PresentHistoryToken;
+
+   /** Same as GLCBPRESENTBUFFERSDATA::pPrivateData */
+   LPVOID pPrivateData;
+} GLPRESENTBUFFERSDATA, *PGLPRESENTBUFFERSDATA;
+
 BOOL APIENTRY
 DrvCopyContext(
    DHGLRC dhrcSource,
@@ -435,6 +542,9 @@ DrvGetProcAddress(
    LPCSTR lpszProc );
 
 BOOL APIENTRY
+DrvPresentBuffers(HDC hdc, PGLPRESENTBUFFERSDATA data);
+
+BOOL APIENTRY
 DrvRealizeLayerPalette(
    HDC hdc,
    INT iLayerPlane,
diff --git a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.c b/src/gallium/state_trackers/wgl/stw_pixelformat.c
index c296744838..7abe5d9f7f 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.c
+++ b/src/gallium/state_trackers/wgl/stw_pixelformat.c
@@ -34,9 +34,9 @@
 
 #include "util/u_debug.h"
 
+#include "stw_icd.h"
 #include "stw_device.h"
 #include "stw_pixelformat.h"
-#include "stw_public.h"
 #include "stw_tls.h"
 
 
@@ -154,8 +154,11 @@ stw_pixelformat_add(
    pfi->pfd.dwFlags = PFD_SUPPORT_OPENGL;
    
    /* TODO: also support non-native pixel formats */
-   pfi->pfd.dwFlags |= PFD_DRAW_TO_WINDOW ;
-   
+   pfi->pfd.dwFlags |= PFD_DRAW_TO_WINDOW;
+
+   /* See http://www.opengl.org/pipeline/article/vol003_7/ */
+   pfi->pfd.dwFlags |= PFD_SUPPORT_COMPOSITION;
+
    if (doublebuffer)
       pfi->pfd.dwFlags |= PFD_DOUBLEBUFFER | PFD_SWAP_COPY;
    
@@ -288,12 +291,12 @@ stw_pixelformat_visual(GLvisual *visual,
 }
 
 
-int
-stw_pixelformat_describe(
+LONG APIENTRY
+DrvDescribePixelFormat(
    HDC hdc,
-   int iPixelFormat,
-   UINT nBytes,
-   LPPIXELFORMATDESCRIPTOR ppfd )
+   INT iPixelFormat,
+   ULONG cjpfd,
+   PIXELFORMATDESCRIPTOR *ppfd )
 {
    uint count;
    uint index;
@@ -306,7 +309,7 @@ stw_pixelformat_describe(
 
    if (ppfd == NULL)
       return count;
-   if (index >= count || nBytes != sizeof( PIXELFORMATDESCRIPTOR ))
+   if (index >= count || cjpfd != sizeof( PIXELFORMATDESCRIPTOR ))
       return 0;
 
    pfi = stw_pixelformat_get_info( index );
@@ -316,6 +319,52 @@ stw_pixelformat_describe(
    return count;
 }
 
+BOOL APIENTRY
+DrvDescribeLayerPlane(
+   HDC hdc,
+   INT iPixelFormat,
+   INT iLayerPlane,
+   UINT nBytes,
+   LPLAYERPLANEDESCRIPTOR plpd )
+{
+   assert(0);
+   return FALSE;
+}
+
+int APIENTRY
+DrvGetLayerPaletteEntries(
+   HDC hdc,
+   INT iLayerPlane,
+   INT iStart,
+   INT cEntries,
+   COLORREF *pcr )
+{
+   assert(0);
+   return 0;
+}
+
+int APIENTRY
+DrvSetLayerPaletteEntries(
+   HDC hdc,
+   INT iLayerPlane,
+   INT iStart,
+   INT cEntries,
+   CONST COLORREF *pcr )
+{
+   assert(0);
+   return 0;
+}
+
+BOOL APIENTRY
+DrvRealizeLayerPalette(
+   HDC hdc,
+   INT iLayerPlane,
+   BOOL bRealize )
+{
+   assert(0);
+   return FALSE;
+}
+
 /* Only used by the wgl code, but have it here to avoid exporting the
  * pixelformat.h functionality.
  */
diff --git a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.h b/src/gallium/state_trackers/wgl/stw_pixelformat.h
index bec429231b..3a690b35ba 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.h
+++ b/src/gallium/state_trackers/wgl/stw_pixelformat.h
@@ -30,6 +30,10 @@
 
 #include <windows.h>
 
+#ifndef PFD_SUPPORT_COMPOSITION
+#define PFD_SUPPORT_COMPOSITION 0x00008000
+#endif
+
 #include "main/mtypes.h"
 
 #include "pipe/p_compiler.h"
@@ -62,4 +66,11 @@ void
 stw_pixelformat_visual(GLvisual *visual, 
                        const struct stw_pixelformat_info *pfi );
 
+int
+stw_pixelformat_choose( HDC hdc,
+                        CONST PIXELFORMATDESCRIPTOR *ppfd );
+
+int
+stw_pixelformat_get(HDC hdc);
+
 #endif /* STW_PIXELFORMAT_H */
diff --git a/src/gallium/state_trackers/wgl/shared/stw_tls.c b/src/gallium/state_trackers/wgl/stw_tls.c
index 4bd6a9289c..4bd6a9289c 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_tls.c
+++ b/src/gallium/state_trackers/wgl/stw_tls.c
diff --git a/src/gallium/state_trackers/wgl/shared/stw_tls.h b/src/gallium/state_trackers/wgl/stw_tls.h
index fbf8b1cbee..fbf8b1cbee 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_tls.h
+++ b/src/gallium/state_trackers/wgl/stw_tls.h
diff --git a/src/gallium/state_trackers/wgl/wgl/stw_wgl.c b/src/gallium/state_trackers/wgl/stw_wgl.c
index a131292f7a..bb199fdd25 100644
--- a/src/gallium/state_trackers/wgl/wgl/stw_wgl.c
+++ b/src/gallium/state_trackers/wgl/stw_wgl.c
@@ -28,7 +28,9 @@
 #include <windows.h>
 
 #include "util/u_debug.h"
-#include "shared/stw_public.h"
+#include "stw_icd.h"
+#include "stw_context.h"
+#include "stw_pixelformat.h"
 #include "stw_wgl.h"
 
 
@@ -38,16 +40,16 @@ wglCopyContext(
    HGLRC hglrcDst,
    UINT mask )
 {
-   return stw_copy_context( (UINT_PTR)hglrcSrc, 
-                            (UINT_PTR)hglrcDst, 
-                            mask );
+   return DrvCopyContext( (DHGLRC)(UINT_PTR)hglrcSrc,
+                          (DHGLRC)(UINT_PTR)hglrcDst,
+                          mask );
 }
 
 WINGDIAPI HGLRC APIENTRY
 wglCreateContext(
    HDC hdc )
 {
-   return wglCreateLayerContext(hdc, 0);
+   return (HGLRC) DrvCreateContext(hdc);
 }
 
 WINGDIAPI HGLRC APIENTRY
@@ -55,21 +57,21 @@ wglCreateLayerContext(
    HDC hdc,
    int iLayerPlane )
 {
-   return (HGLRC) stw_create_layer_context( hdc, iLayerPlane );
+   return (HGLRC) DrvCreateLayerContext( hdc, iLayerPlane );
 }
 
 WINGDIAPI BOOL APIENTRY
 wglDeleteContext(
    HGLRC hglrc )
 {
-   return stw_delete_context( (UINT_PTR)hglrc );
+   return DrvDeleteContext((DHGLRC)(UINT_PTR)hglrc );
 }
 
 
 WINGDIAPI HGLRC APIENTRY
 wglGetCurrentContext( VOID )
 {
-   return (HGLRC)stw_get_current_context();
+   return (HGLRC)(UINT_PTR)stw_get_current_context();
 }
 
 WINGDIAPI HDC APIENTRY
@@ -83,7 +85,7 @@ wglMakeCurrent(
    HDC hdc,
    HGLRC hglrc )
 {
-   return stw_make_current( hdc, (UINT_PTR)hglrc );
+   return DrvSetContext( hdc, (DHGLRC)(UINT_PTR)hglrc, NULL ) ? TRUE : FALSE;
 }
 
 
@@ -91,7 +93,7 @@ WINGDIAPI BOOL APIENTRY
 wglSwapBuffers(
    HDC hdc )
 {
-   return stw_swap_buffers( hdc );
+   return DrvSwapBuffers( hdc );
 }
 
 
@@ -100,14 +102,14 @@ wglSwapLayerBuffers(
    HDC hdc,
    UINT fuPlanes )
 {
-   return stw_swap_layer_buffers( hdc, fuPlanes );
+   return DrvSwapLayerBuffers( hdc, fuPlanes );
 }
 
 WINGDIAPI PROC APIENTRY
 wglGetProcAddress(
     LPCSTR lpszProc )
 {
-   return stw_get_proc_address( lpszProc );
+   return DrvGetProcAddress( lpszProc );
 }
 
 
@@ -141,7 +143,7 @@ wglDescribePixelFormat(
    UINT nBytes,
    LPPIXELFORMATDESCRIPTOR ppfd )
 {
-   return stw_pixelformat_describe( hdc, iPixelFormat, nBytes, ppfd );
+   return DrvDescribePixelFormat( hdc, iPixelFormat, nBytes, ppfd );
 }
 
 WINGDIAPI int APIENTRY
@@ -160,7 +162,7 @@ wglSetPixelFormat(
    if (ppfd->nSize != sizeof( PIXELFORMATDESCRIPTOR ))
       return FALSE;
 
-   return stw_pixelformat_set( hdc, iPixelFormat );
+   return DrvSetPixelFormat( hdc, iPixelFormat );
 }
 
 
@@ -186,7 +188,8 @@ wglShareLists(
    HGLRC hglrc1,
    HGLRC hglrc2 )
 {
-   return stw_share_lists( (UINT_PTR)hglrc1, (UINT_PTR)hglrc2);;
+   return DrvShareLists((DHGLRC)(UINT_PTR)hglrc1,
+                        (DHGLRC)(UINT_PTR)hglrc2);
 }
 
 WINGDIAPI BOOL APIENTRY
@@ -264,15 +267,7 @@ wglDescribeLayerPlane(
    UINT nBytes,
    LPLAYERPLANEDESCRIPTOR plpd )
 {
-   (void) hdc;
-   (void) iPixelFormat;
-   (void) iLayerPlane;
-   (void) nBytes;
-   (void) plpd;
-
-   assert( 0 );
-
-   return FALSE;
+   return DrvDescribeLayerPlane(hdc, iPixelFormat, iLayerPlane, nBytes, plpd);
 }
 
 WINGDIAPI int APIENTRY
@@ -283,15 +278,7 @@ wglSetLayerPaletteEntries(
    int cEntries,
    CONST COLORREF *pcr )
 {
-   (void) hdc;
-   (void) iLayerPlane;
-   (void) iStart;
-   (void) cEntries;
-   (void) pcr;
-
-   assert( 0 );
-
-   return 0;
+   return DrvSetLayerPaletteEntries(hdc, iLayerPlane, iStart, cEntries, pcr);
 }
 
 WINGDIAPI int APIENTRY
@@ -302,15 +289,7 @@ wglGetLayerPaletteEntries(
    int cEntries,
    COLORREF *pcr )
 {
-   (void) hdc;
-   (void) iLayerPlane;
-   (void) iStart;
-   (void) cEntries;
-   (void) pcr;
-
-   assert( 0 );
-
-   return 0;
+   return DrvGetLayerPaletteEntries(hdc, iLayerPlane, iStart, cEntries, pcr);
 }
 
 WINGDIAPI BOOL APIENTRY
diff --git a/src/gallium/state_trackers/wgl/wgl/stw_wgl.h b/src/gallium/state_trackers/wgl/stw_wgl.h
index a98179944a..a98179944a 100644
--- a/src/gallium/state_trackers/wgl/wgl/stw_wgl.h
+++ b/src/gallium/state_trackers/wgl/stw_wgl.h
diff --git a/src/gallium/state_trackers/wgl/stw_winsys.h b/src/gallium/state_trackers/wgl/stw_winsys.h
new file mode 100644
index 0000000000..1ead47d6e6
--- /dev/null
+++ b/src/gallium/state_trackers/wgl/stw_winsys.h
@@ -0,0 +1,109 @@
+/**************************************************************************
+ *
+ * Copyright 2008-2009 Vmware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#ifndef STW_WINSYS_H
+#define STW_WINSYS_H
+
+#include <windows.h> /* for HDC */
+
+#include "pipe/p_compiler.h"
+
+struct pipe_screen;
+struct pipe_context;
+struct pipe_surface;
+
+struct stw_shared_surface;
+
+struct stw_winsys
+{
+   struct pipe_screen *
+   (*create_screen)( void );
+
+   struct pipe_context *
+   (*create_context)( struct pipe_screen *screen );
+
+   /**
+    * Present the color buffer to the window associated with the device context.
+    */
+   void
+   (*present)( struct pipe_screen *screen,
+               struct pipe_surface *surf,
+               HDC hDC );
+
+   /**
+    * Locally unique identifier (LUID) of the graphics adapter.
+    *
+    * @sa GLCBPRESENTBUFFERSDATA::AdapterLuid;
+    */
+   boolean
+   (*get_adapter_luid)( struct pipe_screen *screen,
+                        LUID *pAdapterLuid );
+
+   /**
+    * Open a shared surface (optional).
+    *
+    * @sa GLCBPRESENTBUFFERSDATA::hSharedSurface;
+    */
+   struct stw_shared_surface *
+   (*shared_surface_open)(struct pipe_screen *screen,
+                          HANDLE hSharedSurface);
+
+   /**
+    * Open a shared surface (optional).
+    */
+   void
+   (*shared_surface_close)(struct pipe_screen *screen,
+                           struct stw_shared_surface *surface);
+
+   /**
+    * Compose into a shared (optional).
+    *
+    * Blit the color buffer into a shared surface.
+    *
+    * @sa GLPRESENTBUFFERSDATA::PresentHistoryToken.
+    */
+   void
+   (*compose)( struct pipe_screen *screen,
+               struct pipe_surface *src,
+               struct stw_shared_surface *dest,
+               LPCRECT pRect,
+               ULONGLONG PresentHistoryToken );
+};
+
+boolean
+stw_init(const struct stw_winsys *stw_winsys);
+
+boolean
+stw_init_thread(void);
+
+void
+stw_cleanup_thread(void);
+
+void
+stw_cleanup(void);
+
+#endif /* STW_WINSYS_H */
diff --git a/src/gallium/state_trackers/xorg/Makefile b/src/gallium/state_trackers/xorg/Makefile
index 27a1990724..22c107370e 100644
--- a/src/gallium/state_trackers/xorg/Makefile
+++ b/src/gallium/state_trackers/xorg/Makefile
@@ -5,6 +5,8 @@ LIBNAME = xorgtracker
 
 LIBRARY_INCLUDES = \
 	-DHAVE_CONFIG_H \
+	$(shell pkg-config xextproto --atleast-version=7.0.99.1 \
+				&& echo "-DHAVE_XEXTPROTO_71") \
 	$(shell pkg-config --cflags-only-I pixman-1 xorg-server libdrm xproto) \
 	-I$(TOP)/src/gallium/include \
 	-I$(TOP)/src/gallium/auxiliary \
diff --git a/src/gallium/state_trackers/xorg/SConscript b/src/gallium/state_trackers/xorg/SConscript
index 65f55ea378..6165bae7a4 100644
--- a/src/gallium/state_trackers/xorg/SConscript
+++ b/src/gallium/state_trackers/xorg/SConscript
@@ -22,6 +22,7 @@ if 'xorg' in env['statetrackers']:
 		'xorg_exa.c',
 		'xorg_exa_tgsi.c',
 		'xorg_output.c',
+		'xorg_xv.c',
 		]
     )
     Export('st_xorg')
diff --git a/src/gallium/state_trackers/xorg/xorg_composite.c b/src/gallium/state_trackers/xorg/xorg_composite.c
index c708ac3170..e039bb12b6 100644
--- a/src/gallium/state_trackers/xorg/xorg_composite.c
+++ b/src/gallium/state_trackers/xorg/xorg_composite.c
@@ -1,97 +1,115 @@
 #include "xorg_composite.h"
 
+#include "xorg_renderer.h"
 #include "xorg_exa_tgsi.h"
 
 #include "cso_cache/cso_context.h"
 #include "util/u_draw_quad.h"
+#include "util/u_math.h"
 
 #include "pipe/p_inlines.h"
 
+/*XXX also in Xrender.h but the including it here breaks compilition */
+#define XFixedToDouble(f)    (((double) (f)) / 65536.)
+
 struct xorg_composite_blend {
-   int op:8;
+   int op : 8;
 
-   unsigned rgb_src_factor:5;    /**< PIPE_BLENDFACTOR_x */
-   unsigned rgb_dst_factor:5;    /**< PIPE_BLENDFACTOR_x */
+   unsigned alpha_dst : 4;
+   unsigned alpha_src : 4;
 
-   unsigned alpha_src_factor:5;  /**< PIPE_BLENDFACTOR_x */
-   unsigned alpha_dst_factor:5;  /**< PIPE_BLENDFACTOR_x */
+   unsigned rgb_src : 8;    /**< PIPE_BLENDFACTOR_x */
+   unsigned rgb_dst : 8;    /**< PIPE_BLENDFACTOR_x */
 };
 
 #define BLEND_OP_OVER 3
 static const struct xorg_composite_blend xorg_blends[] = {
    { PictOpClear,
-     PIPE_BLENDFACTOR_CONST_COLOR, PIPE_BLENDFACTOR_CONST_ALPHA,
-     PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_ZERO },
-
+     0, 0, PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_ZERO},
    { PictOpSrc,
-     PIPE_BLENDFACTOR_ONE, PIPE_BLENDFACTOR_ONE,
-     PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_ZERO },
-
+     0, 0, PIPE_BLENDFACTOR_ONE, PIPE_BLENDFACTOR_ZERO},
    { PictOpDst,
-     PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_ZERO,
-     PIPE_BLENDFACTOR_ONE, PIPE_BLENDFACTOR_ONE },
-
+     0, 0, PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_ONE},
    { PictOpOver,
-     PIPE_BLENDFACTOR_SRC_ALPHA, PIPE_BLENDFACTOR_ONE,
-     PIPE_BLENDFACTOR_INV_SRC_ALPHA, PIPE_BLENDFACTOR_INV_SRC_ALPHA },
-
+     0, 1, PIPE_BLENDFACTOR_ONE, PIPE_BLENDFACTOR_INV_SRC_ALPHA},
    { PictOpOverReverse,
-     PIPE_BLENDFACTOR_SRC_ALPHA, PIPE_BLENDFACTOR_ONE,
-     PIPE_BLENDFACTOR_INV_SRC_ALPHA, PIPE_BLENDFACTOR_INV_SRC_ALPHA },
+     1, 0, PIPE_BLENDFACTOR_INV_DST_ALPHA, PIPE_BLENDFACTOR_ONE},
+   { PictOpIn,
+     1, 0, PIPE_BLENDFACTOR_DST_ALPHA, PIPE_BLENDFACTOR_ZERO},
+   { PictOpInReverse,
+     0, 1, PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_SRC_ALPHA},
+   { PictOpOut,
+     1, 0, PIPE_BLENDFACTOR_INV_DST_ALPHA, PIPE_BLENDFACTOR_ZERO},
+   { PictOpOutReverse,
+     0, 1, PIPE_BLENDFACTOR_ZERO, PIPE_BLENDFACTOR_INV_SRC_ALPHA},
+   { PictOpAtop,
+     1, 1, PIPE_BLENDFACTOR_DST_ALPHA, PIPE_BLENDFACTOR_INV_SRC_ALPHA},
+   { PictOpAtopReverse,
+     1, 1, PIPE_BLENDFACTOR_INV_DST_ALPHA, PIPE_BLENDFACTOR_SRC_ALPHA},
+   { PictOpXor,
+     1, 1, PIPE_BLENDFACTOR_INV_DST_ALPHA, PIPE_BLENDFACTOR_INV_SRC_ALPHA},
+   { PictOpAdd,
+     0, 0, PIPE_BLENDFACTOR_ONE, PIPE_BLENDFACTOR_ONE},
 };
 
+
 static INLINE void
-pixel_to_float4(PictFormatPtr format,
-                CARD32 pixel, float *color)
+pixel_to_float4(Pixel pixel, float *color)
 {
    CARD32	    r, g, b, a;
 
-   debug_assert(format->type == PictTypeDirect);
-
-   r = (pixel >> format->direct.red) & format->direct.redMask;
-   g = (pixel >> format->direct.green) & format->direct.greenMask;
-   b = (pixel >> format->direct.blue) & format->direct.blueMask;
-   a = (pixel >> format->direct.alpha) & format->direct.alphaMask;
-   color[0] = ((float)r) / ((float)format->direct.redMask);
-   color[1] = ((float)g) / ((float)format->direct.greenMask);
-   color[2] = ((float)b) / ((float)format->direct.blueMask);
-   color[3] = ((float)a) / ((float)format->direct.alphaMask);
+   a = (pixel >> 24) & 0xff;
+   r = (pixel >> 16) & 0xff;
+   g = (pixel >>  8) & 0xff;
+   b = (pixel >>  0) & 0xff;
+   color[0] = ((float)r) / 255.;
+   color[1] = ((float)g) / 255.;
+   color[2] = ((float)b) / 255.;
+   color[3] = ((float)a) / 255.;
 }
 
-struct acceleration_info {
-   int op : 16;
-   int with_mask : 1;
-   int component_alpha : 1;
-};
-static const struct acceleration_info accelerated_ops[] = {
-   {PictOpClear,       1, 0},
-   {PictOpSrc,         1, 0},
-   {PictOpDst,         1, 0},
-   {PictOpOver,        1, 0},
-   {PictOpOverReverse, 1, 0},
-   {PictOpIn,          1, 0},
-   {PictOpInReverse,   1, 0},
-   {PictOpOut,         1, 0},
-   {PictOpOutReverse,  1, 0},
-   {PictOpAtop,        1, 0},
-   {PictOpAtopReverse, 1, 0},
-   {PictOpXor,         1, 0},
-   {PictOpAdd,         1, 0},
-   {PictOpSaturate,    1, 0},
-};
-
-static struct xorg_composite_blend
-blend_for_op(int op)
+static boolean
+blend_for_op(struct xorg_composite_blend *blend,
+             int op, PicturePtr pSrcPicture, PicturePtr pMaskPicture,
+             PicturePtr pDstPicture)
 {
    const int num_blends =
       sizeof(xorg_blends)/sizeof(struct xorg_composite_blend);
    int i;
+   boolean supported = FALSE;
+
+   /* our default in case something goes wrong */
+   *blend = xorg_blends[BLEND_OP_OVER];
 
    for (i = 0; i < num_blends; ++i) {
-      if (xorg_blends[i].op == op)
-         return xorg_blends[i];
+      if (xorg_blends[i].op == op) {
+         *blend = xorg_blends[i];
+         supported = TRUE;
+      }
+   }
+
+   /* If there's no dst alpha channel, adjust the blend op so that we'll treat
+    * it as always 1. */
+   if (pDstPicture &&
+       PICT_FORMAT_A(pDstPicture->format) == 0 && blend->alpha_dst) {
+      if (blend->rgb_src == PIPE_BLENDFACTOR_DST_ALPHA)
+         blend->rgb_src = PIPE_BLENDFACTOR_ONE;
+      else if (blend->rgb_src == PIPE_BLENDFACTOR_INV_DST_ALPHA)
+         blend->rgb_src = PIPE_BLENDFACTOR_ZERO;
+   }
+
+   /* If the source alpha is being used, then we should only be in a case where
+    * the source blend factor is 0, and the source blend value is the mask
+    * channels multiplied by the source picture's alpha. */
+   if (pMaskPicture && pMaskPicture->componentAlpha &&
+       PICT_FORMAT_RGB(pMaskPicture->format) && blend->alpha_src) {
+      if (blend->rgb_dst == PIPE_BLENDFACTOR_SRC_ALPHA) {
+         blend->rgb_dst = PIPE_BLENDFACTOR_SRC_COLOR;
+      } else if (blend->rgb_dst == PIPE_BLENDFACTOR_INV_SRC_ALPHA) {
+         blend->rgb_dst = PIPE_BLENDFACTOR_INV_SRC_COLOR;
+      }
    }
-   return xorg_blends[BLEND_OP_OVER];
+   return supported;
 }
 
 static INLINE int
@@ -112,150 +130,41 @@ render_repeat_to_gallium(int mode)
    return PIPE_TEX_WRAP_REPEAT;
 }
 
-
-static INLINE void
-setup_vertex0(float vertex[2][4], float x, float y,
-              float color[4])
-{
-   vertex[0][0] = x;
-   vertex[0][1] = y;
-   vertex[0][2] = 0.f; /*z*/
-   vertex[0][3] = 1.f; /*w*/
-
-   vertex[1][0] = color[0]; /*r*/
-   vertex[1][1] = color[1]; /*g*/
-   vertex[1][2] = color[2]; /*b*/
-   vertex[1][3] = color[3]; /*a*/
-}
-
-static struct pipe_buffer *
-setup_vertex_data0(struct exa_context *ctx,
-                   int srcX, int srcY, int maskX, int maskY,
-                   int dstX, int dstY, int width, int height)
-{
-   float vertices[4][2][4];
-
-   /* 1st vertex */
-   setup_vertex0(vertices[0], dstX, dstY,
-                 ctx->solid_color);
-   /* 2nd vertex */
-   setup_vertex0(vertices[1], dstX + width, dstY,
-                 ctx->solid_color);
-   /* 3rd vertex */
-   setup_vertex0(vertices[2], dstX + width, dstY + height,
-                 ctx->solid_color);
-   /* 4th vertex */
-   setup_vertex0(vertices[3], dstX, dstY + height,
-                 ctx->solid_color);
-
-   return pipe_user_buffer_create(ctx->ctx->screen,
-                                  vertices,
-                                  sizeof(vertices));
-}
-
-static INLINE void
-setup_vertex1(float vertex[2][4], float x, float y, float s, float t)
-{
-   vertex[0][0] = x;
-   vertex[0][1] = y;
-   vertex[0][2] = 0.f; /*z*/
-   vertex[0][3] = 1.f; /*w*/
-
-   vertex[1][0] = s;   /*s*/
-   vertex[1][1] = t;   /*t*/
-   vertex[1][2] = 0.f; /*r*/
-   vertex[1][3] = 1.f; /*q*/
-}
-
-static struct pipe_buffer *
-setup_vertex_data1(struct exa_context *ctx,
-                   int srcX, int srcY, int maskX, int maskY,
-                   int dstX, int dstY, int width, int height)
+static INLINE boolean
+render_filter_to_gallium(int xrender_filter, int *out_filter)
 {
-   float vertices[4][2][4];
-   float s0, t0, s1, t1;
-   struct pipe_texture *src = ctx->bound_textures[0];
-
-   s0 = srcX / src->width[0];
-   s1 = srcX + width / src->width[0];
-   t0 = srcY / src->height[0];
-   t1 = srcY + height / src->height[0];
-
-   /* 1st vertex */
-   setup_vertex1(vertices[0], dstX, dstY,
-                 s0, t0);
-   /* 2nd vertex */
-   setup_vertex1(vertices[1], dstX + width, dstY,
-                 s1, t0);
-   /* 3rd vertex */
-   setup_vertex1(vertices[2], dstX + width, dstY + height,
-                 s1, t1);
-   /* 4th vertex */
-   setup_vertex1(vertices[3], dstX, dstY + height,
-                 s0, t1);
-
-   return pipe_user_buffer_create(ctx->ctx->screen,
-                                  vertices,
-                                  sizeof(vertices));
-}
 
+   switch (xrender_filter) {
+   case PictFilterNearest:
+      *out_filter = PIPE_TEX_FILTER_NEAREST;
+      break;
+   case PictFilterBilinear:
+      *out_filter = PIPE_TEX_FILTER_LINEAR;
+      break;
+   case PictFilterFast:
+      *out_filter = PIPE_TEX_FILTER_NEAREST;
+      break;
+   case PictFilterGood:
+      *out_filter = PIPE_TEX_FILTER_LINEAR;
+      break;
+   case PictFilterBest:
+      *out_filter = PIPE_TEX_FILTER_LINEAR;
+      break;
+   default:
+      debug_printf("Unkown xrender filter");
+      *out_filter = PIPE_TEX_FILTER_NEAREST;
+      return FALSE;
+   }
 
-static INLINE void
-setup_vertex2(float vertex[3][4], float x, float y,
-              float s0, float t0, float s1, float t1)
-{
-   vertex[0][0] = x;
-   vertex[0][1] = y;
-   vertex[0][2] = 0.f; /*z*/
-   vertex[0][3] = 1.f; /*w*/
-
-   vertex[1][0] = s0;  /*s*/
-   vertex[1][1] = t0;  /*t*/
-   vertex[1][2] = 0.f; /*r*/
-   vertex[1][3] = 1.f; /*q*/
-
-   vertex[2][0] = s1;  /*s*/
-   vertex[2][1] = t1;  /*t*/
-   vertex[2][2] = 0.f; /*r*/
-   vertex[2][3] = 1.f; /*q*/
+   return TRUE;
 }
 
-static struct pipe_buffer *
-setup_vertex_data2(struct exa_context *ctx,
-                   int srcX, int srcY, int maskX, int maskY,
-                   int dstX, int dstY, int width, int height)
+static boolean is_filter_accelerated(PicturePtr pic)
 {
-   float vertices[4][3][4];
-   float st0[4], st1[4];
-   struct pipe_texture *src = ctx->bound_textures[0];
-   struct pipe_texture *mask = ctx->bound_textures[0];
-
-   st0[0] = srcX / src->width[0];
-   st0[1] = srcY / src->height[0];
-   st0[2] = srcX + width / src->width[0];
-   st0[3] = srcY + height / src->height[0];
-
-   st1[0] = maskX / mask->width[0];
-   st1[1] = maskY / mask->height[0];
-   st1[2] = maskX + width / mask->width[0];
-   st1[3] = maskY + height / mask->height[0];
-
-   /* 1st vertex */
-   setup_vertex2(vertices[0], dstX, dstY,
-                 st0[0], st0[1], st1[0], st1[1]);
-   /* 2nd vertex */
-   setup_vertex2(vertices[1], dstX + width, dstY,
-                 st0[2], st0[1], st1[2], st1[1]);
-   /* 3rd vertex */
-   setup_vertex2(vertices[2], dstX + width, dstY + height,
-                 st0[2], st0[3], st1[2], st1[3]);
-   /* 4th vertex */
-   setup_vertex2(vertices[3], dstX, dstY + height,
-                 st0[0], st0[3], st1[0], st1[3]);
-
-   return pipe_user_buffer_create(ctx->ctx->screen,
-                                  vertices,
-                                  sizeof(vertices));
+   int filter;
+   if (pic && !render_filter_to_gallium(pic->filter, &filter))
+       return FALSE;
+   return TRUE;
 }
 
 boolean xorg_composite_accelerated(int op,
@@ -263,125 +172,60 @@ boolean xorg_composite_accelerated(int op,
                                    PicturePtr pMaskPicture,
                                    PicturePtr pDstPicture)
 {
-   unsigned i;
-   unsigned accel_ops_count =
-      sizeof(accelerated_ops)/sizeof(struct acceleration_info);
-
-
-   /*FIXME: currently accel is disabled */
-   return FALSE;
+   ScreenPtr pScreen = pDstPicture->pDrawable->pScreen;
+   ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
+   modesettingPtr ms = modesettingPTR(pScrn);
+   struct xorg_composite_blend blend;
+
+   if (!is_filter_accelerated(pSrcPicture) ||
+       !is_filter_accelerated(pMaskPicture)) {
+      XORG_FALLBACK("Unsupported Xrender filter");
+   }
 
-   if (pSrcPicture) {
-      /* component alpha not supported */
-      if (pSrcPicture->componentAlpha)
-         return FALSE;
-      /* fills not supported */
-      if (pSrcPicture->pSourcePict)
-         return FALSE;
+   if (pSrcPicture->pSourcePict) {
+      if (pSrcPicture->pSourcePict->type != SourcePictTypeSolidFill)
+         XORG_FALLBACK("Gradients not enabled (haven't been well tested)");
    }
 
-   for (i = 0; i < accel_ops_count; ++i) {
-      if (op == accelerated_ops[i].op) {
-         if (pMaskPicture && !accelerated_ops[i].with_mask)
-            return FALSE;
-         return TRUE;
+   if (blend_for_op(&blend, op,
+                    pSrcPicture, pMaskPicture, pDstPicture)) {
+      /* Check for component alpha */
+      if (pMaskPicture && pMaskPicture->componentAlpha &&
+          PICT_FORMAT_RGB(pMaskPicture->format)) {
+         if (blend.alpha_src && blend.rgb_src != PIPE_BLENDFACTOR_ZERO) {
+            XORG_FALLBACK("Component alpha not supported with source "
+                          "alpha and source value blending. (op=%d)",
+                          op);
+         }
       }
+      return TRUE;
    }
-   return FALSE;
-}
-
-static void
-bind_framebuffer_state(struct exa_context *exa, PicturePtr pDstPicture,
-                       struct exa_pixmap_priv *pDst)
-{
-   unsigned i;
-   struct pipe_framebuffer_state state;
-   struct pipe_surface *surface = exa_gpu_surface(exa, pDst);
-   memset(&state, 0, sizeof(struct pipe_framebuffer_state));
-
-   state.width  = pDstPicture->pDrawable->width;
-   state.height = pDstPicture->pDrawable->height;
-
-   state.nr_cbufs = 1;
-   state.cbufs[0] = surface;
-   for (i = 1; i < PIPE_MAX_COLOR_BUFS; ++i)
-      state.cbufs[i] = 0;
-
-   /* currently we don't use depth/stencil */
-   state.zsbuf = 0;
-
-   cso_set_framebuffer(exa->cso, &state);
-}
-
-enum AxisOrientation {
-   Y0_BOTTOM,
-   Y0_TOP
-};
-
-static void
-set_viewport(struct exa_context *exa, int width, int height,
-             enum AxisOrientation orientation)
-{
-   struct pipe_viewport_state viewport;
-   float y_scale = (orientation == Y0_BOTTOM) ? -2.f : 2.f;
-
-   viewport.scale[0] =  width / 2.f;
-   viewport.scale[1] =  height / y_scale;
-   viewport.scale[2] =  1.0;
-   viewport.scale[3] =  1.0;
-   viewport.translate[0] = width / 2.f;
-   viewport.translate[1] = height / 2.f;
-   viewport.translate[2] = 0.0;
-   viewport.translate[3] = 0.0;
-
-   cso_set_viewport(exa->cso, &viewport);
-}
-
-static void
-bind_viewport_state(struct exa_context *exa, PicturePtr pDstPicture)
-{
-   int width = pDstPicture->pDrawable->width;
-   int height = pDstPicture->pDrawable->height;
-
-   set_viewport(exa, width, height, Y0_TOP);
+   XORG_FALLBACK("Unsupported composition operation = %d", op);
 }
 
 static void
 bind_blend_state(struct exa_context *exa, int op,
-                 PicturePtr pSrcPicture, PicturePtr pMaskPicture)
+                 PicturePtr pSrcPicture,
+                 PicturePtr pMaskPicture,
+                 PicturePtr pDstPicture)
 {
-   boolean component_alpha = pSrcPicture->componentAlpha;
    struct xorg_composite_blend blend_opt;
    struct pipe_blend_state blend;
 
-   if (component_alpha) {
-      op = PictOpOver;
-   }
-   blend_opt = blend_for_op(op);
+   blend_for_op(&blend_opt, op, pSrcPicture, pMaskPicture, pDstPicture);
 
    memset(&blend, 0, sizeof(struct pipe_blend_state));
    blend.blend_enable = 1;
-   blend.colormask |= PIPE_MASK_R;
-   blend.colormask |= PIPE_MASK_G;
-   blend.colormask |= PIPE_MASK_B;
-   blend.colormask |= PIPE_MASK_A;
+   blend.colormask |= PIPE_MASK_RGBA;
 
-   blend.rgb_src_factor   = blend_opt.rgb_src_factor;
-   blend.alpha_src_factor = blend_opt.alpha_src_factor;
-   blend.rgb_dst_factor   = blend_opt.rgb_dst_factor;
-   blend.alpha_dst_factor = blend_opt.alpha_dst_factor;
+   blend.rgb_src_factor   = blend_opt.rgb_src;
+   blend.alpha_src_factor = blend_opt.rgb_src;
+   blend.rgb_dst_factor   = blend_opt.rgb_dst;
+   blend.alpha_dst_factor = blend_opt.rgb_dst;
 
-   cso_set_blend(exa->cso, &blend);
+   cso_set_blend(exa->renderer->cso, &blend);
 }
 
-static void
-bind_rasterizer_state(struct exa_context *exa)
-{
-   struct pipe_rasterizer_state raster;
-   memset(&raster, 0, sizeof(struct pipe_rasterizer_state));
-   raster.gl_rasterization_rules = 1;
-   cso_set_rasterizer(exa->cso, &raster);
-}
 
 static void
 bind_shaders(struct exa_context *exa, int op,
@@ -390,14 +234,17 @@ bind_shaders(struct exa_context *exa, int op,
    unsigned vs_traits = 0, fs_traits = 0;
    struct xorg_shader shader;
 
+   exa->has_solid_color = FALSE;
+
    if (pSrcPicture) {
       if (pSrcPicture->pSourcePict) {
          if (pSrcPicture->pSourcePict->type == SourcePictTypeSolidFill) {
             fs_traits |= FS_SOLID_FILL;
             vs_traits |= VS_SOLID_FILL;
-            pixel_to_float4(pSrcPicture->pFormat,
-                            pSrcPicture->pSourcePict->solidFill.color,
+            debug_assert(pSrcPicture->format == PICT_a8r8g8b8);
+            pixel_to_float4(pSrcPicture->pSourcePict->solidFill.color,
                             exa->solid_color);
+            exa->has_solid_color = TRUE;
          } else {
             debug_assert("!gradients not supported");
          }
@@ -410,14 +257,15 @@ bind_shaders(struct exa_context *exa, int op,
    if (pMaskPicture) {
       vs_traits |= VS_MASK;
       fs_traits |= FS_MASK;
+      if (pMaskPicture->componentAlpha)
+         fs_traits |= FS_COMPONENT_ALPHA;
    }
 
-   shader = xorg_shaders_get(exa->shaders, vs_traits, fs_traits);
-   cso_set_vertex_shader_handle(exa->cso, shader.vs);
-   cso_set_fragment_shader_handle(exa->cso, shader.fs);
+   shader = xorg_shaders_get(exa->renderer->shaders, vs_traits, fs_traits);
+   cso_set_vertex_shader_handle(exa->renderer->cso, shader.vs);
+   cso_set_fragment_shader_handle(exa->renderer->cso, shader.fs);
 }
 
-
 static void
 bind_samplers(struct exa_context *exa, int op,
               PicturePtr pSrcPicture, PicturePtr pMaskPicture,
@@ -434,13 +282,24 @@ bind_samplers(struct exa_context *exa, int op,
    memset(&src_sampler, 0, sizeof(struct pipe_sampler_state));
    memset(&mask_sampler, 0, sizeof(struct pipe_sampler_state));
 
+   if ((pSrc && exa->pipe->is_texture_referenced(exa->pipe, pSrc->tex, 0, 0) &
+        PIPE_REFERENCED_FOR_WRITE) ||
+       (pMask && exa->pipe->is_texture_referenced(exa->pipe, pMask->tex, 0, 0) &
+        PIPE_REFERENCED_FOR_WRITE))
+      exa->pipe->flush(exa->pipe, PIPE_FLUSH_RENDER_CACHE, NULL);
+
    if (pSrcPicture && pSrc) {
       unsigned src_wrap = render_repeat_to_gallium(
          pSrcPicture->repeatType);
+      int filter;
+
+      render_filter_to_gallium(pSrcPicture->filter, &filter);
+
       src_sampler.wrap_s = src_wrap;
       src_sampler.wrap_t = src_wrap;
-      src_sampler.min_img_filter = PIPE_TEX_MIPFILTER_NEAREST;
-      src_sampler.mag_img_filter = PIPE_TEX_MIPFILTER_NEAREST;
+      src_sampler.min_img_filter = filter;
+      src_sampler.mag_img_filter = filter;
+      src_sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NEAREST;
       src_sampler.normalized_coords = 1;
       samplers[0] = &src_sampler;
       exa->bound_textures[0] = pSrc->tex;
@@ -450,19 +309,24 @@ bind_samplers(struct exa_context *exa, int op,
    if (pMaskPicture && pMask) {
       unsigned mask_wrap = render_repeat_to_gallium(
          pMaskPicture->repeatType);
+      int filter;
+
+      render_filter_to_gallium(pMaskPicture->filter, &filter);
+
       mask_sampler.wrap_s = mask_wrap;
       mask_sampler.wrap_t = mask_wrap;
-      mask_sampler.min_img_filter = PIPE_TEX_MIPFILTER_NEAREST;
-      mask_sampler.mag_img_filter = PIPE_TEX_MIPFILTER_NEAREST;
+      mask_sampler.min_img_filter = filter;
+      mask_sampler.mag_img_filter = filter;
+      src_sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NEAREST;
       mask_sampler.normalized_coords = 1;
       samplers[1] = &mask_sampler;
       exa->bound_textures[1] = pMask->tex;
       ++exa->num_bound_samplers;
    }
 
-   cso_set_samplers(exa->cso, exa->num_bound_samplers,
+   cso_set_samplers(exa->renderer->cso, exa->num_bound_samplers,
                     (const struct pipe_sampler_state **)samplers);
-   cso_set_sampler_textures(exa->cso, exa->num_bound_samplers,
+   cso_set_sampler_textures(exa->renderer->cso, exa->num_bound_samplers,
                             exa->bound_textures);
 }
 
@@ -475,18 +339,8 @@ setup_vs_constant_buffer(struct exa_context *exa,
       2.f/width, 2.f/height, 1, 1,
       -1, -1, 0, 0
    };
-   struct pipe_constant_buffer *cbuf = &exa->vs_const_buffer;
-
-   pipe_buffer_reference(&cbuf->buffer, NULL);
-   cbuf->buffer = pipe_buffer_create(exa->ctx->screen, 16,
-                                     PIPE_BUFFER_USAGE_CONSTANT,
-                                     param_bytes);
-
-   if (cbuf->buffer) {
-      pipe_buffer_write(exa->ctx->screen, cbuf->buffer,
-                        0, param_bytes, vs_consts);
-   }
-   exa->ctx->set_constant_buffer(exa->ctx, PIPE_SHADER_VERTEX, 0, cbuf);
+   renderer_set_constants(exa->renderer, PIPE_SHADER_VERTEX,
+                          vs_consts, param_bytes);
 }
 
 
@@ -494,33 +348,61 @@ static void
 setup_fs_constant_buffer(struct exa_context *exa)
 {
    const int param_bytes = 4 * sizeof(float);
-   float fs_consts[8] = {
+   const float fs_consts[8] = {
       0, 0, 0, 1,
    };
-   struct pipe_constant_buffer *cbuf = &exa->fs_const_buffer;
-
-   pipe_buffer_reference(&cbuf->buffer, NULL);
-   cbuf->buffer = pipe_buffer_create(exa->ctx->screen, 16,
-                                     PIPE_BUFFER_USAGE_CONSTANT,
-                                     param_bytes);
-
-   if (cbuf->buffer) {
-      pipe_buffer_write(exa->ctx->screen, cbuf->buffer,
-                        0, param_bytes, fs_consts);
-   }
-   exa->ctx->set_constant_buffer(exa->ctx, PIPE_SHADER_FRAGMENT, 0, cbuf);
+   renderer_set_constants(exa->renderer, PIPE_SHADER_FRAGMENT,
+                          fs_consts, param_bytes);
 }
 
 static void
-setup_constant_buffers(struct exa_context *exa, PicturePtr pDstPicture)
+setup_constant_buffers(struct exa_context *exa, struct exa_pixmap_priv *pDst)
 {
-   int width = pDstPicture->pDrawable->width;
-   int height = pDstPicture->pDrawable->height;
+   int width = pDst->tex->width[0];
+   int height = pDst->tex->height[0];
 
    setup_vs_constant_buffer(exa, width, height);
    setup_fs_constant_buffer(exa);
 }
 
+static INLINE boolean matrix_from_pict_transform(PictTransform *trans, float *matrix)
+{
+   if (!trans)
+      return FALSE;
+
+   matrix[0] = XFixedToDouble(trans->matrix[0][0]);
+   matrix[1] = XFixedToDouble(trans->matrix[0][1]);
+   matrix[2] = XFixedToDouble(trans->matrix[0][2]);
+
+   matrix[3] = XFixedToDouble(trans->matrix[1][0]);
+   matrix[4] = XFixedToDouble(trans->matrix[1][1]);
+   matrix[5] = XFixedToDouble(trans->matrix[1][2]);
+
+   matrix[6] = XFixedToDouble(trans->matrix[2][0]);
+   matrix[7] = XFixedToDouble(trans->matrix[2][1]);
+   matrix[8] = XFixedToDouble(trans->matrix[2][2]);
+
+   return TRUE;
+}
+
+static void
+setup_transforms(struct  exa_context *exa,
+                 PicturePtr pSrcPicture, PicturePtr pMaskPicture)
+{
+   PictTransform *src_t = NULL;
+   PictTransform *mask_t = NULL;
+
+   if (pSrcPicture)
+      src_t = pSrcPicture->transform;
+   if (pMaskPicture)
+      mask_t = pMaskPicture->transform;
+
+   exa->transform.has_src  =
+      matrix_from_pict_transform(src_t, exa->transform.src);
+   exa->transform.has_mask =
+      matrix_from_pict_transform(mask_t, exa->transform.mask);
+}
+
 boolean xorg_composite_bind_state(struct exa_context *exa,
                                   int op,
                                   PicturePtr pSrcPicture,
@@ -530,17 +412,18 @@ boolean xorg_composite_bind_state(struct exa_context *exa,
                                   struct exa_pixmap_priv *pMask,
                                   struct exa_pixmap_priv *pDst)
 {
-   bind_framebuffer_state(exa, pDstPicture, pDst);
-   bind_viewport_state(exa, pDstPicture);
-   bind_blend_state(exa, op, pSrcPicture, pMaskPicture);
-   bind_rasterizer_state(exa);
+   renderer_bind_framebuffer(exa->renderer, pDst);
+   renderer_bind_viewport(exa->renderer, pDst);
+   bind_blend_state(exa, op, pSrcPicture, pMaskPicture, pDstPicture);
+   renderer_bind_rasterizer(exa->renderer);
    bind_shaders(exa, op, pSrcPicture, pMaskPicture);
    bind_samplers(exa, op, pSrcPicture, pMaskPicture,
                  pDstPicture, pSrc, pMask, pDst);
+   setup_constant_buffers(exa, pDst);
 
-   setup_constant_buffers(exa, pDstPicture);
+   setup_transforms(exa, pSrcPicture, pMaskPicture);
 
-   return FALSE;
+   return TRUE;
 }
 
 void xorg_composite(struct exa_context *exa,
@@ -548,37 +431,67 @@ void xorg_composite(struct exa_context *exa,
                     int srcX, int srcY, int maskX, int maskY,
                     int dstX, int dstY, int width, int height)
 {
-   struct pipe_context *pipe = exa->ctx;
-   struct pipe_buffer *buf = 0;
-
    if (exa->num_bound_samplers == 0 ) { /* solid fill */
-      buf = setup_vertex_data0(exa,
-                               srcX, srcY, maskX, maskY,
-                               dstX, dstY, width, height);
-   } else if (exa->num_bound_samplers == 1 ) /* src */
-      buf = setup_vertex_data1(exa,
-                               srcX, srcY, maskX, maskY,
-                               dstX, dstY, width, height);
-   else if (exa->num_bound_samplers == 2) /* src + mask */
-      buf = setup_vertex_data2(exa,
-                               srcX, srcY, maskX, maskY,
-                               dstX, dstY, width, height);
-   else if (exa->num_bound_samplers == 3) { /* src + mask + dst */
-      debug_assert(!"src/mask/dst not handled right now");
+      renderer_draw_solid_rect(exa->renderer,
+                               dstX, dstY, dstX + width, dstY + height,
+                               exa->solid_color);
+   } else {
+      int pos[6] = {srcX, srcY, maskX, maskY, dstX, dstY};
+      float *src_matrix = NULL;
+      float *mask_matrix = NULL;
+
+      if (exa->transform.has_src)
+         src_matrix = exa->transform.src;
+      if (exa->transform.has_mask)
+         mask_matrix = exa->transform.mask;
+
+      renderer_draw_textures(exa->renderer,
+                             pos, width, height,
+                             exa->bound_textures,
+                             exa->num_bound_samplers,
+                             src_matrix, mask_matrix);
+   }
+}
+
+boolean xorg_solid_bind_state(struct exa_context *exa,
+                              struct exa_pixmap_priv *pixmap,
+                              Pixel fg)
+{
+   unsigned vs_traits, fs_traits;
+   struct xorg_shader shader;
+
+   pixel_to_float4(fg, exa->solid_color);
+   exa->has_solid_color = TRUE;
+
 #if 0
-      buf = setup_vertex_data2(exa,
-                               srcX, srcY, maskX, maskY,
-                               dstX, dstY, width, height);
+   debug_printf("Color Pixel=(%d, %d, %d, %d), RGBA=(%f, %f, %f, %f)\n",
+                (fg >> 24) & 0xff, (fg >> 16) & 0xff,
+                (fg >> 8) & 0xff,  (fg >> 0) & 0xff,
+                exa->solid_color[0], exa->solid_color[1],
+                exa->solid_color[2], exa->solid_color[3]);
 #endif
-   }
 
-   if (buf) {
-      util_draw_vertex_buffer(pipe, buf, 0,
-                              PIPE_PRIM_TRIANGLE_FAN,
-                              4,  /* verts */
-                              1 + exa->num_bound_samplers); /* attribs/vert */
+   vs_traits = VS_SOLID_FILL;
+   fs_traits = FS_SOLID_FILL;
 
-      pipe_buffer_reference(&buf, NULL);
-   }
+   renderer_bind_framebuffer(exa->renderer, pixmap);
+   renderer_bind_viewport(exa->renderer, pixmap);
+   renderer_bind_rasterizer(exa->renderer);
+   bind_blend_state(exa, PictOpSrc, NULL, NULL, NULL);
+   setup_constant_buffers(exa, pixmap);
+
+   shader = xorg_shaders_get(exa->renderer->shaders, vs_traits, fs_traits);
+   cso_set_vertex_shader_handle(exa->renderer->cso, shader.vs);
+   cso_set_fragment_shader_handle(exa->renderer->cso, shader.fs);
+
+   return TRUE;
+}
+
+void xorg_solid(struct exa_context *exa,
+                struct exa_pixmap_priv *pixmap,
+                int x0, int y0, int x1, int y1)
+{
+   renderer_draw_solid_rect(exa->renderer,
+                            x0, y0, x1, y1, exa->solid_color);
 }
 
diff --git a/src/gallium/state_trackers/xorg/xorg_composite.h b/src/gallium/state_trackers/xorg/xorg_composite.h
index 17dfcb199e..236addf1ce 100644
--- a/src/gallium/state_trackers/xorg/xorg_composite.h
+++ b/src/gallium/state_trackers/xorg/xorg_composite.h
@@ -22,4 +22,11 @@ void xorg_composite(struct exa_context *exa,
                     int srcX, int srcY, int maskX, int maskY,
                     int dstX, int dstY, int width, int height);
 
+boolean xorg_solid_bind_state(struct exa_context *exa,
+                              struct exa_pixmap_priv *pixmap,
+                              Pixel fg);
+void xorg_solid(struct exa_context *exa,
+                struct exa_pixmap_priv *pixmap,
+                int x0, int y0, int x1, int y1);
+
 #endif
diff --git a/src/gallium/state_trackers/xorg/xorg_crtc.c b/src/gallium/state_trackers/xorg/xorg_crtc.c
index fe08bde9ef..85b9162d4c 100644
--- a/src/gallium/state_trackers/xorg/xorg_crtc.c
+++ b/src/gallium/state_trackers/xorg/xorg_crtc.c
@@ -42,8 +42,12 @@
 #include "xorg_tracker.h"
 #include "xf86Modes.h"
 
+#ifdef HAVE_XEXTPROTO_71
+#include <X11/extensions/dpmsconst.h>
+#else
 #define DPMS_SERVER
 #include <X11/extensions/dpms.h>
+#endif
 
 #include "pipe/p_inlines.h"
 #include "util/u_rect.h"
@@ -60,8 +64,6 @@ struct crtc_private
 static void
 crtc_dpms(xf86CrtcPtr crtc, int mode)
 {
-    //ScrnInfoPtr pScrn = crtc->scrn;
-
     switch (mode) {
     case DPMSModeOn:
     case DPMSModeStandby:
@@ -73,44 +75,29 @@ crtc_dpms(xf86CrtcPtr crtc, int mode)
 }
 
 static Bool
-crtc_lock(xf86CrtcPtr crtc)
-{
-    return FALSE;
-}
-
-static void
-crtc_unlock(xf86CrtcPtr crtc)
-{
-}
-
-static void
-crtc_prepare(xf86CrtcPtr crtc)
-{
-}
-
-static void
-crtc_commit(xf86CrtcPtr crtc)
-{
-}
-
-static Bool
-crtc_mode_fixup(xf86CrtcPtr crtc, DisplayModePtr mode,
-		DisplayModePtr adjusted_mode)
-{
-    return TRUE;
-}
-
-static void
-crtc_mode_set(xf86CrtcPtr crtc, DisplayModePtr mode,
-	      DisplayModePtr adjusted_mode, int x, int y)
+crtc_set_mode_major(xf86CrtcPtr crtc, DisplayModePtr mode,
+		    Rotation rotation, int x, int y)
 {
     xf86CrtcConfigPtr config = XF86_CRTC_CONFIG_PTR(crtc->scrn);
     modesettingPtr ms = modesettingPTR(crtc->scrn);
-    xf86OutputPtr output = config->output[config->compat_output];
-    drmModeConnectorPtr drm_connector = output->driver_private;
+    xf86OutputPtr output = NULL;
+    drmModeConnectorPtr drm_connector;
     struct crtc_private *crtcp = crtc->driver_private;
     drmModeCrtcPtr drm_crtc = crtcp->drm_crtc;
     drmModeModeInfo drm_mode;
+    int i, ret;
+
+    for (i = 0; i < config->num_output; output = NULL, i++) {
+	output = config->output[i];
+
+	if (output->crtc == crtc)
+	    break;
+    }
+
+    if (!output)
+	return FALSE;
+
+    drm_connector = output->driver_private;
 
     drm_mode.clock = mode->Clock;
     drm_mode.hdisplay = mode->HDisplay;
@@ -129,17 +116,19 @@ crtc_mode_set(xf86CrtcPtr crtc, DisplayModePtr mode,
 	xf86SetModeDefaultName(mode);
     strncpy(drm_mode.name, mode->name, DRM_DISPLAY_MODE_LEN);
 
-    drmModeSetCrtc(ms->fd, drm_crtc->crtc_id, ms->fb_id, x, y,
-		   &drm_connector->connector_id, 1, &drm_mode);
-}
+    ret = drmModeSetCrtc(ms->fd, drm_crtc->crtc_id, ms->fb_id, x, y,
+			 &drm_connector->connector_id, 1, &drm_mode);
 
-#if 0
-static void
-crtc_load_lut(xf86CrtcPtr crtc)
-{
-    //ScrnInfoPtr pScrn = crtc->scrn;
+    if (ret)
+	return FALSE;
+
+    crtc->x = x;
+    crtc->y = y;
+    crtc->mode = *mode;
+    crtc->rotation = rotation;
+
+    return TRUE;
 }
-#endif
 
 static void
 crtc_gamma_set(xf86CrtcPtr crtc, CARD16 * red, CARD16 * green, CARD16 * blue,
@@ -150,37 +139,37 @@ crtc_gamma_set(xf86CrtcPtr crtc, CARD16 * red, CARD16 * green, CARD16 * blue,
 static void *
 crtc_shadow_allocate(xf86CrtcPtr crtc, int width, int height)
 {
-    //ScrnInfoPtr pScrn = crtc->scrn;
-
     return NULL;
 }
 
 static PixmapPtr
 crtc_shadow_create(xf86CrtcPtr crtc, void *data, int width, int height)
 {
-    //ScrnInfoPtr pScrn = crtc->scrn;
-
     return NULL;
 }
 
 static void
 crtc_shadow_destroy(xf86CrtcPtr crtc, PixmapPtr rotate_pixmap, void *data)
 {
-    //ScrnInfoPtr pScrn = crtc->scrn;
 }
 
+/*
+ * Cursor functions
+ */
+
 static void
-crtc_destroy(xf86CrtcPtr crtc)
+crtc_set_cursor_colors(xf86CrtcPtr crtc, int bg, int fg)
 {
-    struct crtc_private *crtcp = crtc->driver_private;
+}
 
-    if (crtcp->cursor_tex)
-	pipe_texture_reference(&crtcp->cursor_tex, NULL);
+static void
+crtc_set_cursor_position(xf86CrtcPtr crtc, int x, int y)
+{
+    modesettingPtr ms = modesettingPTR(crtc->scrn);
+    struct crtc_private *crtcp = crtc->driver_private;
 
-    drmModeFreeCrtc(crtcp->drm_crtc);
-    xfree(crtcp);
+    drmModeMoveCursor(ms->fd, crtcp->drm_crtc->crtc_id, x, y);
 }
-
 static void
 crtc_load_cursor_argb(xf86CrtcPtr crtc, CARD32 * image)
 {
@@ -226,15 +215,6 @@ crtc_load_cursor_argb(xf86CrtcPtr crtc, CARD32 * image)
 }
 
 static void
-crtc_set_cursor_position(xf86CrtcPtr crtc, int x, int y)
-{
-    modesettingPtr ms = modesettingPTR(crtc->scrn);
-    struct crtc_private *crtcp = crtc->driver_private;
-
-    drmModeMoveCursor(ms->fd, crtcp->drm_crtc->crtc_id, x, y);
-}
-
-static void
 crtc_show_cursor(xf86CrtcPtr crtc)
 {
     modesettingPtr ms = modesettingPTR(crtc->scrn);
@@ -254,29 +234,6 @@ crtc_hide_cursor(xf86CrtcPtr crtc)
     drmModeSetCursor(ms->fd, crtcp->drm_crtc->crtc_id, 0, 0, 0);
 }
 
-static const xf86CrtcFuncsRec crtc_funcs = {
-    .dpms = crtc_dpms,
-    .save = NULL,
-    .restore = NULL,
-    .lock = crtc_lock,
-    .unlock = crtc_unlock,
-    .mode_fixup = crtc_mode_fixup,
-    .prepare = crtc_prepare,
-    .mode_set = crtc_mode_set,
-    .commit = crtc_commit,
-    .gamma_set = crtc_gamma_set,
-    .shadow_create = crtc_shadow_create,
-    .shadow_allocate = crtc_shadow_allocate,
-    .shadow_destroy = crtc_shadow_destroy,
-    .set_cursor_position = crtc_set_cursor_position,
-    .show_cursor = crtc_show_cursor,
-    .hide_cursor = crtc_hide_cursor,
-    .load_cursor_image = NULL,	       /* lets convert to argb only */
-    .set_cursor_colors = NULL,	       /* using argb only */
-    .load_cursor_argb = crtc_load_cursor_argb,
-    .destroy = crtc_destroy,
-};
-
 void
 crtc_cursor_destroy(xf86CrtcPtr crtc)
 {
@@ -287,6 +244,40 @@ crtc_cursor_destroy(xf86CrtcPtr crtc)
     }
 }
 
+/*
+ * Misc functions
+ */
+
+static void
+crtc_destroy(xf86CrtcPtr crtc)
+{
+    struct crtc_private *crtcp = crtc->driver_private;
+
+    if (crtcp->cursor_tex)
+	pipe_texture_reference(&crtcp->cursor_tex, NULL);
+
+    drmModeFreeCrtc(crtcp->drm_crtc);
+    xfree(crtcp);
+}
+
+static const xf86CrtcFuncsRec crtc_funcs = {
+    .dpms = crtc_dpms,
+    .set_mode_major = crtc_set_mode_major,
+
+    .set_cursor_colors = crtc_set_cursor_colors,
+    .set_cursor_position = crtc_set_cursor_position,
+    .show_cursor = crtc_show_cursor,
+    .hide_cursor = crtc_hide_cursor,
+    .load_cursor_argb = crtc_load_cursor_argb,
+
+    .shadow_create = crtc_shadow_create,
+    .shadow_allocate = crtc_shadow_allocate,
+    .shadow_destroy = crtc_shadow_destroy,
+
+    .gamma_set = crtc_gamma_set,
+    .destroy = crtc_destroy,
+};
+
 void
 crtc_init(ScrnInfoPtr pScrn)
 {
@@ -305,6 +296,7 @@ crtc_init(ScrnInfoPtr pScrn)
 
     for (c = 0; c < res->count_crtcs; c++) {
 	drm_crtc = drmModeGetCrtc(ms->fd, res->crtcs[c]);
+
 	if (!drm_crtc)
 	    continue;
 
@@ -321,7 +313,6 @@ crtc_init(ScrnInfoPtr pScrn)
 	crtcp->drm_crtc = drm_crtc;
 
 	crtc->driver_private = crtcp;
-
     }
 
   out:
diff --git a/src/gallium/state_trackers/xorg/xorg_dri2.c b/src/gallium/state_trackers/xorg/xorg_dri2.c
index 6431a0fe25..c41a7cd639 100644
--- a/src/gallium/state_trackers/xorg/xorg_dri2.c
+++ b/src/gallium/state_trackers/xorg/xorg_dri2.c
@@ -81,11 +81,14 @@ driDoCreateBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int format)
     case DRI2BufferStencil:
 #if defined(DRI2INFOREC_VERSION) && DRI2INFOREC_VERSION > 2
     case DRI2BufferDepthStencil:
+#else
+    /* Works on old X servers because sanity checking is for the weak */
+    case 9:
+#endif
 	if (exa_priv->depth_stencil_tex &&
 	    !pf_is_depth_stencil(exa_priv->depth_stencil_tex->format))
 	    exa_priv->depth_stencil_tex = NULL;
         /* Fall through */
-#endif
     case DRI2BufferDepth:
 	if (exa_priv->depth_stencil_tex)
 	    pipe_texture_reference(&tex, exa_priv->depth_stencil_tex);
@@ -118,6 +121,7 @@ driDoCreateBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int format)
     }
 
     if (!tex) {
+	exaMoveInPixmap(private->pPixmap);
 	xorg_exa_set_shared_usage(private->pPixmap);
 	pScreen->ModifyPixmapHeader(private->pPixmap, 0, 0, 0, 0, 0, NULL);
 	tex = xorg_exa_get_texture(private->pPixmap);
diff --git a/src/gallium/state_trackers/xorg/xorg_driver.c b/src/gallium/state_trackers/xorg/xorg_driver.c
index 643b6b3b9e..847647c1e4 100644
--- a/src/gallium/state_trackers/xorg/xorg_driver.c
+++ b/src/gallium/state_trackers/xorg/xorg_driver.c
@@ -144,20 +144,22 @@ static Bool
 CreateFrontBuffer(ScrnInfoPtr pScrn)
 {
     modesettingPtr ms = modesettingPTR(pScrn);
-    ScreenPtr pScreen = pScrn->pScreen;
-    PixmapPtr rootPixmap = pScreen->GetScreenPixmap(pScreen);
     unsigned handle, stride;
+    struct pipe_texture *tex;
 
     ms->noEvict = TRUE;
-    xorg_exa_set_displayed_usage(rootPixmap);
-    pScreen->ModifyPixmapHeader(rootPixmap,
-				pScrn->virtualX, pScrn->virtualY,
-				pScrn->depth, pScrn->bitsPerPixel,
-				pScrn->displayWidth * pScrn->bitsPerPixel / 8,
-				NULL);
-    ms->noEvict = FALSE;
 
-    handle = xorg_exa_get_pixmap_handle(rootPixmap, &stride);
+    tex = xorg_exa_create_root_texture(pScrn, pScrn->virtualX, pScrn->virtualY,
+				       pScrn->depth, pScrn->bitsPerPixel);
+
+    if (!tex)
+	return FALSE;
+
+    if (!ms->api->local_handle_from_texture(ms->api, ms->screen,
+					    tex,
+					    &stride,
+					    &handle))
+	return FALSE;
 
     drmModeAddFB(ms->fd,
 		 pScrn->virtualX,
@@ -166,12 +168,39 @@ CreateFrontBuffer(ScrnInfoPtr pScrn)
 		 pScrn->bitsPerPixel,
 		 stride,
 		 handle,
-		 &ms->fb_id);
+                 &ms->fb_id);
 
     pScrn->frameX0 = 0;
     pScrn->frameY0 = 0;
     AdjustFrame(pScrn->scrnIndex, pScrn->frameX0, pScrn->frameY0, 0);
 
+    pipe_texture_reference(&ms->root_texture, tex);
+    pipe_texture_reference(&tex, NULL);
+    return TRUE;
+}
+
+static Bool
+BindTextureToRoot(ScrnInfoPtr pScrn)
+{
+    modesettingPtr ms = modesettingPTR(pScrn);
+    ScreenPtr pScreen = pScrn->pScreen;
+    struct pipe_texture *check;
+    PixmapPtr rootPixmap;
+
+    rootPixmap = pScreen->GetScreenPixmap(pScreen);
+
+    xorg_exa_set_displayed_usage(rootPixmap);
+    xorg_exa_set_shared_usage(rootPixmap);
+    xorg_exa_set_texture(rootPixmap, ms->root_texture);
+    if (!pScreen->ModifyPixmapHeader(rootPixmap, -1, -1, -1, -1, -1, NULL))
+	FatalError("Couldn't adjust screen pixmap\n");
+
+    check = xorg_exa_get_texture(rootPixmap);
+    if (ms->root_texture != check)
+	FatalError("Created new root texture\n");
+
+    pipe_texture_reference(&check, NULL);
+
     return TRUE;
 }
 
@@ -179,10 +208,9 @@ static Bool
 crtc_resize(ScrnInfoPtr pScrn, int width, int height)
 {
     modesettingPtr ms = modesettingPTR(pScrn);
-    //ScreenPtr pScreen = pScrn->pScreen;
-    //PixmapPtr rootPixmap = pScreen->GetScreenPixmap(pScreen);
-    //Bool fbAccessDisabled;
-    //CARD8 *fbstart;
+    unsigned handle, stride;
+    PixmapPtr rootPixmap;
+    ScreenPtr pScreen = pScrn->pScreen;
 
     if (width == pScrn->virtualX && height == pScrn->virtualY)
 	return TRUE;
@@ -192,13 +220,40 @@ crtc_resize(ScrnInfoPtr pScrn, int width, int height)
     pScrn->virtualX = width;
     pScrn->virtualY = height;
 
+    /*
+     * Remove the old framebuffer & texture.
+     */
+    drmModeRmFB(ms->fd, ms->fb_id);
+    pipe_texture_reference(&ms->root_texture, NULL);
+
+
+    rootPixmap = pScreen->GetScreenPixmap(pScreen);
+    if (!pScreen->ModifyPixmapHeader(rootPixmap, width, height, -1, -1, -1, NULL))
+	return FALSE;
+
+    /* takes one ref */
+    ms->root_texture = xorg_exa_get_texture(rootPixmap);
+
+    if (!ms->api->local_handle_from_texture(ms->api, ms->screen,
+					    ms->root_texture,
+					    &stride,
+					    &handle))
+	FatalError("Could not get handle and stride from texture\n");
+
+    drmModeAddFB(ms->fd,
+		 pScrn->virtualX,
+		 pScrn->virtualY,
+		 pScrn->depth,
+		 pScrn->bitsPerPixel,
+		 stride,
+		 handle,
+                 &ms->fb_id);
+
     /* HW dependent - FIXME */
     pScrn->displayWidth = pScrn->virtualX;
 
-    drmModeRmFB(ms->fd, ms->fb_id);
-
     /* now create new frontbuffer */
-    return CreateFrontBuffer(pScrn);
+    return CreateFrontBuffer(pScrn) && BindTextureToRoot(pScrn);
 }
 
 static const xf86CrtcConfigFuncsRec crtc_config_funcs = {
@@ -206,6 +261,37 @@ static const xf86CrtcConfigFuncsRec crtc_config_funcs = {
 };
 
 static Bool
+InitDRM(ScrnInfoPtr pScrn)
+{
+    modesettingPtr ms = modesettingPTR(pScrn);
+
+    /* deal with server regeneration */
+    if (ms->fd < 0) {
+	char *BusID;
+
+	BusID = xalloc(64);
+	sprintf(BusID, "PCI:%d:%d:%d",
+		((ms->PciInfo->domain << 8) | ms->PciInfo->bus),
+		ms->PciInfo->dev, ms->PciInfo->func
+	    );
+
+	ms->fd = drmOpen(NULL, BusID);
+
+	if (ms->fd < 0)
+	    return FALSE;
+    }
+
+    if (!ms->api) {
+	ms->api = drm_api_create();
+
+	if (!ms->api)
+	    return FALSE;
+    }
+
+    return TRUE;
+}
+
+static Bool
 PreInit(ScrnInfoPtr pScrn, int flags)
 {
     xf86CrtcConfigPtr xf86_config;
@@ -213,7 +299,6 @@ PreInit(ScrnInfoPtr pScrn, int flags)
     rgb defaultWeight = { 0, 0, 0 };
     EntityInfoPtr pEnt;
     EntPtr msEnt = NULL;
-    char *BusID;
     int max_width, max_height;
 
     if (pScrn->numEntities != 1)
@@ -262,16 +347,9 @@ PreInit(ScrnInfoPtr pScrn, int flags)
 	}
     }
 
-    BusID = xalloc(64);
-    sprintf(BusID, "PCI:%d:%d:%d",
-	    ((ms->PciInfo->domain << 8) | ms->PciInfo->bus),
-	    ms->PciInfo->dev, ms->PciInfo->func
-	);
-
-    ms->api = drm_api_create();
-    ms->fd = drmOpen(NULL, BusID);
-
-    if (ms->fd < 0)
+    ms->fd = -1;
+    ms->api = NULL;
+    if (!InitDRM(pScrn))
 	return FALSE;
 
     pScrn->monitor = pScrn->confScreen->monitor;
@@ -429,7 +507,6 @@ CreateScreenResources(ScreenPtr pScreen)
     modesettingPtr ms = modesettingPTR(pScrn);
     PixmapPtr rootPixmap;
     Bool ret;
-    unsigned handle, stride;
 
     ms->noEvict = TRUE;
 
@@ -437,29 +514,14 @@ CreateScreenResources(ScreenPtr pScreen)
     ret = pScreen->CreateScreenResources(pScreen);
     pScreen->CreateScreenResources = CreateScreenResources;
 
-    rootPixmap = pScreen->GetScreenPixmap(pScreen);
-
-    xorg_exa_set_displayed_usage(rootPixmap);
-    xorg_exa_set_shared_usage(rootPixmap);
-    if (!pScreen->ModifyPixmapHeader(rootPixmap, -1, -1, -1, -1, -1, NULL))
-	FatalError("Couldn't adjust screen pixmap\n");
+    BindTextureToRoot(pScrn);
 
     ms->noEvict = FALSE;
 
-    handle = xorg_exa_get_pixmap_handle(rootPixmap, &stride);
-
-    drmModeAddFB(ms->fd,
-		 pScrn->virtualX,
-		 pScrn->virtualY,
-		 pScrn->depth,
-		 pScrn->bitsPerPixel,
-		 stride,
-		 handle,
-                 &ms->fb_id);
-
     AdjustFrame(pScrn->scrnIndex, pScrn->frameX0, pScrn->frameY0, 0);
 
 #ifdef DRM_MODE_FEATURE_DIRTYFB
+    rootPixmap = pScreen->GetScreenPixmap(pScreen);
     ms->damage = DamageCreate(NULL, NULL, DamageReportNone, TRUE,
                               pScreen, rootPixmap);
 
@@ -472,6 +534,8 @@ CreateScreenResources(ScreenPtr pScreen)
                   "Failed to create screen damage record\n");
        return FALSE;
     }
+#else
+    (void)rootPixmap;
 #endif
 
     return ret;
@@ -484,21 +548,8 @@ ScreenInit(int scrnIndex, ScreenPtr pScreen, int argc, char **argv)
     modesettingPtr ms = modesettingPTR(pScrn);
     VisualPtr visual;
 
-    /* deal with server regeneration */
-    if (ms->fd < 0) {
-	char *BusID;
-
-	BusID = xalloc(64);
-	sprintf(BusID, "PCI:%d:%d:%d",
-		((ms->PciInfo->domain << 8) | ms->PciInfo->bus),
-		ms->PciInfo->dev, ms->PciInfo->func
-	    );
-
-	ms->fd = drmOpen(NULL, BusID);
-
-	if (ms->fd < 0)
-	    return FALSE;
-    }
+    if (!InitDRM(pScrn))
+	return FALSE;
 
     if (!ms->screen) {
 	ms->screen = ms->api->create_screen(ms->api, ms->fd, NULL);
@@ -558,6 +609,9 @@ ScreenInit(int scrnIndex, ScreenPtr pScreen, int argc, char **argv)
     xf86SetBlackWhitePixels(pScreen);
 
     ms->exa = xorg_exa_init(pScrn);
+    ms->debug_fallback = debug_get_bool_option("XORG_DEBUG_FALLBACK", TRUE);
+
+    xorg_init_video(pScreen);
 
     miInitializeBackingStore(pScreen);
     xf86SetBackingStore(pScreen);
@@ -607,8 +661,8 @@ AdjustFrame(int scrnIndex, int x, int y, int flags)
     xf86CrtcPtr crtc = output->crtc;
 
     if (crtc && crtc->enabled) {
-	crtc->funcs->mode_set(crtc, pScrn->currentMode, pScrn->currentMode, x,
-			      y);
+	crtc->funcs->set_mode_major(crtc, pScrn->currentMode,
+				    RR_Rotate_0, x, y);
 	crtc->x = output->initial_x + x;
 	crtc->y = output->initial_y + y;
     }
@@ -681,8 +735,11 @@ EnterVT(int scrnIndex, int flags)
 	SaveHWState(pScrn);
     }
 
-    if (!flags)			       /* signals startup as we'll do this in CreateScreenResources */
-	CreateFrontBuffer(pScrn);
+    if (!CreateFrontBuffer(pScrn))
+	return FALSE;
+
+    if (!flags && !BindTextureToRoot(pScrn))
+	return FALSE;
 
     if (!xf86SetDesiredModes(pScrn))
 	return FALSE;
@@ -722,11 +779,15 @@ CloseScreen(int scrnIndex, ScreenPtr pScreen)
     }
 #endif
 
+    pipe_texture_reference(&ms->root_texture, NULL);
+
     if (ms->exa)
 	xorg_exa_close(pScrn);
 
+    if (ms->api && ms->api->destroy)
 	ms->api->destroy(ms->api);
-	ms->api = NULL;
+    ms->api = NULL;
+
     drmClose(ms->fd);
     ms->fd = -1;
 
diff --git a/src/gallium/state_trackers/xorg/xorg_exa.c b/src/gallium/state_trackers/xorg/xorg_exa.c
index a17a71f23a..b83d97bdb6 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa.c
+++ b/src/gallium/state_trackers/xorg/xorg_exa.c
@@ -43,10 +43,12 @@
 #include "pipe/p_state.h"
 #include "pipe/p_inlines.h"
 
-#include "cso_cache/cso_context.h"
-
 #include "util/u_rect.h"
 
+#define DEBUG_PRINT 0
+#define DEBUG_SOLID 0
+#define ACCEL_ENABLED TRUE
+
 /*
  * Helper functions
  */
@@ -72,6 +74,9 @@ exa_get_pipe_format(int depth, enum pipe_format *format, int *bbp)
 	assert(*bbp == 16);
 	break;
     case 8:
+	*format = PIPE_FORMAT_A8_UNORM;
+	assert(*bbp == 8);
+	break;
     case 4:
     case 1:
 	*format = PIPE_FORMAT_A8R8G8B8_UNORM; /* bad bad bad */
@@ -82,6 +87,17 @@ exa_get_pipe_format(int depth, enum pipe_format *format, int *bbp)
     }
 }
 
+static void
+xorg_exa_common_done(struct exa_context *exa)
+{
+   exa->copy.src = NULL;
+   exa->copy.dst = NULL;
+   exa->transform.has_src = FALSE;
+   exa->transform.has_mask = FALSE;
+   exa->has_solid_color = FALSE;
+   exa->num_bound_samplers = 0;
+}
+
 /*
  * Static exported EXA functions
  */
@@ -89,12 +105,21 @@ exa_get_pipe_format(int depth, enum pipe_format *format, int *bbp)
 static void
 ExaWaitMarker(ScreenPtr pScreen, int marker)
 {
+   ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
+   modesettingPtr ms = modesettingPTR(pScrn);
+   struct exa_context *exa = ms->exa;
+
+#if 0
+   xorg_exa_flush(exa, PIPE_FLUSH_RENDER_CACHE, NULL);
+#else
+   xorg_exa_finish(exa);
+#endif
 }
 
 static int
 ExaMarkSync(ScreenPtr pScreen)
 {
-    return 1;
+   return 1;
 }
 
 static Bool
@@ -111,9 +136,9 @@ ExaDownloadFromScreen(PixmapPtr pPix, int x,  int y, int w,  int h, char *dst,
     if (!priv || !priv->tex)
 	return FALSE;
 
-    if (exa->ctx->is_texture_referenced(exa->ctx, priv->tex, 0, 0) &
+    if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) &
 	PIPE_REFERENCED_FOR_WRITE)
-	exa->ctx->flush(exa->ctx, 0, NULL);
+	exa->pipe->flush(exa->pipe, 0, NULL);
 
     transfer = exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0,
 					   PIPE_TRANSFER_READ, x, y, w, h);
@@ -176,22 +201,33 @@ ExaPrepareAccess(PixmapPtr pPix, int index)
     if (!priv->tex)
 	return FALSE;
 
-    if (priv->map_count++ == 0)
+    if (priv->map_count == 0)
     {
-	if (exa->ctx->is_texture_referenced(exa->ctx, priv->tex, 0, 0) &
+	if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) &
 	    PIPE_REFERENCED_FOR_WRITE)
-	    exa->ctx->flush(exa->ctx, 0, NULL);
+	    exa->pipe->flush(exa->pipe, 0, NULL);
 
 	priv->map_transfer =
 	    exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0,
+#ifdef EXA_MIXED_PIXMAPS
+					PIPE_TRANSFER_MAP_DIRECTLY |
+#endif
 					PIPE_TRANSFER_READ_WRITE,
 					0, 0, priv->tex->width[0], priv->tex->height[0]);
+	if (!priv->map_transfer)
+#ifdef EXA_MIXED_PIXMAPS
+	    return FALSE;
+#else
+	    FatalError("failed to create transfer\n");
+#endif
 
 	pPix->devPrivate.ptr =
 	    exa->scrn->transfer_map(exa->scrn, priv->map_transfer);
 	pPix->devKind = priv->map_transfer->stride;
     }
 
+    priv->map_count++;
+
     return TRUE;
 }
 
@@ -231,15 +267,17 @@ ExaDone(PixmapPtr pPixmap)
     if (!priv)
 	return;
 
-    if (priv->src_surf)
-	exa->scrn->tex_surface_destroy(priv->src_surf);
-    priv->src_surf = NULL;
+    xorg_exa_common_done(exa);
 }
 
 static void
 ExaDoneComposite(PixmapPtr pPixmap)
 {
+   ScrnInfoPtr pScrn = xf86Screens[pPixmap->drawable.pScreen->myNum];
+   modesettingPtr ms = modesettingPTR(pScrn);
+   struct exa_context *exa = ms->exa;
 
+   xorg_exa_common_done(exa);
 }
 
 static Bool
@@ -250,24 +288,32 @@ ExaPrepareSolid(PixmapPtr pPixmap, int alu, Pixel planeMask, Pixel fg)
     struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pPixmap);
     struct exa_context *exa = ms->exa;
 
-    if (pPixmap->drawable.depth < 15)
-	return FALSE;
-
-    if (!EXA_PM_IS_SOLID(&pPixmap->drawable, planeMask))
-	return FALSE;
+#if DEBUG_PRINT
+    debug_printf("ExaPrepareSolid(0x%x)\n", fg);
+#endif
+    if (!exa->pipe)
+	XORG_FALLBACK("accle not enabled");
 
     if (!priv || !priv->tex)
-	return FALSE;
+	XORG_FALLBACK("%s", !priv ? "!priv" : "!priv->tex");
+
+    if (!EXA_PM_IS_SOLID(&pPixmap->drawable, planeMask))
+	XORG_FALLBACK("planeMask is not solid");
 
     if (alu != GXcopy)
-	return FALSE;
+	XORG_FALLBACK("not GXcopy");
 
-    if (!exa->ctx || !exa->ctx->surface_fill)
-	return FALSE;
+    if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format,
+                                        priv->tex->target,
+                                        PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) {
+	XORG_FALLBACK("format %s", pf_name(priv->tex->format));
+    }
 
-    priv->color = fg;
+#if DEBUG_SOLID
+    fg = 0xffff0000;
+#endif
 
-    return TRUE;
+    return ACCEL_ENABLED && xorg_solid_bind_state(exa, priv, fg);
 }
 
 static void
@@ -277,12 +323,51 @@ ExaSolid(PixmapPtr pPixmap, int x0, int y0, int x1, int y1)
     modesettingPtr ms = modesettingPTR(pScrn);
     struct exa_context *exa = ms->exa;
     struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pPixmap);
-    struct pipe_surface *surf = exa_gpu_surface(exa, priv);
 
-    exa->ctx->surface_fill(exa->ctx, surf, x0, y0, x1 - x0, y1 - y0,
-			   priv->color);
+#if DEBUG_PRINT
+    debug_printf("\tExaSolid(%d, %d, %d, %d)\n", x0, y0, x1, y1);
+#endif
 
-    exa->scrn->tex_surface_destroy(surf);
+#if 0
+    if (x0 == 0 && y0 == 0 &&
+        x1 == priv->tex->width[0] &&
+        y1 == priv->tex->height[0]) {
+       exa->ctx->clear(exa->pipe, PIPE_CLEAR_COLOR,
+                       exa->solid_color, 1., 0);
+    } else
+#endif
+
+#if DEBUG_SOLID
+       exa->solid_color[0] = 0.f;
+       exa->solid_color[1] = 1.f;
+       exa->solid_color[2] = 0.f;
+       exa->solid_color[3] = 1.f;
+    xorg_solid(exa, priv, 0, 0, 1024, 768);
+       exa->solid_color[0] = 1.f;
+       exa->solid_color[1] = 0.f;
+       exa->solid_color[2] = 0.f;
+       exa->solid_color[3] = 1.f;
+       xorg_solid(exa, priv, 0, 0, 300, 300);
+       xorg_solid(exa, priv, 300, 300, 350, 350);
+       xorg_solid(exa, priv, 350, 350, 500, 500);
+
+       xorg_solid(exa, priv,
+               priv->tex->width[0] - 10,
+               priv->tex->height[0] - 10,
+               priv->tex->width[0],
+               priv->tex->height[0]);
+
+    exa->solid_color[0] = 0.f;
+    exa->solid_color[1] = 0.f;
+    exa->solid_color[2] = 1.f;
+    exa->solid_color[3] = 1.f;
+
+    exa->has_solid_color = FALSE;
+    ExaPrepareCopy(pPixmap, pPixmap, 0, 0, GXcopy, 0xffffffff);
+    ExaCopy(pPixmap, 0, 0, 50, 50, 500, 500);
+#else
+    xorg_solid(exa, priv, x0, y0, x1, y1) ;
+#endif
 }
 
 static Bool
@@ -295,42 +380,59 @@ ExaPrepareCopy(PixmapPtr pSrcPixmap, PixmapPtr pDstPixmap, int xdir,
     struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDstPixmap);
     struct exa_pixmap_priv *src_priv = exaGetPixmapDriverPrivate(pSrcPixmap);
 
-    if (alu != GXcopy)
-	return FALSE;
+#if DEBUG_PRINT
+    debug_printf("ExaPrepareCopy\n");
+#endif
+    if (!exa->pipe)
+	XORG_FALLBACK("accle not enabled");
 
-    if (pSrcPixmap->drawable.depth < 15 || pDstPixmap->drawable.depth < 15)
-	return FALSE;
+    if (!priv || !priv->tex)
+	XORG_FALLBACK("pDst %s", !priv ? "!priv" : "!priv->tex");
+
+    if (!src_priv || !src_priv->tex)
+	XORG_FALLBACK("pSrc %s", !src_priv ? "!priv" : "!priv->tex");
 
     if (!EXA_PM_IS_SOLID(&pSrcPixmap->drawable, planeMask))
-	return FALSE;
+	XORG_FALLBACK("planeMask is not solid");
 
-    if (!priv || !src_priv)
-	return FALSE;
+    if (alu != GXcopy)
+	XORG_FALLBACK("alu not GXcopy");
 
-    if (!priv->tex || !src_priv->tex)
-	return FALSE;
+    if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format,
+                                        priv->tex->target,
+                                        PIPE_TEXTURE_USAGE_RENDER_TARGET, 0))
+	XORG_FALLBACK("pDst format %s", pf_name(priv->tex->format));
 
-    if (!exa->ctx || !exa->ctx->surface_copy)
-	return FALSE;
+    if (!exa->scrn->is_format_supported(exa->scrn, src_priv->tex->format,
+                                        src_priv->tex->target,
+                                        PIPE_TEXTURE_USAGE_SAMPLER, 0))
+	XORG_FALLBACK("pSrc format %s", pf_name(src_priv->tex->format));
 
-    priv->src_surf = exa_gpu_surface(exa, src_priv);
+    exa->copy.src = src_priv;
+    exa->copy.dst = priv;
 
-    return TRUE;
+    return ACCEL_ENABLED;
 }
 
 static void
 ExaCopy(PixmapPtr pDstPixmap, int srcX, int srcY, int dstX, int dstY,
 	int width, int height)
 {
-    ScrnInfoPtr pScrn = xf86Screens[pDstPixmap->drawable.pScreen->myNum];
-    modesettingPtr ms = modesettingPTR(pScrn);
-    struct exa_context *exa = ms->exa;
-    struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDstPixmap);
-    struct pipe_surface *surf = exa_gpu_surface(exa, priv);
+   ScrnInfoPtr pScrn = xf86Screens[pDstPixmap->drawable.pScreen->myNum];
+   modesettingPtr ms = modesettingPTR(pScrn);
+   struct exa_context *exa = ms->exa;
+   struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDstPixmap);
+
+#if DEBUG_PRINT
+   debug_printf("\tExaCopy(srcx=%d, srcy=%d, dstX=%d, dstY=%d, w=%d, h=%d)\n",
+                srcX, srcY, dstX, dstY, width, height);
+#endif
+
+   debug_assert(priv == exa->copy.dst);
 
-    exa->ctx->surface_copy(exa->ctx, surf, dstX, dstY, priv->src_surf,
-			   srcX, srcY, width, height);
-    exa->scrn->tex_surface_destroy(surf);
+   renderer_copy_pixmap(exa->renderer, exa->copy.dst, dstX, dstY,
+                        exa->copy.src, srcX, srcY,
+                        width, height);
 }
 
 static Bool
@@ -341,11 +443,51 @@ ExaPrepareComposite(int op, PicturePtr pSrcPicture,
    ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
    modesettingPtr ms = modesettingPTR(pScrn);
    struct exa_context *exa = ms->exa;
+   struct exa_pixmap_priv *priv;
 
-   return xorg_composite_bind_state(exa, op, pSrcPicture, pMaskPicture,
+#if DEBUG_PRINT
+   debug_printf("ExaPrepareComposite(%d, src=0x%p, mask=0x%p, dst=0x%p)\n",
+                op, pSrcPicture, pMaskPicture, pDstPicture);
+#endif
+   if (!exa->pipe)
+      XORG_FALLBACK("accle not enabled");
+
+   priv = exaGetPixmapDriverPrivate(pDst);
+   if (!priv || !priv->tex)
+      XORG_FALLBACK("pDst %s", !priv ? "!priv" : "!priv->tex");
+
+   if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format,
+                                       priv->tex->target,
+                                       PIPE_TEXTURE_USAGE_RENDER_TARGET, 0))
+      XORG_FALLBACK("pDst format: %s", pf_name(priv->tex->format));
+
+   if (pSrc) {
+      priv = exaGetPixmapDriverPrivate(pSrc);
+      if (!priv || !priv->tex)
+         XORG_FALLBACK("pSrc %s", !priv ? "!priv" : "!priv->tex");
+
+      if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format,
+                                          priv->tex->target,
+                                          PIPE_TEXTURE_USAGE_SAMPLER, 0))
+         XORG_FALLBACK("pSrc format: %s", pf_name(priv->tex->format));
+   }
+
+   if (pMask) {
+      priv = exaGetPixmapDriverPrivate(pMask);
+      if (!priv || !priv->tex)
+         XORG_FALLBACK("pMask %s", !priv ? "!priv" : "!priv->tex");
+
+      if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format,
+                                          priv->tex->target,
+                                          PIPE_TEXTURE_USAGE_SAMPLER, 0))
+         XORG_FALLBACK("pMask format: %s", pf_name(priv->tex->format));
+   }
+
+   return ACCEL_ENABLED &&
+          xorg_composite_bind_state(exa, op, pSrcPicture, pMaskPicture,
                                     pDstPicture,
-                                    exaGetPixmapDriverPrivate(pSrc),
-                                    exaGetPixmapDriverPrivate(pMask),
+                                    pSrc ? exaGetPixmapDriverPrivate(pSrc) : NULL,
+                                    pMask ? exaGetPixmapDriverPrivate(pMask) : NULL,
                                     exaGetPixmapDriverPrivate(pDst));
 }
 
@@ -358,6 +500,10 @@ ExaComposite(PixmapPtr pDst, int srcX, int srcY, int maskX, int maskY,
    struct exa_context *exa = ms->exa;
    struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDst);
 
+#if DEBUG_PRINT
+   debug_printf("\tExaComposite(src[%d,%d], mask=[%d, %d], dst=[%d, %d], dim=[%d, %d])\n", srcX, srcY, maskX, maskY, dstX, dstY, width, height);
+#endif
+
    xorg_composite(exa, priv, srcX, srcY, maskX, maskY,
                   dstX, dstY, width, height);
 }
@@ -367,10 +513,15 @@ ExaCheckComposite(int op,
 		  PicturePtr pSrcPicture, PicturePtr pMaskPicture,
 		  PicturePtr pDstPicture)
 {
-   return xorg_composite_accelerated(op,
-                                     pSrcPicture,
-                                     pMaskPicture,
-                                     pDstPicture);
+   boolean accelerated = xorg_composite_accelerated(op,
+                                                    pSrcPicture,
+                                                    pMaskPicture,
+                                                    pDstPicture);
+#if DEBUG_PRINT
+   debug_printf("ExaCheckComposite(%d, %p, %p, %p) = %d\n",
+                op, pSrcPicture, pMaskPicture, pDstPicture, accelerated);
+#endif
+   return ACCEL_ENABLED && accelerated;
 }
 
 static void *
@@ -389,14 +540,11 @@ static void
 ExaDestroyPixmap(ScreenPtr pScreen, void *dPriv)
 {
     struct exa_pixmap_priv *priv = (struct exa_pixmap_priv *)dPriv;
-    ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
-    modesettingPtr ms = modesettingPTR(pScrn);
 
     if (!priv)
 	return;
 
-    if (priv->tex)
-	ms->screen->texture_destroy(priv->tex);
+    pipe_texture_reference(&priv->tex, NULL);
 
     xfree(priv);
 }
@@ -516,64 +664,43 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height,
          priv->tex->height[0] != height ||
          priv->tex_flags != priv->flags)) {
 	struct pipe_texture *texture = NULL;
-
-#ifdef DRM_MODE_FEATURE_DIRTYFB
-	if (priv->flags)
-#endif
-	{
-	    struct pipe_texture template;
-
-	    memset(&template, 0, sizeof(template));
-	    template.target = PIPE_TEXTURE_2D;
-	    exa_get_pipe_format(depth, &template.format, &bitsPerPixel);
-	    pf_get_block(template.format, &template.block);
-	    template.width[0] = width;
-	    template.height[0] = height;
-	    template.depth[0] = 1;
-	    template.last_level = 0;
-	    template.tex_usage = PIPE_TEXTURE_USAGE_RENDER_TARGET | priv->flags;
-	    priv->tex_flags = priv->flags;
-	    texture = exa->scrn->texture_create(exa->scrn, &template);
-
-	    if (priv->tex) {
-		struct pipe_surface *dst_surf;
-
-		dst_surf = exa->scrn->get_tex_surface(exa->scrn, texture, 0, 0, 0,
-						      PIPE_BUFFER_USAGE_GPU_WRITE);
-		priv->src_surf = exa_gpu_surface(exa, priv);
-		exa->ctx->surface_copy(exa->ctx, dst_surf, 0, 0, priv->src_surf,
-				       0, 0, min(width, texture->width[0]),
-				       min(height, texture->height[0]));
-		exa->scrn->tex_surface_destroy(dst_surf);
-		exa->scrn->tex_surface_destroy(priv->src_surf);
-		priv->src_surf = NULL;
-	    } else if (pPixmap->devPrivate.ptr) {
-		struct pipe_transfer *transfer;
-
-		if (priv->map_count != 0)
-		     FatalError("doing ExaModifyPixmapHeader on mapped buffer\n");
-
-		transfer =
-		    exa->scrn->get_tex_transfer(exa->scrn, texture, 0, 0, 0,
-						PIPE_TRANSFER_WRITE,
-						0, 0, width, height);
-		util_copy_rect(exa->scrn->transfer_map(exa->scrn, transfer),
-			       &texture->block, transfer->stride, 0, 0,
-			       width, height, pPixmap->devPrivate.ptr,
-			       pPixmap->devKind, 0, 0);
-		exa->scrn->transfer_unmap(exa->scrn, transfer);
-		exa->scrn->tex_transfer_destroy(transfer);
-	    }
-	}
-#ifdef DRM_MODE_FEATURE_DIRTYFB
-	else {
-	    xfree(pPixmap->devPrivate.ptr);
-	    pPixmap->devPrivate.ptr = xalloc(pPixmap->drawable.height *
-					     pPixmap->devKind);
+	struct pipe_texture template;
+
+	memset(&template, 0, sizeof(template));
+	template.target = PIPE_TEXTURE_2D;
+	exa_get_pipe_format(depth, &template.format, &bitsPerPixel);
+	pf_get_block(template.format, &template.block);
+	template.width[0] = width;
+	template.height[0] = height;
+	template.depth[0] = 1;
+	template.last_level = 0;
+	template.tex_usage = PIPE_TEXTURE_USAGE_RENDER_TARGET | priv->flags;
+	priv->tex_flags = priv->flags;
+	texture = exa->scrn->texture_create(exa->scrn, &template);
+
+	if (priv->tex) {
+	    struct pipe_surface *dst_surf;
+	    struct pipe_surface *src_surf;
+
+	    dst_surf = exa->scrn->get_tex_surface(
+		exa->scrn, texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE);
+	    src_surf = xorg_gpu_surface(exa->pipe->screen, priv);
+        if (exa->pipe->surface_copy) {
+            exa->pipe->surface_copy(exa->pipe, dst_surf, 0, 0, src_surf,
+                        0, 0, min(width, texture->width[0]),
+                        min(height, texture->height[0]));
+        } else {
+            util_surface_copy(exa->pipe, FALSE, dst_surf, 0, 0, src_surf,
+                        0, 0, min(width, texture->width[0]),
+                        min(height, texture->height[0]));
+        }
+	    exa->scrn->tex_surface_destroy(dst_surf);
+	    exa->scrn->tex_surface_destroy(src_surf);
 	}
-#endif
 
 	pipe_texture_reference(&priv->tex, texture);
+	/* the texture we create has one reference */
+	pipe_texture_reference(&texture, NULL);
     }
 
     return TRUE;
@@ -588,31 +715,60 @@ xorg_exa_get_texture(PixmapPtr pPixmap)
    return tex;
 }
 
+Bool
+xorg_exa_set_texture(PixmapPtr pPixmap, struct  pipe_texture *tex)
+{
+    struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pPixmap);
+
+    int mask = PIPE_TEXTURE_USAGE_PRIMARY | PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
+
+    if (!priv)
+	return FALSE;
+
+    if (pPixmap->drawable.width != tex->width[0] ||
+	pPixmap->drawable.height != tex->height[0])
+	return FALSE;
+
+    pipe_texture_reference(&priv->tex, tex);
+    priv->tex_flags = tex->tex_usage & mask;
+
+    return TRUE;
+}
+
+struct pipe_texture *
+xorg_exa_create_root_texture(ScrnInfoPtr pScrn,
+			     int width, int height,
+			     int depth, int bitsPerPixel)
+{
+    modesettingPtr ms = modesettingPTR(pScrn);
+    struct exa_context *exa = ms->exa;
+    struct pipe_texture template;
+
+    memset(&template, 0, sizeof(template));
+    template.target = PIPE_TEXTURE_2D;
+    exa_get_pipe_format(depth, &template.format, &bitsPerPixel);
+    pf_get_block(template.format, &template.block);
+    template.width[0] = width;
+    template.height[0] = height;
+    template.depth[0] = 1;
+    template.last_level = 0;
+    template.tex_usage |= PIPE_TEXTURE_USAGE_RENDER_TARGET;
+    template.tex_usage |= PIPE_TEXTURE_USAGE_PRIMARY;
+    template.tex_usage |= PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
+
+    return exa->scrn->texture_create(exa->scrn, &template);
+}
+
 void
 xorg_exa_close(ScrnInfoPtr pScrn)
 {
    modesettingPtr ms = modesettingPTR(pScrn);
    struct exa_context *exa = ms->exa;
-   struct pipe_constant_buffer *vsbuf = &exa->vs_const_buffer;
-   struct pipe_constant_buffer *fsbuf = &exa->fs_const_buffer;
-
-   if (exa->shaders) {
-      xorg_shaders_destroy(exa->shaders);
-   }
 
-   if (vsbuf && vsbuf->buffer)
-      pipe_buffer_reference(&vsbuf->buffer, NULL);
+   renderer_destroy(exa->renderer);
 
-   if (fsbuf && fsbuf->buffer)
-      pipe_buffer_reference(&fsbuf->buffer, NULL);
-
-   if (exa->cso) {
-      cso_release_all(exa->cso);
-      cso_destroy_context(exa->cso);
-   }
-
-   if (exa->ctx)
-      exa->ctx->destroy(exa->ctx);
+   if (exa->pipe)
+      exa->pipe->destroy(exa->pipe);
 
    exaDriverFini(pScrn->pScreen);
    xfree(exa);
@@ -645,6 +801,12 @@ xorg_exa_init(ScrnInfoPtr pScrn)
    pExa->pixmapOffsetAlign = 0;
    pExa->pixmapPitchAlign  = 1;
    pExa->flags             = EXA_OFFSCREEN_PIXMAPS | EXA_HANDLES_PIXMAPS;
+#ifdef EXA_SUPPORTS_PREPARE_AUX
+   pExa->flags            |= EXA_SUPPORTS_PREPARE_AUX;
+#endif
+#ifdef EXA_MIXED_PIXMAPS
+   pExa->flags            |= EXA_MIXED_PIXMAPS;
+#endif
    pExa->maxX              = 8191; /* FIXME */
    pExa->maxY              = 8191; /* FIXME */
 
@@ -674,12 +836,11 @@ xorg_exa_init(ScrnInfoPtr pScrn)
    }
 
    exa->scrn = ms->screen;
-   exa->ctx = ms->api->create_context(ms->api, exa->scrn);
+   exa->pipe = ms->api->create_context(ms->api, exa->scrn);
    /* Share context with DRI */
-   ms->ctx = exa->ctx;
+   ms->ctx = exa->pipe;
 
-   exa->cso = cso_create_context(exa->ctx);
-   exa->shaders = xorg_shaders_create(exa);
+   exa->renderer = renderer_create(exa->pipe);
 
    return (void *)exa;
 
@@ -690,11 +851,27 @@ out_err:
 }
 
 struct pipe_surface *
-exa_gpu_surface(struct exa_context *exa, struct exa_pixmap_priv *priv)
+xorg_gpu_surface(struct pipe_screen *scrn, struct exa_pixmap_priv *priv)
 {
-   return exa->scrn->get_tex_surface(exa->scrn, priv->tex, 0, 0, 0,
-                                     PIPE_BUFFER_USAGE_GPU_READ |
-                                     PIPE_BUFFER_USAGE_GPU_WRITE);
+   return scrn->get_tex_surface(scrn, priv->tex, 0, 0, 0,
+                                PIPE_BUFFER_USAGE_GPU_READ |
+                                PIPE_BUFFER_USAGE_GPU_WRITE);
+
+}
+
+void xorg_exa_flush(struct exa_context *exa, uint pipeFlushFlags,
+                    struct pipe_fence_handle **fence)
+{
+   exa->pipe->flush(exa->pipe, pipeFlushFlags, fence);
+}
+
+void xorg_exa_finish(struct exa_context *exa)
+{
+   struct pipe_fence_handle *fence = NULL;
+
+   xorg_exa_flush(exa, PIPE_FLUSH_RENDER_CACHE, &fence);
 
+   exa->pipe->screen->fence_finish(exa->pipe->screen, fence, 0);
+   exa->pipe->screen->fence_reference(exa->pipe->screen, &fence, NULL);
 }
 
diff --git a/src/gallium/state_trackers/xorg/xorg_exa.h b/src/gallium/state_trackers/xorg/xorg_exa.h
index 5b515be139..45f88d9404 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa.h
+++ b/src/gallium/state_trackers/xorg/xorg_exa.h
@@ -14,20 +14,29 @@ struct xorg_shaders;
 struct exa_context
 {
    ExaDriverPtr pExa;
-   struct pipe_context *ctx;
+   struct pipe_context *pipe;
    struct pipe_screen *scrn;
-   struct cso_context *cso;
-   struct xorg_shaders *shaders;
-
-   struct pipe_constant_buffer vs_const_buffer;
-   struct pipe_constant_buffer fs_const_buffer;
+   struct xorg_renderer *renderer;
 
    struct pipe_texture *bound_textures[MAX_EXA_SAMPLERS];
    int num_bound_samplers;
 
    float solid_color[4];
-};
+   boolean has_solid_color;
 
+   /* float[9] projective matrix bound to pictures */
+   struct {
+      float    src[9];
+      float   mask[9];
+      boolean has_src;
+      boolean has_mask;
+   } transform;
+
+   struct {
+      struct exa_pixmap_priv *src;
+      struct exa_pixmap_priv *dst;
+   } copy;
+};
 
 struct exa_pixmap_priv
 {
@@ -36,15 +45,25 @@ struct exa_pixmap_priv
 
    struct pipe_texture *tex;
    struct pipe_texture *depth_stencil_tex;
-   unsigned int color;
-   struct pipe_surface *src_surf; /* for copies */
 
    struct pipe_transfer *map_transfer;
    unsigned map_count;
 };
 
+#define XORG_FALLBACK(s, arg...)                              \
+do {                                                          \
+   if (ms->debug_fallback) {                                  \
+      xf86DrvMsg(pScrn->scrnIndex, X_INFO,                    \
+                 "%s fallback " s "\n", __FUNCTION__, ##arg); \
+   }                                                          \
+   return FALSE;                                              \
+} while(0)
+
 struct pipe_surface *
-exa_gpu_surface(struct exa_context *exa, struct exa_pixmap_priv *priv);
+xorg_gpu_surface(struct pipe_screen *scrn, struct exa_pixmap_priv *priv);
 
+void xorg_exa_flush(struct exa_context *exa, uint pipeFlushFlags,
+                    struct pipe_fence_handle **fence);
+void xorg_exa_finish(struct exa_context *exa);
 
 #endif
diff --git a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c
index cfee10c3b3..041f4f96dc 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c
+++ b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c
@@ -44,33 +44,26 @@
  */
 
 struct xorg_shaders {
-   struct exa_context *exa;
+   struct xorg_renderer *r;
 
    struct cso_hash *vs_hash;
    struct cso_hash *fs_hash;
 };
 
-static const char over_op[] =
-   "SUB TEMP[3], CONST[0].wwww, TEMP[1].wwww\n"
-   "MUL TEMP[3], TEMP[0], TEMP[3]\n"
-   "ADD TEMP[0], TEMP[3], TEMP[0]\n";
-
-
-static INLINE void
-create_preamble(struct ureg_program *ureg)
-{
-}
-
-
 static INLINE void
 src_in_mask(struct ureg_program *ureg,
             struct ureg_dst dst,
             struct ureg_src src,
-            struct ureg_src mask)
+            struct ureg_src mask,
+            boolean component_alpha)
 {
-   /* MUL dst, src, mask.wwww */
-   ureg_MUL(ureg, dst, src,
-            ureg_scalar(mask, TGSI_SWIZZLE_W));
+   if (component_alpha) {
+      ureg_MUL(ureg, dst, src, mask);
+   }
+   else {
+      ureg_MUL(ureg, dst, src,
+               ureg_scalar(mask, TGSI_SWIZZLE_W));
+   }
 }
 
 static struct ureg_src
@@ -79,8 +72,7 @@ vs_normalize_coords(struct ureg_program *ureg, struct ureg_src coords,
 {
    struct ureg_dst tmp = ureg_DECL_temporary(ureg);
    struct ureg_src ret;
-   ureg_MUL(ureg, tmp, coords, const0);
-   ureg_ADD(ureg, tmp, ureg_src(tmp), const1);
+   ureg_MAD(ureg, tmp, coords, const0, const1);
    ret = ureg_src(tmp);
    ureg_release_temporary(ureg, tmp);
    return ret;
@@ -241,42 +233,39 @@ create_vs(struct pipe_context *pipe,
    boolean is_fill = vs_traits & VS_FILL;
    boolean is_composite = vs_traits & VS_COMPOSITE;
    boolean has_mask = vs_traits & VS_MASK;
+   unsigned input_slot = 0;
 
    ureg = ureg_create(TGSI_PROCESSOR_VERTEX);
    if (ureg == NULL)
       return 0;
 
-   const0 = ureg_DECL_constant(ureg);
-   const1 = ureg_DECL_constant(ureg);
+   const0 = ureg_DECL_constant(ureg, 0);
+   const1 = ureg_DECL_constant(ureg, 1);
 
    /* it has to be either a fill or a composite op */
    debug_assert(is_fill ^ is_composite);
 
-   src = ureg_DECL_vs_input(ureg,
-                            TGSI_SEMANTIC_POSITION, 0);
+   src = ureg_DECL_vs_input(ureg, input_slot++);
    dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_POSITION, 0);
    src = vs_normalize_coords(ureg, src,
                              const0, const1);
    ureg_MOV(ureg, dst, src);
 
-
    if (is_composite) {
-      src = ureg_DECL_vs_input(ureg,
-                               TGSI_SEMANTIC_GENERIC, 1);
-      dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 1);
+      src = ureg_DECL_vs_input(ureg, input_slot++);
+      dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 0);
       ureg_MOV(ureg, dst, src);
    }
+
    if (is_fill) {
-      src = ureg_DECL_vs_input(ureg,
-                               TGSI_SEMANTIC_COLOR, 1);
-      dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, 1);
+      src = ureg_DECL_vs_input(ureg, input_slot++);
+      dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, 0);
       ureg_MOV(ureg, dst, src);
    }
 
    if (has_mask) {
-      src = ureg_DECL_vs_input(ureg,
-                               TGSI_SEMANTIC_GENERIC, 2);
-      dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_POSITION, 2);
+      src = ureg_DECL_vs_input(ureg, input_slot++);
+      dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 1);
       ureg_MOV(ureg, dst, src);
    }
 
@@ -300,6 +289,7 @@ create_fs(struct pipe_context *pipe,
    boolean is_solid   = fs_traits & FS_SOLID_FILL;
    boolean is_lingrad = fs_traits & FS_LINGRAD_FILL;
    boolean is_radgrad = fs_traits & FS_RADGRAD_FILL;
+   boolean is_comp_alpha = fs_traits & FS_COMPONENT_ALPHA;
 
    ureg = ureg_create(TGSI_PROCESSOR_FRAGMENT);
    if (ureg == NULL)
@@ -315,11 +305,11 @@ create_fs(struct pipe_context *pipe,
    if (is_composite) {
       src_sampler = ureg_DECL_sampler(ureg, 0);
       src_input = ureg_DECL_fs_input(ureg,
-                                     TGSI_SEMANTIC_POSITION,
+                                     TGSI_SEMANTIC_GENERIC,
                                      0,
                                      TGSI_INTERPOLATE_PERSPECTIVE);
-   }
-   if (is_fill) {
+   } else {
+      debug_assert(is_fill);
       if (is_solid)
          src_input = ureg_DECL_fs_input(ureg,
                                         TGSI_SEMANTIC_COLOR,
@@ -335,7 +325,7 @@ create_fs(struct pipe_context *pipe,
    if (has_mask) {
       mask_sampler = ureg_DECL_sampler(ureg, 1);
       mask_pos = ureg_DECL_fs_input(ureg,
-                                    TGSI_SEMANTIC_POSITION,
+                                    TGSI_SEMANTIC_GENERIC,
                                     1,
                                     TGSI_INTERPOLATE_PERSPECTIVE);
    }
@@ -370,11 +360,11 @@ create_fs(struct pipe_context *pipe,
          else
             src = out;
 
-         coords = ureg_DECL_constant(ureg);
-         const0124 = ureg_DECL_constant(ureg);
-         matrow0 = ureg_DECL_constant(ureg);
-         matrow1 = ureg_DECL_constant(ureg);
-         matrow2 = ureg_DECL_constant(ureg);
+         coords = ureg_DECL_constant(ureg, 0);
+         const0124 = ureg_DECL_constant(ureg, 1);
+         matrow0 = ureg_DECL_constant(ureg, 2);
+         matrow1 = ureg_DECL_constant(ureg, 3);
+         matrow2 = ureg_DECL_constant(ureg, 4);
 
          if (is_lingrad) {
             linear_gradient(ureg, src,
@@ -396,7 +386,7 @@ create_fs(struct pipe_context *pipe,
       ureg_TEX(ureg, mask,
                TGSI_TEXTURE_2D, mask_pos, mask_sampler);
       /* src IN mask */
-      src_in_mask(ureg, out, ureg_src(src), ureg_src(mask));
+      src_in_mask(ureg, out, ureg_src(src), ureg_src(mask), is_comp_alpha);
       ureg_release_temporary(ureg, mask);
    }
 
@@ -405,11 +395,11 @@ create_fs(struct pipe_context *pipe,
    return ureg_create_shader_and_destroy(ureg, pipe);
 }
 
-struct xorg_shaders * xorg_shaders_create(struct exa_context *exa)
+struct xorg_shaders * xorg_shaders_create(struct xorg_renderer *r)
 {
    struct xorg_shaders *sc = CALLOC_STRUCT(xorg_shaders);
 
-   sc->exa = exa;
+   sc->r = r;
    sc->vs_hash = cso_hash_create();
    sc->fs_hash = cso_hash_create();
 
@@ -436,9 +426,9 @@ cache_destroy(struct cso_context *cso,
 
 void xorg_shaders_destroy(struct xorg_shaders *sc)
 {
-   cache_destroy(sc->exa->cso, sc->vs_hash,
+   cache_destroy(sc->r->cso, sc->vs_hash,
                  PIPE_SHADER_VERTEX);
-   cache_destroy(sc->exa->cso, sc->fs_hash,
+   cache_destroy(sc->r->cso, sc->fs_hash,
                  PIPE_SHADER_FRAGMENT);
 
    free(sc);
@@ -470,12 +460,12 @@ struct xorg_shader xorg_shaders_get(struct xorg_shaders *sc,
                                     unsigned vs_traits,
                                     unsigned fs_traits)
 {
-   struct xorg_shader shader = {0};
+   struct xorg_shader shader = { NULL, NULL };
    void *vs, *fs;
 
-   vs = shader_from_cache(sc->exa->ctx, PIPE_SHADER_VERTEX,
+   vs = shader_from_cache(sc->r->pipe, PIPE_SHADER_VERTEX,
                           sc->vs_hash, vs_traits);
-   fs = shader_from_cache(sc->exa->ctx, PIPE_SHADER_FRAGMENT,
+   fs = shader_from_cache(sc->r->pipe, PIPE_SHADER_FRAGMENT,
                           sc->fs_hash, fs_traits);
 
    debug_assert(vs && fs);
diff --git a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.h b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.h
index 1535a0c8c3..c290d44e8f 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.h
+++ b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.h
@@ -1,7 +1,7 @@
 #ifndef XORG_EXA_TGSI_H
 #define XORG_EXA_TGSI_H
 
-#include "xorg_exa.h"
+#include "xorg_renderer.h"
 
 enum xorg_vs_traits {
    VS_COMPOSITE        = 1 << 0,
@@ -23,7 +23,8 @@ enum xorg_fs_traits {
    FS_RADGRAD_FILL     = 1 << 4,
    FS_FILL             = (FS_SOLID_FILL |
                           FS_LINGRAD_FILL |
-                          FS_RADGRAD_FILL)
+                          FS_RADGRAD_FILL),
+   FS_COMPONENT_ALPHA  = 1 << 5
 };
 
 struct xorg_shader {
@@ -33,7 +34,7 @@ struct xorg_shader {
 
 struct xorg_shaders;
 
-struct xorg_shaders *xorg_shaders_create(struct exa_context *exa);
+struct xorg_shaders *xorg_shaders_create(struct xorg_renderer *renderer);
 void xorg_shaders_destroy(struct xorg_shaders *shaders);
 
 struct xorg_shader xorg_shaders_get(struct xorg_shaders *shaders,
diff --git a/src/gallium/state_trackers/xorg/xorg_output.c b/src/gallium/state_trackers/xorg/xorg_output.c
index 950af942f5..bfeddc5e11 100644
--- a/src/gallium/state_trackers/xorg/xorg_output.c
+++ b/src/gallium/state_trackers/xorg/xorg_output.c
@@ -42,8 +42,12 @@
 #include <sys/stat.h>
 #include <sys/types.h>
 
+#ifdef HAVE_XEXTPROTO_71
+#include <X11/extensions/dpmsconst.h>
+#else
 #define DPMS_SERVER
 #include <X11/extensions/dpms.h>
+#endif
 
 #include "X11/Xatom.h"
 
@@ -52,66 +56,29 @@
 static char *connector_enum_list[] = {
     "Unknown",
     "VGA",
-    "DVI-I",
-    "DVI-D",
-    "DVI-A",
+    "DVI",
+    "DVI",
+    "DVI",
     "Composite",
     "SVIDEO",
     "LVDS",
-    "Component",
-    "9-pin DIN",
-    "DisplayPort",
-    "HDMI Type A",
-    "HDMI Type B",
+    "CTV",
+    "DIN",
+    "DP",
+    "HDMI",
+    "HDMI",
 };
 
 static void
-dpms(xf86OutputPtr output, int mode)
-{
-}
-
-static void
-save(xf86OutputPtr output)
-{
-}
-
-static void
-restore(xf86OutputPtr output)
-{
-}
-
-static int
-mode_valid(xf86OutputPtr output, DisplayModePtr pMode)
-{
-    return MODE_OK;
-}
-
-static Bool
-mode_fixup(xf86OutputPtr output, DisplayModePtr mode,
-	   DisplayModePtr adjusted_mode)
-{
-    return TRUE;
-}
-
-static void
-prepare(xf86OutputPtr output)
-{
-    dpms(output, DPMSModeOff);
-}
-
-static void
-mode_set(xf86OutputPtr output, DisplayModePtr mode,
-	 DisplayModePtr adjusted_mode)
+create_resources(xf86OutputPtr output)
 {
+#ifdef RANDR_12_INTERFACE
+#endif /* RANDR_12_INTERFACE */
 }
 
 static void
-commit(xf86OutputPtr output)
+dpms(xf86OutputPtr output, int mode)
 {
-    dpms(output, DPMSModeOn);
-
-    if (output->scrn->pScreen != NULL)
-	xf86_reload_cursors(output->scrn->pScreen);
 }
 
 static xf86OutputStatus
@@ -167,17 +134,10 @@ get_modes(xf86OutputPtr output)
     return modes;
 }
 
-static void
-destroy(xf86OutputPtr output)
-{
-    drmModeFreeConnector(output->driver_private);
-}
-
-static void
-create_resources(xf86OutputPtr output)
+static int
+mode_valid(xf86OutputPtr output, DisplayModePtr pMode)
 {
-#ifdef RANDR_12_INTERFACE
-#endif /* RANDR_12_INTERFACE */
+    return MODE_OK;
 }
 
 #ifdef RANDR_12_INTERFACE
@@ -196,36 +156,26 @@ get_property(xf86OutputPtr output, Atom property)
 }
 #endif /* RANDR_13_INTERFACE */
 
-#ifdef RANDR_GET_CRTC_INTERFACE
-static xf86CrtcPtr
-get_crtc(xf86OutputPtr output)
+static void
+destroy(xf86OutputPtr output)
 {
-    return NULL;
+    drmModeFreeConnector(output->driver_private);
 }
-#endif
 
 static const xf86OutputFuncsRec output_funcs = {
     .create_resources = create_resources,
-    .dpms = dpms,
-    .save = save,
-    .restore = restore,
-    .mode_valid = mode_valid,
-    .mode_fixup = mode_fixup,
-    .prepare = prepare,
-    .mode_set = mode_set,
-    .commit = commit,
-    .detect = detect,
-    .get_modes = get_modes,
 #ifdef RANDR_12_INTERFACE
     .set_property = set_property,
 #endif
 #ifdef RANDR_13_INTERFACE
     .get_property = get_property,
 #endif
+    .dpms = dpms,
+    .detect = detect,
+
+    .get_modes = get_modes,
+    .mode_valid = mode_valid,
     .destroy = destroy,
-#ifdef RANDR_GET_CRTC_INTERFACE
-    .get_crtc = get_crtc,
-#endif
 };
 
 void
@@ -236,7 +186,7 @@ output_init(ScrnInfoPtr pScrn)
     drmModeResPtr res;
     drmModeConnectorPtr drm_connector = NULL;
     drmModeEncoderPtr drm_encoder = NULL;
-    char *name;
+    char name[32];
     int c, v, p;
 
     res = drmModeGetResources(ms->fd);
@@ -269,7 +219,10 @@ output_init(ScrnInfoPtr pScrn)
 	(void)v;
 #endif
 
-	name = connector_enum_list[drm_connector->connector_type];
+	snprintf(name, 32, "%s%d",
+		 connector_enum_list[drm_connector->connector_type],
+		 drm_connector->connector_type_id);
+
 
 	output = xf86OutputCreate(pScrn, &output_funcs, name);
 	if (!output)
diff --git a/src/gallium/state_trackers/xorg/xorg_renderer.c b/src/gallium/state_trackers/xorg/xorg_renderer.c
new file mode 100644
index 0000000000..393f3fac3e
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xorg_renderer.c
@@ -0,0 +1,841 @@
+#include "xorg_exa.h"
+#include "xorg_renderer.h"
+
+#include "xorg_exa_tgsi.h"
+
+#include "cso_cache/cso_context.h"
+#include "util/u_draw_quad.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "util/u_rect.h"
+
+#include "pipe/p_inlines.h"
+
+#include <math.h>
+
+enum AxisOrientation {
+   Y0_BOTTOM,
+   Y0_TOP
+};
+
+#define floatsEqual(x, y) (fabs(x - y) <= 0.00001f * MIN2(fabs(x), fabs(y)))
+#define floatIsZero(x) (floatsEqual((x) + 1, 1))
+
+static INLINE boolean is_affine(float *matrix)
+{
+   return floatIsZero(matrix[2]) && floatIsZero(matrix[5])
+      && floatsEqual(matrix[8], 1);
+}
+static INLINE void map_point(float *mat, float x, float y,
+                             float *out_x, float *out_y)
+{
+   if (!mat) {
+      *out_x = x;
+      *out_y = y;
+      return;
+   }
+
+   *out_x = mat[0]*x + mat[3]*y + mat[6];
+   *out_y = mat[1]*x + mat[4]*y + mat[7];
+   if (!is_affine(mat)) {
+      float w = 1/(mat[2]*x + mat[5]*y + mat[8]);
+      *out_x *= w;
+      *out_y *= w;
+   }
+}
+
+static void
+renderer_init_state(struct xorg_renderer *r)
+{
+   struct pipe_depth_stencil_alpha_state dsa;
+
+   /* set common initial clip state */
+   memset(&dsa, 0, sizeof(struct pipe_depth_stencil_alpha_state));
+   cso_set_depth_stencil_alpha(r->cso, &dsa);
+}
+
+
+static INLINE void
+setup_vertex0(float vertex[2][4], float x, float y,
+              float color[4])
+{
+   vertex[0][0] = x;
+   vertex[0][1] = y;
+   vertex[0][2] = 0.f; /*z*/
+   vertex[0][3] = 1.f; /*w*/
+
+   vertex[1][0] = color[0]; /*r*/
+   vertex[1][1] = color[1]; /*g*/
+   vertex[1][2] = color[2]; /*b*/
+   vertex[1][3] = color[3]; /*a*/
+}
+
+static INLINE void
+setup_vertex1(float vertex[2][4], float x, float y, float s, float t)
+{
+   vertex[0][0] = x;
+   vertex[0][1] = y;
+   vertex[0][2] = 0.f; /*z*/
+   vertex[0][3] = 1.f; /*w*/
+
+   vertex[1][0] = s;   /*s*/
+   vertex[1][1] = t;   /*t*/
+   vertex[1][2] = 0.f; /*r*/
+   vertex[1][3] = 1.f; /*q*/
+}
+
+static struct pipe_buffer *
+setup_vertex_data1(struct xorg_renderer *r,
+                   int srcX, int srcY,  int dstX, int dstY,
+                   int width, int height,
+                   struct pipe_texture *src, float *src_matrix)
+{
+   float s0, t0, s1, t1, stmp, ttmp;
+
+   s0 = srcX / src->width[0];
+   s1 = srcX + width / src->width[0];
+   t0 = srcY / src->height[0];
+   t1 = srcY + height / src->height[0];
+
+   if (src_matrix) {
+      /* 1st vertex */
+      map_point(src_matrix, s0, t0, &stmp, &ttmp);
+      setup_vertex1(r->vertices2[0], dstX, dstY, stmp, ttmp);
+      /* 2nd vertex */
+      map_point(src_matrix, s1, t0, &stmp, &ttmp);
+      setup_vertex1(r->vertices2[1], dstX + width, dstY, stmp, ttmp);
+      /* 3rd vertex */
+      map_point(src_matrix, s1, t1, &stmp, &ttmp);
+      setup_vertex1(r->vertices2[2], dstX + width, dstY + height, stmp, ttmp);
+      /* 4th vertex */
+      map_point(src_matrix, s0, t1, &stmp, &ttmp);
+      setup_vertex1(r->vertices2[3], dstX, dstY + height, stmp, ttmp);
+   } else {
+      /* 1st vertex */
+      setup_vertex1(r->vertices2[0], dstX, dstY, s0, t0);
+      /* 2nd vertex */
+      setup_vertex1(r->vertices2[1], dstX + width, dstY, s1, t0);
+      /* 3rd vertex */
+      setup_vertex1(r->vertices2[2], dstX + width, dstY + height, s1, t1);
+      /* 4th vertex */
+      setup_vertex1(r->vertices2[3], dstX, dstY + height, s0, t1);
+   }
+
+   return pipe_user_buffer_create(r->pipe->screen,
+                                  r->vertices2,
+                                  sizeof(r->vertices2));
+}
+
+static struct pipe_buffer *
+setup_vertex_data_tex(struct xorg_renderer *r,
+                      float x0, float y0, float x1, float y1,
+                      float s0, float t0, float s1, float t1,
+                      float z)
+{
+   /* 1st vertex */
+   setup_vertex1(r->vertices2[0], x0, y0, s0, t0);
+   /* 2nd vertex */
+   setup_vertex1(r->vertices2[1], x1, y0, s1, t0);
+   /* 3rd vertex */
+   setup_vertex1(r->vertices2[2], x1, y1, s1, t1);
+   /* 4th vertex */
+   setup_vertex1(r->vertices2[3], x0, y1, s0, t1);
+
+   return pipe_user_buffer_create(r->pipe->screen,
+                                  r->vertices2,
+                                  sizeof(r->vertices2));
+}
+
+static INLINE void
+setup_vertex2(float vertex[3][4], float x, float y,
+              float s0, float t0, float s1, float t1)
+{
+   vertex[0][0] = x;
+   vertex[0][1] = y;
+   vertex[0][2] = 0.f; /*z*/
+   vertex[0][3] = 1.f; /*w*/
+
+   vertex[1][0] = s0;  /*s*/
+   vertex[1][1] = t0;  /*t*/
+   vertex[1][2] = 0.f; /*r*/
+   vertex[1][3] = 1.f; /*q*/
+
+   vertex[2][0] = s1;  /*s*/
+   vertex[2][1] = t1;  /*t*/
+   vertex[2][2] = 0.f; /*r*/
+   vertex[2][3] = 1.f; /*q*/
+}
+
+static struct pipe_buffer *
+setup_vertex_data2(struct xorg_renderer *r,
+                   int srcX, int srcY, int maskX, int maskY,
+                   int dstX, int dstY, int width, int height,
+                   struct pipe_texture *src,
+                   struct pipe_texture *mask,
+                   float *src_matrix, float *mask_matrix)
+{
+   float st0[4], st1[4];
+   float pt0[2], pt1[2];
+
+   st0[0] = srcX / src->width[0];
+   st0[1] = srcY / src->height[0];
+   st0[2] = srcX + width / src->width[0];
+   st0[3] = srcY + height / src->height[0];
+
+   st1[0] = maskX / mask->width[0];
+   st1[1] = maskY / mask->height[0];
+   st1[2] = maskX + width / mask->width[0];
+   st1[3] = maskY + height / mask->height[0];
+
+   if (src_matrix || mask_matrix) {
+      /* 1st vertex */
+      map_point(src_matrix, st0[0], st0[1],
+                pt0 + 0, pt0 + 1);
+      map_point(mask_matrix, st1[0], st1[1],
+                pt1 + 0, pt1 + 1);
+      setup_vertex2(r->vertices3[0], dstX, dstY,
+                    pt0[0], pt0[1], pt1[0], pt1[1]);
+      /* 2nd vertex */
+      map_point(src_matrix, st0[2], st0[1],
+                pt0 + 0, pt0 + 1);
+      map_point(mask_matrix, st1[2], st1[1],
+                pt1 + 0, pt1 + 1);
+      setup_vertex2(r->vertices3[1], dstX + width, dstY,
+                    pt0[0], pt0[1], pt1[0], pt1[1]);
+      /* 3rd vertex */
+      map_point(src_matrix, st0[2], st0[3],
+                pt0 + 0, pt0 + 1);
+      map_point(mask_matrix, st1[2], st1[3],
+                pt1 + 0, pt1 + 1);
+      setup_vertex2(r->vertices3[2], dstX + width, dstY + height,
+                    pt0[0], pt0[1], pt1[0], pt1[1]);
+      /* 4th vertex */
+      map_point(src_matrix, st0[0], st0[3],
+                pt0 + 0, pt0 + 1);
+      map_point(mask_matrix, st1[0], st1[3],
+                pt1 + 0, pt1 + 1);
+      setup_vertex2(r->vertices3[3], dstX, dstY + height,
+                    pt0[0], pt0[1], pt1[0], pt1[1]);
+   } else {
+      /* 1st vertex */
+      setup_vertex2(r->vertices3[0], dstX, dstY,
+                    st0[0], st0[1], st1[0], st1[1]);
+      /* 2nd vertex */
+      setup_vertex2(r->vertices3[1], dstX + width, dstY,
+                    st0[2], st0[1], st1[2], st1[1]);
+      /* 3rd vertex */
+      setup_vertex2(r->vertices3[2], dstX + width, dstY + height,
+                    st0[2], st0[3], st1[2], st1[3]);
+      /* 4th vertex */
+      setup_vertex2(r->vertices3[3], dstX, dstY + height,
+                    st0[0], st0[3], st1[0], st1[3]);
+   }
+
+   return pipe_user_buffer_create(r->pipe->screen,
+                                  r->vertices3,
+                                  sizeof(r->vertices3));
+}
+
+
+
+static void
+set_viewport(struct xorg_renderer *r, int width, int height,
+             enum AxisOrientation orientation)
+{
+   struct pipe_viewport_state viewport;
+   float y_scale = (orientation == Y0_BOTTOM) ? -2.f : 2.f;
+
+   viewport.scale[0] =  width / 2.f;
+   viewport.scale[1] =  height / y_scale;
+   viewport.scale[2] =  1.0;
+   viewport.scale[3] =  1.0;
+   viewport.translate[0] = width / 2.f;
+   viewport.translate[1] = height / 2.f;
+   viewport.translate[2] = 0.0;
+   viewport.translate[3] = 0.0;
+
+   cso_set_viewport(r->cso, &viewport);
+}
+
+
+
+struct xorg_renderer * renderer_create(struct pipe_context *pipe)
+{
+   struct xorg_renderer *renderer = CALLOC_STRUCT(xorg_renderer);
+
+   renderer->pipe = pipe;
+   renderer->cso = cso_create_context(pipe);
+   renderer->shaders = xorg_shaders_create(renderer);
+
+   renderer_init_state(renderer);
+
+   return renderer;
+}
+
+void renderer_destroy(struct xorg_renderer *r)
+{
+   struct pipe_constant_buffer *vsbuf = &r->vs_const_buffer;
+   struct pipe_constant_buffer *fsbuf = &r->fs_const_buffer;
+
+   if (vsbuf && vsbuf->buffer)
+      pipe_buffer_reference(&vsbuf->buffer, NULL);
+
+   if (fsbuf && fsbuf->buffer)
+      pipe_buffer_reference(&fsbuf->buffer, NULL);
+
+   if (r->shaders) {
+      xorg_shaders_destroy(r->shaders);
+      r->shaders = NULL;
+   }
+
+   if (r->cso) {
+      cso_release_all(r->cso);
+      cso_destroy_context(r->cso);
+      r->cso = NULL;
+   }
+}
+
+void renderer_bind_framebuffer(struct xorg_renderer *r,
+                               struct exa_pixmap_priv *priv)
+{
+   unsigned i;
+   struct pipe_framebuffer_state state;
+   struct pipe_surface *surface = xorg_gpu_surface(r->pipe->screen, priv);
+   memset(&state, 0, sizeof(struct pipe_framebuffer_state));
+
+   state.width  = priv->tex->width[0];
+   state.height = priv->tex->height[0];
+
+   state.nr_cbufs = 1;
+   state.cbufs[0] = surface;
+   for (i = 1; i < PIPE_MAX_COLOR_BUFS; ++i)
+      state.cbufs[i] = 0;
+
+   /* currently we don't use depth/stencil */
+   state.zsbuf = 0;
+
+   cso_set_framebuffer(r->cso, &state);
+
+   /* we do fire and forget for the framebuffer, this is the forget part */
+   pipe_surface_reference(&surface, NULL);
+}
+
+void renderer_bind_viewport(struct xorg_renderer *r,
+                            struct exa_pixmap_priv *dst)
+{
+   int width = dst->tex->width[0];
+   int height = dst->tex->height[0];
+
+   /*debug_printf("Bind viewport (%d, %d)\n", width, height);*/
+
+   set_viewport(r, width, height, Y0_TOP);
+}
+
+void renderer_bind_rasterizer(struct xorg_renderer *r)
+{
+   struct pipe_rasterizer_state raster;
+
+   /* XXX: move to renderer_init_state? */
+   memset(&raster, 0, sizeof(struct pipe_rasterizer_state));
+   raster.gl_rasterization_rules = 1;
+   cso_set_rasterizer(r->cso, &raster);
+}
+
+void renderer_set_constants(struct xorg_renderer *r,
+                            int shader_type,
+                            const float *params,
+                            int param_bytes)
+{
+   struct pipe_constant_buffer *cbuf =
+      (shader_type == PIPE_SHADER_VERTEX) ? &r->vs_const_buffer :
+      &r->fs_const_buffer;
+
+   pipe_buffer_reference(&cbuf->buffer, NULL);
+   cbuf->buffer = pipe_buffer_create(r->pipe->screen, 16,
+                                     PIPE_BUFFER_USAGE_CONSTANT,
+                                     param_bytes);
+
+   if (cbuf->buffer) {
+      pipe_buffer_write(r->pipe->screen, cbuf->buffer,
+                        0, param_bytes, params);
+   }
+   r->pipe->set_constant_buffer(r->pipe, shader_type, 0, cbuf);
+}
+
+static void
+setup_vs_constant_buffer(struct xorg_renderer *r,
+                         int width, int height)
+{
+   const int param_bytes = 8 * sizeof(float);
+   float vs_consts[8] = {
+      2.f/width, 2.f/height, 1, 1,
+      -1, -1, 0, 0
+   };
+   renderer_set_constants(r, PIPE_SHADER_VERTEX,
+                          vs_consts, param_bytes);
+}
+
+static void
+setup_fs_constant_buffer(struct xorg_renderer *r)
+{
+   const int param_bytes = 4 * sizeof(float);
+   const float fs_consts[8] = {
+      0, 0, 0, 1,
+   };
+   renderer_set_constants(r, PIPE_SHADER_FRAGMENT,
+                          fs_consts, param_bytes);
+}
+
+static INLINE void shift_rectx(float coords[4],
+                               const float *bounds,
+                               const float shift)
+{
+   coords[0] += shift;
+   coords[2] -= shift;
+   if (bounds) {
+      coords[2] = MIN2(coords[2], bounds[2]);
+      /* bound x/y + width/height */
+      if ((coords[0] + coords[2]) > (bounds[0] + bounds[2])) {
+         coords[2] = (bounds[0] + bounds[2]) - coords[0];
+      }
+   }
+}
+
+static INLINE void shift_recty(float coords[4],
+                               const float *bounds,
+                               const float shift)
+{
+   coords[1] += shift;
+   coords[3] -= shift;
+   if (bounds) {
+      coords[3] = MIN2(coords[3], bounds[3]);
+      if ((coords[1] + coords[3]) > (bounds[1] + bounds[3])) {
+         coords[3] = (bounds[1] + bounds[3]) - coords[1];
+      }
+   }
+}
+
+static INLINE void bound_rect(float coords[4],
+                              const float bounds[4],
+                              float shift[4])
+{
+   /* if outside the bounds */
+   if (coords[0] > (bounds[0] + bounds[2]) ||
+       coords[1] > (bounds[1] + bounds[3]) ||
+       (coords[0] + coords[2]) < bounds[0] ||
+       (coords[1] + coords[3]) < bounds[1]) {
+      coords[0] = 0.f;
+      coords[1] = 0.f;
+      coords[2] = 0.f;
+      coords[3] = 0.f;
+      shift[0] = 0.f;
+      shift[1] = 0.f;
+      return;
+   }
+
+   /* bound x */
+   if (coords[0] < bounds[0]) {
+      shift[0] = bounds[0] - coords[0];
+      coords[2] -= shift[0];
+      coords[0] = bounds[0];
+   } else
+      shift[0] = 0.f;
+
+   /* bound y */
+   if (coords[1] < bounds[1]) {
+      shift[1] = bounds[1] - coords[1];
+      coords[3] -= shift[1];
+      coords[1] = bounds[1];
+   } else
+      shift[1] = 0.f;
+
+   shift[2] = bounds[2] - coords[2];
+   shift[3] = bounds[3] - coords[3];
+   /* bound width/height */
+   coords[2] = MIN2(coords[2], bounds[2]);
+   coords[3] = MIN2(coords[3], bounds[3]);
+
+   /* bound x/y + width/height */
+   if ((coords[0] + coords[2]) > (bounds[0] + bounds[2])) {
+      coords[2] = (bounds[0] + bounds[2]) - coords[0];
+   }
+   if ((coords[1] + coords[3]) > (bounds[1] + bounds[3])) {
+      coords[3] = (bounds[1] + bounds[3]) - coords[1];
+   }
+
+   /* if outside the bounds */
+   if ((coords[0] + coords[2]) < bounds[0] ||
+       (coords[1] + coords[3]) < bounds[1]) {
+      coords[0] = 0.f;
+      coords[1] = 0.f;
+      coords[2] = 0.f;
+      coords[3] = 0.f;
+      return;
+   }
+}
+
+static INLINE void sync_size(float *src_loc, float *dst_loc)
+{
+   src_loc[2] = MIN2(src_loc[2], dst_loc[2]);
+   src_loc[3] = MIN2(src_loc[3], dst_loc[3]);
+   dst_loc[2] = src_loc[2];
+   dst_loc[3] = src_loc[3];
+}
+
+static void renderer_copy_texture(struct xorg_renderer *r,
+                                  struct pipe_texture *src,
+                                  float sx1, float sy1,
+                                  float sx2, float sy2,
+                                  struct pipe_texture *dst,
+                                  float dx1, float dy1,
+                                  float dx2, float dy2)
+{
+   struct pipe_context *pipe = r->pipe;
+   struct pipe_screen *screen = pipe->screen;
+   struct pipe_buffer *buf;
+   struct pipe_surface *dst_surf = screen->get_tex_surface(
+      screen, dst, 0, 0, 0,
+      PIPE_BUFFER_USAGE_GPU_WRITE);
+   struct pipe_framebuffer_state fb;
+   float s0, t0, s1, t1;
+   struct xorg_shader shader;
+
+   assert(src->width[0] != 0);
+   assert(src->height[0] != 0);
+   assert(dst->width[0] != 0);
+   assert(dst->height[0] != 0);
+
+#if 1
+   s0 = sx1 / src->width[0];
+   s1 = sx2 / src->width[0];
+   t0 = sy1 / src->height[0];
+   t1 = sy2 / src->height[0];
+#else
+   s0 = 0;
+   s1 = 1;
+   t0 = 0;
+   t1 = 1;
+#endif
+
+#if 0
+   debug_printf("copy texture src=[%f, %f, %f, %f], dst=[%f, %f, %f, %f], tex=[%f, %f, %f, %f]\n",
+                sx1, sy1, sx2, sy2, dx1, dy1, dx2, dy2,
+                s0, t0, s1, t1);
+#endif
+
+   assert(screen->is_format_supported(screen, dst_surf->format,
+                                      PIPE_TEXTURE_2D,
+                                      PIPE_TEXTURE_USAGE_RENDER_TARGET,
+                                      0));
+
+   /* save state (restored below) */
+   cso_save_blend(r->cso);
+   cso_save_samplers(r->cso);
+   cso_save_sampler_textures(r->cso);
+   cso_save_framebuffer(r->cso);
+   cso_save_fragment_shader(r->cso);
+   cso_save_vertex_shader(r->cso);
+
+   cso_save_viewport(r->cso);
+
+
+   /* set misc state we care about */
+   {
+      struct pipe_blend_state blend;
+      memset(&blend, 0, sizeof(blend));
+      blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE;
+      blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE;
+      blend.rgb_dst_factor = PIPE_BLENDFACTOR_ZERO;
+      blend.alpha_dst_factor = PIPE_BLENDFACTOR_ZERO;
+      blend.colormask = PIPE_MASK_RGBA;
+      cso_set_blend(r->cso, &blend);
+   }
+
+   /* sampler */
+   {
+      struct pipe_sampler_state sampler;
+      memset(&sampler, 0, sizeof(sampler));
+      sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+      sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+      sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+      sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+      sampler.min_img_filter = PIPE_TEX_FILTER_NEAREST;
+      sampler.mag_img_filter = PIPE_TEX_FILTER_NEAREST;
+      sampler.normalized_coords = 1;
+      cso_single_sampler(r->cso, 0, &sampler);
+      cso_single_sampler_done(r->cso);
+   }
+
+   set_viewport(r, dst_surf->width, dst_surf->height, Y0_TOP);
+
+   /* texture */
+   cso_set_sampler_textures(r->cso, 1, &src);
+
+   renderer_bind_rasterizer(r);
+
+   /* shaders */
+   shader = xorg_shaders_get(r->shaders,
+                             VS_COMPOSITE,
+                             FS_COMPOSITE);
+   cso_set_vertex_shader_handle(r->cso, shader.vs);
+   cso_set_fragment_shader_handle(r->cso, shader.fs);
+
+   /* drawing dest */
+   memset(&fb, 0, sizeof(fb));
+   fb.width = dst_surf->width;
+   fb.height = dst_surf->height;
+   fb.nr_cbufs = 1;
+   fb.cbufs[0] = dst_surf;
+   {
+      int i;
+      for (i = 1; i < PIPE_MAX_COLOR_BUFS; ++i)
+         fb.cbufs[i] = 0;
+   }
+   cso_set_framebuffer(r->cso, &fb);
+   setup_vs_constant_buffer(r, fb.width, fb.height);
+   setup_fs_constant_buffer(r);
+
+   /* draw quad */
+   buf = setup_vertex_data_tex(r,
+                               dx1, dy1,
+                               dx2, dy2,
+                               s0, t0, s1, t1,
+                               0.0f);
+
+   if (buf) {
+      util_draw_vertex_buffer(r->pipe, buf, 0,
+                              PIPE_PRIM_TRIANGLE_FAN,
+                              4,  /* verts */
+                              2); /* attribs/vert */
+
+      pipe_buffer_reference(&buf, NULL);
+   }
+
+   /* restore state we changed */
+   cso_restore_blend(r->cso);
+   cso_restore_samplers(r->cso);
+   cso_restore_sampler_textures(r->cso);
+   cso_restore_framebuffer(r->cso);
+   cso_restore_vertex_shader(r->cso);
+   cso_restore_fragment_shader(r->cso);
+   cso_restore_viewport(r->cso);
+
+   pipe_surface_reference(&dst_surf, NULL);
+}
+
+static struct pipe_texture *
+create_sampler_texture(struct xorg_renderer *r,
+                       struct pipe_texture *src)
+{
+   enum pipe_format format;
+   struct pipe_context *pipe = r->pipe;
+   struct pipe_screen *screen = pipe->screen;
+   struct pipe_texture *pt;
+   struct pipe_texture templ;
+
+   pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL);
+
+   /* the coming in texture should already have that invariance */
+   debug_assert(screen->is_format_supported(screen, src->format,
+                                            PIPE_TEXTURE_2D,
+                                            PIPE_TEXTURE_USAGE_SAMPLER, 0));
+
+   format = src->format;
+
+   memset(&templ, 0, sizeof(templ));
+   templ.target = PIPE_TEXTURE_2D;
+   templ.format = format;
+   templ.last_level = 0;
+   templ.width[0] = src->width[0];
+   templ.height[0] = src->height[0];
+   templ.depth[0] = 1;
+   pf_get_block(format, &templ.block);
+   templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
+
+   pt = screen->texture_create(screen, &templ);
+
+   debug_assert(!pt || pipe_is_referenced(&pt->reference));
+
+   if (!pt)
+      return NULL;
+
+   {
+      /* copy source framebuffer surface into texture */
+      struct pipe_surface *ps_read = screen->get_tex_surface(
+         screen, src, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ);
+      struct pipe_surface *ps_tex = screen->get_tex_surface(
+         screen, pt, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE );
+      if (pipe->surface_copy) {
+         pipe->surface_copy(pipe,
+                ps_tex, /* dest */
+                0, 0, /* destx/y */
+                ps_read,
+                0, 0, src->width[0], src->height[0]);
+      } else {
+          util_surface_copy(pipe, FALSE,
+                ps_tex, /* dest */
+                0, 0, /* destx/y */
+                ps_read,
+                0, 0, src->width[0], src->height[0]);
+      }
+      pipe_surface_reference(&ps_read, NULL);
+      pipe_surface_reference(&ps_tex, NULL);
+   }
+
+   return pt;
+}
+
+
+void renderer_copy_pixmap(struct xorg_renderer *r,
+                          struct exa_pixmap_priv *dst_priv, int dx, int dy,
+                          struct exa_pixmap_priv *src_priv, int sx, int sy,
+                          int width, int height)
+{
+   float dst_loc[4], src_loc[4];
+   float dst_bounds[4], src_bounds[4];
+   float src_shift[4], dst_shift[4], shift[4];
+   struct pipe_texture *dst = dst_priv->tex;
+   struct pipe_texture *src = src_priv->tex;
+
+   if (r->pipe->is_texture_referenced(r->pipe, src, 0, 0) &
+       PIPE_REFERENCED_FOR_WRITE)
+      r->pipe->flush(r->pipe, PIPE_FLUSH_RENDER_CACHE, NULL);
+
+   dst_loc[0] = dx;
+   dst_loc[1] = dy;
+   dst_loc[2] = width;
+   dst_loc[3] = height;
+   dst_bounds[0] = 0.f;
+   dst_bounds[1] = 0.f;
+   dst_bounds[2] = dst->width[0];
+   dst_bounds[3] = dst->height[0];
+
+   src_loc[0] = sx;
+   src_loc[1] = sy;
+   src_loc[2] = width;
+   src_loc[3] = height;
+   src_bounds[0] = 0.f;
+   src_bounds[1] = 0.f;
+   src_bounds[2] = src->width[0];
+   src_bounds[3] = src->height[0];
+
+   bound_rect(src_loc, src_bounds, src_shift);
+   bound_rect(dst_loc, dst_bounds, dst_shift);
+   shift[0] = src_shift[0] - dst_shift[0];
+   shift[1] = src_shift[1] - dst_shift[1];
+
+   if (shift[0] < 0)
+      shift_rectx(src_loc, src_bounds, -shift[0]);
+   else
+      shift_rectx(dst_loc, dst_bounds, shift[0]);
+
+   if (shift[1] < 0)
+      shift_recty(src_loc, src_bounds, -shift[1]);
+   else
+      shift_recty(dst_loc, dst_bounds, shift[1]);
+
+   sync_size(src_loc, dst_loc);
+
+   if (src_loc[2] >= 0 && src_loc[3] >= 0 &&
+       dst_loc[2] >= 0 && dst_loc[3] >= 0) {
+      struct pipe_texture *temp_src = src;
+
+      if (src == dst)
+         temp_src = create_sampler_texture(r, src);
+
+      renderer_copy_texture(r,
+                            temp_src,
+                            src_loc[0],
+                            src_loc[1],
+                            src_loc[0] + src_loc[2],
+                            src_loc[1] + src_loc[3],
+                            dst,
+                            dst_loc[0],
+                            dst_loc[1],
+                            dst_loc[0] + dst_loc[2],
+                            dst_loc[1] + dst_loc[3]);
+
+      if (src == dst)
+         pipe_texture_reference(&temp_src, NULL);
+   }
+}
+
+void renderer_draw_solid_rect(struct xorg_renderer *r,
+                              int x0, int y0,
+                              int x1, int y1,
+                              float *color)
+{
+   struct pipe_context *pipe = r->pipe;
+   struct pipe_buffer *buf = 0;
+
+   /*
+   debug_printf("solid rect[(%d, %d), (%d, %d)], rgba[%f, %f, %f, %f]\n",
+   x0, y0, x1, y1, color[0], color[1], color[2], color[3]);*/
+   /* 1st vertex */
+   setup_vertex0(r->vertices2[0], x0, y0, color);
+   /* 2nd vertex */
+   setup_vertex0(r->vertices2[1], x1, y0, color);
+   /* 3rd vertex */
+   setup_vertex0(r->vertices2[2], x1, y1, color);
+   /* 4th vertex */
+   setup_vertex0(r->vertices2[3], x0, y1, color);
+
+   buf = pipe_user_buffer_create(pipe->screen,
+                                 r->vertices2,
+                                 sizeof(r->vertices2));
+
+
+   if (buf) {
+      util_draw_vertex_buffer(pipe, buf, 0,
+                              PIPE_PRIM_TRIANGLE_FAN,
+                              4,  /* verts */
+                              2); /* attribs/vert */
+
+      pipe_buffer_reference(&buf, NULL);
+   }
+}
+
+void renderer_draw_textures(struct xorg_renderer *r,
+                            int *pos,
+                            int width, int height,
+                            struct pipe_texture **textures,
+                            int num_textures,
+                            float *src_matrix, float *mask_matrix)
+{
+   struct pipe_context *pipe = r->pipe;
+   struct pipe_buffer *buf = 0;
+
+   switch(num_textures) {
+   case 1:
+      buf = setup_vertex_data1(r,
+                               pos[0], pos[1], /* src */
+                               pos[4], pos[5], /* dst */
+                               width, height,
+                               textures[0], src_matrix);
+      break;
+   case 2:
+      buf = setup_vertex_data2(r,
+                               pos[0], pos[1], /* src */
+                               pos[2], pos[3], /* mask */
+                               pos[4], pos[5], /* dst */
+                               width, height,
+                               textures[0], textures[1],
+                               src_matrix, mask_matrix);
+      break;
+   default:
+      debug_assert(!"Unsupported number of textures");
+      break;
+   }
+
+   if (buf) {
+      int num_attribs = 1; /*pos*/
+      num_attribs += num_textures;
+
+      util_draw_vertex_buffer(pipe, buf, 0,
+                              PIPE_PRIM_TRIANGLE_FAN,
+                              4,  /* verts */
+                              num_attribs); /* attribs/vert */
+
+      pipe_buffer_reference(&buf, NULL);
+   }
+}
diff --git a/src/gallium/state_trackers/xorg/xorg_renderer.h b/src/gallium/state_trackers/xorg/xorg_renderer.h
new file mode 100644
index 0000000000..3e37c9aa93
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xorg_renderer.h
@@ -0,0 +1,55 @@
+#ifndef XORG_RENDERER_H
+#define XORG_RENDERER_H
+
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+struct xorg_shaders;
+struct exa_pixmap_priv;
+
+struct xorg_renderer {
+   struct pipe_context *pipe;
+
+   struct cso_context *cso;
+   struct xorg_shaders *shaders;
+
+   struct pipe_constant_buffer vs_const_buffer;
+   struct pipe_constant_buffer fs_const_buffer;
+
+   /* we should combine these two */
+   float vertices2[4][2][4];
+   float vertices3[4][3][4];
+};
+
+struct xorg_renderer *renderer_create(struct pipe_context *pipe);
+void renderer_destroy(struct xorg_renderer *renderer);
+
+void renderer_bind_framebuffer(struct xorg_renderer *r,
+                               struct exa_pixmap_priv *priv);
+void renderer_bind_viewport(struct xorg_renderer *r,
+                            struct exa_pixmap_priv *dst);
+void renderer_bind_rasterizer(struct xorg_renderer *r);
+void renderer_set_constants(struct xorg_renderer *r,
+                            int shader_type,
+                            const float *buffer,
+                            int size);
+void renderer_copy_pixmap(struct xorg_renderer *r,
+                          struct exa_pixmap_priv *dst_priv, int dx, int dy,
+                          struct exa_pixmap_priv *src_priv, int sx, int sy,
+                          int width, int height);
+
+void renderer_draw_solid_rect(struct xorg_renderer *r,
+                              int x0, int y0,
+                              int x1, int y1,
+                              float *color);
+
+void renderer_draw_textures(struct xorg_renderer *r,
+                            int *pos,
+                            int width, int height,
+                            struct pipe_texture **textures,
+                            int num_textures,
+                            float *src_matrix,
+                            float *mask_matrix);
+
+
+#endif
diff --git a/src/gallium/state_trackers/xorg/xorg_tracker.h b/src/gallium/state_trackers/xorg/xorg_tracker.h
index b1ab783a15..6130cf6621 100644
--- a/src/gallium/state_trackers/xorg/xorg_tracker.h
+++ b/src/gallium/state_trackers/xorg/xorg_tracker.h
@@ -51,6 +51,8 @@
 
 #define DRV_ERROR(msg)	xf86DrvMsg(pScrn->scrnIndex, X_ERROR, msg);
 
+struct exa_context;
+
 typedef struct
 {
     int lastInstance;
@@ -90,10 +92,12 @@ typedef struct _modesettingRec
     struct pipe_context *ctx;
     boolean d_depth_bits_last;
     boolean ds_depth_bits_last;
+    struct pipe_texture *root_texture;
 
     /* exa */
-    void *exa;
+    struct exa_context *exa;
     Bool noEvict;
+    Bool debug_fallback;
 
 #ifdef DRM_MODE_FEATURE_DIRTYFB
     DamagePtr damage;
@@ -118,6 +122,14 @@ xorg_exa_set_displayed_usage(PixmapPtr pPixmap);
 int
 xorg_exa_set_shared_usage(PixmapPtr pPixmap);
 
+Bool
+xorg_exa_set_texture(PixmapPtr pPixmap, struct  pipe_texture *tex);
+
+struct pipe_texture *
+xorg_exa_create_root_texture(ScrnInfoPtr pScrn,
+			     int width, int height,
+			     int depth, int bpp);
+
 void *
 xorg_exa_init(ScrnInfoPtr pScrn);
 
@@ -151,5 +163,11 @@ crtc_cursor_destroy(xf86CrtcPtr crtc);
 void
 output_init(ScrnInfoPtr pScrn);
 
+/***********************************************************************
+ * xorg_xv.c
+ */
+void
+xorg_init_video(ScreenPtr pScreen);
+
 
 #endif /* _XORG_TRACKER_H_ */
diff --git a/src/gallium/state_trackers/xorg/xorg_xv.c b/src/gallium/state_trackers/xorg/xorg_xv.c
new file mode 100644
index 0000000000..efac9275b2
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xorg_xv.c
@@ -0,0 +1,574 @@
+#include "xorg_tracker.h"
+
+#include <xf86xv.h>
+#include <X11/extensions/Xv.h>
+#include <fourcc.h>
+
+#include "xorg_exa.h"
+#include "xorg_renderer.h"
+
+#include "pipe/p_screen.h"
+#include "pipe/p_inlines.h"
+
+/*XXX get these from pipe's texture limits */
+#define IMAGE_MAX_WIDTH		2048
+#define IMAGE_MAX_HEIGHT	2048
+
+#define RES_720P_X 1280
+#define RES_720P_Y 720
+
+
+/* The ITU-R BT.601 conversion matrix for SDTV. */
+static const float bt_601[] = {
+    1.0, 0.0, 1.4075,
+    1.0, -0.3455, -0.7169,
+    1.0, 1.7790, 0.
+};
+
+/* The ITU-R BT.709 conversion matrix for HDTV. */
+static const float bt_709[] = {
+    1.0, 0.0, 1.581,
+    1.0, -0.1881, -0.47,
+    1.0, 1.8629, 0.
+};
+
+#define MAKE_ATOM(a) MakeAtom(a, sizeof(a) - 1, TRUE)
+
+static Atom xvBrightness, xvContrast;
+
+#define NUM_TEXTURED_ATTRIBUTES 2
+static XF86AttributeRec TexturedAttributes[NUM_TEXTURED_ATTRIBUTES] = {
+   {XvSettable | XvGettable, -128, 127, "XV_BRIGHTNESS"},
+   {XvSettable | XvGettable, 0, 255, "XV_CONTRAST"}
+};
+
+#define NUM_FORMATS 3
+static XF86VideoFormatRec Formats[NUM_FORMATS] = {
+   {15, TrueColor}, {16, TrueColor}, {24, TrueColor}
+};
+
+static XF86VideoEncodingRec DummyEncoding[1] = {
+   {
+      0,
+      "XV_IMAGE",
+      IMAGE_MAX_WIDTH, IMAGE_MAX_HEIGHT,
+      {1, 1}
+   }
+};
+
+#define NUM_IMAGES 2
+static XF86ImageRec Images[NUM_IMAGES] = {
+   XVIMAGE_UYVY,
+   XVIMAGE_YUY2,
+};
+
+struct xorg_xv_port_priv {
+   struct xorg_renderer *r;
+
+   RegionRec clip;
+
+   int brightness;
+   int contrast;
+
+   int current_set;
+   /* juggle two sets of seperate Y, U and V
+    * textures */
+   struct pipe_texture *yuv[2][3];
+};
+
+
+static void
+stop_video(ScrnInfoPtr pScrn, pointer data, Bool shutdown)
+{
+   struct xorg_xv_port_priv *priv = (struct xorg_xv_port_priv *)data;
+
+   REGION_EMPTY(pScrn->pScreen, &priv->clip);
+}
+
+static int
+set_port_attribute(ScrnInfoPtr pScrn,
+                   Atom attribute, INT32 value, pointer data)
+{
+   struct xorg_xv_port_priv *priv = (struct xorg_xv_port_priv *)data;
+
+   if (attribute == xvBrightness) {
+      if ((value < -128) || (value > 127))
+         return BadValue;
+      priv->brightness = value;
+   } else if (attribute == xvContrast) {
+      if ((value < 0) || (value > 255))
+         return BadValue;
+      priv->contrast = value;
+   } else
+      return BadMatch;
+
+   return Success;
+}
+
+static int
+get_port_attribute(ScrnInfoPtr pScrn,
+                   Atom attribute, INT32 * value, pointer data)
+{
+   struct xorg_xv_port_priv *priv = (struct xorg_xv_port_priv *)data;
+
+   if (attribute == xvBrightness)
+      *value = priv->brightness;
+   else if (attribute == xvContrast)
+      *value = priv->contrast;
+   else
+      return BadMatch;
+
+   return Success;
+}
+
+static void
+query_best_size(ScrnInfoPtr pScrn,
+                Bool motion,
+                short vid_w, short vid_h,
+                short drw_w, short drw_h,
+                unsigned int *p_w, unsigned int *p_h, pointer data)
+{
+   if (vid_w > (drw_w << 1))
+      drw_w = vid_w >> 1;
+   if (vid_h > (drw_h << 1))
+      drw_h = vid_h >> 1;
+
+   *p_w = drw_w;
+   *p_h = drw_h;
+}
+
+static INLINE struct pipe_texture *
+create_component_texture(struct pipe_context *pipe,
+                         int width, int height)
+{
+   struct pipe_screen *screen = pipe->screen;
+   struct pipe_texture *tex = 0;
+   struct pipe_texture templ;
+
+   memset(&templ, 0, sizeof(templ));
+   templ.target = PIPE_TEXTURE_2D;
+   templ.format = PIPE_FORMAT_L8_UNORM;
+   templ.last_level = 0;
+   templ.width[0] = width;
+   templ.height[0] = height;
+   templ.depth[0] = 1;
+   pf_get_block(PIPE_FORMAT_L8_UNORM, &templ.block);
+   templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
+
+   tex = screen->texture_create(screen, &templ);
+
+   return tex;
+}
+
+static int
+check_yuv_textures(struct xorg_xv_port_priv *priv,  int width, int height)
+{
+   struct pipe_texture **dst = priv->yuv[priv->current_set];
+   if (!dst[0] ||
+       dst[0]->width[0] != width ||
+       dst[0]->height[0] != height) {
+      pipe_texture_reference(&dst[0], NULL);
+   }
+   if (!dst[1] ||
+       dst[1]->width[0] != width ||
+       dst[1]->height[0] != height) {
+      pipe_texture_reference(&dst[1], NULL);
+   }
+   if (!dst[2] ||
+       dst[2]->width[0] != width ||
+       dst[2]->height[0] != height) {
+      pipe_texture_reference(&dst[2], NULL);
+   }
+
+   if (!dst[0])
+      dst[0] = create_component_texture(priv->r->pipe, width, height);
+
+   if (!dst[1])
+      dst[1] = create_component_texture(priv->r->pipe, width, height);
+
+   if (!dst[2])
+      dst[2] = create_component_texture(priv->r->pipe, width, height);
+
+   if (!dst[0] || !dst[1] || !dst[2])
+      return BadAlloc;
+
+   return Success;
+}
+
+static void
+copy_packed_data(ScrnInfoPtr pScrn,
+                 struct xorg_xv_port_priv *port,
+                 int id,
+                 unsigned char *buf,
+                 int srcPitch,
+                 int left,
+                 int top,
+                 int w, int h)
+{
+   unsigned char *src;
+   int i, j;
+   struct pipe_texture **dst = port->yuv[port->current_set];
+   struct pipe_transfer *ytrans, *utrans, *vtrans;
+   struct pipe_screen *screen = port->r->pipe->screen;
+   char *ymap, *vmap, *umap;
+   unsigned char y1, y2, u, v;
+   int yidx, uidx, vidx;
+
+   src = buf + (top * srcPitch) + (left << 1);
+
+   ytrans = screen->get_tex_transfer(screen, dst[0],
+                                     0, 0, 0,
+                                     PIPE_TRANSFER_WRITE,
+                                     left, top, w, h);
+   utrans = screen->get_tex_transfer(screen, dst[1],
+                                     0, 0, 0,
+                                     PIPE_TRANSFER_WRITE,
+                                     left, top, w, h);
+   vtrans = screen->get_tex_transfer(screen, dst[2],
+                                     0, 0, 0,
+                                     PIPE_TRANSFER_WRITE,
+                                     left, top, w, h);
+
+   ymap = screen->transfer_map(screen, ytrans);
+   umap = screen->transfer_map(screen, utrans);
+   vmap = screen->transfer_map(screen, vtrans);
+
+   switch (id) {
+   case FOURCC_YV12: {
+      int y_array_size = w * h;
+      for (i = 0; i < w; ++i) {
+         for (j = 0; i < h; ++j) {
+            /*XXX use src? */
+            y1  = buf[j*w + i];
+            u   = buf[(j/2) * (w/2) + i/2 + y_array_size];
+            v   = buf[(j/2) * (w/2) + i/2 + y_array_size + y_array_size/4];
+            ymap[yidx++] = y1;
+            umap[uidx++] = u;
+            vmap[vidx++] = v;
+         }
+      }
+   }
+      break;
+   case FOURCC_YUY2:
+      for (j = 0; j < h; ++j) {
+         for (i = 0; i < w; ++i) {
+            /* extracting two pixels */
+            y1 = buf[0];
+            u  = buf[1];
+            y2 = buf[2];
+            v  = buf[3];
+
+            ymap[yidx++] = y1;
+            ymap[yidx++] = y2;
+            umap[uidx++] = u;
+            umap[uidx++] = u;
+            vmap[vidx++] = v;
+            vmap[vidx++] = v;
+         }
+      }
+      break;
+   default:
+      debug_assert(!"Unsupported yuv format!");
+      break;
+   }
+
+   screen->transfer_unmap(screen, ytrans);
+   screen->transfer_unmap(screen, utrans);
+   screen->transfer_unmap(screen, vtrans);
+   screen->tex_transfer_destroy(ytrans);
+   screen->tex_transfer_destroy(utrans);
+   screen->tex_transfer_destroy(vtrans);
+}
+
+static void
+setup_video_constants(struct xorg_renderer *r, boolean hdtv)
+{
+   struct pipe_context *pipe = r->pipe;
+   const int param_bytes = 9 * sizeof(float);
+   struct pipe_constant_buffer *cbuf = &r->vs_const_buffer;
+
+   pipe_buffer_reference(&cbuf->buffer, NULL);
+   cbuf->buffer = pipe_buffer_create(pipe->screen, 16,
+                                     PIPE_BUFFER_USAGE_CONSTANT,
+                                     param_bytes);
+
+   if (cbuf->buffer) {
+      const float *video_constants = (hdtv) ? bt_709 : bt_601;
+
+      pipe_buffer_write(pipe->screen, cbuf->buffer,
+                        0, param_bytes, video_constants);
+   }
+   pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, cbuf);
+}
+
+static int
+display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
+              RegionPtr dstRegion,
+              short width, short height,
+              int x1, int y1, int x2, int y2,
+              short src_w, short src_h, short drw_w, short drw_h,
+              PixmapPtr pPixmap)
+{
+   BoxPtr pbox;
+   int nbox;
+   int dxo, dyo;
+   Bool hdtv;
+   float tc0[2], tc1[2], tc2[2];
+
+   hdtv = ((src_w >= RES_720P_X) && (src_h >= RES_720P_Y));
+   setup_video_constants(pPriv->r, hdtv);
+
+   REGION_TRANSLATE(pScrn->pScreen, dstRegion, -pPixmap->screen_x,
+                    -pPixmap->screen_y);
+
+   dxo = dstRegion->extents.x1;
+   dyo = dstRegion->extents.y1;
+
+   pbox = REGION_RECTS(dstRegion);
+   nbox = REGION_NUM_RECTS(dstRegion);
+
+   while (nbox--) {
+      int box_x1 = pbox->x1;
+      int box_y1 = pbox->y1;
+      int box_x2 = pbox->x2;
+      int box_y2 = pbox->y2;
+
+      tc0[0] = (double) (box_x1 - dxo) / (double) drw_w; /* u0 */
+      tc0[1] = (double) (box_y1 - dyo) / (double) drw_h; /* v0 */
+      tc1[0] = (double) (box_x2 - dxo) / (double) drw_w; /* u1 */
+      tc1[1] = tc0[1];
+      tc2[0] = tc0[0];
+      tc2[1] = (double) (box_y2 - dyo) / (double) drw_h; /* v1 */
+
+#if 0
+      x = box_x1;
+      y = box_y1;
+      w = box_x2 - box_x1;
+      h = box_y2 - box_y1;
+
+      pbox++;
+      draw_yuv(pScrn, x, y, w, h, &src, 1, FALSE,
+               0, tc0, tc1, tc2, 1,
+               pPriv->conversionData);
+#endif
+   }
+   DamageDamageRegion(&pPixmap->drawable, dstRegion);
+   return TRUE;
+}
+
+static int
+put_image(ScrnInfoPtr pScrn,
+          short src_x, short src_y,
+          short drw_x, short drw_y,
+          short src_w, short src_h,
+          short drw_w, short drw_h,
+          int id, unsigned char *buf,
+          short width, short height,
+          Bool sync, RegionPtr clipBoxes, pointer data,
+          DrawablePtr pDraw)
+{
+   struct xorg_xv_port_priv *pPriv = (struct xorg_xv_port_priv *) data;
+   ScreenPtr pScreen = screenInfo.screens[pScrn->scrnIndex];
+   PixmapPtr pPixmap;
+   INT32 x1, x2, y1, y2;
+   int srcPitch;
+   BoxRec dstBox;
+   int ret;
+
+   /* Clip */
+   x1 = src_x;
+   x2 = src_x + src_w;
+   y1 = src_y;
+   y2 = src_y + src_h;
+
+   dstBox.x1 = drw_x;
+   dstBox.x2 = drw_x + drw_w;
+   dstBox.y1 = drw_y;
+   dstBox.y2 = drw_y + drw_h;
+
+   if (!xf86XVClipVideoHelper(&dstBox, &x1, &x2, &y1, &y2, clipBoxes,
+			      width, height))
+      return Success;
+
+   switch (id) {
+   case FOURCC_UYVY:
+   case FOURCC_YUY2:
+   default:
+      srcPitch = width << 1;
+      break;
+   }
+
+   ret = check_yuv_textures(pPriv, width, height);
+
+   if (ret)
+      return ret;
+
+   copy_packed_data(pScrn, pPriv, id, buf, srcPitch,
+                    src_x, src_y, width, height);
+
+   if (pDraw->type == DRAWABLE_WINDOW) {
+      pPixmap = (*pScreen->GetWindowPixmap)((WindowPtr)pDraw);
+   } else {
+      pPixmap = (PixmapPtr)pDraw;
+   }
+
+   display_video(pScrn, pPriv, id, clipBoxes, width, height,
+                 x1, y1, x2, y2,
+                 src_w, src_h, drw_w, drw_h, pPixmap);
+
+   pPriv->current_set = (pPriv->current_set + 1) & 1;
+   return Success;
+}
+
+static int
+query_image_attributes(ScrnInfoPtr pScrn,
+                       int id,
+                       unsigned short *w, unsigned short *h,
+                       int *pitches, int *offsets)
+{
+   int size;
+
+   if (*w > IMAGE_MAX_WIDTH)
+      *w = IMAGE_MAX_WIDTH;
+   if (*h > IMAGE_MAX_HEIGHT)
+      *h = IMAGE_MAX_HEIGHT;
+
+   *w = (*w + 1) & ~1;
+   if (offsets)
+      offsets[0] = 0;
+
+   switch (id) {
+   case FOURCC_UYVY:
+   case FOURCC_YUY2:
+   default:
+      size = *w << 1;
+      if (pitches)
+	 pitches[0] = size;
+      size *= *h;
+      break;
+   }
+
+   return size;
+}
+
+static struct xorg_xv_port_priv *
+port_priv_create(struct xorg_renderer *r)
+{
+   struct xorg_xv_port_priv *priv = NULL;
+
+   priv = calloc(1, sizeof(struct xorg_xv_port_priv));
+
+   if (!priv)
+      return NULL;
+
+   priv->r = r;
+
+   REGION_NULL(pScreen, &priv->clip);
+
+   debug_assert(priv && priv->r);
+
+   return priv;
+}
+
+static XF86VideoAdaptorPtr
+xorg_setup_textured_adapter(ScreenPtr pScreen)
+{
+   ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
+   modesettingPtr ms = modesettingPTR(pScrn);
+   XF86VideoAdaptorPtr adapt;
+   XF86AttributePtr attrs;
+   DevUnion *dev_unions;
+   int nports = 16, i;
+   int nattributes;
+
+   nattributes = NUM_TEXTURED_ATTRIBUTES;
+
+   debug_assert(ms->exa);
+   debug_assert(ms->exa->renderer);
+
+   adapt = calloc(1, sizeof(XF86VideoAdaptorRec));
+   dev_unions = calloc(nports, sizeof(DevUnion));
+   attrs = calloc(nattributes, sizeof(XF86AttributeRec));
+   if (adapt == NULL || dev_unions == NULL || attrs == NULL) {
+      free(adapt);
+      free(dev_unions);
+      free(attrs);
+      return NULL;
+   }
+
+   adapt->type = XvWindowMask | XvInputMask | XvImageMask;
+   adapt->flags = 0;
+   adapt->name = "Gallium3D Textured Video";
+   adapt->nEncodings = 1;
+   adapt->pEncodings = DummyEncoding;
+   adapt->nFormats = NUM_FORMATS;
+   adapt->pFormats = Formats;
+   adapt->nPorts = 0;
+   adapt->pPortPrivates = dev_unions;
+   adapt->nAttributes = nattributes;
+   adapt->pAttributes = attrs;
+   memcpy(attrs, TexturedAttributes, nattributes * sizeof(XF86AttributeRec));
+   adapt->nImages = NUM_IMAGES;
+   adapt->pImages = Images;
+   adapt->PutVideo = NULL;
+   adapt->PutStill = NULL;
+   adapt->GetVideo = NULL;
+   adapt->GetStill = NULL;
+   adapt->StopVideo = stop_video;
+   adapt->SetPortAttribute = set_port_attribute;
+   adapt->GetPortAttribute = get_port_attribute;
+   adapt->QueryBestSize = query_best_size;
+   adapt->PutImage = put_image;
+   adapt->QueryImageAttributes = query_image_attributes;
+
+   for (i = 0; i < nports; i++) {
+      struct xorg_xv_port_priv *priv =
+         port_priv_create(ms->exa->renderer);
+
+      adapt->pPortPrivates[i].ptr = (pointer) (priv);
+      adapt->nPorts++;
+   }
+
+   return adapt;
+}
+
+void
+xorg_init_video(ScreenPtr pScreen)
+{
+   ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
+   /*modesettingPtr ms = modesettingPTR(pScrn);*/
+   XF86VideoAdaptorPtr *adaptors, *new_adaptors = NULL;
+   XF86VideoAdaptorPtr textured_adapter;
+   int num_adaptors;
+
+   num_adaptors = xf86XVListGenericAdaptors(pScrn, &adaptors);
+   new_adaptors = malloc((num_adaptors + 1) * sizeof(XF86VideoAdaptorPtr *));
+   if (new_adaptors == NULL)
+      return;
+
+   memcpy(new_adaptors, adaptors, num_adaptors * sizeof(XF86VideoAdaptorPtr));
+   adaptors = new_adaptors;
+
+   /* Add the adaptors supported by our hardware.  First, set up the atoms
+    * that will be used by both output adaptors.
+    */
+   xvBrightness = MAKE_ATOM("XV_BRIGHTNESS");
+   xvContrast = MAKE_ATOM("XV_CONTRAST");
+
+   textured_adapter = xorg_setup_textured_adapter(pScreen);
+
+   debug_assert(textured_adapter);
+
+   if (textured_adapter) {
+      adaptors[num_adaptors++] = textured_adapter;
+   }
+
+   if (num_adaptors) {
+      xf86XVScreenInit(pScreen, adaptors, num_adaptors);
+   } else {
+      xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
+                 "Disabling Xv because no adaptors could be initialized.\n");
+   }
+
+   free(adaptors);
+}
diff --git a/src/gallium/state_trackers/xorg/xvmc/Makefile b/src/gallium/state_trackers/xorg/xvmc/Makefile
new file mode 100644
index 0000000000..126dc6d58f
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/Makefile
@@ -0,0 +1,16 @@
+TOP = ../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = xvmctracker
+
+LIBRARY_INCLUDES = \
+	$(shell pkg-config --cflags-only-I xvmc) \
+	-I$(TOP)/src/gallium/winsys/g3dvl
+
+C_SOURCES = block.c \
+            surface.c \
+            context.c \
+            subpicture.c \
+            attributes.c
+
+include ../../../Makefile.template
diff --git a/src/gallium/state_trackers/xorg/xvmc/SConscript b/src/gallium/state_trackers/xorg/xvmc/SConscript
new file mode 100644
index 0000000000..cb25d68bd8
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/SConscript
@@ -0,0 +1,27 @@
+#######################################################################
+# SConscript for xvmc state_tracker
+
+Import('*')
+
+if 'xorg/xvmc' in env['statetrackers']:
+
+    env = env.Clone()
+    
+    env.Append(CPPPATH = [
+	'#/src/gallium/include',
+	'#/src/gallium/auxiliary',
+	'#/src/gallium/winsys/g3dvl',
+    ])
+
+    env.ParseConfig('pkg-config --cflags --libs xvmc')
+
+    st_xvmc = env.ConvenienceLibrary(
+	target = 'st_xvmc',
+	source = [ 'block.c',
+		'surface.c',
+		'context.c',
+		'subpicture.c',
+		'attributes.c',
+		]
+    )
+    Export('st_xvmc')
diff --git a/src/gallium/drivers/i965simple/brw_draw.h b/src/gallium/state_trackers/xorg/xvmc/attributes.c
index 62fe0d5d0e..79a67838e6 100644
--- a/src/gallium/drivers/i965simple/brw_draw.h
+++ b/src/gallium/state_trackers/xorg/xvmc/attributes.c
@@ -1,6 +1,6 @@
- /**************************************************************************
+/**************************************************************************
  * 
- * Copyright 2005 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2009 Younes Manton.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,31 +25,22 @@
  * 
  **************************************************************************/
 
-#ifndef BRW_DRAW_H
-#define BRW_DRAW_H
-
-#include "pipe/p_context.h"
-
-struct brw_context;
-
-
-
-void brw_init_draw_functions( struct brw_context *brw );
-
-
-boolean brw_upload_vertices( struct brw_context *brw,
-			       unsigned min_index,
-			       unsigned max_index );
-
-boolean brw_upload_indices(struct brw_context *brw,
-                           const struct pipe_buffer *index_buffer,
-                           int ib_size, int start, int count);
-
-boolean brw_upload_vertex_buffers( struct brw_context *brw );
-boolean brw_upload_vertex_elements( struct brw_context *brw );
-
-unsigned brw_translate_surface_format( unsigned id );
-
-
-
-#endif
+#include <assert.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/Xvlib.h>
+#include <X11/extensions/XvMClib.h>
+
+XvAttribute* XvMCQueryAttributes(Display *dpy, XvMCContext *context, int *number)
+{
+   return NULL;
+}
+
+Status XvMCSetAttribute(Display *dpy, XvMCContext *context, Atom attribute, int value)
+{
+   return BadImplementation;
+}
+
+Status XvMCGetAttribute(Display *dpy, XvMCContext *context, Atom attribute, int *value)
+{
+   return BadImplementation;
+}
diff --git a/src/gallium/state_trackers/xorg/xvmc/block.c b/src/gallium/state_trackers/xorg/xvmc/block.c
new file mode 100644
index 0000000000..5102375fcf
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/block.c
@@ -0,0 +1,88 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <assert.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMClib.h>
+#include <util/u_memory.h>
+#include "xvmc_private.h"
+
+Status XvMCCreateBlocks(Display *dpy, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks)
+{
+   assert(dpy);
+
+   if (!context)
+      return XvMCBadContext;
+   if (num_blocks == 0)
+      return BadValue;
+
+   assert(blocks);
+
+   blocks->context_id = context->context_id;
+   blocks->num_blocks = num_blocks;
+   blocks->blocks = MALLOC(BLOCK_SIZE_BYTES * num_blocks);
+   blocks->privData = NULL;
+
+   return Success;
+}
+
+Status XvMCDestroyBlocks(Display *dpy, XvMCBlockArray *blocks)
+{
+   assert(dpy);
+   assert(blocks);
+   FREE(blocks->blocks);
+
+   return Success;
+}
+
+Status XvMCCreateMacroBlocks(Display *dpy, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks)
+{
+   assert(dpy);
+
+   if (!context)
+      return XvMCBadContext;
+   if (num_blocks == 0)
+      return BadValue;
+
+   assert(blocks);
+
+   blocks->context_id = context->context_id;
+   blocks->num_blocks = num_blocks;
+   blocks->macro_blocks = MALLOC(sizeof(XvMCMacroBlock) * num_blocks);
+   blocks->privData = NULL;
+
+   return Success;
+}
+
+Status XvMCDestroyMacroBlocks(Display *dpy, XvMCMacroBlockArray *blocks)
+{
+   assert(dpy);
+   assert(blocks);
+   FREE(blocks->macro_blocks);
+
+   return Success;
+}
diff --git a/src/gallium/state_trackers/xorg/xvmc/context.c b/src/gallium/state_trackers/xorg/xvmc/context.c
new file mode 100644
index 0000000000..942692d1bb
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/context.c
@@ -0,0 +1,252 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <assert.h>
+#include <X11/Xlibint.h>
+#include <X11/extensions/XvMClib.h>
+#include <pipe/p_screen.h>
+#include <pipe/p_video_context.h>
+#include <pipe/p_video_state.h>
+#include <pipe/p_state.h>
+#include <vl_winsys.h>
+#include <util/u_memory.h>
+#include <util/u_debug.h>
+#include <vl/vl_csc.h>
+#include "xvmc_private.h"
+
+static Status Validate(Display *dpy, XvPortID port, int surface_type_id,
+                       unsigned int width, unsigned int height, int flags,
+                       bool *found_port, int *screen, int *chroma_format,
+                       int *mc_type, int *surface_flags)
+{
+   bool found_surface = false;
+   XvAdaptorInfo *adaptor_info;
+   unsigned int num_adaptors;
+   int num_types;
+   unsigned int max_width, max_height;
+   Status ret;
+
+   assert(dpy);
+   assert(found_port);
+   assert(screen);
+   assert(chroma_format);
+   assert(mc_type);
+   assert(surface_flags);
+
+   *found_port = false;
+
+   for (unsigned int i = 0; i < XScreenCount(dpy); ++i) {
+      ret = XvQueryAdaptors(dpy, XRootWindow(dpy, i), &num_adaptors, &adaptor_info);
+      if (ret != Success)
+         return ret;
+
+      for (unsigned int j = 0; j < num_adaptors && !*found_port; ++j) {
+         for (unsigned int k = 0; k < adaptor_info[j].num_ports && !*found_port; ++k) {
+            XvMCSurfaceInfo *surface_info;
+
+            if (adaptor_info[j].base_id + k != port)
+               continue;
+
+            *found_port = true;
+
+            surface_info = XvMCListSurfaceTypes(dpy, adaptor_info[j].base_id, &num_types);
+            if (!surface_info) {
+               XvFreeAdaptorInfo(adaptor_info);
+               return BadAlloc;
+            }
+
+            for (unsigned int l = 0; l < num_types && !found_surface; ++l) {
+               if (surface_info[l].surface_type_id != surface_type_id)
+                  continue;
+
+               found_surface = true;
+               max_width = surface_info[l].max_width;
+               max_height = surface_info[l].max_height;
+               *chroma_format = surface_info[l].chroma_format;
+               *mc_type = surface_info[l].mc_type;
+               *surface_flags = surface_info[l].flags;
+               *screen = i;
+            }
+
+            XFree(surface_info);
+         }
+      }
+
+      XvFreeAdaptorInfo(adaptor_info);
+   }
+
+   if (!*found_port)
+      return XvBadPort;
+   if (!found_surface)
+      return BadMatch;
+   if (width > max_width || height > max_height)
+      return BadValue;
+   if (flags != XVMC_DIRECT && flags != 0)
+      return BadValue;
+
+   return Success;
+}
+
+static enum pipe_video_profile ProfileToPipe(int xvmc_profile)
+{
+   if (xvmc_profile & XVMC_MPEG_1)
+      assert(0);
+   if (xvmc_profile & XVMC_MPEG_2)
+      return PIPE_VIDEO_PROFILE_MPEG2_MAIN;
+   if (xvmc_profile & XVMC_H263)
+      assert(0);
+   if (xvmc_profile & XVMC_MPEG_4)
+      assert(0);
+	
+   assert(0);
+
+   return -1;
+}
+
+static enum pipe_video_chroma_format FormatToPipe(int xvmc_format)
+{
+   switch (xvmc_format) {
+      case XVMC_CHROMA_FORMAT_420:
+         return PIPE_VIDEO_CHROMA_FORMAT_420;
+      case XVMC_CHROMA_FORMAT_422:
+         return PIPE_VIDEO_CHROMA_FORMAT_422;
+      case XVMC_CHROMA_FORMAT_444:
+         return PIPE_VIDEO_CHROMA_FORMAT_444;
+      default:
+         assert(0);
+   }
+
+   return -1;
+}
+
+Status XvMCCreateContext(Display *dpy, XvPortID port, int surface_type_id,
+                         int width, int height, int flags, XvMCContext *context)
+{
+   bool found_port;
+   int scrn;
+   int chroma_format;
+   int mc_type;
+   int surface_flags;
+   Status ret;
+   struct pipe_screen *screen;
+   struct pipe_video_context *vpipe;
+   XvMCContextPrivate *context_priv;
+   float csc[16];
+
+   assert(dpy);
+
+   if (!context)
+      return XvMCBadContext;
+
+   ret = Validate(dpy, port, surface_type_id, width, height, flags,
+                  &found_port, &scrn, &chroma_format, &mc_type, &surface_flags);
+
+   /* Success and XvBadPort have the same value */
+   if (ret != Success || !found_port)
+      return ret;
+
+   /* XXX: Current limits */
+   if (chroma_format != XVMC_CHROMA_FORMAT_420) {
+      debug_printf("[XvMCg3dvl] Cannot decode requested surface type. Unsupported chroma format.\n");
+      return BadImplementation;
+   }
+   if (mc_type != (XVMC_MOCOMP | XVMC_MPEG_2)) {
+      debug_printf("[XvMCg3dvl] Cannot decode requested surface type. Non-MPEG2/Mocomp acceleration unsupported.\n");
+      return BadImplementation;
+   }
+   if (!(surface_flags & XVMC_INTRA_UNSIGNED)) {
+      debug_printf("[XvMCg3dvl] Cannot decode requested surface type. Signed intra unsupported.\n");
+      return BadImplementation;
+   }
+
+   context_priv = CALLOC(1, sizeof(XvMCContextPrivate));
+   if (!context_priv)
+      return BadAlloc;
+
+   /* TODO: Reuse screen if process creates another context */
+   screen = vl_screen_create(dpy, scrn);
+
+   if (!screen) {
+      FREE(context_priv);
+      return BadAlloc;
+   }
+
+   vpipe = vl_video_create(screen, ProfileToPipe(mc_type),
+                           FormatToPipe(chroma_format), width, height);
+
+   if (!vpipe) {
+      screen->destroy(screen);
+      FREE(context_priv);
+      return BadAlloc;
+   }
+
+   /* TODO: Define some Xv attribs to allow users to specify color standard, procamp */
+   vl_csc_get_matrix
+   (
+      debug_get_bool_option("G3DVL_NO_CSC", FALSE) ?
+      VL_CSC_COLOR_STANDARD_IDENTITY : VL_CSC_COLOR_STANDARD_BT_601,
+      NULL, true, csc
+   );
+   vpipe->set_csc_matrix(vpipe, csc);
+
+   context_priv->vpipe = vpipe;
+
+   context->context_id = XAllocID(dpy);
+   context->surface_type_id = surface_type_id;
+   context->width = width;
+   context->height = height;
+   context->flags = flags;
+   context->port = port;
+   context->privData = context_priv;
+	
+   SyncHandle();
+
+   return Success;
+}
+
+Status XvMCDestroyContext(Display *dpy, XvMCContext *context)
+{
+   struct pipe_screen *screen;
+   struct pipe_video_context *vpipe;
+   XvMCContextPrivate *context_priv;
+
+   assert(dpy);
+
+   if (!context || !context->privData)
+      return XvMCBadContext;
+
+   context_priv = context->privData;
+   vpipe = context_priv->vpipe;
+   pipe_surface_reference(&context_priv->backbuffer, NULL);
+   screen = vpipe->screen;
+   vpipe->destroy(vpipe);
+   screen->destroy(screen);
+   FREE(context_priv);
+   context->privData = NULL;
+
+   return Success;
+}
diff --git a/src/gallium/state_trackers/xorg/xvmc/subpicture.c b/src/gallium/state_trackers/xorg/xvmc/subpicture.c
new file mode 100644
index 0000000000..69898d5fcd
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/subpicture.c
@@ -0,0 +1,195 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <assert.h>
+#include <X11/Xlibint.h>
+#include <X11/extensions/XvMClib.h>
+
+Status XvMCCreateSubpicture(Display *dpy, XvMCContext *context, XvMCSubpicture *subpicture,
+                            unsigned short width, unsigned short height, int xvimage_id)
+{
+   assert(dpy);
+
+   if (!context)
+      return XvMCBadContext;
+
+   assert(subpicture);
+
+   /*if (width > || height > )
+      return BadValue;*/
+
+   /*if (xvimage_id != )
+      return BadMatch;*/
+
+   subpicture->subpicture_id = XAllocID(dpy);
+   subpicture->context_id = context->context_id;
+   subpicture->xvimage_id = xvimage_id;
+   subpicture->width = width;
+   subpicture->height = height;
+   subpicture->num_palette_entries = 0;
+   subpicture->entry_bytes = 0;
+   subpicture->component_order[0] = 0;
+   subpicture->component_order[1] = 0;
+   subpicture->component_order[2] = 0;
+   subpicture->component_order[3] = 0;
+   /* TODO: subpicture->privData = ;*/
+
+   SyncHandle();
+
+   return Success;
+}
+
+Status XvMCClearSubpicture(Display *dpy, XvMCSubpicture *subpicture, short x, short y,
+                           unsigned short width, unsigned short height, unsigned int color)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   /* TODO: Assert clear rect is within bounds? Or clip? */
+
+   return Success;
+}
+
+Status XvMCCompositeSubpicture(Display *dpy, XvMCSubpicture *subpicture, XvImage *image,
+                               short srcx, short srcy, unsigned short width, unsigned short height,
+                               short dstx, short dsty)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   assert(image);
+
+   if (subpicture->xvimage_id != image->id)
+      return BadMatch;
+
+   /* TODO: Assert rects are within bounds? Or clip? */
+
+   return Success;
+}
+
+Status XvMCDestroySubpicture(Display *dpy, XvMCSubpicture *subpicture)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   return BadImplementation;
+}
+
+Status XvMCSetSubpicturePalette(Display *dpy, XvMCSubpicture *subpicture, unsigned char *palette)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   assert(palette);
+
+   /* We don't support paletted subpictures */
+   return BadMatch;
+}
+
+Status XvMCBlendSubpicture(Display *dpy, XvMCSurface *target_surface, XvMCSubpicture *subpicture,
+                           short subx, short suby, unsigned short subw, unsigned short subh,
+                           short surfx, short surfy, unsigned short surfw, unsigned short surfh)
+{
+   assert(dpy);
+
+   if (!target_surface)
+      return XvMCBadSurface;
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   if (target_surface->context_id != subpicture->context_id)
+      return BadMatch;
+
+   /* TODO: Assert rects are within bounds? Or clip? */
+   return Success;
+}
+
+Status XvMCBlendSubpicture2(Display *dpy, XvMCSurface *source_surface, XvMCSurface *target_surface,
+                            XvMCSubpicture *subpicture, short subx, short suby, unsigned short subw, unsigned short subh,
+                            short surfx, short surfy, unsigned short surfw, unsigned short surfh)
+{
+   assert(dpy);
+
+   if (!source_surface || !target_surface)
+      return XvMCBadSurface;
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   if (source_surface->context_id != subpicture->context_id)
+      return BadMatch;
+
+   if (source_surface->context_id != subpicture->context_id)
+      return BadMatch;
+
+   /* TODO: Assert rects are within bounds? Or clip? */
+   return Success;
+}
+
+Status XvMCSyncSubpicture(Display *dpy, XvMCSubpicture *subpicture)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   return Success;
+}
+
+Status XvMCFlushSubpicture(Display *dpy, XvMCSubpicture *subpicture)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   return Success;
+}
+
+Status XvMCGetSubpictureStatus(Display *dpy, XvMCSubpicture *subpicture, int *status)
+{
+   assert(dpy);
+
+   if (!subpicture)
+      return XvMCBadSubpicture;
+
+   assert(status);
+
+   /* TODO */
+   *status = 0;
+
+   return Success;
+}
diff --git a/src/gallium/state_trackers/xorg/xvmc/surface.c b/src/gallium/state_trackers/xorg/xvmc/surface.c
new file mode 100644
index 0000000000..bf9038f356
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/surface.c
@@ -0,0 +1,409 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <assert.h>
+#include <X11/Xlibint.h>
+#include <pipe/p_video_context.h>
+#include <pipe/p_video_state.h>
+#include <pipe/p_state.h>
+#include <util/u_memory.h>
+#include "xvmc_private.h"
+
+static enum pipe_mpeg12_macroblock_type TypeToPipe(int xvmc_mb_type)
+{
+   if (xvmc_mb_type & XVMC_MB_TYPE_INTRA)
+      return PIPE_MPEG12_MACROBLOCK_TYPE_INTRA;
+   if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_FORWARD)
+      return PIPE_MPEG12_MACROBLOCK_TYPE_FWD;
+   if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_BACKWARD)
+      return PIPE_MPEG12_MACROBLOCK_TYPE_BKWD;
+   if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD))
+      return PIPE_MPEG12_MACROBLOCK_TYPE_BI;
+
+   assert(0);
+
+   return -1;
+}
+
+static enum pipe_mpeg12_picture_type PictureToPipe(int xvmc_pic)
+{
+   switch (xvmc_pic) {
+      case XVMC_TOP_FIELD:
+         return PIPE_MPEG12_PICTURE_TYPE_FIELD_TOP;
+      case XVMC_BOTTOM_FIELD:
+         return PIPE_MPEG12_PICTURE_TYPE_FIELD_BOTTOM;
+      case XVMC_FRAME_PICTURE:
+         return PIPE_MPEG12_PICTURE_TYPE_FRAME;
+      default:
+         assert(0);
+   }
+
+   return -1;
+}
+
+static enum pipe_mpeg12_motion_type MotionToPipe(int xvmc_motion_type, int xvmc_dct_type)
+{
+   switch (xvmc_motion_type) {
+      case XVMC_PREDICTION_FRAME:
+         return xvmc_dct_type == XVMC_DCT_TYPE_FIELD ?
+            PIPE_MPEG12_MOTION_TYPE_16x8 : PIPE_MPEG12_MOTION_TYPE_FRAME;
+      case XVMC_PREDICTION_FIELD:
+         return PIPE_MPEG12_MOTION_TYPE_FIELD;
+      case XVMC_PREDICTION_DUAL_PRIME:
+         return PIPE_MPEG12_MOTION_TYPE_DUALPRIME;
+      default:
+         assert(0);
+   }
+
+   return -1;
+}
+
+static bool
+CreateOrResizeBackBuffer(struct pipe_video_context *vpipe, unsigned int width, unsigned int height,
+                         struct pipe_surface **backbuffer)
+{
+   struct pipe_texture template;
+   struct pipe_texture *tex;
+
+   assert(vpipe);
+
+   if (*backbuffer) {
+      if ((*backbuffer)->width != width || (*backbuffer)->height != height)
+         pipe_surface_reference(backbuffer, NULL);
+      else
+         return true;
+   }
+
+   memset(&template, 0, sizeof(struct pipe_texture));
+   template.target = PIPE_TEXTURE_2D;
+   /* XXX: Needs to match the drawable's format? */
+   template.format = PIPE_FORMAT_X8R8G8B8_UNORM;
+   template.last_level = 0;
+   template.width[0] = width;
+   template.height[0] = height;
+   template.depth[0] = 1;
+   pf_get_block(template.format, &template.block);
+   template.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
+
+   tex = vpipe->screen->texture_create(vpipe->screen, &template);
+   if (!tex)
+      return false;
+
+   *backbuffer = vpipe->screen->get_tex_surface(vpipe->screen, tex, 0, 0, 0,
+                                                PIPE_BUFFER_USAGE_GPU_READ |
+                                                PIPE_BUFFER_USAGE_GPU_WRITE);
+   pipe_texture_reference(&tex, NULL);
+
+   if (!*backbuffer)
+      return false;
+
+   /* Clear the backbuffer in case the video doesn't cover the whole window */
+   /* FIXME: Need to clear every time a frame moves and leaves dirty rects */
+   vpipe->clear_surface(vpipe, 0, 0, width, height, 0, *backbuffer);
+
+   return true;
+}
+
+static void
+MacroBlocksToPipe(const XvMCMacroBlockArray *xvmc_macroblocks,
+                  const XvMCBlockArray *xvmc_blocks,
+                  unsigned int first_macroblock,
+                  unsigned int num_macroblocks,
+                  struct pipe_mpeg12_macroblock *pipe_macroblocks)
+{
+   unsigned int i, j, k, l;
+   XvMCMacroBlock *xvmc_mb;
+
+   assert(xvmc_macroblocks);
+   assert(xvmc_blocks);
+   assert(pipe_macroblocks);
+   assert(num_macroblocks);
+
+   xvmc_mb = xvmc_macroblocks->macro_blocks + first_macroblock;
+
+   for (i = 0; i < num_macroblocks; ++i) {
+      pipe_macroblocks->base.codec = PIPE_VIDEO_CODEC_MPEG12;
+      pipe_macroblocks->mbx = xvmc_mb->x;
+      pipe_macroblocks->mby = xvmc_mb->y;
+      pipe_macroblocks->mb_type = TypeToPipe(xvmc_mb->macroblock_type);
+      if (pipe_macroblocks->mb_type != PIPE_MPEG12_MACROBLOCK_TYPE_INTRA)
+         pipe_macroblocks->mo_type = MotionToPipe(xvmc_mb->motion_type, xvmc_mb->dct_type);
+      /* Get rid of Valgrind 'undefined' warnings */
+      else
+         pipe_macroblocks->mo_type = -1;
+      pipe_macroblocks->dct_type = xvmc_mb->dct_type == XVMC_DCT_TYPE_FIELD ?
+         PIPE_MPEG12_DCT_TYPE_FIELD : PIPE_MPEG12_DCT_TYPE_FRAME;
+
+      for (j = 0; j < 2; ++j)
+         for (k = 0; k < 2; ++k)
+            for (l = 0; l < 2; ++l)
+               pipe_macroblocks->pmv[j][k][l] = xvmc_mb->PMV[j][k][l];
+
+      pipe_macroblocks->cbp = xvmc_mb->coded_block_pattern;
+      pipe_macroblocks->blocks = xvmc_blocks->blocks + xvmc_mb->index * BLOCK_SIZE_SAMPLES;
+
+      ++pipe_macroblocks;
+      ++xvmc_mb;
+   }
+}
+
+Status XvMCCreateSurface(Display *dpy, XvMCContext *context, XvMCSurface *surface)
+{
+   XvMCContextPrivate *context_priv;
+   struct pipe_video_context *vpipe;
+   XvMCSurfacePrivate *surface_priv;
+   struct pipe_video_surface *vsfc;
+
+   assert(dpy);
+
+   if (!context)
+      return XvMCBadContext;
+   if (!surface)
+      return XvMCBadSurface;
+
+   context_priv = context->privData;
+   vpipe = context_priv->vpipe;
+
+   surface_priv = CALLOC(1, sizeof(XvMCSurfacePrivate));
+   if (!surface_priv)
+      return BadAlloc;
+
+   vsfc = vpipe->screen->video_surface_create(vpipe->screen, vpipe->chroma_format,
+                                              vpipe->width, vpipe->height);
+   if (!vsfc) {
+      FREE(surface_priv);
+      return BadAlloc;
+   }
+
+   surface_priv->pipe_vsfc = vsfc;
+   surface_priv->context = context;
+
+   surface->surface_id = XAllocID(dpy);
+   surface->context_id = context->context_id;
+   surface->surface_type_id = context->surface_type_id;
+   surface->width = context->width;
+   surface->height = context->height;
+   surface->privData = surface_priv;
+
+   SyncHandle();
+
+   return Success;
+}
+
+Status XvMCRenderSurface(Display *dpy, XvMCContext *context, unsigned int picture_structure,
+                         XvMCSurface *target_surface, XvMCSurface *past_surface, XvMCSurface *future_surface,
+                         unsigned int flags, unsigned int num_macroblocks, unsigned int first_macroblock,
+                         XvMCMacroBlockArray *macroblocks, XvMCBlockArray *blocks
+)
+{
+   struct pipe_video_context *vpipe;
+   struct pipe_surface *t_vsfc;
+   struct pipe_surface *p_vsfc;
+   struct pipe_surface *f_vsfc;
+   XvMCContextPrivate *context_priv;
+   XvMCSurfacePrivate *target_surface_priv;
+   XvMCSurfacePrivate *past_surface_priv;
+   XvMCSurfacePrivate *future_surface_priv;
+   struct pipe_mpeg12_macroblock pipe_macroblocks[num_macroblocks];
+
+   assert(dpy);
+
+   if (!context || !context->privData)
+      return XvMCBadContext;
+   if (!target_surface || !target_surface->privData)
+      return XvMCBadSurface;
+
+   if (picture_structure != XVMC_TOP_FIELD &&
+       picture_structure != XVMC_BOTTOM_FIELD &&
+       picture_structure != XVMC_FRAME_PICTURE)
+      return BadValue;
+   /* Bkwd pred equivalent to fwd (past && !future) */
+   if (future_surface && !past_surface)
+      return BadMatch;
+
+   assert(context->context_id == target_surface->context_id);
+   assert(!past_surface || context->context_id == past_surface->context_id);
+   assert(!future_surface || context->context_id == future_surface->context_id);
+
+   assert(macroblocks);
+   assert(blocks);
+
+   assert(macroblocks->context_id == context->context_id);
+   assert(blocks->context_id == context->context_id);
+
+   assert(flags == 0 || flags == XVMC_SECOND_FIELD);
+
+   target_surface_priv = target_surface->privData;
+   past_surface_priv = past_surface ? past_surface->privData : NULL;
+   future_surface_priv = future_surface ? future_surface->privData : NULL;
+
+   assert(target_surface_priv->context == context);
+   assert(!past_surface || past_surface_priv->context == context);
+   assert(!future_surface || future_surface_priv->context == context);
+
+   context_priv = context->privData;
+   vpipe = context_priv->vpipe;
+
+   t_vsfc = target_surface_priv->pipe_vsfc;
+   p_vsfc = past_surface ? past_surface_priv->pipe_vsfc : NULL;
+   f_vsfc = future_surface ? future_surface_priv->pipe_vsfc : NULL;
+
+   MacroBlocksToPipe(macroblocks, blocks, first_macroblock,
+                     num_macroblocks, pipe_macroblocks);
+
+   vpipe->set_decode_target(vpipe, t_vsfc);
+   vpipe->decode_macroblocks(vpipe, p_vsfc, f_vsfc, num_macroblocks,
+                             &pipe_macroblocks->base, target_surface_priv->render_fence);
+
+   return Success;
+}
+
+Status XvMCFlushSurface(Display *dpy, XvMCSurface *surface)
+{
+   assert(dpy);
+
+   if (!surface)
+      return XvMCBadSurface;
+
+   return Success;
+}
+
+Status XvMCSyncSurface(Display *dpy, XvMCSurface *surface)
+{
+   assert(dpy);
+
+   if (!surface)
+      return XvMCBadSurface;
+
+   return Success;
+}
+
+Status XvMCPutSurface(Display *dpy, XvMCSurface *surface, Drawable drawable,
+                      short srcx, short srcy, unsigned short srcw, unsigned short srch,
+                      short destx, short desty, unsigned short destw, unsigned short desth,
+                      int flags)
+{
+   Window root;
+   int x, y;
+   unsigned int width, height;
+   unsigned int border_width;
+   unsigned int depth;
+   struct pipe_video_context *vpipe;
+   XvMCSurfacePrivate *surface_priv;
+   XvMCContextPrivate *context_priv;
+   XvMCContext *context;
+   struct pipe_video_rect src_rect = {srcx, srcy, srcw, srch};
+   struct pipe_video_rect dst_rect = {destx, desty, destw, desth};
+
+   assert(dpy);
+
+   if (!surface || !surface->privData)
+      return XvMCBadSurface;
+
+   if (XGetGeometry(dpy, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable)
+      return BadDrawable;
+
+   assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE);
+   assert(srcx + srcw - 1 < surface->width);
+   assert(srcy + srch - 1 < surface->height);
+   /*
+    * Some apps (mplayer) hit these asserts because they call
+    * this function after the window has been resized by the WM
+    * but before they've handled the corresponding XEvent and
+    * know about the new dimensions. The output should be clipped
+    * until the app updates destw and desth.
+    */
+   /*
+   assert(destx + destw - 1 < width);
+   assert(desty + desth - 1 < height);
+    */
+
+   surface_priv = surface->privData;
+   context = surface_priv->context;
+   context_priv = context->privData;
+   vpipe = context_priv->vpipe;
+
+   if (!CreateOrResizeBackBuffer(vpipe, width, height, &context_priv->backbuffer))
+      return BadAlloc;
+
+   vpipe->render_picture(vpipe, surface_priv->pipe_vsfc, PictureToPipe(flags), &src_rect,
+                         context_priv->backbuffer, &dst_rect, surface_priv->disp_fence);
+
+   vl_video_bind_drawable(vpipe, drawable);
+	
+   vpipe->screen->flush_frontbuffer
+   (
+      vpipe->screen,
+      context_priv->backbuffer,
+      vpipe->priv
+   );
+
+   return Success;
+}
+
+Status XvMCGetSurfaceStatus(Display *dpy, XvMCSurface *surface, int *status)
+{
+   assert(dpy);
+
+   if (!surface)
+      return XvMCBadSurface;
+
+   assert(status);
+
+   *status = 0;
+
+   return Success;
+}
+
+Status XvMCDestroySurface(Display *dpy, XvMCSurface *surface)
+{
+   XvMCSurfacePrivate *surface_priv;
+
+   assert(dpy);
+
+   if (!surface || !surface->privData)
+      return XvMCBadSurface;
+
+   surface_priv = surface->privData;
+   pipe_video_surface_reference(&surface_priv->pipe_vsfc, NULL);
+   FREE(surface_priv);
+   surface->privData = NULL;
+
+   return Success;
+}
+
+Status XvMCHideSurface(Display *dpy, XvMCSurface *surface)
+{
+   assert(dpy);
+
+   if (!surface || !surface->privData)
+      return XvMCBadSurface;
+
+   /* No op, only for overlaid rendering */
+
+   return Success;
+}
diff --git a/src/xvmc/tests/.gitignore b/src/gallium/state_trackers/xorg/xvmc/tests/.gitignore
index e1d2f9023d..e1d2f9023d 100644
--- a/src/xvmc/tests/.gitignore
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/.gitignore
diff --git a/src/gallium/state_trackers/xorg/xvmc/tests/Makefile b/src/gallium/state_trackers/xorg/xvmc/tests/Makefile
new file mode 100644
index 0000000000..c875dd7605
--- /dev/null
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/Makefile
@@ -0,0 +1,28 @@
+TOP = ../../../../../..
+include $(TOP)/configs/current
+
+LIBS = -lXvMCW -lXvMC -lXv -lX11
+
+#############################################
+
+.PHONY: default clean
+
+default: test_context test_surface test_blocks test_rendering xvmc_bench
+
+test_context: test_context.o testlib.o
+	$(CC) $(LDFLAGS) -o $@ $^ $(LIBS)
+
+test_surface: test_surface.o testlib.o
+	$(CC) $(LDFLAGS) -o $@ $^ $(LIBS)
+
+test_blocks: test_blocks.o testlib.o
+	$(CC) $(LDFLAGS) -o $@ $^ $(LIBS)
+
+test_rendering: test_rendering.o testlib.o
+	$(CC) $(LDFLAGS) -o $@ $^ $(LIBS)
+
+xvmc_bench: xvmc_bench.o testlib.o
+	$(CC) $(LDFLAGS) -o $@ $^ $(LIBS)
+
+clean:
+	$(RM) -rf *.o test_context test_surface test_blocks test_rendering xvmc_bench
diff --git a/src/xvmc/tests/test_blocks.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_blocks.c
index dc80adfa65..994e3ca4d1 100644
--- a/src/xvmc/tests/test_blocks.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_blocks.c
@@ -1,3 +1,30 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #include <assert.h>
 #include <error.h>
 #include "testlib.h"
diff --git a/src/xvmc/tests/test_context.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_context.c
index 53f7449cd0..3da957c933 100644
--- a/src/xvmc/tests/test_context.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_context.c
@@ -1,3 +1,30 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #include <assert.h>
 #include <error.h>
 #include "testlib.h"
diff --git a/src/xvmc/tests/test_rendering.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_rendering.c
index 1e7467a3aa..6058783a79 100644
--- a/src/xvmc/tests/test_rendering.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_rendering.c
@@ -1,3 +1,30 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #include <assert.h>
 #include <stdio.h>
 #include <string.h>
@@ -23,6 +50,9 @@
 #define DEFAULT_OUTPUT_HEIGHT		INPUT_HEIGHT
 #define DEFAULT_ACCEPTABLE_ERR		0.01
 
+void ParseArgs(int argc, char **argv, unsigned int *output_width, unsigned int *output_height, double *acceptable_error, int *prompt);
+void Gradient(short *block, unsigned int start, unsigned int stop, int horizontal);
+
 void ParseArgs(int argc, char **argv, unsigned int *output_width, unsigned int *output_height, double *acceptable_error, int *prompt)
 {
 	int fail = 0;
diff --git a/src/xvmc/tests/test_surface.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_surface.c
index 25ebdcc4fc..b65eb265c0 100644
--- a/src/xvmc/tests/test_surface.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_surface.c
@@ -1,3 +1,30 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #include <assert.h>
 #include <error.h>
 #include "testlib.h"
@@ -6,7 +33,7 @@ int main(int argc, char **argv)
 {
 	const unsigned int	width = 16, height = 16;
 	const unsigned int	mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2};
-	
+
 	Display			*display;
 	XvPortID		port_num;
 	int			surface_type_id;
@@ -14,9 +41,9 @@ int main(int argc, char **argv)
 	int			colorkey;
 	XvMCContext		context;
 	XvMCSurface		surface = {0};
-	
+
 	display = XOpenDisplay(NULL);
-	
+
 	if (!GetPort
 	(
 		display,
@@ -34,15 +61,15 @@ int main(int argc, char **argv)
 		XCloseDisplay(display);
 		error(1, 0, "Error, unable to find a good port.\n");
 	}
-	
+
 	if (is_overlay)
 	{
 		Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0);
 		XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey);
 	}
-	
+
 	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success);
-	
+
 	/* Test NULL context */
 	assert(XvMCCreateSurface(display, NULL, &surface) == XvMCBadContext);
 	/* Test NULL surface */
@@ -61,12 +88,11 @@ int main(int argc, char **argv)
 	assert(XvMCDestroySurface(display, &surface) == Success);
 	/* Test NULL surface */
 	assert(XvMCDestroySurface(display, NULL) == XvMCBadSurface);
-	
+
 	assert(XvMCDestroyContext(display, &context) == Success);
-	
+
 	XvUngrabPort(display, port_num, CurrentTime);
 	XCloseDisplay(display);
-	
+
 	return 0;
 }
-
diff --git a/src/xvmc/tests/testlib.c b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.c
index 59a03ca813..142c09bb59 100644
--- a/src/xvmc/tests/testlib.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.c
@@ -1,3 +1,30 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #include "testlib.h"
 #include <stdio.h>
 
diff --git a/src/gallium/drivers/i965simple/brw_flush.c b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.h
index e6001c30d9..0438e52928 100644
--- a/src/gallium/drivers/i965simple/brw_flush.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2009 Younes Manton.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,49 +25,45 @@
  * 
  **************************************************************************/
 
-/* Author:
- *    Keith Whitwell <keith@tungstengraphics.com>
- */
-
-
-#include "pipe/p_defines.h"
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_batch.h"
-
+#ifndef testlib_h
+#define testlib_h
 
-static void brw_flush( struct pipe_context *pipe,
-                       unsigned flags,
-                       struct pipe_fence_handle **fence )
-{
-   struct brw_context *brw = brw_context(pipe);
+/*
+#define TEST(pred, doc)	test(pred, #pred, doc, __FILE__, __LINE__)
 
-   /* Do we need to emit an MI_FLUSH command to flush the hardware
-    * caches?
-    */
-   if (flags & (PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_TEXTURE_CACHE)) {
-      struct brw_mi_flush flush;
+void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line);
+*/
 
-      memset(&flush, 0, sizeof(flush));      
-      flush.opcode = CMD_MI_FLUSH;
+#include <sys/time.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMClib.h>
 
-      if (!(flags & PIPE_FLUSH_RENDER_CACHE))
-	 flush.flags |= BRW_INHIBIT_FLUSH_RENDER_CACHE;
-
-      if (flags & PIPE_FLUSH_TEXTURE_CACHE)
-	 flush.flags |= BRW_FLUSH_READ_CACHE;
-
-      BRW_BATCH_STRUCT(brw, &flush);
-   }
-
-   /* If there are no flags, just flush pending commands to hardware:
-    */
-   FLUSH_BATCH( fence );
-}
+/*
+ * display: IN			A valid X display
+ * width, height: IN		Surface size that the port must display
+ * chroma_format: IN		Chroma format that the port must display
+ * mc_types, num_mc_types: IN	List of MC types that the port must support, first port that matches the first mc_type will be returned
+ * port_id: OUT			Your port's ID
+ * surface_type_id: OUT		Your port's surface ID
+ * is_overlay: OUT		If 1, port uses overlay surfaces, you need to set a colorkey
+ * intra_unsigned: OUT		If 1, port uses unsigned values for intra-coded blocks
+ */
+int GetPort
+(
+	Display *display,
+	unsigned int width,
+	unsigned int height,
+	unsigned int chroma_format,
+	const unsigned int *mc_types,
+	unsigned int num_mc_types,
+	XvPortID *port_id,
+	int *surface_type_id,
+	unsigned int *is_overlay,
+	unsigned int *intra_unsigned
+);
 
+unsigned int align(unsigned int value, unsigned int alignment);
 
+int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y);
 
-void brw_init_flush_functions( struct brw_context *brw )
-{
-   brw->pipe.flush = brw_flush;
-}
+#endif
diff --git a/src/xvmc/tests/xvmc_bench.c b/src/gallium/state_trackers/xorg/xvmc/tests/xvmc_bench.c
index d5a39ecf17..bf94d85623 100644
--- a/src/xvmc/tests/xvmc_bench.c
+++ b/src/gallium/state_trackers/xorg/xvmc/tests/xvmc_bench.c
@@ -1,3 +1,30 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #include <assert.h>
 #include <stdio.h>
 #include <string.h>
@@ -32,6 +59,8 @@ struct Config
 	unsigned int reps;
 };
 
+void ParseArgs(int argc, char **argv, struct Config *config);
+
 void ParseArgs(int argc, char **argv, struct Config *config)
 {
 	int fail = 0;
diff --git a/src/gallium/state_trackers/wgl/shared/stw_winsys.h b/src/gallium/state_trackers/xorg/xvmc/xvmc_private.h
index c0bf82c9ed..42337631ca 100644
--- a/src/gallium/state_trackers/wgl/shared/stw_winsys.h
+++ b/src/gallium/state_trackers/xorg/xvmc/xvmc_private.h
@@ -1,8 +1,8 @@
 /**************************************************************************
- *
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * 
+ * Copyright 2009 Younes Manton.
  * All Rights Reserved.
- *
+ * 
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -10,11 +10,11 @@
  * distribute, sub license, and/or sell copies of the Software, and to
  * permit persons to whom the Software is furnished to do so, subject to
  * the following conditions:
- *
+ * 
  * The above copyright notice and this permission notice (including the
  * next paragraph) shall be included in all copies or substantial portions
  * of the Software.
- *
+ * 
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
@@ -22,44 +22,37 @@
  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
+ * 
  **************************************************************************/
 
-#ifndef STW_WINSYS_H
-#define STW_WINSYS_H
+#ifndef xvmc_private_h
+#define xvmc_private_h
 
-#include <windows.h> /* for HDC */
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMClib.h>
 
-#include "pipe/p_compiler.h"
+#define BLOCK_SIZE_SAMPLES 64
+#define BLOCK_SIZE_BYTES (BLOCK_SIZE_SAMPLES * 2)
 
-struct pipe_screen;
-struct pipe_context;
+struct pipe_video_context;
 struct pipe_surface;
+struct pipe_fence_handle;
 
-struct stw_winsys
+typedef struct
 {
-   struct pipe_screen *
-   (*create_screen)( void );
+	struct pipe_video_context *vpipe;
+	struct pipe_surface *backbuffer;
+} XvMCContextPrivate;
 
-   struct pipe_context *
-   (*create_context)( struct pipe_screen *screen );
-
-   void
-   (*flush_frontbuffer)( struct pipe_screen *screen,
-                         struct pipe_surface *surf,
-                         HDC hDC );
-};
-
-boolean
-stw_init(const struct stw_winsys *stw_winsys);
-
-boolean
-stw_init_thread(void);
-
-void
-stw_cleanup_thread(void);
-
-void
-stw_cleanup(void);
-
-#endif /* STW_WINSYS_H */
+typedef struct
+{
+	struct pipe_video_surface *pipe_vsfc;
+	struct pipe_fence_handle *render_fence;
+	struct pipe_fence_handle *disp_fence;
+	
+	/* Some XvMC functions take a surface but not a context,
+	   so we keep track of which context each surface belongs to. */
+	XvMCContext *context;
+} XvMCSurfacePrivate;
+
+#endif /* xvmc_private_h */
diff --git a/src/gallium/winsys/drm/intel/dri/Makefile b/src/gallium/winsys/drm/intel/dri/Makefile
index 5e212b62a4..c0ecd9680e 100644
--- a/src/gallium/winsys/drm/intel/dri/Makefile
+++ b/src/gallium/winsys/drm/intel/dri/Makefile
@@ -9,7 +9,7 @@ PIPE_DRIVERS = \
 	$(TOP)/src/gallium/drivers/trace/libtrace.a \
 	$(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \
 	$(TOP)/src/gallium/drivers/identity/libidentity.a \
-	$(TOP)/src/gallium/drivers/i915simple/libi915simple.a
+	$(TOP)/src/gallium/drivers/i915/libi915.a
 
 
 DRIVER_SOURCES =
diff --git a/src/gallium/winsys/drm/intel/dri/SConscript b/src/gallium/winsys/drm/intel/dri/SConscript
index 6c00861f51..b1b654d9f8 100644
--- a/src/gallium/winsys/drm/intel/dri/SConscript
+++ b/src/gallium/winsys/drm/intel/dri/SConscript
@@ -8,12 +8,13 @@ drivers = [
     st_dri,
     inteldrm,
     softpipe,
-    i915simple,
+    i915,
     trace,
 ]
 
-env.SharedLibrary(
+env.LoadableModule(
     target ='i915_dri.so',
     source = COMMON_GALLIUM_SOURCES,
     LIBS = drivers + mesa + auxiliaries + env['LIBS'],
+    SHLIBPREFIX = '',
 )
diff --git a/src/gallium/winsys/drm/intel/egl/Makefile b/src/gallium/winsys/drm/intel/egl/Makefile
index 490baded66..1397e9f729 100644
--- a/src/gallium/winsys/drm/intel/egl/Makefile
+++ b/src/gallium/winsys/drm/intel/egl/Makefile
@@ -9,7 +9,7 @@ PIPE_DRIVERS = \
 	$(GALLIUMDIR)/winsys/drm/intel/gem/libinteldrm.a \
 	$(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \
 	$(TOP)/src/gallium/drivers/trace/libtrace.a \
-	$(TOP)/src/gallium/drivers/i915simple/libi915simple.a
+	$(TOP)/src/gallium/drivers/i915/libi915.a
 
 DRIVER_SOURCES =
 
diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c
index 4c5a1d2ea8..9ed570ff6e 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c
+++ b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c
@@ -4,9 +4,10 @@
 #include "intel_drm_winsys.h"
 #include "util/u_memory.h"
 
-#include "i915simple/i915_context.h"
-#include "i915simple/i915_screen.h"
+#include "i915/i915_context.h"
+#include "i915/i915_screen.h"
 
+#include "trace/tr_drm.h"
 
 /*
  * Helper functions
@@ -40,6 +41,7 @@ intel_drm_buffer_from_handle(struct intel_drm_winsys *idws,
                              const char* name, unsigned handle)
 {
    struct intel_drm_buffer *buf = CALLOC_STRUCT(intel_drm_buffer);
+   uint32_t tile = 0, swizzle = 0;
 
    if (!buf)
       return NULL;
@@ -52,6 +54,10 @@ intel_drm_buffer_from_handle(struct intel_drm_winsys *idws,
    if (!buf->bo)
       goto err;
 
+   drm_intel_bo_get_tiling(buf->bo, &tile, &swizzle);
+   if (tile != INTEL_TILE_NONE)
+      buf->map_gtt = TRUE;
+
    return (struct intel_buffer *)buf;
 
 err:
@@ -166,6 +172,7 @@ intel_drm_create_screen(struct drm_api *api, int drmFD,
    idws->base.destroy = intel_drm_winsys_destroy;
 
    idws->pools.gem = drm_intel_bufmgr_gem_init(idws->fd, idws->max_batch_size);
+   drm_intel_bufmgr_gem_enable_reuse(idws->pools.gem);
 
    idws->softpipe = FALSE;
    idws->dump_cmd = debug_get_bool_option("INTEL_DUMP_CMD", FALSE);
@@ -198,5 +205,5 @@ struct drm_api intel_drm_api =
 struct drm_api *
 drm_api_create()
 {
-   return &intel_drm_api;
+   return trace_drm_create(&intel_drm_api);
 }
diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_batchbuffer.c b/src/gallium/winsys/drm/intel/gem/intel_drm_batchbuffer.c
index 5ca3ad9762..5b4dafc8e4 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_drm_batchbuffer.c
+++ b/src/gallium/winsys/drm/intel/gem/intel_drm_batchbuffer.c
@@ -13,6 +13,9 @@
 #define INTEL_BATCH_CLIPRECTS    0x2
 
 #undef INTEL_RUN_SYNC
+#undef INTEL_MAP_BATCHBUFFER
+#undef INTEL_MAP_GTT
+#define INTEL_ALWAYS_FLUSH
 
 struct intel_drm_batchbuffer
 {
@@ -33,6 +36,7 @@ static void
 intel_drm_batchbuffer_reset(struct intel_drm_batchbuffer *batch)
 {
    struct intel_drm_winsys *idws = intel_drm_winsys(batch->base.iws);
+   int ret;
 
    if (batch->bo)
       drm_intel_bo_unreference(batch->bo);
@@ -40,8 +44,18 @@ intel_drm_batchbuffer_reset(struct intel_drm_batchbuffer *batch)
                                   "gallium3d_batchbuffer",
                                   batch->actual_size,
                                   4096);
-   drm_intel_bo_map(batch->bo, TRUE);
+
+#ifdef INTEL_MAP_BATCHBUFFER
+#ifdef INTEL_MAP_GTT
+   ret = drm_intel_gem_bo_map_gtt(batch->bo);
+#else
+   ret = drm_intel_bo_map(batch->bo, TRUE);
+#endif
+   assert(ret == 0);
    batch->base.map = batch->bo->virtual;
+#else
+   (void)ret;
+#endif
 
    memset(batch->base.map, 0, batch->actual_size);
    batch->base.ptr = batch->base.map;
@@ -55,7 +69,13 @@ intel_drm_batchbuffer_create(struct intel_winsys *iws)
    struct intel_drm_winsys *idws = intel_drm_winsys(iws);
    struct intel_drm_batchbuffer *batch = CALLOC_STRUCT(intel_drm_batchbuffer);
 
+   batch->actual_size = idws->max_batch_size;
+
+#ifdef INTEL_MAP_BATCHBUFFER
    batch->base.map = NULL;
+#else
+   batch->base.map = MALLOC(batch->actual_size);
+#endif
    batch->base.ptr = NULL;
    batch->base.size = 0;
 
@@ -64,8 +84,6 @@ intel_drm_batchbuffer_create(struct intel_winsys *iws)
 
    batch->base.iws = iws;
 
-   batch->actual_size = idws->max_batch_size;
-
    intel_drm_batchbuffer_reset(batch);
 
    return &batch->base;
@@ -140,23 +158,32 @@ intel_drm_batchbuffer_flush(struct intel_batchbuffer *ibatch,
    used = batch->base.ptr - batch->base.map;
    assert((used & 3) == 0);
 
-   if (used & 4) {
-      // MI_FLUSH | FLUSH_MAP_CACHE;
-      intel_batchbuffer_dword(ibatch, (0x0<<29)|(0x4<<23)|(1<<0));
-      // MI_NOOP
-      intel_batchbuffer_dword(ibatch, (0x0<<29)|(0x0<<23));
-      // MI_BATCH_BUFFER_END;
-      intel_batchbuffer_dword(ibatch, (0x0<<29)|(0xA<<23));
-   } else {
-      //MI_FLUSH | FLUSH_MAP_CACHE;
-      intel_batchbuffer_dword(ibatch, (0x0<<29)|(0x4<<23)|(1<<0));
-      // MI_BATCH_BUFFER_END;
-      intel_batchbuffer_dword(ibatch, (0x0<<29)|(0xA<<23));
+
+#ifdef INTEL_ALWAYS_FLUSH
+   /* MI_FLUSH | FLUSH_MAP_CACHE */
+   intel_batchbuffer_dword(ibatch, (0x4<<23)|(1<<0));
+   used += 4;
+#endif
+
+   if ((used & 4) == 0) {
+      /* MI_NOOP */
+      intel_batchbuffer_dword(ibatch, 0);
    }
+   /* MI_BATCH_BUFFER_END */
+   intel_batchbuffer_dword(ibatch, (0xA<<23));
 
    used = batch->base.ptr - batch->base.map;
+   assert((used & 4) == 0);
 
+#ifdef INTEL_MAP_BATCHBUFFER
+#ifdef INTEL_MAP_GTT
+   drm_intel_gem_bo_unmap_gtt(batch->bo);
+#else
    drm_intel_bo_unmap(batch->bo);
+#endif
+#else
+   drm_intel_bo_subdata(batch->bo, 0, used, batch->base.map);
+#endif
 
    /* Do the sending to HW */
    ret = drm_intel_bo_exec(batch->bo, used, NULL, 0, 0);
@@ -202,7 +229,10 @@ intel_drm_batchbuffer_destroy(struct intel_batchbuffer *ibatch)
    if (batch->bo)
       drm_intel_bo_unreference(batch->bo);
 
-   free(batch);
+#ifndef INTEL_MAP_BATCHBUFFER
+   FREE(batch->base.map);
+#endif
+   FREE(batch);
 }
 
 void intel_drm_winsys_init_batchbuffer_functions(struct intel_drm_winsys *idws)
diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c b/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c
index e017cd2e98..ac4dd6e00e 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c
+++ b/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c
@@ -28,6 +28,7 @@ intel_drm_buffer_create(struct intel_winsys *iws,
    } else if (type == INTEL_NEW_VERTEX) {
       name = "gallium3d_vertex";
       pool = idws->pools.gem;
+      buf->map_gtt = TRUE;
    } else if (type == INTEL_NEW_SCANOUT) {
       name = "gallium3d_scanout";
       pool = idws->pools.gem;
@@ -57,11 +58,17 @@ intel_drm_buffer_set_fence_reg(struct intel_winsys *iws,
                                unsigned stride,
                                enum intel_buffer_tile tile)
 {
+   struct intel_drm_buffer *buf = intel_drm_buffer(buffer);
    assert(I915_TILING_NONE == INTEL_TILE_NONE);
    assert(I915_TILING_X == INTEL_TILE_X);
    assert(I915_TILING_Y == INTEL_TILE_Y);
 
-   return drm_intel_bo_set_tiling(intel_bo(buffer), &tile, stride);
+   if (tile != INTEL_TILE_NONE) {
+      assert(buf->map_count == 0);
+      buf->map_gtt = TRUE;
+   }
+
+   return drm_intel_bo_set_tiling(buf->bo, &tile, stride);
 }
 
 static void *
@@ -109,6 +116,18 @@ intel_drm_buffer_unmap(struct intel_winsys *iws,
       drm_intel_bo_unmap(intel_bo(buffer));
 }
 
+static int
+intel_drm_buffer_write(struct intel_winsys *iws,
+                       struct intel_buffer *buffer,
+                       size_t offset,
+                       size_t size,
+                       const void *data)
+{
+   struct intel_drm_buffer *buf = intel_drm_buffer(buffer);
+
+   return drm_intel_bo_subdata(buf->bo, offset, size, (void*)data);
+}
+
 static void
 intel_drm_buffer_destroy(struct intel_winsys *iws,
                          struct intel_buffer *buffer)
@@ -130,5 +149,6 @@ intel_drm_winsys_init_buffer_functions(struct intel_drm_winsys *idws)
    idws->base.buffer_set_fence_reg = intel_drm_buffer_set_fence_reg;
    idws->base.buffer_map = intel_drm_buffer_map;
    idws->base.buffer_unmap = intel_drm_buffer_unmap;
+   idws->base.buffer_write = intel_drm_buffer_write;
    idws->base.buffer_destroy = intel_drm_buffer_destroy;
 }
diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_winsys.h b/src/gallium/winsys/drm/intel/gem/intel_drm_winsys.h
index 415c45feea..b4a60563ef 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_drm_winsys.h
+++ b/src/gallium/winsys/drm/intel/gem/intel_drm_winsys.h
@@ -2,7 +2,7 @@
 #ifndef INTEL_DRM_WINSYS_H
 #define INTEL_DRM_WINSYS_H
 
-#include "i915simple/intel_batchbuffer.h"
+#include "i915/intel_batchbuffer.h"
 
 #include "drm.h"
 #include "intel_bufmgr.h"
diff --git a/src/gallium/winsys/drm/intel/xorg/Makefile b/src/gallium/winsys/drm/intel/xorg/Makefile
index 9e56853b02..14c2462524 100644
--- a/src/gallium/winsys/drm/intel/xorg/Makefile
+++ b/src/gallium/winsys/drm/intel/xorg/Makefile
@@ -18,7 +18,7 @@ INCLUDES = \
 LIBS = \
 	$(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a \
 	$(TOP)/src/gallium/winsys/drm/intel/gem/libinteldrm.a \
-	$(TOP)/src/gallium/drivers/i915simple/libi915simple.a \
+	$(TOP)/src/gallium/drivers/i915/libi915.a \
 	$(TOP)/src/gallium/drivers/trace/libtrace.a \
 	$(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \
 	$(GALLIUM_AUXILIARIES)
diff --git a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
index 091cbbcfed..f512c0e5f3 100644
--- a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
+++ b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
@@ -112,7 +112,7 @@ nouveau_drm_create_screen(struct drm_api *api, int fd,
 		return NULL;
 	}
 
-	if (arg->mode == DRM_CREATE_DRI1) {
+	if (arg && arg->mode == DRM_CREATE_DRI1) {
 		struct nouveau_dri *nvdri = dri1->ddx_info;
 		enum pipe_format format;
 
@@ -197,6 +197,7 @@ nouveau_drm_pt_from_name(struct drm_api *api, struct pipe_screen *pscreen,
 			 unsigned stride, unsigned handle)
 {
 	struct nouveau_device *dev = nouveau_screen(pscreen)->device;
+	struct pipe_texture *pt;
 	struct pipe_buffer *pb;
 	int ret;
 
@@ -218,7 +219,9 @@ nouveau_drm_pt_from_name(struct drm_api *api, struct pipe_screen *pscreen,
 	pb->usage = PIPE_BUFFER_USAGE_GPU_READ_WRITE |
 		    PIPE_BUFFER_USAGE_CPU_READ_WRITE;
 	pb->size = nouveau_bo(pb)->size;
-	return pscreen->texture_blanket(pscreen, templ, &stride, pb);
+	pt = pscreen->texture_blanket(pscreen, templ, &stride, pb);
+	pipe_buffer_reference(&pb, NULL);
+	return pt;
 }
 
 static boolean
@@ -245,6 +248,7 @@ nouveau_drm_handle_from_pt(struct drm_api *api, struct pipe_screen *pscreen,
 		return false;
 
 	*handle = mt->bo->handle;
+	*stride = mt->base.nblocksx[0] * mt->base.block.size;
 	return true;
 }
 
diff --git a/src/gallium/winsys/drm/nouveau/xorg/Makefile b/src/gallium/winsys/drm/nouveau/xorg/Makefile
new file mode 100644
index 0000000000..f0d3b337e8
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/xorg/Makefile
@@ -0,0 +1,61 @@
+TARGET     = modesetting_drv.so
+CFILES     = $(wildcard ./*.c)
+OBJECTS    = $(patsubst ./%.c,./%.o,$(CFILES))
+TOP        = ../../../../../..
+
+include $(TOP)/configs/current
+
+INCLUDES = \
+	$(shell pkg-config --cflags-only-I pixman-1 xorg-server libdrm xproto) \
+	-I../gem \
+	-I$(TOP)/src/gallium/include \
+	-I$(TOP)/src/gallium/drivers \
+	-I$(TOP)/src/gallium/auxiliary \
+	-I$(TOP)/src/mesa \
+	-I$(TOP)/include \
+	-I$(TOP)/src/egl/main
+
+LIBS = \
+	$(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a \
+	$(TOP)/src/gallium/winsys/drm/nouveau/drm/libnouveaudrm.a \
+	$(TOP)/src/gallium/drivers/nv04/libnv04.a \
+	$(TOP)/src/gallium/drivers/nv10/libnv10.a \
+	$(TOP)/src/gallium/drivers/nv20/libnv20.a \
+	$(TOP)/src/gallium/drivers/nv30/libnv30.a \
+	$(TOP)/src/gallium/drivers/nv40/libnv40.a \
+	$(TOP)/src/gallium/drivers/nv50/libnv50.a \
+	$(TOP)/src/gallium/drivers/nouveau/libnouveau.a \
+	$(GALLIUM_AUXILIARIES)
+
+DRIVER_DEFINES = \
+	-DHAVE_CONFIG_H
+
+
+#############################################
+
+
+
+all default: $(TARGET)
+
+$(TARGET): $(OBJECTS) Makefile $(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a $(LIBS)
+	$(TOP)/bin/mklib -noprefix -o $@ \
+	$(OBJECTS) $(LIBS) $(shell pkg-config --libs libdrm) -ldrm_nouveau
+
+clean:
+	rm -rf $(OBJECTS) $(TARGET)
+
+install:
+	$(INSTALL) -d $(DESTDIR)/$(XORG_DRIVER_INSTALL_DIR)
+	$(MINSTALL) -m 755 $(TARGET) $(DESTDIR)/$(XORG_DRIVER_INSTALL_DIR)
+
+
+##############################################
+
+
+.c.o:
+	$(CC) -c $(CFLAGS) $(INCLUDES) $(DRIVER_DEFINES) $< -o $@
+
+
+##############################################
+
+.PHONY	= all clean install
diff --git a/src/gallium/winsys/drm/nouveau/xorg/nouveau_xorg.c b/src/gallium/winsys/drm/nouveau/xorg/nouveau_xorg.c
new file mode 100644
index 0000000000..a669b3080a
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/xorg/nouveau_xorg.c
@@ -0,0 +1,149 @@
+/*
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ * Author: Alan Hourihane <alanh@tungstengraphics.com>
+ * Author: Jakob Bornecrantz <wallbraker@gmail.com>
+ *
+ */
+
+#include "../../../../state_trackers/xorg/xorg_winsys.h"
+
+static void nouveau_xorg_identify(int flags);
+static Bool nouveau_xorg_pci_probe(DriverPtr driver, int entity_num,
+				   struct pci_device *device,
+				   intptr_t match_data);
+
+static const struct pci_id_match nouveau_xorg_device_match[] = {
+    { 0x10de, PCI_MATCH_ANY, PCI_MATCH_ANY, PCI_MATCH_ANY,
+      0x00030000, 0x00ffffff, 0 },
+    { 0x12d2, PCI_MATCH_ANY, PCI_MATCH_ANY, PCI_MATCH_ANY,
+      0x00030000, 0x00ffffff, 0 },
+    {0, 0, 0},
+};
+
+static SymTabRec nouveau_xorg_chipsets[] = {
+    {PCI_MATCH_ANY, "NVIDIA Graphics Device"},
+    {-1, NULL}
+};
+
+static PciChipsets nouveau_xorg_pci_devices[] = {
+    {PCI_MATCH_ANY, PCI_MATCH_ANY, NULL},
+    {-1, -1, NULL}
+};
+
+static XF86ModuleVersionInfo nouveau_xorg_version = {
+    "modesetting",
+    MODULEVENDORSTRING,
+    MODINFOSTRING1,
+    MODINFOSTRING2,
+    XORG_VERSION_CURRENT,
+    0, 1, 0, /* major, minor, patch */
+    ABI_CLASS_VIDEODRV,
+    ABI_VIDEODRV_VERSION,
+    MOD_CLASS_VIDEODRV,
+    {0, 0, 0, 0}
+};
+
+/*
+ * Xorg driver exported structures
+ */
+
+_X_EXPORT DriverRec modesetting = {
+    1,
+    "modesetting",
+    nouveau_xorg_identify,
+    NULL,
+    xorg_tracker_available_options,
+    NULL,
+    0,
+    NULL,
+    nouveau_xorg_device_match,
+    nouveau_xorg_pci_probe
+};
+
+static MODULESETUPPROTO(nouveau_xorg_setup);
+
+_X_EXPORT XF86ModuleData modesettingModuleData = {
+    &nouveau_xorg_version,
+    nouveau_xorg_setup,
+    NULL
+};
+
+/*
+ * Xorg driver functions
+ */
+
+static pointer
+nouveau_xorg_setup(pointer module, pointer opts, int *errmaj, int *errmin)
+{
+    static Bool setupDone = 0;
+
+    /* This module should be loaded only once, but check to be sure.
+     */
+    if (!setupDone) {
+	setupDone = 1;
+	xf86AddDriver(&modesetting, module, HaveDriverFuncs);
+
+	/*
+	 * The return value must be non-NULL on success even though there
+	 * is no TearDownProc.
+	 */
+	return (pointer) 1;
+    } else {
+	if (errmaj)
+	    *errmaj = LDR_ONCEONLY;
+	return NULL;
+    }
+}
+
+static void
+nouveau_xorg_identify(int flags)
+{
+    xf86PrintChipsets("modesetting", "Driver for Modesetting Kernel Drivers",
+		      nouveau_xorg_chipsets);
+}
+
+static Bool
+nouveau_xorg_pci_probe(DriverPtr driver,
+	  int entity_num, struct pci_device *device, intptr_t match_data)
+{
+    ScrnInfoPtr scrn = NULL;
+    EntityInfoPtr entity;
+
+    scrn = xf86ConfigPciEntity(scrn, 0, entity_num, nouveau_xorg_pci_devices,
+			       NULL, NULL, NULL, NULL, NULL);
+    if (scrn != NULL) {
+	scrn->driverVersion = 1;
+	scrn->driverName = "i915";
+	scrn->name = "modesetting";
+	scrn->Probe = NULL;
+
+	entity = xf86GetEntityInfo(entity_num);
+
+	/* Use all the functions from the xorg tracker */
+	xorg_tracker_set_functions(scrn);
+    }
+    return scrn != NULL;
+}
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
index 07551e7cd1..7bf23cba23 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
@@ -32,6 +32,8 @@
 
 #include "radeon_buffer.h"
 
+#include "radeon_bo_gem.h"
+
 static const char *radeon_get_name(struct pipe_winsys *ws)
 {
     return "Radeon/GEM+KMS";
@@ -99,6 +101,7 @@ static struct pipe_buffer *radeon_surface_buffer_create(struct pipe_winsys *ws,
                                                         unsigned height,
                                                         enum pipe_format format,
                                                         unsigned usage,
+                                                        unsigned tex_usage,
                                                         unsigned *stride)
 {
     struct pipe_format_block block;
@@ -134,8 +137,11 @@ static void *radeon_buffer_map(struct pipe_winsys *ws,
         (struct radeon_pipe_buffer*)buffer;
     int write = 0;
 
-    if (!(flags & PIPE_BUFFER_USAGE_DONTBLOCK)) {
-        radeon_bo_wait(radeon_buffer->bo);
+    if (flags & PIPE_BUFFER_USAGE_DONTBLOCK) {
+        uint32_t domain;
+
+        if (radeon_bo_is_busy(radeon_buffer->bo, &domain))
+            return NULL;
     }
     if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) {
         write = 1;
@@ -187,7 +193,6 @@ static void radeon_flush_frontbuffer(struct pipe_winsys *pipe_winsys,
 struct radeon_winsys* radeon_pipe_winsys(int fd)
 {
     struct radeon_winsys* radeon_ws;
-    struct radeon_bo_manager* bom;
 
     radeon_ws = CALLOC_STRUCT(radeon_winsys);
     if (radeon_ws == NULL) {
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.c b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
index 47376a0f07..69f14e54f2 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
@@ -38,7 +38,7 @@ struct pipe_screen* radeon_create_screen(struct drm_api* api,
 {
     struct radeon_winsys* winsys = radeon_pipe_winsys(drmFB);
 
-    if (getenv("RADEON_SOFTPIPE")) {
+    if (debug_get_bool_option("RADEON_SOFTPIPE", FALSE)) {
         return softpipe_create_screen((struct pipe_winsys*)winsys);
     } else {
         struct r300_winsys* r300 = radeon_create_r300_winsys(drmFB, winsys);
@@ -51,7 +51,7 @@ struct pipe_screen* radeon_create_screen(struct drm_api* api,
 struct pipe_context* radeon_create_context(struct drm_api* api,
                                            struct pipe_screen* screen)
 {
-    if (getenv("RADEON_SOFTPIPE")) {
+    if (debug_get_bool_option("RADEON_SOFTPIPE", FALSE)) {
         return radeon_create_softpipe(screen->winsys);
     } else {
         return r300_create_context(screen,
@@ -98,7 +98,7 @@ struct pipe_buffer* radeon_buffer_from_handle(struct drm_api* api,
     return &radeon_buffer->base;
 }
 
-struct pipe_texture*
+static struct pipe_texture*
 radeon_texture_from_shared_handle(struct drm_api *api,
                                   struct pipe_screen *screen,
                                   struct pipe_texture *templ,
@@ -107,29 +107,36 @@ radeon_texture_from_shared_handle(struct drm_api *api,
                                   unsigned handle)
 {
     struct pipe_buffer *buffer;
+    struct pipe_texture *blanket;
 
     buffer = radeon_buffer_from_handle(api, screen, name, handle);
     if (!buffer) {
         return NULL;
     }
 
-    return screen->texture_blanket(screen, templ, &stride, buffer);
+    blanket = screen->texture_blanket(screen, templ, &stride, buffer);
+
+    pipe_buffer_reference(&buffer, NULL);
+
+    return blanket;
 }
 
-boolean radeon_shared_handle_from_texture(struct drm_api *api,
-                                          struct pipe_screen *screen,
-                                          struct pipe_texture *texture,
-                                          unsigned *stride,
-                                          unsigned *handle)
+static boolean radeon_shared_handle_from_texture(struct drm_api *api,
+                                                 struct pipe_screen *screen,
+                                                 struct pipe_texture *texture,
+                                                 unsigned *stride,
+                                                 unsigned *handle)
 {
     int retval, fd;
     struct drm_gem_flink flink;
     struct radeon_pipe_buffer* radeon_buffer;
-    struct pipe_buffer* buffer = &radeon_buffer->base;
-    if (!radeon_buffer_from_texture(api, texture, buffer, stride)) {
+    struct pipe_buffer *buffer;
+
+    if (!radeon_buffer_from_texture(api, texture, &buffer, stride)) {
         return FALSE;
     }
 
+    radeon_buffer = (struct radeon_pipe_buffer*)buffer;
     if (!radeon_buffer->flinked) {
         fd = ((struct radeon_winsys*)screen->winsys)->priv->fd;
 
@@ -150,11 +157,11 @@ boolean radeon_shared_handle_from_texture(struct drm_api *api,
     return TRUE;
 }
 
-boolean radeon_local_handle_from_texture(struct drm_api *api,
-                                         struct pipe_screen *screen,
-                                         struct pipe_texture *texture,
-                                         unsigned *stride,
-                                         unsigned *handle)
+static boolean radeon_local_handle_from_texture(struct drm_api *api,
+                                                struct pipe_screen *screen,
+                                                struct pipe_texture *texture,
+                                                unsigned *stride,
+                                                unsigned *handle)
 {
     struct pipe_buffer *buffer;
     if (!radeon_buffer_from_texture(api, texture, &buffer, stride)) {
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.h b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
index 88a5c82b28..9a789ec1a4 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
@@ -37,6 +37,7 @@
 #include "pipe/p_screen.h"
 
 #include "trace/tr_drm.h"
+#include "util/u_debug.h"
 #include "util/u_memory.h"
 
 #include "state_tracker/drm_api.h"
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_r300.c b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
index d723876221..7ea5d1fb4e 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_r300.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
@@ -22,6 +22,17 @@
 
 #include "radeon_r300.h"
 
+static void radeon_r300_set_flush_cb(struct r300_winsys *winsys,
+				     void (*flush_cb)(void *),
+				     void *data)
+{
+    struct radeon_winsys_priv* priv =
+        (struct radeon_winsys_priv*)winsys->radeon_winsys;
+
+    radeon_cs_space_set_flush(priv->cs, flush_cb,
+			      data);
+}
+
 static boolean radeon_r300_add_buffer(struct r300_winsys* winsys,
                                       struct pipe_buffer* pbuffer,
                                       uint32_t rd,
@@ -95,6 +106,13 @@ static void radeon_r300_write_cs_reloc(struct r300_winsys* winsys,
     }
 }
 
+static void radeon_r300_reset_bos(struct r300_winsys *winsys)
+{
+    struct radeon_winsys_priv* priv =
+        (struct radeon_winsys_priv*)winsys->radeon_winsys;
+    radeon_cs_space_reset_bos(priv->cs);
+}
+
 static void radeon_r300_end_cs(struct r300_winsys* winsys,
                                const char* file,
                                const char* function,
@@ -119,9 +137,6 @@ static void radeon_r300_flush_cs(struct r300_winsys* winsys)
         radeon_cs_print(priv->cs, stderr);
     }
 
-    /* Clean out BOs. */
-    radeon_cs_space_reset_bos(priv->cs);
-
     /* Reset CS.
      * Someday, when we care about performance, we should really find a way
      * to rotate between two or three CS objects so that the GPU can be
@@ -137,7 +152,7 @@ static void do_ioctls(struct r300_winsys* winsys, int fd)
     int target = 0;
     int retval;
 
-    info.value = &target;
+    info.value = (unsigned long)&target;
 
     /* First, get the number of pixel pipes */
     info.request = RADEON_INFO_NUM_GB_PIPES;
@@ -149,6 +164,16 @@ static void do_ioctls(struct r300_winsys* winsys, int fd)
     }
     winsys->gb_pipes = target;
 
+    /* get Z pipes */
+    info.request = RADEON_INFO_NUM_Z_PIPES;
+    retval = drmCommandWriteRead(fd, DRM_RADEON_INFO, &info, sizeof(info));
+    if (retval) {
+        fprintf(stderr, "%s: Failed to get GB pipe count, "
+                "error number %d\n", __FUNCTION__, retval);
+        exit(1);
+    }
+    winsys->z_pipes = target;
+
     /* Then, get PCI ID */
     info.request = RADEON_INFO_DEVICE_ID;
     retval = drmCommandWriteRead(fd, DRM_RADEON_INFO, &info, sizeof(info));
@@ -203,6 +228,8 @@ radeon_create_r300_winsys(int fd, struct radeon_winsys* old_winsys)
     winsys->write_cs_reloc = radeon_r300_write_cs_reloc;
     winsys->end_cs = radeon_r300_end_cs;
     winsys->flush_cs = radeon_r300_flush_cs;
+    winsys->reset_bos = radeon_r300_reset_bos;
+    winsys->set_flush_cb = radeon_r300_set_flush_cb;
 
     memcpy(winsys, old_winsys, sizeof(struct radeon_winsys));
 
diff --git a/src/gallium/winsys/drm/radeon/xorg/Makefile b/src/gallium/winsys/drm/radeon/xorg/Makefile
index 0241625f69..9fa16dab24 100644
--- a/src/gallium/winsys/drm/radeon/xorg/Makefile
+++ b/src/gallium/winsys/drm/radeon/xorg/Makefile
@@ -20,6 +20,8 @@ LIBS = \
 	$(GALLIUMDIR)/state_trackers/xorg/libxorgtracker.a \
 	$(GALLIUMDIR)/winsys/drm/radeon/core/libradeonwinsys.a \
 	$(TOP)/src/gallium/drivers/r300/libr300.a \
+	$(TOP)/src/gallium/drivers/trace/libtrace.a \
+	$(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \
 	$(GALLIUM_AUXILIARIES)
 
 #############################################
diff --git a/src/gallium/winsys/g3dvl/Makefile b/src/gallium/winsys/g3dvl/Makefile
new file mode 100644
index 0000000000..424ddea87a
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/Makefile
@@ -0,0 +1,11 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+SUBDIRS = $(GALLIUM_WINSYS_DIRS)
+
+default install clean:
+	@for dir in $(SUBDIRS) ; do \
+		if [ -d $$dir ] ; then \
+			(cd $$dir && $(MAKE) $@) || exit 1; \
+		fi \
+	done
diff --git a/src/gallium/winsys/g3dvl/vl_winsys.h b/src/gallium/winsys/g3dvl/vl_winsys.h
index c83db28dd9..22119f9559 100644
--- a/src/gallium/winsys/g3dvl/vl_winsys.h
+++ b/src/gallium/winsys/g3dvl/vl_winsys.h
@@ -1,14 +1,50 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #ifndef vl_winsys_h
 #define vl_winsys_h
 
 #include <X11/Xlib.h>
+#include <pipe/p_defines.h>
+#include <pipe/p_format.h>
 
-struct pipe_context;
+struct pipe_screen;
+struct pipe_video_context;
 
-struct pipe_context* create_pipe_context(Display *display, int screen);
-int destroy_pipe_context(struct pipe_context *pipe);
-int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable);
-int unbind_pipe_drawable(struct pipe_context *pipe);
+struct pipe_screen*
+vl_screen_create(Display *display, int screen);
 
-#endif
+struct pipe_video_context*
+vl_video_create(struct pipe_screen *screen,
+                enum pipe_video_profile profile,
+                enum pipe_video_chroma_format chroma_format,
+                unsigned width, unsigned height);
 
+Drawable
+vl_video_bind_drawable(struct pipe_video_context *vpipe, Drawable drawable);
+
+#endif
diff --git a/src/gallium/winsys/g3dvl/xlib/Makefile b/src/gallium/winsys/g3dvl/xlib/Makefile
new file mode 100644
index 0000000000..cf765ef51a
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/xlib/Makefile
@@ -0,0 +1,74 @@
+# This makefile produces a "stand-alone" libXvMCg3dvl.so which is
+# based on Xlib (no DRI HW acceleration)
+
+TOP = ../../../../..
+include $(TOP)/configs/current
+
+XVMC_MAJOR = 1
+XVMC_MINOR = 0
+XVMC_LIB = XvMCg3dvl
+XVMC_LIB_NAME = lib$(XVMC_LIB).so
+XVMC_LIB_DEPS = $(EXTRA_LIB_PATH) -lXvMC -lXv -lX11 -lm
+
+INCLUDES = -I$(TOP)/src/gallium/include \
+           -I$(TOP)/src/gallium/auxiliary \
+           -I$(TOP)/src/gallium/drivers \
+           -I$(TOP)/src/gallium/winsys/g3dvl
+
+DEFINES += -DGALLIUM_SOFTPIPE \
+	   -DGALLIUM_TRACE
+
+SOURCES = xsp_winsys.c
+
+# XXX: Hack, if we include libxvmctracker.a in LIBS none of the symbols are
+# pulled in by the linker because xsp_winsys.c doesn't refer to them
+OBJECTS = $(SOURCES:.c=.o) $(TOP)/src/gallium/state_trackers/xorg/xvmc/*.o
+
+LIBS = $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \
+       $(TOP)/src/gallium/auxiliary/vl/libvl.a \
+       $(TOP)/src/gallium/auxiliary/tgsi/libtgsi.a \
+       $(TOP)/src/gallium/auxiliary/draw/libdraw.a \
+       $(TOP)/src/gallium/auxiliary/translate/libtranslate.a \
+       $(TOP)/src/gallium/auxiliary/cso_cache/libcso_cache.a \
+       $(TOP)/src/gallium/auxiliary/rtasm/librtasm.a \
+       $(TOP)/src/gallium/auxiliary/util/libutil.a
+
+.c.o:
+	$(CC) -c $(INCLUDES) $(DEFINES) $(CFLAGS) $< -o $@
+
+.S.o:
+	$(CC) -c $(INCLUDES) $(DEFINES) $(CFLAGS) $< -o $@
+
+.PHONY: default $(TOP)/$(LIB_DIR)/gallium clean
+
+default: depend $(TOP)/$(LIB_DIR)/gallium $(TOP)/$(LIB_DIR)/gallium/$(XVMC_LIB_NAME)
+
+$(TOP)/$(LIB_DIR)/gallium:
+	@mkdir -p $(TOP)/$(LIB_DIR)/gallium
+
+# Make the libXvMCg3dvl.so library
+$(TOP)/$(LIB_DIR)/gallium/$(XVMC_LIB_NAME): $(OBJECTS) $(LIBS) Makefile
+	$(MKLIB) -o $(XVMC_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
+		-major $(XVMC_MAJOR) -minor $(XVMC_MINOR) $(MKLIB_OPTIONS) \
+		-install $(TOP)/$(LIB_DIR)/gallium -id $(INSTALL_LIB_DIR)/lib$(XVMC_LIB).1.dylib \
+		$(XVMC_LIB_DEPS) $(OBJECTS) $(LIBS)
+
+depend: $(SOURCES) Makefile
+	$(RM) depend
+	touch depend
+	$(MKDEP) $(MKDEP_OPTIONS) $(DEFINES) $(INCLUDES) $(SOURCES)
+
+#install: default
+#	$(INSTALL) -d $(INSTALL_DIR)/include/GL
+#	$(INSTALL) -d $(INSTALL_DIR)/$(LIB_DIR)
+#	$(INSTALL) -m 644 $(TOP)/include/GL/*.h $(INSTALL_DIR)/include/GL
+#	@if [ -e $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) ]; then \
+#		$(INSTALL) $(TOP)/$(LIB_DIR)/libGL* $(INSTALL_DIR)/$(LIB_DIR); \
+#	fi
+
+clean: Makefile
+	$(RM) $(TOP)/$(LIB_DIR)/gallium/$(XVMC_LIB_NAME)
+	$(RM) *.o *~
+	$(RM) depend depend.bak
+
+-include depend
diff --git a/src/gallium/winsys/g3dvl/xlib/xsp_winsys.c b/src/gallium/winsys/g3dvl/xlib/xsp_winsys.c
new file mode 100644
index 0000000000..0e5f5a587b
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/xlib/xsp_winsys.c
@@ -0,0 +1,331 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 Younes Manton.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <vl_winsys.h>
+#include <X11/Xutil.h>
+#include <pipe/internal/p_winsys_screen.h>
+#include <pipe/p_state.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <util/u_math.h>
+#include <softpipe/sp_winsys.h>
+#include <softpipe/sp_video_context.h>
+#include <softpipe/sp_texture.h>
+
+/* pipe_winsys implementation */
+
+struct xsp_pipe_winsys
+{
+   struct pipe_winsys base;
+   Display *display;
+   int screen;
+   XImage *fbimage;
+};
+
+struct xsp_context
+{
+   Drawable drawable;
+
+   void (*pipe_destroy)(struct pipe_video_context *vpipe);
+};
+
+struct xsp_buffer
+{
+   struct pipe_buffer base;
+   boolean is_user_buffer;
+   void *data;
+   void *mapped_data;
+};
+
+static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size)
+{
+   struct xsp_buffer *buffer;
+
+   assert(pws);
+
+   buffer = calloc(1, sizeof(struct xsp_buffer));
+   pipe_reference_init(&buffer->base.reference, 1);
+   buffer->base.alignment = alignment;
+   buffer->base.usage = usage;
+   buffer->base.size = size;
+   buffer->data = align_malloc(size, alignment);
+
+   return (struct pipe_buffer*)buffer;
+}
+
+static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size)
+{
+   struct xsp_buffer *buffer;
+
+   assert(pws);
+
+   buffer = calloc(1, sizeof(struct xsp_buffer));
+   pipe_reference_init(&buffer->base.reference, 1);
+   buffer->base.size = size;
+   buffer->is_user_buffer = TRUE;
+   buffer->data = data;
+
+   return (struct pipe_buffer*)buffer;
+}
+
+static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags)
+{
+   struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
+
+   assert(pws);
+   assert(buffer);
+
+   xsp_buf->mapped_data = xsp_buf->data;
+
+   return xsp_buf->mapped_data;
+}
+
+static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer)
+{
+   struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
+
+   assert(pws);
+   assert(buffer);
+
+   xsp_buf->mapped_data = NULL;
+}
+
+static void xsp_buffer_destroy(struct pipe_buffer *buffer)
+{
+   struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
+
+   assert(buffer);
+
+   if (!xsp_buf->is_user_buffer)
+      align_free(xsp_buf->data);
+
+   free(xsp_buf);
+}
+
+static struct pipe_buffer* xsp_surface_buffer_create
+(
+   struct pipe_winsys *pws,
+   unsigned width,
+   unsigned height,
+   enum pipe_format format,
+   unsigned usage,
+   unsigned tex_usage,
+   unsigned *stride
+)
+{
+   const unsigned int ALIGNMENT = 1;
+   struct pipe_format_block block;
+   unsigned nblocksx, nblocksy;
+
+   pf_get_block(format, &block);
+   nblocksx = pf_get_nblocksx(&block, width);
+   nblocksy = pf_get_nblocksy(&block, height);
+   *stride = align(nblocksx * block.size, ALIGNMENT);
+
+   return pws->buffer_create(pws, ALIGNMENT, usage,
+                             *stride * nblocksy);
+}
+
+static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence)
+{
+   assert(pws);
+   assert(ptr);
+   assert(fence);
+}
+
+static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag)
+{
+   assert(pws);
+   assert(fence);
+
+   return 0;
+}
+
+static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag)
+{
+   assert(pws);
+   assert(fence);
+
+   return 0;
+}
+
+static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private)
+{
+   struct xsp_pipe_winsys *xsp_winsys;
+   struct xsp_context *xsp_context;
+
+   assert(pws);
+   assert(surface);
+   assert(context_private);
+
+   xsp_winsys = (struct xsp_pipe_winsys*)pws;
+   xsp_context = (struct xsp_context*)context_private;
+   xsp_winsys->fbimage->width = surface->width;
+   xsp_winsys->fbimage->height = surface->height;
+   xsp_winsys->fbimage->bytes_per_line = surface->width * (xsp_winsys->fbimage->bits_per_pixel >> 3);
+   xsp_winsys->fbimage->data = (char*)((struct xsp_buffer *)softpipe_texture(surface->texture)->buffer)->data + surface->offset;
+
+   XPutImage
+   (
+      xsp_winsys->display, xsp_context->drawable,
+      XDefaultGC(xsp_winsys->display, xsp_winsys->screen),
+      xsp_winsys->fbimage, 0, 0, 0, 0,
+      surface->width, surface->height
+   );
+   XFlush(xsp_winsys->display);
+}
+
+static const char* xsp_get_name(struct pipe_winsys *pws)
+{
+   assert(pws);
+   return "X11 SoftPipe";
+}
+
+static void xsp_destroy(struct pipe_winsys *pws)
+{
+   struct xsp_pipe_winsys *xsp_winsys = (struct xsp_pipe_winsys*)pws;
+
+   assert(pws);
+
+   /* XDestroyImage() wants to free the data as well */
+   xsp_winsys->fbimage->data = NULL;
+
+   XDestroyImage(xsp_winsys->fbimage);
+   FREE(xsp_winsys);
+}
+
+/* Called through pipe_video_context::destroy() */
+static void xsp_pipe_destroy(struct pipe_video_context *vpipe)
+{
+   struct xsp_context *xsp_context;
+
+   assert(vpipe);
+
+   xsp_context = vpipe->priv;
+
+   /* Call the original destroy */
+   xsp_context->pipe_destroy(vpipe);
+
+   FREE(xsp_context);
+}
+
+/* Show starts here */
+
+Drawable
+vl_video_bind_drawable(struct pipe_video_context *vpipe, Drawable drawable)
+{
+   struct xsp_context *xsp_context;
+   Drawable old_drawable;
+
+   assert(vpipe);
+
+   xsp_context = vpipe->priv;
+   old_drawable = xsp_context->drawable;
+   xsp_context->drawable = drawable;
+
+   return old_drawable;
+}
+
+struct pipe_screen*
+vl_screen_create(Display *display, int screen)
+{
+   struct xsp_pipe_winsys *xsp_winsys;
+
+   assert(display);
+
+   xsp_winsys = CALLOC_STRUCT(xsp_pipe_winsys);
+   if (!xsp_winsys)
+      return NULL;
+
+   xsp_winsys->base.buffer_create = xsp_buffer_create;
+   xsp_winsys->base.user_buffer_create = xsp_user_buffer_create;
+   xsp_winsys->base.buffer_map = xsp_buffer_map;
+   xsp_winsys->base.buffer_unmap = xsp_buffer_unmap;
+   xsp_winsys->base.buffer_destroy = xsp_buffer_destroy;
+   xsp_winsys->base.surface_buffer_create = xsp_surface_buffer_create;
+   xsp_winsys->base.fence_reference = xsp_fence_reference;
+   xsp_winsys->base.fence_signalled = xsp_fence_signalled;
+   xsp_winsys->base.fence_finish = xsp_fence_finish;
+   xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer;
+   xsp_winsys->base.get_name = xsp_get_name;
+   xsp_winsys->base.destroy = xsp_destroy;
+   xsp_winsys->display = display;
+   xsp_winsys->screen = screen;
+   xsp_winsys->fbimage = XCreateImage
+   (
+      display,
+      XDefaultVisual(display, screen),
+      XDefaultDepth(display, screen),
+      ZPixmap,
+      0,
+      NULL,
+      0, /* Don't know the width and height until flush_frontbuffer */
+      0,
+      32,
+      0
+   );
+
+   if (!xsp_winsys->fbimage) {
+      FREE(xsp_winsys);
+      return NULL;
+   }
+
+   XInitImage(xsp_winsys->fbimage);
+
+   return softpipe_create_screen(&xsp_winsys->base);
+}
+
+struct pipe_video_context*
+vl_video_create(struct pipe_screen *screen,
+                enum pipe_video_profile profile,
+                enum pipe_video_chroma_format chroma_format,
+                unsigned width, unsigned height)
+{
+   struct pipe_video_context *vpipe;
+   struct xsp_context *xsp_context;
+
+   assert(screen);
+   assert(width && height);
+
+   vpipe = sp_video_create(screen, profile, chroma_format, width, height);
+   if (!vpipe)
+      return NULL;
+
+   xsp_context = CALLOC_STRUCT(xsp_context);
+   if (!xsp_context) {
+      vpipe->destroy(vpipe);
+      return NULL;
+   }
+
+   /* Override this so we can free our xsp_context when the pipe is freed */
+   xsp_context->pipe_destroy = vpipe->destroy;
+   vpipe->destroy = xsp_pipe_destroy;
+
+   vpipe->priv = xsp_context;
+
+   return vpipe;
+}
diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c
deleted file mode 100644
index 698c2856a4..0000000000
--- a/src/gallium/winsys/g3dvl/xsp_winsys.c
+++ /dev/null
@@ -1,290 +0,0 @@
-#include "vl_winsys.h"
-#include <X11/Xutil.h>
-#include <pipe/internal/p_winsys_screen.h>
-#include <pipe/p_state.h>
-#include <pipe/p_inlines.h>
-#include <util/u_memory.h>
-#include <util/u_math.h>
-#include <softpipe/sp_winsys.h>
-#include <softpipe/sp_texture.h>
-
-/* pipe_winsys implementation */
-
-struct xsp_pipe_winsys
-{
-	struct pipe_winsys	base;
-	XImage			fbimage;
-};
-
-struct xsp_context
-{
-	Display			*display;
-	int			screen;
-	Drawable		drawable;
-	int			drawable_bound;
-};
-
-struct xsp_buffer
-{
-	struct pipe_buffer	base;
-	boolean			is_user_buffer;
-	void			*data;
-	void			*mapped_data;
-};
-
-static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size)
-{
-	struct xsp_buffer *buffer;
-
-	assert(pws);
-
-	buffer = calloc(1, sizeof(struct xsp_buffer));
-	pipe_reference_init(&buffer->base.reference, 1);
-	buffer->base.alignment = alignment;
-	buffer->base.usage = usage;
-	buffer->base.size = size;
-	buffer->data = align_malloc(size, alignment);
-
-	return (struct pipe_buffer*)buffer;
-}
-
-static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size)
-{
-	struct xsp_buffer *buffer;
-
-	assert(pws);
-
-	buffer = calloc(1, sizeof(struct xsp_buffer));
-	pipe_reference_init(&buffer->base.reference, 1);
-	buffer->base.size = size;
-	buffer->is_user_buffer = TRUE;
-	buffer->data = data;
-
-	return (struct pipe_buffer*)buffer;
-}
-
-static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags)
-{
-	struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
-
-	assert(pws);
-	assert(buffer);
-
-	xsp_buf->mapped_data = xsp_buf->data;
-
-	return xsp_buf->mapped_data;
-}
-
-static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer)
-{
-	struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
-
-	assert(pws);
-	assert(buffer);
-
-	xsp_buf->mapped_data = NULL;
-}
-
-static void xsp_buffer_destroy(struct pipe_winsys *pws, struct pipe_buffer *buffer)
-{
-	struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
-
-	assert(pws);
-	assert(buffer);
-
-	if (!xsp_buf->is_user_buffer)
-		align_free(xsp_buf->data);
-
-	free(xsp_buf);
-}
-
-static struct pipe_buffer* xsp_surface_buffer_create
-(
-	struct pipe_winsys *pws,
-	unsigned width,
-	unsigned height,
-	enum pipe_format format,
-	unsigned usage,
-	unsigned *stride
-)
-{
-	const unsigned int ALIGNMENT = 1;
-	struct pipe_format_block block;
-	unsigned nblocksx, nblocksy;
-
-	pf_get_block(format, &block);
-	nblocksx = pf_get_nblocksx(&block, width);
-	nblocksy = pf_get_nblocksy(&block, height);
-	*stride = align(nblocksx * block.size, ALIGNMENT);
-
-	return pws->buffer_create(pws, ALIGNMENT,
-				  usage,
-				  *stride * nblocksy);
-}
-
-static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence)
-{
-	assert(pws);
-	assert(ptr);
-	assert(fence);
-}
-
-static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag)
-{
-	assert(pws);
-	assert(fence);
-
-	return 0;
-}
-
-static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag)
-{
-	assert(pws);
-	assert(fence);
-
-	return 0;
-}
-
-static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private)
-{
-	struct xsp_pipe_winsys	*xsp_winsys;
-	struct xsp_context	*xsp_context;
-
-	assert(pws);
-	assert(surface);
-	assert(context_private);
-
-	xsp_winsys = (struct xsp_pipe_winsys*)pws;
-	xsp_context = (struct xsp_context*)context_private;
-
-	if (!xsp_context->drawable_bound)
-		return;
-
-	xsp_winsys->fbimage.width = surface->width;
-	xsp_winsys->fbimage.height = surface->height;
-	xsp_winsys->fbimage.bytes_per_line = surface->width * (xsp_winsys->fbimage.bits_per_pixel >> 3);
-	xsp_winsys->fbimage.data = ((struct xsp_buffer *)softpipe_texture(surface->texture)->buffer)->data + surface->offset;
-
-	XPutImage
-	(
-		xsp_context->display,
-		xsp_context->drawable,
-		XDefaultGC(xsp_context->display, xsp_context->screen),
-		&xsp_winsys->fbimage,
-		0,
-		0,
-		0,
-		0,
-		surface->width,
-		surface->height
-	);
-	XFlush(xsp_context->display);
-}
-
-static const char* xsp_get_name(struct pipe_winsys *pws)
-{
-	assert(pws);
-	return "X11 SoftPipe";
-}
-
-/* Show starts here */
-
-int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable)
-{
-	struct xsp_context *xsp_context;
-
-	assert(pipe);
-
-	xsp_context = pipe->priv;
-	xsp_context->drawable = drawable;
-	xsp_context->drawable_bound = 1;
-
-	return 0;
-}
-
-int unbind_pipe_drawable(struct pipe_context *pipe)
-{
-	struct xsp_context *xsp_context;
-
-	assert(pipe);
-
-	xsp_context = pipe->priv;
-	xsp_context->drawable_bound = 0;
-
-	return 0;
-}
-
-struct pipe_context* create_pipe_context(Display *display, int screen)
-{
-	struct xsp_pipe_winsys	*xsp_winsys;
-	struct xsp_context	*xsp_context;
-	struct pipe_screen	*sp_screen;
-	struct pipe_context	*sp_pipe;
-
-	assert(display);
-
-	xsp_winsys = calloc(1, sizeof(struct xsp_pipe_winsys));
-	xsp_winsys->base.buffer_create = xsp_buffer_create;
-	xsp_winsys->base.user_buffer_create = xsp_user_buffer_create;
-	xsp_winsys->base.buffer_map = xsp_buffer_map;
-	xsp_winsys->base.buffer_unmap = xsp_buffer_unmap;
-	xsp_winsys->base.buffer_destroy = xsp_buffer_destroy;
-	xsp_winsys->base.surface_buffer_create = xsp_surface_buffer_create;
-	xsp_winsys->base.fence_reference = xsp_fence_reference;
-	xsp_winsys->base.fence_signalled = xsp_fence_signalled;
-	xsp_winsys->base.fence_finish = xsp_fence_finish;
-	xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer;
-	xsp_winsys->base.get_name = xsp_get_name;
-
-	{
-		/* XXX: Can't use the returned XImage* directly,
-		since we don't have control over winsys destruction
-		and we wouldn't be able to free it */
-		XImage *template = XCreateImage
-		(
-			display,
-			XDefaultVisual(display, XDefaultScreen(display)),
-			XDefaultDepth(display, XDefaultScreen(display)),
-			ZPixmap,
-			0,
-			NULL,
-			0,	/* Don't know the width and height until flush_frontbuffer */
-			0,
-			32,
-			0
-		);
-
-		memcpy(&xsp_winsys->fbimage, template, sizeof(XImage));
-		XInitImage(&xsp_winsys->fbimage);
-
-		XDestroyImage(template);
-	}
-
-	sp_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys);
-	sp_pipe = softpipe_create(sp_screen);
-
-	xsp_context = calloc(1, sizeof(struct xsp_context));
-	xsp_context->display = display;
-	xsp_context->screen = screen;
-
-	sp_pipe->priv = xsp_context;
-
-	return sp_pipe;
-}
-
-int destroy_pipe_context(struct pipe_context *pipe)
-{
-	struct pipe_screen *screen;
-	struct pipe_winsys *winsys;
-
-	assert(pipe);
-
-	screen = pipe->screen;
-	winsys = pipe->winsys;
-	free(pipe->priv);
-	pipe->destroy(pipe);
-	screen->destroy(screen);
-	free(winsys);
-
-	return 0;
-}
diff --git a/src/gallium/winsys/gdi/SConscript b/src/gallium/winsys/gdi/SConscript
index 86eb9ef55e..8f556daf04 100644
--- a/src/gallium/winsys/gdi/SConscript
+++ b/src/gallium/winsys/gdi/SConscript
@@ -5,35 +5,45 @@ Import('*')
 
 if env['platform'] == 'windows':
 
-	env = env.Clone()
-
-	env.Append(CPPPATH = [
-		'#src/gallium/state_trackers/wgl',
-	])
-
-	env.Append(LIBS = [
-		'gdi32',
-		'user32',
-		'kernel32',
-		'ws2_32',
-	])
-
-	sources = [
-		'gdi_softpipe_winsys.c',
-	]
-	
-	if env['gcc']:
-		sources += ['#src/gallium/state_trackers/wgl/opengl32.mingw.def']
-	else:
-		sources += ['#src/gallium/state_trackers/wgl/opengl32.def']
-		
-	drivers = [
-		trace,
-		softpipe,
-	]
-
-	env.SharedLibrary(
-		target ='opengl32',
-		source = sources,
-		LIBS = wgl + glapi + mesa + drivers + auxiliaries + env['LIBS'],
-	)
+    env = env.Clone()
+
+    env.Append(CPPPATH = [
+        '#src/gallium/state_trackers/wgl',
+    ])
+
+    env.Append(LIBS = [
+        'gdi32',
+        'user32',
+        'kernel32',
+        'ws2_32',
+    ])
+
+    sources = []
+    drivers = []
+
+    if 'softpipe' in env['drivers']:
+        sources = ['gdi_softpipe_winsys.c']
+        drivers = [softpipe]
+
+    if 'llvmpipe' in env['drivers']:
+        env.Tool('llvm')
+        if 'LLVM_VERSION' in env:
+            sources = ['gdi_llvmpipe_winsys.c']
+            drivers = [llvmpipe]
+
+    if not sources or not drivers:
+        print 'warning: softpipe or llvmpipe not selected, gdi winsys disabled'
+        Return()
+    
+    if env['gcc']:
+        sources += ['#src/gallium/state_trackers/wgl/opengl32.mingw.def']
+    else:
+        sources += ['#src/gallium/state_trackers/wgl/opengl32.def']
+        
+    drivers += [trace]
+
+    env.SharedLibrary(
+        target ='opengl32',
+        source = sources,
+        LIBS = wgl + glapi + mesa + drivers + auxiliaries + env['LIBS'],
+    )
diff --git a/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c b/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c
new file mode 100644
index 0000000000..e8bc0f55ac
--- /dev/null
+++ b/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c
@@ -0,0 +1,288 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * LLVMpipe support.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#include <windows.h>
+
+#include "pipe/p_format.h"
+#include "pipe/p_context.h"
+#include "pipe/p_inlines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "llvmpipe/lp_winsys.h"
+#include "llvmpipe/lp_texture.h"
+#include "stw_winsys.h"
+
+
+struct gdi_llvmpipe_displaytarget
+{
+   enum pipe_format format;
+   struct pipe_format_block block;
+   unsigned width;
+   unsigned height;
+   unsigned stride;
+
+   unsigned size;
+
+   void *data;
+
+   BITMAPINFO bmi;
+};
+
+
+/** Cast wrapper */
+static INLINE struct gdi_llvmpipe_displaytarget *
+gdi_llvmpipe_displaytarget( struct llvmpipe_displaytarget *buf )
+{
+   return (struct gdi_llvmpipe_displaytarget *)buf;
+}
+
+
+static boolean
+gdi_llvmpipe_is_displaytarget_format_supported( struct llvmpipe_winsys *ws,
+                                                enum pipe_format format )
+{
+   switch(format) {
+   case PIPE_FORMAT_X8R8G8B8_UNORM:
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+      return TRUE;
+
+   /* TODO: Support other formats possible with BMPs, as described in 
+    * http://msdn.microsoft.com/en-us/library/dd183376(VS.85).aspx */
+      
+   default:
+      return FALSE;
+   }
+}
+
+
+static void *
+gdi_llvmpipe_displaytarget_map(struct llvmpipe_winsys *ws,
+                               struct llvmpipe_displaytarget *dt,
+                               unsigned flags )
+{
+   struct gdi_llvmpipe_displaytarget *gdt = gdi_llvmpipe_displaytarget(dt);
+
+   return gdt->data;
+}
+
+
+static void
+gdi_llvmpipe_displaytarget_unmap(struct llvmpipe_winsys *ws,
+                                 struct llvmpipe_displaytarget *dt )
+{
+
+}
+
+
+static void
+gdi_llvmpipe_displaytarget_destroy(struct llvmpipe_winsys *winsys,
+                                   struct llvmpipe_displaytarget *dt)
+{
+   struct gdi_llvmpipe_displaytarget *gdt = gdi_llvmpipe_displaytarget(dt);
+
+   align_free(gdt->data);
+   FREE(gdt);
+}
+
+
+/**
+ * Round n up to next multiple.
+ */
+static INLINE unsigned
+round_up(unsigned n, unsigned multiple)
+{
+   return (n + multiple - 1) & ~(multiple - 1);
+}
+
+
+static struct llvmpipe_displaytarget *
+gdi_llvmpipe_displaytarget_create(struct llvmpipe_winsys *winsys,
+                                  enum pipe_format format,
+                                  unsigned width, unsigned height,
+                                  unsigned alignment,
+                                  unsigned *stride)
+{
+   struct gdi_llvmpipe_displaytarget *gdt;
+   unsigned cpp;
+   unsigned bpp;
+   
+   gdt = CALLOC_STRUCT(gdi_llvmpipe_displaytarget);
+   if(!gdt)
+      goto no_gdt;
+
+   gdt->format = format;
+   gdt->width = width;
+   gdt->height = height;
+
+   bpp = pf_get_bits(format);
+   cpp = pf_get_size(format);
+   
+   gdt->stride = round_up(width * cpp, alignment);
+   gdt->size = gdt->stride * height;
+   
+   gdt->data = align_malloc(gdt->size, alignment);
+   if(!gdt->data)
+      goto no_data;
+
+   gdt->bmi.bmiHeader.biSize = sizeof(BITMAPINFOHEADER);
+   gdt->bmi.bmiHeader.biWidth = gdt->stride / cpp;
+   gdt->bmi.bmiHeader.biHeight= -(long)height;
+   gdt->bmi.bmiHeader.biPlanes = 1;
+   gdt->bmi.bmiHeader.biBitCount = bpp;
+   gdt->bmi.bmiHeader.biCompression = BI_RGB;
+   gdt->bmi.bmiHeader.biSizeImage = 0;
+   gdt->bmi.bmiHeader.biXPelsPerMeter = 0;
+   gdt->bmi.bmiHeader.biYPelsPerMeter = 0;
+   gdt->bmi.bmiHeader.biClrUsed = 0;
+   gdt->bmi.bmiHeader.biClrImportant = 0;
+
+   *stride = gdt->stride;
+   return (struct llvmpipe_displaytarget *)gdt;
+
+no_data:
+   FREE(gdt);
+no_gdt:
+   return NULL;
+}
+
+
+static void
+gdi_llvmpipe_displaytarget_display(struct llvmpipe_winsys *winsys, 
+                                   struct llvmpipe_displaytarget *dt,
+                                   void *context_private)
+{
+   assert(0);
+}
+
+
+static void
+gdi_llvmpipe_destroy(struct llvmpipe_winsys *winsys)
+{
+   FREE(winsys);
+}
+
+
+static struct pipe_screen *
+gdi_llvmpipe_screen_create(void)
+{
+   static struct llvmpipe_winsys *winsys;
+   struct pipe_screen *screen;
+
+   winsys = CALLOC_STRUCT(llvmpipe_winsys);
+   if(!winsys)
+      goto no_winsys;
+
+   winsys->destroy = gdi_llvmpipe_destroy;
+   winsys->is_displaytarget_format_supported = gdi_llvmpipe_is_displaytarget_format_supported;
+   winsys->displaytarget_create = gdi_llvmpipe_displaytarget_create;
+   winsys->displaytarget_map = gdi_llvmpipe_displaytarget_map;
+   winsys->displaytarget_unmap = gdi_llvmpipe_displaytarget_unmap;
+   winsys->displaytarget_display = gdi_llvmpipe_displaytarget_display;
+   winsys->displaytarget_destroy = gdi_llvmpipe_displaytarget_destroy;
+
+   screen = llvmpipe_create_screen(winsys);
+   if(!screen)
+      goto no_screen;
+
+   return screen;
+   
+no_screen:
+   FREE(winsys);
+no_winsys:
+   return NULL;
+}
+
+
+static struct pipe_context *
+gdi_llvmpipe_context_create(struct pipe_screen *screen)
+{
+   return llvmpipe_create(screen);
+}
+
+
+static void
+gdi_llvmpipe_present(struct pipe_screen *screen,
+                     struct pipe_surface *surface,
+                     HDC hDC)
+{
+    struct llvmpipe_texture *texture;
+    struct gdi_llvmpipe_displaytarget *gdt;
+
+    texture = llvmpipe_texture(surface->texture);
+    gdt = gdi_llvmpipe_displaytarget(texture->dt);
+
+    StretchDIBits(hDC,
+                  0, 0, gdt->width, gdt->height,
+                  0, 0, gdt->width, gdt->height,
+                  gdt->data, &gdt->bmi, 0, SRCCOPY);
+}
+
+
+static const struct stw_winsys stw_winsys = {
+   &gdi_llvmpipe_screen_create,
+   &gdi_llvmpipe_context_create,
+   &gdi_llvmpipe_present,
+   NULL, /* get_adapter_luid */
+   NULL, /* shared_surface_open */
+   NULL, /* shared_surface_close */
+   NULL  /* compose */
+};
+
+
+BOOL WINAPI
+DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpReserved)
+{
+   switch (fdwReason) {
+   case DLL_PROCESS_ATTACH:
+      if (!stw_init(&stw_winsys)) {
+         return FALSE;
+      }
+      return stw_init_thread();
+
+   case DLL_THREAD_ATTACH:
+      return stw_init_thread();
+
+   case DLL_THREAD_DETACH:
+      stw_cleanup_thread();
+      break;
+
+   case DLL_PROCESS_DETACH:
+      stw_cleanup_thread();
+      stw_cleanup();
+      break;
+   }
+   return TRUE;
+}
diff --git a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c
index 33826524d7..5e0ccf32f4 100644
--- a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c
+++ b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c
@@ -46,7 +46,7 @@
 #include "util/u_memory.h"
 #include "softpipe/sp_winsys.h"
 #include "softpipe/sp_texture.h"
-#include "shared/stw_winsys.h"
+#include "stw_winsys.h"
 
 
 struct gdi_softpipe_buffer
@@ -166,6 +166,7 @@ gdi_softpipe_surface_buffer_create(struct pipe_winsys *winsys,
                                    unsigned width, unsigned height,
                                    enum pipe_format format,
                                    unsigned usage,
+                                   unsigned tex_usage,
                                    unsigned *stride)
 {
    const unsigned alignment = 64;
@@ -268,9 +269,9 @@ gdi_softpipe_context_create(struct pipe_screen *screen)
 
 
 static void
-gdi_softpipe_flush_frontbuffer(struct pipe_screen *screen,
-                               struct pipe_surface *surface,
-                               HDC hDC)
+gdi_softpipe_present(struct pipe_screen *screen,
+                     struct pipe_surface *surface,
+                     HDC hDC)
 {
     struct softpipe_texture *texture;
     struct gdi_softpipe_buffer *buffer;
@@ -303,7 +304,11 @@ gdi_softpipe_flush_frontbuffer(struct pipe_screen *screen,
 static const struct stw_winsys stw_winsys = {
    &gdi_softpipe_screen_create,
    &gdi_softpipe_context_create,
-   &gdi_softpipe_flush_frontbuffer
+   &gdi_softpipe_present,
+   NULL, /* get_adapter_luid */
+   NULL, /* shared_surface_open */
+   NULL, /* shared_surface_close */
+   NULL  /* compose */
 };
 
 
diff --git a/src/gallium/winsys/xlib/Makefile b/src/gallium/winsys/xlib/Makefile
index 3a1945d92c..3dc38a78e4 100644
--- a/src/gallium/winsys/xlib/Makefile
+++ b/src/gallium/winsys/xlib/Makefile
@@ -31,9 +31,6 @@ DEFINES += \
 XLIB_WINSYS_SOURCES = \
 	xlib.c \
 	xlib_cell.c \
-	xlib_brw_aub.c \
-	xlib_brw_context.c \
-	xlib_brw_screen.c \
 	xlib_llvmpipe.c \
 	xlib_softpipe.c \
 	xlib_trace.c 
diff --git a/src/gallium/winsys/xlib/SConscript b/src/gallium/winsys/xlib/SConscript
index 467d595d33..dfe550f733 100644
--- a/src/gallium/winsys/xlib/SConscript
+++ b/src/gallium/winsys/xlib/SConscript
@@ -5,7 +5,7 @@ Import('*')
 
 if env['platform'] == 'linux' \
         and 'mesa' in env['statetrackers'] \
-        and set(('softpipe', 'llvmpipe', 'i915simple', 'trace')).intersection(env['drivers']) \
+        and set(('softpipe', 'llvmpipe', 'i915', 'trace')).intersection(env['drivers']) \
         and not env['dri']:
 
     env = env.Clone()
@@ -36,15 +36,6 @@ if env['platform'] == 'linux' \
             env.Tool('udis86')
             sources += ['xlib_llvmpipe.c']
             drivers += [llvmpipe]
-
-    if 'i965simple' in env['drivers']:
-        env.Append(CPPDEFINES = 'GALLIUM_I965SIMPLE')
-        sources += [
-            'xlib_brw_aub.c',
-            'xlib_brw_context.c',
-            'xlib_brw_screen.c',
-        ]
-        drivers += [i965simple]
         
     if 'cell' in env['drivers']:
         env.Append(CPPDEFINES = 'GALLIUM_CELL')
diff --git a/src/gallium/winsys/xlib/xlib.c b/src/gallium/winsys/xlib/xlib.c
index 4b71cf7ec3..163cc8863c 100644
--- a/src/gallium/winsys/xlib/xlib.c
+++ b/src/gallium/winsys/xlib/xlib.c
@@ -43,7 +43,6 @@
 
 enum mode {
    MODE_TRACE,
-   MODE_BRW,
    MODE_CELL,
    MODE_LLVMPIPE,
    MODE_SOFTPIPE
@@ -55,9 +54,6 @@ static enum mode get_mode()
    if (getenv("XMESA_TRACE"))
       return MODE_TRACE;
 
-   if (getenv("XMESA_BRW"))
-      return MODE_BRW;
-
 #ifdef GALLIUM_CELL
    if (!getenv("GALLIUM_NOCELL")) 
       return MODE_CELL;
@@ -82,11 +78,6 @@ static void _init( void )
       xmesa_set_driver( &xlib_trace_driver );
 #endif
       break;
-   case MODE_BRW:
-#if defined(GALLIUM_BRW)
-      xmesa_set_driver( &xlib_brw_driver );
-#endif
-      break;
    case MODE_CELL:
 #if defined(GALLIUM_CELL)
       xmesa_set_driver( &xlib_cell_driver );
diff --git a/src/gallium/winsys/xlib/xlib.h b/src/gallium/winsys/xlib/xlib.h
index 347d45f4d6..f0855035f7 100644
--- a/src/gallium/winsys/xlib/xlib.h
+++ b/src/gallium/winsys/xlib/xlib.h
@@ -9,7 +9,6 @@ extern struct xm_driver xlib_trace_driver;
 extern struct xm_driver xlib_softpipe_driver;
 extern struct xm_driver xlib_llvmpipe_driver;
 extern struct xm_driver xlib_cell_driver;
-extern struct xm_driver xlib_brw_driver;
 
 
 #endif
diff --git a/src/gallium/winsys/xlib/xlib_brw.h b/src/gallium/winsys/xlib/xlib_brw.h
deleted file mode 100644
index be2dd147db..0000000000
--- a/src/gallium/winsys/xlib/xlib_brw.h
+++ /dev/null
@@ -1,30 +0,0 @@
-#ifndef XLIB_BRW_H
-#define XLIB_BRW_H
-
-struct pipe_winsys;
-struct pipe_buffer;
-struct pipe_surface;
-struct xmesa_buffer;
-
-unsigned xlib_brw_get_buffer_offset( struct pipe_winsys *pws,
-                                     struct pipe_buffer *buf,
-                                     unsigned access_flags );
-
-void xlib_brw_buffer_subdata_typed( struct pipe_winsys *pws,
-                                    struct pipe_buffer *buf,
-                                    unsigned long offset, 
-                                    unsigned long size, 
-                                    const void *data,
-                                    unsigned data_type );
-
-
-
-void xlib_brw_commands_aub(struct pipe_winsys *winsys,
-                           unsigned *cmds,
-                           unsigned nr_dwords);
-
-struct pipe_context *
-xlib_create_brw_context( struct pipe_screen *screen,
-                         void *unused );
-
-#endif
diff --git a/src/gallium/winsys/xlib/xlib_brw_aub.c b/src/gallium/winsys/xlib/xlib_brw_aub.c
deleted file mode 100644
index b6bd849ef2..0000000000
--- a/src/gallium/winsys/xlib/xlib_brw_aub.c
+++ /dev/null
@@ -1,399 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include <stdio.h>
-#include <stdlib.h>
-#include "xlib_brw_aub.h"
-#include "pipe/p_context.h"
-#include "pipe/p_state.h"
-#include "util/u_debug.h"
-#include "util/u_memory.h"
-#include "softpipe/sp_texture.h"
-
-
-struct brw_aubfile {
-   FILE *file;
-   unsigned next_free_page;
-};
-
-
-extern char *__progname;
-
-
-struct aub_file_header {
-   unsigned int instruction_type;
-   unsigned int pad0:16;
-   unsigned int minor:8;
-   unsigned int major:8;
-   unsigned char application[8*4];
-   unsigned int day:8;
-   unsigned int month:8;
-   unsigned int year:16;
-   unsigned int timezone:8;
-   unsigned int second:8;
-   unsigned int minute:8;
-   unsigned int hour:8;
-   unsigned int comment_length:16;   
-   unsigned int pad1:16;
-};
-
-struct aub_block_header {
-   unsigned int instruction_type;
-   unsigned int operation:8;
-   unsigned int type:8;
-   unsigned int address_space:8;
-   unsigned int pad0:8;
-   unsigned int general_state_type:8;
-   unsigned int surface_state_type:8;
-   unsigned int pad1:16;
-   unsigned int address;
-   unsigned int length;
-};
-
-struct aub_dump_bmp {
-   unsigned int instruction_type;
-   unsigned int xmin:16;
-   unsigned int ymin:16;
-   unsigned int pitch:16;
-   unsigned int bpp:8;
-   unsigned int format:8;
-   unsigned int xsize:16;
-   unsigned int ysize:16;
-   unsigned int addr;
-   unsigned int unknown;
-};
-
-enum bh_operation {
-   BH_COMMENT,
-   BH_DATA_WRITE,
-   BH_COMMAND_WRITE,
-   BH_MMI0_WRITE32,
-   BH_END_SCENE,
-   BH_CONFIG_MEMORY_MAP,
-   BH_MAX_OPERATION
-};
-
-enum command_write_type {
-   CW_HWB_RING = 1,
-   CW_PRIMARY_RING_A,
-   CW_PRIMARY_RING_B,		/* XXX - disagreement with listaub! */
-   CW_PRIMARY_RING_C,
-   CW_MAX_TYPE
-};
-
-enum memory_map_type {
-   MM_DEFAULT,
-   MM_DYNAMIC,
-   MM_MAX_TYPE
-};
-
-enum address_space {
-   ADDR_GTT,
-   ADDR_LOCAL,
-   ADDR_MAIN,
-   ADDR_MAX
-};
-
-
-#define AUB_FILE_HEADER 0xe085000b
-#define AUB_BLOCK_HEADER 0xe0c10003
-#define AUB_DUMP_BMP 0xe09e0004
-
-/* Registers to control page table
- */
-#define PGETBL_CTL       0x2020
-#define PGETBL_ENABLED   0x1
-
-#define NR_GTT_ENTRIES  65536	/* 256 mb */
-
-#define FAIL										\
-do {											\
-   fprintf(stderr, "failed to write aub data at %s/%d\n", __FUNCTION__, __LINE__);	\
-   exit(1);										\
-} while (0)
-
-
-/* Emit the headers at the top of each aubfile.  Initialize the GTT.
- */
-static void init_aubfile( FILE *aub_file )
-{   
-   struct aub_file_header fh;
-   struct aub_block_header bh;
-   unsigned int data;
-
-   static int nr;
-   
-   nr++;
-
-   /* Emit the aub header:
-    */
-   memset(&fh, 0, sizeof(fh));
-
-   fh.instruction_type = AUB_FILE_HEADER;
-   fh.minor = 0x0;
-   fh.major = 0x7;
-   memcpy(fh.application, __progname, sizeof(fh.application));
-   fh.day = (nr>>24) & 0xff;
-   fh.month = 0x0;
-   fh.year = 0x0;
-   fh.timezone = 0x0;
-   fh.second = nr & 0xff;
-   fh.minute = (nr>>8) & 0xff;
-   fh.hour = (nr>>16) & 0xff;
-   fh.comment_length = 0x0;   
-
-   if (fwrite(&fh, sizeof(fh), 1, aub_file) < 0) 
-      FAIL;
-         
-   /* Setup the GTT starting at main memory address zero (!):
-    */
-   memset(&bh, 0, sizeof(bh));
-   
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_MMI0_WRITE32;
-   bh.type = 0x0;
-   bh.address_space = ADDR_GTT;	/* ??? */
-   bh.general_state_type = 0x0;
-   bh.surface_state_type = 0x0;
-   bh.address = PGETBL_CTL;
-   bh.length = 0x4;
-
-   if (fwrite(&bh, sizeof(bh), 1, aub_file) < 0) 
-      FAIL;
-
-   data = 0x0 | PGETBL_ENABLED;
-
-   if (fwrite(&data, sizeof(data), 1, aub_file) < 0) 
-      FAIL;
-}
-
-
-static void init_aub_gtt( struct brw_aubfile *aubfile,
-			  unsigned start_offset, 
-			  unsigned size )
-{
-   FILE *aub_file = aubfile->file;
-   struct aub_block_header bh;
-   unsigned int i;
-
-   assert(start_offset + size < NR_GTT_ENTRIES * 4096);
-
-
-   memset(&bh, 0, sizeof(bh));
-   
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_DATA_WRITE;
-   bh.type = 0x0;
-   bh.address_space = ADDR_MAIN;
-   bh.general_state_type = 0x0;
-   bh.surface_state_type = 0x0;
-   bh.address =  start_offset / 4096 * 4;
-   bh.length = size / 4096 * 4;
-
-   if (fwrite(&bh, sizeof(bh), 1, aub_file) < 0) 
-      FAIL;
-
-   for (i = 0; i < size / 4096; i++) {
-      unsigned data = aubfile->next_free_page | 1;
-
-      aubfile->next_free_page += 4096;
-
-      if (fwrite(&data, sizeof(data), 1, aub_file) < 0) 
-	 FAIL;
-   }
-
-}
-
-static void write_block_header( FILE *aub_file,
-				struct aub_block_header *bh,
-				const unsigned *data,
-				unsigned sz )
-{
-   sz = (sz + 3) & ~3;
-
-   if (fwrite(bh, sizeof(*bh), 1, aub_file) < 0) 
-      FAIL;
-
-   if (fwrite(data, sz, 1, aub_file) < 0) 
-      FAIL;
-
-   fflush(aub_file);
-}
-
-
-static void write_dump_bmp( FILE *aub_file,
-			    struct aub_dump_bmp *db )
-{
-   if (fwrite(db, sizeof(*db), 1, aub_file) < 0) 
-      FAIL;
-
-   fflush(aub_file);
-}
-
-
-
-void brw_aub_gtt_data( struct brw_aubfile *aubfile,
-		       unsigned offset,
-		       const void *data,
-		       unsigned sz,
-		       unsigned type,
-		       unsigned state_type )
-{
-   struct aub_block_header bh;
-
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_DATA_WRITE;
-   bh.type = type;
-   bh.address_space = ADDR_GTT;
-   bh.pad0 = 0;
-
-   if (type == DW_GENERAL_STATE) {
-      bh.general_state_type = state_type;
-      bh.surface_state_type = 0;
-   }
-   else {
-      bh.general_state_type = 0;
-      bh.surface_state_type = state_type;
-   }
-
-   bh.pad1 = 0;
-   bh.address = offset;
-   bh.length = sz;
-
-   write_block_header(aubfile->file, &bh, data, sz);
-}
-
-
-
-void brw_aub_gtt_cmds( struct brw_aubfile *aubfile,
-		       unsigned offset,
-		       const void *data,
-		       unsigned sz )
-{
-   struct aub_block_header bh;   
-   unsigned type = CW_PRIMARY_RING_A;
-   
-
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_COMMAND_WRITE;
-   bh.type = type;
-   bh.address_space = ADDR_GTT;
-   bh.pad0 = 0;
-   bh.general_state_type = 0;
-   bh.surface_state_type = 0;
-   bh.pad1 = 0;
-   bh.address = offset;
-   bh.length = sz;
-
-   write_block_header(aubfile->file, &bh, data, sz);
-}
-
-void brw_aub_dump_bmp( struct brw_aubfile *aubfile,
-		       struct pipe_surface *surface,
-		       unsigned gtt_offset )
-{
-   struct aub_dump_bmp db;
-   unsigned format;
-
-   assert(surface->texture->block.width == 1);
-   assert(surface->texture->block.height == 1);
-   
-   if (surface->texture->block.size == 4)
-      format = 0x7;
-   else
-      format = 0x3;
-
-   db.instruction_type = AUB_DUMP_BMP;
-   db.xmin = 0;
-   db.ymin = 0;
-   db.format = format;
-   db.bpp = surface->texture->block.size * 8;
-   db.pitch = softpipe_texture(surface->texture)->stride[surface->level] /
-      surface->texture->block.size;
-   db.xsize = surface->width;
-   db.ysize = surface->height;
-   db.addr = gtt_offset;
-   db.unknown = /* surface->tiled ? 0x4 : */ 0x0;
-
-   write_dump_bmp(aubfile->file, &db);
-}
-
-
-
-struct brw_aubfile *brw_aubfile_create( void )
-{
-   struct brw_aubfile *aubfile = CALLOC_STRUCT(brw_aubfile);
-   char filename[80];
-   int val;
-   static int i = 0;
-
-   i++;
-
-   if (getenv("INTEL_AUBFILE")) {
-      val = snprintf(filename, sizeof(filename), "%s%d.aub", getenv("INTEL_AUBFILE"), i%4);
-      debug_printf("--> Aub file: %s\n", filename);
-      aubfile->file = fopen(filename, "w");
-   }
-   else {
-      val = snprintf(filename, sizeof(filename), "%s.aub", __progname);
-      if (val < 0 || val > sizeof(filename)) 
-	 strcpy(filename, "default.aub");   
-   
-      debug_printf("--> Aub file: %s\n", filename);
-      aubfile->file = fopen(filename, "w");
-   }
-
-   if (!aubfile->file) {
-      debug_printf("couldn't open aubfile\n");
-      exit(1);
-   }
-
-   init_aubfile(aubfile->file);
-
-   /* The GTT is located starting address zero in main memory.  Pages
-    * to populate the gtt start after this point.
-    */
-   aubfile->next_free_page = (NR_GTT_ENTRIES * 4 + 4095) & ~4095;
-
-   /* More or less correspond with all the agp regions mapped by the
-    * driver:
-    */
-   init_aub_gtt(aubfile, 0, 4096*4);
-   init_aub_gtt(aubfile, AUB_BUF_START, AUB_BUF_SIZE);
-
-   return aubfile;
-}
-
-void brw_aub_destroy( struct brw_aubfile *aubfile )
-{
-   fclose(aubfile->file);
-   FREE(aubfile);
-}
diff --git a/src/gallium/winsys/xlib/xlib_brw_aub.h b/src/gallium/winsys/xlib/xlib_brw_aub.h
deleted file mode 100644
index f5c60c7be2..0000000000
--- a/src/gallium/winsys/xlib/xlib_brw_aub.h
+++ /dev/null
@@ -1,114 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#ifndef BRW_AUB_H
-#define BRW_AUB_H
-
-/* We set up this region, buffers may be allocated here:
- */
-#define AUB_BUF_START (4096*4)
-#define AUB_BUF_SIZE  (8*1024*1024)
-
-struct intel_context;
-struct pipe_surface;
-
-struct brw_aubfile *brw_aubfile_create( void );
-
-void brw_aub_destroy( struct brw_aubfile *aubfile );
-
-void brw_aub_gtt_data( struct brw_aubfile *aubfile,
-		       unsigned offset,
-		       const void *data,
-		       unsigned sz,
-		       unsigned type,
-		       unsigned state_type );
-
-void brw_aub_gtt_cmds( struct brw_aubfile *aubfile,
-		       unsigned offset,
-		       const void *data,
-		       unsigned sz );
-
-void brw_aub_dump_bmp( struct brw_aubfile *aubfile,
-		       struct pipe_surface *surface,
-		       unsigned gtt_offset );
-
-
-enum data_write_type {
-   DW_NOTYPE,
-   DW_BATCH_BUFFER,
-   DW_BIN_BUFFER,
-   DW_BIN_POINTER_LIST,
-   DW_SLOW_STATE_BUFFER,
-   DW_VERTEX_BUFFER,
-   DW_2D_MAP,
-   DW_CUBE_MAP,
-   DW_INDIRECT_STATE_BUFFER,
-   DW_VOLUME_MAP,
-   DW_1D_MAP,
-   DW_CONSTANT_BUFFER,
-   DW_CONSTANT_URB_ENTRY,
-   DW_INDEX_BUFFER,
-   DW_GENERAL_STATE,
-   DW_SURFACE_STATE,
-   DW_MEDIA_OBJECT_INDIRECT_DATA,
-   DW_MAX_TYPE
-};
-
-enum data_write_general_state_type {
-   DWGS_NOTYPE,
-   DWGS_VERTEX_SHADER_STATE,
-   DWGS_GEOMETRY_SHADER_STATE ,
-   DWGS_CLIPPER_STATE,
-   DWGS_STRIPS_FANS_STATE,
-   DWGS_WINDOWER_IZ_STATE,
-   DWGS_COLOR_CALC_STATE,
-   DWGS_CLIPPER_VIEWPORT_STATE,	/* was 0x7 */
-   DWGS_STRIPS_FANS_VIEWPORT_STATE,
-   DWGS_COLOR_CALC_VIEWPORT_STATE, /* was 0x9 */
-   DWGS_SAMPLER_STATE,
-   DWGS_KERNEL_INSTRUCTIONS,
-   DWGS_SCRATCH_SPACE,
-   DWGS_SAMPLER_DEFAULT_COLOR,
-   DWGS_INTERFACE_DESCRIPTOR,
-   DWGS_VLD_STATE,
-   DWGS_VFE_STATE,
-   DWGS_MAX_TYPE
-};
-
-enum data_write_surface_state_type {
-   DWSS_NOTYPE,
-   DWSS_BINDING_TABLE_STATE,
-   DWSS_SURFACE_STATE,
-   DWSS_MAX_TYPE
-};
-
-
-#endif
diff --git a/src/gallium/winsys/xlib/xlib_brw_context.c b/src/gallium/winsys/xlib/xlib_brw_context.c
deleted file mode 100644
index 09599507f4..0000000000
--- a/src/gallium/winsys/xlib/xlib_brw_context.c
+++ /dev/null
@@ -1,209 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell
- *   Brian Paul
- */
-
-
-//#include "glxheader.h"
-//#include "xmesaP.h"
-
-#include "pipe/internal/p_winsys_screen.h"
-#include "pipe/p_inlines.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "i965simple/brw_winsys.h"
-#include "xlib_brw_aub.h"
-#include "xlib_brw.h"
-
-
-
-
-#define XBCWS_BATCHBUFFER_SIZE 1024
-
-
-/* The backend to the brw driver (ie struct brw_winsys) is actually a
- * per-context entity.
- */
-struct xlib_brw_context_winsys {
-   struct brw_winsys brw_context_winsys;   /**< batch buffer funcs */
-   struct aub_context *aub;
-                         
-   struct pipe_winsys *pipe_winsys;
-
-   unsigned batch_data[XBCWS_BATCHBUFFER_SIZE];
-   unsigned batch_nr;
-   unsigned batch_size;
-   unsigned batch_alloc;
-};
-
-
-/* Turn a brw_winsys into an xlib_brw_context_winsys:
- */
-static inline struct xlib_brw_context_winsys *
-xlib_brw_context_winsys( struct brw_winsys *sws )
-{
-   return (struct xlib_brw_context_winsys *)sws;
-}
-
-
-/* Simple batchbuffer interface:
- */
-
-static unsigned *xbcws_batch_start( struct brw_winsys *sws,
-					 unsigned dwords,
-					 unsigned relocs )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-
-   if (xbcws->batch_size < xbcws->batch_nr + dwords)
-      return NULL;
-
-   xbcws->batch_alloc = xbcws->batch_nr + dwords;
-   return (void *)1;			/* not a valid pointer! */
-}
-
-static void xbcws_batch_dword( struct brw_winsys *sws,
-				    unsigned dword )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-
-   assert(xbcws->batch_nr < xbcws->batch_alloc);
-   xbcws->batch_data[xbcws->batch_nr++] = dword;
-}
-
-static void xbcws_batch_reloc( struct brw_winsys *sws,
-			     struct pipe_buffer *buf,
-			     unsigned access_flags,
-			     unsigned delta )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-
-   assert(xbcws->batch_nr < xbcws->batch_alloc);
-   xbcws->batch_data[xbcws->batch_nr++] = 
-      ( xlib_brw_get_buffer_offset( NULL, buf, access_flags ) +
-        delta );
-}
-
-static void xbcws_batch_end( struct brw_winsys *sws )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-
-   assert(xbcws->batch_nr <= xbcws->batch_alloc);
-   xbcws->batch_alloc = 0;
-}
-
-static void xbcws_batch_flush( struct brw_winsys *sws,
-				    struct pipe_fence_handle **fence )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-   assert(xbcws->batch_nr <= xbcws->batch_size);
-
-   if (xbcws->batch_nr) {
-      xlib_brw_commands_aub( xbcws->pipe_winsys,
-                             xbcws->batch_data,
-                             xbcws->batch_nr );
-   }
-
-   xbcws->batch_nr = 0;
-}
-
-  
-
-/* Really a per-device function, just pass through:
- */
-static unsigned xbcws_get_buffer_offset( struct brw_winsys *sws,
-                                         struct pipe_buffer *buf,
-                                         unsigned access_flags )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-
-   return xlib_brw_get_buffer_offset( xbcws->pipe_winsys,
-                                      buf,
-                                      access_flags );
-}
-
-
-/* Really a per-device function, just pass through:
- */
-static void xbcws_buffer_subdata_typed( struct brw_winsys *sws,
-                                       struct pipe_buffer *buf,
-                                       unsigned long offset, 
-                                       unsigned long size, 
-                                       const void *data,
-                                       unsigned data_type )
-{
-   struct xlib_brw_context_winsys *xbcws = xlib_brw_context_winsys(sws);
-
-   xlib_brw_buffer_subdata_typed( xbcws->pipe_winsys,
-                                  buf,
-                                  offset,
-                                  size,
-                                  data,
-                                  data_type );
-}
-
-
-/**
- * Create i965 hardware rendering context, but plugged into a
- * dump-to-aubfile backend.
- */
-struct pipe_context *
-xlib_create_brw_context( struct pipe_screen *screen,
-                         void *unused )
-{
-   struct xlib_brw_context_winsys *xbcws = CALLOC_STRUCT( xlib_brw_context_winsys );
-   
-   /* Fill in this struct with callbacks that i965simple will need to
-    * communicate with the window system, buffer manager, etc. 
-    */
-   xbcws->brw_context_winsys.batch_start = xbcws_batch_start;
-   xbcws->brw_context_winsys.batch_dword = xbcws_batch_dword;
-   xbcws->brw_context_winsys.batch_reloc = xbcws_batch_reloc;
-   xbcws->brw_context_winsys.batch_end = xbcws_batch_end;
-   xbcws->brw_context_winsys.batch_flush = xbcws_batch_flush;
-   xbcws->brw_context_winsys.buffer_subdata_typed = xbcws_buffer_subdata_typed;
-   xbcws->brw_context_winsys.get_buffer_offset = xbcws_get_buffer_offset;
-
-   xbcws->pipe_winsys = screen->winsys; /* redundant */
-
-   xbcws->batch_size = XBCWS_BATCHBUFFER_SIZE;
-
-   /* Create the i965simple context:
-    */
-#ifdef GALLIUM_CELL
-   return NULL;
-#else
-   return brw_create( screen,
-		      &xbcws->brw_context_winsys,
-		      0 );
-#endif
-}
diff --git a/src/gallium/winsys/xlib/xlib_brw_screen.c b/src/gallium/winsys/xlib/xlib_brw_screen.c
deleted file mode 100644
index ef545796f3..0000000000
--- a/src/gallium/winsys/xlib/xlib_brw_screen.c
+++ /dev/null
@@ -1,469 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell
- *   Brian Paul
- */
-
-
-//#include "state_trackers/xlib/glxheader.h"
-//#include "state_trackers/xlib/xmesaP.h"
-
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-#include "i965simple/brw_winsys.h"
-#include "i965simple/brw_screen.h"
-#include "i965simple/brw_context.h"
-
-
-#include "xlib_brw_aub.h"
-#include "xlib_brw.h"
-#include "xlib.h"
-
-static struct pipe_buffer *
-buffer_from_surface(struct pipe_surface *surface)
-{
-   struct brw_texture *texture = (struct brw_texture *)surface;
-   return texture->buffer;
-}
-
-struct aub_buffer {
-   struct pipe_reference reference;
-   char *data;
-   unsigned offset;
-   unsigned size;
-   unsigned map_count;
-   boolean dump_on_unmap;
-};
-
-
-
-struct aub_pipe_winsys {
-   struct pipe_winsys winsys;
-
-   struct brw_aubfile *aubfile;
-
-   /* This is simple, isn't it:
-    */
-   char *pool;
-   unsigned size;
-   unsigned used;
-};
-
-
-/* Turn a pipe winsys into an aub/pipe winsys:
- */
-static inline struct aub_pipe_winsys *
-aub_pipe_winsys( struct pipe_winsys *winsys )
-{
-   return (struct aub_pipe_winsys *)winsys;
-}
-
-
-
-static INLINE struct aub_buffer *
-aub_bo( struct pipe_buffer *bo )
-{
-   return (struct aub_buffer *)bo;
-}
-
-static INLINE struct pipe_buffer *
-pipe_bo( struct aub_buffer *bo )
-{
-   return (struct pipe_buffer *)bo;
-}
-
-
-
-
-static void *aub_buffer_map(struct pipe_winsys *winsys, 
-			      struct pipe_buffer *buf,
-			      unsigned flags )
-{
-   struct aub_buffer *sbo = aub_bo(buf);
-
-   assert(sbo->data);
-
-   if (flags & PIPE_BUFFER_USAGE_CPU_WRITE)
-      sbo->dump_on_unmap = 1;
-
-   sbo->map_count++;
-   return sbo->data;
-}
-
-static void aub_buffer_unmap(struct pipe_winsys *winsys, 
-			       struct pipe_buffer *buf)
-{
-   struct aub_pipe_winsys *iws = aub_pipe_winsys(winsys);
-   struct aub_buffer *sbo = aub_bo(buf);
-
-   sbo->map_count--;
-
-   if (sbo->map_count == 0 &&
-       sbo->dump_on_unmap) {
-
-      sbo->dump_on_unmap = 0;
-
-      brw_aub_gtt_data( iws->aubfile, 
-			sbo->offset,
-			sbo->data,
-			sbo->size,
-			0,
-			0);
-   }
-}
-
-
-static void
-aub_buffer_destroy(struct pipe_buffer *buf)
-{
-   free(buf);
-}
-
-
-
-void xlib_brw_commands_aub(struct pipe_winsys *winsys,
-			unsigned *cmds,
-			unsigned nr_dwords)
-{
-   struct aub_pipe_winsys *iws = aub_pipe_winsys(winsys);
-   unsigned size = nr_dwords * 4;
-
-   assert(iws->used + size < iws->size);
-
-   brw_aub_gtt_cmds( iws->aubfile, 
-		     AUB_BUF_START + iws->used,
-		     cmds,
-		     nr_dwords * sizeof(int) );
-
-   iws->used += align(size, 4096);
-}
-
-
-/* XXX: fix me:
- */
-static struct aub_pipe_winsys *global_winsys = NULL;
-
-
-
-
-/* Pipe has no concept of pools.  We choose the tex/region pool
- * for all buffers.
- */
-static struct pipe_buffer *
-aub_buffer_create(struct pipe_winsys *winsys,
-                  unsigned alignment,
-                  unsigned usage,
-                  unsigned size)
-{
-   struct aub_pipe_winsys *iws = aub_pipe_winsys(winsys);
-   struct aub_buffer *sbo = CALLOC_STRUCT(aub_buffer);
-
-   pipe_reference_init(&sbo->reference, 1);
-
-   /* Could reuse buffers that are not referenced in current
-    * batchbuffer.  Can't do that atm, so always reallocate:
-    */
-   assert(iws->used + size < iws->size);
-   sbo->data = iws->pool + iws->used;
-   sbo->offset = AUB_BUF_START + iws->used;
-   iws->used += align(size, 4096);
-
-   sbo->size = size;
-
-   return pipe_bo(sbo);
-}
-
-
-static struct pipe_buffer *
-aub_user_buffer_create(struct pipe_winsys *winsys, void *ptr, unsigned bytes)
-{
-   struct aub_buffer *sbo;
-
-   /* Lets hope this is meant for upload, not as a result!  
-    */
-   sbo = aub_bo(aub_buffer_create( winsys, 0, 0, 0 ));
-
-   sbo->data = ptr;
-   sbo->size = bytes;
-
-   return pipe_bo(sbo);
-}
-
-
-/* The state tracker (should!) keep track of whether the fake
- * frontbuffer has been touched by any rendering since the last time
- * we copied its contents to the real frontbuffer.  Our task is easy:
- */
-static void
-aub_flush_frontbuffer( struct pipe_winsys *winsys,
-                       struct pipe_surface *surface,
-                       void *context_private)
-{
-//   struct aub_pipe_winsys *iws = aub_pipe_winsys(winsys);
-   brw_aub_dump_bmp( global_winsys->aubfile, 
-		     surface,
-		     aub_bo(buffer_from_surface(surface))->offset );
-}
-
-
-/**
- * Round n up to next multiple.
- */
-static INLINE unsigned
-round_up(unsigned n, unsigned multiple)
-{
-   return (n + multiple - 1) & ~(multiple - 1);
-}
-
-static struct pipe_buffer *
-aub_i915_surface_buffer_create(struct pipe_winsys *winsys,
-                               unsigned width, unsigned height,
-                               enum pipe_format format,
-                               unsigned usage,
-                               unsigned tex_usage,
-                               unsigned *stride)
-{
-   const unsigned alignment = 64;
-   struct pipe_format_block block;
-   unsigned nblocksx, nblocksy;
-
-   pf_get_block(format, &block);
-   nblocksx = pf_get_nblocksx(&block, width);
-   nblocksy = pf_get_nblocksy(&block, height);
-   *stride = round_up(nblocksx * block.size, alignment);
-
-   return winsys->buffer_create(winsys, alignment,
-                                usage,
-                                *stride * nblocksy);
-}
-
-
-static const char *
-aub_get_name( struct pipe_winsys *winsys )
-{
-   return "Aub/xlib";
-}
-
-static void
-xlib_brw_destroy_pipe_winsys_aub( struct pipe_winsys *winsys )
-
-{
-   struct aub_pipe_winsys *iws = aub_pipe_winsys(winsys);
-   brw_aub_destroy(iws->aubfile);
-   free(iws->pool);
-   free(iws);
-}
-
-
-
-static struct pipe_winsys *
-xlib_create_brw_winsys( void )
-{
-   struct aub_pipe_winsys *iws = CALLOC_STRUCT( aub_pipe_winsys );
-   
-   /* Fill in this struct with callbacks that pipe will need to
-    * communicate with the window system, buffer manager, etc. 
-    *
-    * Pipe would be happy with a malloc based memory manager, but
-    * the SwapBuffers implementation in this winsys driver requires
-    * that rendering be done to an appropriate _DriBufferObject.  
-    */
-   iws->winsys.buffer_create = aub_buffer_create;
-   iws->winsys.user_buffer_create = aub_user_buffer_create;
-   iws->winsys.buffer_map = aub_buffer_map;
-   iws->winsys.buffer_unmap = aub_buffer_unmap;
-   iws->winsys.buffer_destroy = aub_buffer_destroy;
-   iws->winsys.flush_frontbuffer = aub_flush_frontbuffer;
-   iws->winsys.get_name = aub_get_name;
-   iws->winsys.destroy = xlib_brw_destroy_pipe_winsys_aub;
-
-   iws->winsys.surface_buffer_create = aub_i915_surface_buffer_create;
-
-   iws->aubfile = brw_aubfile_create();
-   iws->size = AUB_BUF_SIZE;
-   iws->pool = malloc(AUB_BUF_SIZE);
-
-   /* HACK: static copy of this pointer:
-    */
-   assert(global_winsys == NULL);
-   global_winsys = iws;
-
-   return &iws->winsys;
-}
-
-
-static struct pipe_screen *
-xlib_create_brw_screen( void )
-{
-#ifndef GALLIUM_CELL
-   struct pipe_winsys *winsys;
-   struct pipe_screen *screen;
-
-   winsys = xlib_create_brw_winsys();
-   if (winsys == NULL)
-      return NULL;
-
-   screen = brw_create_screen(winsys, 0/* XXX pci_id */);
-   if (screen == NULL)
-      goto fail;
-
-   return screen;
-
-fail:
-   if (winsys)
-      winsys->destroy( winsys );
-
-#endif
-   return NULL;
-}
-
-
-/* These per-screen functions are acually made available to the driver
- * through the brw_winsys (per-context) entity.
- */
-unsigned xlib_brw_get_buffer_offset( struct pipe_winsys *pws,
-                                     struct pipe_buffer *buf,
-                                     unsigned access_flags )
-{
-   return aub_bo(buf)->offset;
-}
-
-void xlib_brw_buffer_subdata_typed( struct pipe_winsys *pws,
-                                    struct pipe_buffer *buf,
-                                    unsigned long offset, 
-                                    unsigned long size, 
-                                    const void *data,
-                                    unsigned data_type )
-{
-   unsigned aub_type = DW_GENERAL_STATE;
-   unsigned aub_sub_type = 0;
-
-   switch (data_type) {
-   case BRW_CC_VP:
-      aub_sub_type = DWGS_COLOR_CALC_VIEWPORT_STATE;
-      break;
-   case BRW_CC_UNIT:
-      aub_sub_type = DWGS_COLOR_CALC_STATE;
-      break;
-   case BRW_WM_PROG:
-      aub_sub_type = DWGS_KERNEL_INSTRUCTIONS;
-      break;
-   case BRW_SAMPLER_DEFAULT_COLOR:
-      aub_sub_type = DWGS_SAMPLER_DEFAULT_COLOR;
-      break;
-   case BRW_SAMPLER:
-      aub_sub_type = DWGS_SAMPLER_STATE;
-      break;
-   case BRW_WM_UNIT:
-      aub_sub_type = DWGS_WINDOWER_IZ_STATE;
-      break;
-   case BRW_SF_PROG:
-      aub_sub_type = DWGS_KERNEL_INSTRUCTIONS;
-      break;
-   case BRW_SF_VP:
-      aub_sub_type = DWGS_STRIPS_FANS_VIEWPORT_STATE;
-      break;
-   case BRW_SF_UNIT:
-      aub_sub_type = DWGS_STRIPS_FANS_STATE;
-      break;
-   case BRW_VS_UNIT:
-      aub_sub_type = DWGS_VERTEX_SHADER_STATE;
-      break;
-   case BRW_VS_PROG:
-      aub_sub_type = DWGS_KERNEL_INSTRUCTIONS;
-      break;
-   case BRW_GS_UNIT:
-      aub_sub_type = DWGS_GEOMETRY_SHADER_STATE;
-      break;
-   case BRW_GS_PROG:
-      aub_sub_type = DWGS_KERNEL_INSTRUCTIONS;
-      break;
-   case BRW_CLIP_VP:
-      aub_sub_type = DWGS_CLIPPER_VIEWPORT_STATE;
-      break;
-   case BRW_CLIP_UNIT:
-      aub_sub_type = DWGS_CLIPPER_STATE;
-      break;
-   case BRW_CLIP_PROG:
-      aub_sub_type = DWGS_KERNEL_INSTRUCTIONS;
-      break;
-   case BRW_SS_SURFACE:
-      aub_type = DW_SURFACE_STATE;
-      aub_sub_type = DWSS_SURFACE_STATE; 
-      break;
-   case BRW_SS_SURF_BIND:
-      aub_type = DW_SURFACE_STATE;
-      aub_sub_type = DWSS_BINDING_TABLE_STATE; 
-      break;
-   case BRW_CONSTANT_BUFFER:
-      aub_type = DW_CONSTANT_URB_ENTRY;
-      aub_sub_type = 0; 
-      break;
-
-   default:
-      assert(0);
-      break;
-   }
-
-   {
-      struct aub_pipe_winsys *iws = aub_pipe_winsys(pws);
-      struct aub_buffer *sbo = aub_bo(buf);
-
-      assert(sbo->size > offset + size);
-      memcpy(sbo->data + offset, data, size);
-
-      brw_aub_gtt_data( iws->aubfile, 
-                        sbo->offset + offset,
-                        sbo->data + offset,
-                        size,
-                        aub_type,
-                        aub_sub_type );
-   }
-}
- 
-
-static void
-xlib_brw_display_surface(struct xmesa_buffer *b, 
-                         struct pipe_surface *surf)
-{
-   brw_aub_dump_bmp( global_winsys->aubfile, 
-		     surf,
-		     aub_bo(buffer_from_surface(surf))->offset );
-}
-
-
-struct xm_driver xlib_brw_driver = 
-{
-   .create_pipe_screen = xlib_create_brw_screen,
-   .create_pipe_context = xlib_create_brw_context,
-   .display_surface = xlib_brw_display_surface,
-};
diff --git a/src/gallium/winsys/xlib/xlib_softpipe.c b/src/gallium/winsys/xlib/xlib_softpipe.c
index 67fea023a3..260b39e2a0 100644
--- a/src/gallium/winsys/xlib/xlib_softpipe.c
+++ b/src/gallium/winsys/xlib/xlib_softpipe.c
@@ -75,9 +75,6 @@ struct xmesa_pipe_winsys
 {
    struct pipe_winsys base;
 /*   struct xmesa_visual *xm_visual; */
-#ifdef USE_XSHM
-   int shm;
-#endif
 };
 
 
@@ -93,11 +90,6 @@ xm_buffer( struct pipe_buffer *buf )
 /**
  * X Shared Memory Image extension code
  */
-#ifdef USE_XSHM
-#define XSHM_ENABLED(b) ((b)->shm)
-#else
-#define XSHM_ENABLED(b) 0
-#endif
 
 #ifdef USE_XSHM
 
@@ -116,23 +108,23 @@ mesaHandleXError(Display *dpy, XErrorEvent *event)
 }
 
 
-static GLboolean alloc_shm(struct xm_buffer *buf, unsigned size)
+static char *alloc_shm(struct xm_buffer *buf, unsigned size)
 {
    XShmSegmentInfo *const shminfo = & buf->shminfo;
 
    shminfo->shmid = shmget(IPC_PRIVATE, size, IPC_CREAT|0777);
    if (shminfo->shmid < 0) {
-      return GL_FALSE;
+      return NULL;
    }
 
    shminfo->shmaddr = (char *) shmat(shminfo->shmid, 0, 0);
    if (shminfo->shmaddr == (char *) -1) {
       shmctl(shminfo->shmid, IPC_RMID, 0);
-      return GL_FALSE;
+      return NULL;
    }
 
    shminfo->readOnly = False;
-   return GL_TRUE;
+   return shminfo->shmaddr;
 }
 
 
@@ -258,25 +250,30 @@ xlib_softpipe_display_surface(struct xmesa_buffer *b,
       return;
 
 #ifdef USE_XSHM
-   if (XSHM_ENABLED(xm_buf) && (xm_buf->tempImage == NULL)) {
-      assert(surf->texture->block.width == 1);
-      assert(surf->texture->block.height == 1);
-      alloc_shm_ximage(xm_buf, b, spt->stride[surf->level] /
-                       surf->texture->block.size, surf->height);
-   }
-#endif
+   if (xm_buf->shm)
+   {
+      if (xm_buf->tempImage == NULL) 
+      {
+         assert(surf->texture->block.width == 1);
+         assert(surf->texture->block.height == 1);
+         alloc_shm_ximage(xm_buf, b, spt->stride[surf->level] /
+                          surf->texture->block.size, surf->height);
+      }
 
-   ximage = (XSHM_ENABLED(xm_buf)) ? xm_buf->tempImage : b->tempImage;
-   ximage->data = xm_buf->data;
+      ximage = xm_buf->tempImage;
+      ximage->data = xm_buf->data;
 
-   /* display image in Window */
-#ifdef USE_XSHM
-   if (XSHM_ENABLED(xm_buf)) {
+      /* _debug_printf("XSHM\n"); */
       XShmPutImage(b->xm_visual->display, b->drawable, b->gc,
                    ximage, 0, 0, 0, 0, surf->width, surf->height, False);
-   } else
+   }
+   else
 #endif
    {
+      /* display image in Window */
+      ximage = b->tempImage;
+      ximage->data = xm_buf->data;
+
       /* check that the XImage has been previously initialized */
       assert(ximage->format);
       assert(ximage->bitmap_unit);
@@ -286,6 +283,7 @@ xlib_softpipe_display_surface(struct xmesa_buffer *b,
       ximage->height = surf->height;
       ximage->bytes_per_line = spt->stride[surf->level];
 
+      /* _debug_printf("XPUT\n"); */
       XPutImage(b->xm_visual->display, b->drawable, b->gc,
                 ximage, 0, 0, 0, 0, surf->width, surf->height);
    }
@@ -322,21 +320,6 @@ xm_buffer_create(struct pipe_winsys *pws,
                  unsigned size)
 {
    struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer);
-#ifdef USE_XSHM
-   struct xmesa_pipe_winsys *xpws = (struct xmesa_pipe_winsys *) pws;
-
-   buffer->shminfo.shmid = -1;
-   buffer->shminfo.shmaddr = (char *) -1;
-
-   if (xpws->shm && (usage & PIPE_BUFFER_USAGE_PIXEL) != 0) {
-      buffer->shm = xpws->shm;
-
-      if (alloc_shm(buffer, size)) {
-         buffer->data = buffer->shminfo.shmaddr;
-         buffer->shm = 1;
-      }
-   }
-#endif
 
    pipe_reference_init(&buffer->base.reference, 1);
    buffer->base.alignment = alignment;
@@ -363,9 +346,6 @@ xm_user_buffer_create(struct pipe_winsys *pws, void *ptr, unsigned bytes)
    buffer->base.size = bytes;
    buffer->userBuffer = TRUE;
    buffer->data = ptr;
-#ifdef USE_XSHM
-   buffer->shm = 0;
-#endif
 
    return &buffer->base;
 }
@@ -381,16 +361,44 @@ xm_surface_buffer_create(struct pipe_winsys *winsys,
 {
    const unsigned alignment = 64;
    struct pipe_format_block block;
-   unsigned nblocksx, nblocksy;
+   unsigned nblocksx, nblocksy, size;
 
    pf_get_block(format, &block);
    nblocksx = pf_get_nblocksx(&block, width);
    nblocksy = pf_get_nblocksy(&block, height);
    *stride = align(nblocksx * block.size, alignment);
+   size = *stride * nblocksy;
+
+#ifdef USE_XSHM
+   if (!debug_get_bool_option("XLIB_NO_SHM", FALSE))
+   {
+      struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer);
+
+      pipe_reference_init(&buffer->base.reference, 1);
+      buffer->base.alignment = alignment;
+      buffer->base.usage = usage;
+      buffer->base.size = size;
+      buffer->userBuffer = FALSE;
+      buffer->shminfo.shmid = -1;
+      buffer->shminfo.shmaddr = (char *) -1;
+      buffer->shm = TRUE;
+         
+      buffer->data = alloc_shm(buffer, size);
+      if (!buffer->data)
+         goto out;
+
+      return &buffer->base;
+         
+   out:
+      if (buffer)
+         FREE(buffer);
+   }
+#endif
+   
 
    return winsys->buffer_create(winsys, alignment,
                                 usage,
-                                *stride * nblocksy);
+                                size);
 }
 
 
diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c
index a9f3c8e727..4d21b033f7 100644
--- a/src/mesa/drivers/common/driverfuncs.c
+++ b/src/mesa/drivers/common/driverfuncs.c
@@ -56,6 +56,7 @@
 #include "swrast/swrast.h"
 
 #include "driverfuncs.h"
+#include "meta.h"
 
 
 
@@ -100,12 +101,12 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    driver->TexSubImage2D = _mesa_store_texsubimage2d;
    driver->TexSubImage3D = _mesa_store_texsubimage3d;
    driver->GetTexImage = _mesa_get_teximage;
-   driver->CopyTexImage1D = _swrast_copy_teximage1d;
-   driver->CopyTexImage2D = _swrast_copy_teximage2d;
-   driver->CopyTexSubImage1D = _swrast_copy_texsubimage1d;
-   driver->CopyTexSubImage2D = _swrast_copy_texsubimage2d;
-   driver->CopyTexSubImage3D = _swrast_copy_texsubimage3d;
-   driver->GenerateMipmap = _mesa_generate_mipmap;
+   driver->CopyTexImage1D = _mesa_meta_CopyTexImage1D;
+   driver->CopyTexImage2D = _mesa_meta_CopyTexImage2D;
+   driver->CopyTexSubImage1D = _mesa_meta_CopyTexSubImage1D;
+   driver->CopyTexSubImage2D = _mesa_meta_CopyTexSubImage2D;
+   driver->CopyTexSubImage3D = _mesa_meta_CopyTexSubImage3D;
+   driver->GenerateMipmap = _mesa_meta_GenerateMipmap;
    driver->TestProxyTexImage = _mesa_test_proxy_teximage;
    driver->CompressedTexImage1D = _mesa_store_compressed_teximage1d;
    driver->CompressedTexImage2D = _mesa_store_compressed_teximage2d;
@@ -124,15 +125,14 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    driver->UnmapTexture = NULL;
    driver->TextureMemCpy = _mesa_memcpy; 
    driver->IsTextureResident = NULL;
-   driver->PrioritizeTexture = NULL;
    driver->ActiveTexture = NULL;
    driver->UpdateTexturePalette = NULL;
 
    /* imaging */
-   driver->CopyColorTable = _swrast_CopyColorTable;
-   driver->CopyColorSubTable = _swrast_CopyColorSubTable;
-   driver->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
-   driver->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
+   driver->CopyColorTable = _mesa_meta_CopyColorTable;
+   driver->CopyColorSubTable = _mesa_meta_CopyColorSubTable;
+   driver->CopyConvolutionFilter1D = _mesa_meta_CopyConvolutionFilter1D;
+   driver->CopyConvolutionFilter2D = _mesa_meta_CopyConvolutionFilter2D;
 
    /* Vertex/fragment programs */
    driver->BindProgram = NULL;
@@ -182,7 +182,6 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    driver->TexGen = NULL;
    driver->TexEnv = NULL;
    driver->TexParameter = NULL;
-   driver->TextureMatrix = NULL;
    driver->Viewport = NULL;
 
    /* vertex arrays */
diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c
index b6c6ef70fd..b38f22e130 100644
--- a/src/mesa/drivers/common/meta.c
+++ b/src/mesa/drivers/common/meta.c
@@ -37,18 +37,25 @@
 #include "main/arrayobj.h"
 #include "main/blend.h"
 #include "main/bufferobj.h"
+#include "main/buffers.h"
+#include "main/colortab.h"
+#include "main/convolve.h"
 #include "main/depth.h"
 #include "main/enable.h"
+#include "main/fbobject.h"
 #include "main/image.h"
 #include "main/macros.h"
 #include "main/matrix.h"
+#include "main/mipmap.h"
 #include "main/polygon.h"
 #include "main/readpix.h"
 #include "main/scissor.h"
 #include "main/shaders.h"
+#include "main/state.h"
 #include "main/stencil.h"
 #include "main/texobj.h"
 #include "main/texenv.h"
+#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texparam.h"
 #include "main/texstate.h"
@@ -60,6 +67,33 @@
 #include "drivers/common/meta.h"
 
 
+/** Return offset in bytes of the field within a vertex struct */
+#define OFFSET(FIELD) ((void *) offsetof(struct vertex, FIELD))
+
+
+/**
+ * Flags passed to _mesa_meta_begin().
+ */
+/*@{*/
+#define META_ALL              ~0x0
+#define META_ALPHA_TEST        0x1
+#define META_BLEND             0x2  /**< includes logicop */
+#define META_COLOR_MASK        0x4
+#define META_DEPTH_TEST        0x8
+#define META_FOG              0x10
+#define META_PIXEL_STORE      0x20
+#define META_PIXEL_TRANSFER   0x40
+#define META_RASTERIZATION    0x80
+#define META_SCISSOR         0x100
+#define META_SHADER          0x200
+#define META_STENCIL_TEST    0x400
+#define META_TRANSFORM       0x800 /**< modelview, projection, clip planes */
+#define META_TEXTURE        0x1000
+#define META_VERTEX         0x2000
+#define META_VIEWPORT       0x4000
+/*@}*/
+
+
 /**
  * State which we may save/restore across meta ops.
  * XXX this may be incomplete...
@@ -87,6 +121,17 @@ struct save_state
    /** META_PIXEL_STORE */
    struct gl_pixelstore_attrib Pack, Unpack;
 
+   /** META_PIXEL_TRANSFER */
+   GLfloat RedBias, RedScale;
+   GLfloat GreenBias, GreenScale;
+   GLfloat BlueBias, BlueScale;
+   GLfloat AlphaBias, AlphaScale;
+   GLfloat DepthBias, DepthScale;
+   GLboolean MapColorFlag;
+   GLboolean Convolution1DEnabled;
+   GLboolean Convolution2DEnabled;
+   GLboolean Separable2DEnabled;
+
    /** META_RASTERIZATION */
    GLenum FrontPolygonMode, BackPolygonMode;
    GLboolean PolygonOffset;
@@ -138,6 +183,24 @@ struct save_state
 
 
 /**
+ * Temporary texture used for glBlitFramebuffer, glDrawPixels, etc.
+ * This is currently shared by all the meta ops.  But we could create a
+ * separate one for each of glDrawPixel, glBlitFramebuffer, glCopyPixels, etc.
+ */
+struct temp_texture
+{
+   GLuint TexObj;
+   GLenum Target;         /**< GL_TEXTURE_2D or GL_TEXTURE_RECTANGLE */
+   GLsizei MinSize;       /**< Min texture size to allocate */
+   GLsizei MaxSize;       /**< Max possible texture size */
+   GLboolean NPOT;        /**< Non-power of two size OK? */
+   GLsizei Width, Height; /**< Current texture size */
+   GLenum IntFormat;
+   GLfloat Sright, Ttop;  /**< right, top texcoords */
+};
+
+
+/**
  * State for glBlitFramebufer()
  */
 struct blit_state
@@ -181,19 +244,24 @@ struct drawpix_state
 
 
 /**
- * Temporary texture used for glBlitFramebuffer, glDrawPixels, etc.
- * This is currently shared by all the meta ops.  But we could create a
- * separate one for each of glDrawPixel, glBlitFramebuffer, glCopyPixels, etc.
+ * State for glBitmap()
  */
-struct temp_texture
+struct bitmap_state
 {
-   GLuint TexObj;
-   GLenum Target;         /**< GL_TEXTURE_2D or GL_TEXTURE_RECTANGLE */
-   GLsizei MaxSize;       /**< Max possible texture size */
-   GLboolean NPOT;        /**< Non-power of two size OK? */
-   GLsizei Width, Height; /**< Current texture size */
-   GLenum IntFormat;
-   GLfloat Sright, Ttop;  /**< right, top texcoords */
+   GLuint ArrayObj;
+   GLuint VBO;
+   struct temp_texture Tex;  /**< separate texture from other meta ops */
+};
+
+
+/**
+ * State for _mesa_meta_generate_mipmap()
+ */
+struct gen_mipmap_state
+{
+   GLuint ArrayObj;
+   GLuint VBO;
+   GLuint FBO;
 };
 
 
@@ -206,15 +274,12 @@ struct gl_meta_state
 
    struct temp_texture TempTex;
 
-   struct blit_state Blit;    /**< For _mesa_meta_blit_framebuffer() */
-   struct clear_state Clear;  /**< For _mesa_meta_clear() */
-   struct copypix_state CopyPix;  /**< For _mesa_meta_copy_pixels() */
-   struct drawpix_state DrawPix;  /**< For _mesa_meta_draw_pixels() */
-
-   /* other possible meta-ops:
-    * glBitmap()
-    * glGenerateMipmap()
-    */
+   struct blit_state Blit;    /**< For _mesa_meta_BlitFramebuffer() */
+   struct clear_state Clear;  /**< For _mesa_meta_Clear() */
+   struct copypix_state CopyPix;  /**< For _mesa_meta_CopyPixels() */
+   struct drawpix_state DrawPix;  /**< For _mesa_meta_DrawPixels() */
+   struct bitmap_state Bitmap;    /**< For _mesa_meta_Bitmap() */
+   struct gen_mipmap_state Mipmap;    /**< For _mesa_meta_GenerateMipmap() */
 };
 
 
@@ -264,16 +329,16 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state)
    if (state & META_ALPHA_TEST) {
       save->AlphaEnabled = ctx->Color.AlphaEnabled;
       if (ctx->Color.AlphaEnabled)
-         _mesa_Disable(GL_ALPHA_TEST);
+         _mesa_set_enable(ctx, GL_ALPHA_TEST, GL_FALSE);
    }
 
    if (state & META_BLEND) {
       save->BlendEnabled = ctx->Color.BlendEnabled;
       if (ctx->Color.BlendEnabled)
-         _mesa_Disable(GL_BLEND);
+         _mesa_set_enable(ctx, GL_BLEND, GL_FALSE);
       save->ColorLogicOpEnabled = ctx->Color.ColorLogicOpEnabled;
       if (ctx->Color.ColorLogicOpEnabled)
-         _mesa_Disable(GL_COLOR_LOGIC_OP);
+         _mesa_set_enable(ctx, GL_COLOR_LOGIC_OP, GL_FALSE);
    }
 
    if (state & META_COLOR_MASK) {
@@ -288,7 +353,7 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state)
    if (state & META_DEPTH_TEST) {
       save->Depth = ctx->Depth; /* struct copy */
       if (ctx->Depth.Test)
-         _mesa_Disable(GL_DEPTH_TEST);
+         _mesa_set_enable(ctx, GL_DEPTH_TEST, GL_FALSE);
    }
 
    if (state & META_FOG) {
@@ -304,6 +369,35 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state)
       ctx->Unpack = ctx->DefaultPacking;
    }
 
+   if (state & META_PIXEL_TRANSFER) {
+      save->RedScale = ctx->Pixel.RedScale;
+      save->RedBias = ctx->Pixel.RedBias;
+      save->GreenScale = ctx->Pixel.GreenScale;
+      save->GreenBias = ctx->Pixel.GreenBias;
+      save->BlueScale = ctx->Pixel.BlueScale;
+      save->BlueBias = ctx->Pixel.BlueBias;
+      save->AlphaScale = ctx->Pixel.AlphaScale;
+      save->AlphaBias = ctx->Pixel.AlphaBias;
+      save->MapColorFlag = ctx->Pixel.MapColorFlag;
+      save->Convolution1DEnabled = ctx->Pixel.Convolution1DEnabled;
+      save->Convolution2DEnabled = ctx->Pixel.Convolution2DEnabled;
+      save->Separable2DEnabled = ctx->Pixel.Separable2DEnabled;
+      ctx->Pixel.RedScale = 1.0F;
+      ctx->Pixel.RedBias = 0.0F;
+      ctx->Pixel.GreenScale = 1.0F;
+      ctx->Pixel.GreenBias = 0.0F;
+      ctx->Pixel.BlueScale = 1.0F;
+      ctx->Pixel.BlueBias = 0.0F;
+      ctx->Pixel.AlphaScale = 1.0F;
+      ctx->Pixel.AlphaBias = 0.0F;
+      ctx->Pixel.MapColorFlag = GL_FALSE;
+      ctx->Pixel.Convolution1DEnabled = GL_FALSE;
+      ctx->Pixel.Convolution2DEnabled = GL_FALSE;
+      ctx->Pixel.Separable2DEnabled = GL_FALSE;
+      /* XXX more state */
+      ctx->NewState |=_NEW_PIXEL;
+   }
+
    if (state & META_RASTERIZATION) {
       save->FrontPolygonMode = ctx->Polygon.FrontMode;
       save->BackPolygonMode = ctx->Polygon.BackMode;
@@ -345,7 +439,7 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state)
    if (state & META_STENCIL_TEST) {
       save->Stencil = ctx->Stencil; /* struct copy */
       if (ctx->Stencil.Enabled)
-         _mesa_Disable(GL_STENCIL_TEST);
+         _mesa_set_enable(ctx, GL_STENCIL_TEST, GL_FALSE);
       /* NOTE: other stencil state not reset */
    }
 
@@ -499,6 +593,23 @@ _mesa_meta_end(GLcontext *ctx)
       ctx->Unpack = save->Unpack;
    }
 
+   if (state & META_PIXEL_TRANSFER) {
+      ctx->Pixel.RedScale = save->RedScale;
+      ctx->Pixel.RedBias = save->RedBias;
+      ctx->Pixel.GreenScale = save->GreenScale;
+      ctx->Pixel.GreenBias = save->GreenBias;
+      ctx->Pixel.BlueScale = save->BlueScale;
+      ctx->Pixel.BlueBias = save->BlueBias;
+      ctx->Pixel.AlphaScale = save->AlphaScale;
+      ctx->Pixel.AlphaBias = save->AlphaBias;
+      ctx->Pixel.MapColorFlag = save->MapColorFlag;
+      ctx->Pixel.Convolution1DEnabled = save->Convolution1DEnabled;
+      ctx->Pixel.Convolution2DEnabled = save->Convolution2DEnabled;
+      ctx->Pixel.Separable2DEnabled = save->Separable2DEnabled;
+      /* XXX more state */
+      ctx->NewState |=_NEW_PIXEL;
+   }
+
    if (state & META_RASTERIZATION) {
       _mesa_PolygonMode(GL_FRONT, save->FrontPolygonMode);
       _mesa_PolygonMode(GL_BACK, save->BackPolygonMode);
@@ -669,8 +780,35 @@ _mesa_meta_end(GLcontext *ctx)
 
 
 /**
- * Return pointer to temp_texture info.  This does some one-time init
- * if needed.
+ * One-time init for a temp_texture object.
+ * Choose tex target, compute max tex size, etc.
+ */
+static void
+init_temp_texture(GLcontext *ctx, struct temp_texture *tex)
+{
+   /* prefer texture rectangle */
+   if (ctx->Extensions.NV_texture_rectangle) {
+      tex->Target = GL_TEXTURE_RECTANGLE;
+      tex->MaxSize = ctx->Const.MaxTextureRectSize;
+      tex->NPOT = GL_TRUE;
+   }
+   else {
+      /* use 2D texture, NPOT if possible */
+      tex->Target = GL_TEXTURE_2D;
+      tex->MaxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
+      tex->NPOT = ctx->Extensions.ARB_texture_non_power_of_two;
+   }
+   tex->MinSize = 16;  /* 16 x 16 at least */
+   assert(tex->MaxSize > 0);
+
+   _mesa_GenTextures(1, &tex->TexObj);
+   _mesa_BindTexture(tex->Target, tex->TexObj);
+}
+
+
+/**
+ * Return pointer to temp_texture info for non-bitmap ops.
+ * This does some one-time init if needed.
  */
 static struct temp_texture *
 get_temp_texture(GLcontext *ctx)
@@ -678,24 +816,25 @@ get_temp_texture(GLcontext *ctx)
    struct temp_texture *tex = &ctx->Meta->TempTex;
 
    if (!tex->TexObj) {
-      /* do one-time init */
+      init_temp_texture(ctx, tex);
+   }
 
-      /* prefer texture rectangle */
-      if (ctx->Extensions.NV_texture_rectangle) {
-         tex->Target = GL_TEXTURE_RECTANGLE;
-         tex->MaxSize = ctx->Const.MaxTextureRectSize;
-         tex->NPOT = GL_TRUE;
-      }
-      else {
-         /* use 2D texture, NPOT if possible */
-         tex->Target = GL_TEXTURE_2D;
-         tex->MaxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
-         tex->NPOT = ctx->Extensions.ARB_texture_non_power_of_two;
-      }
-      assert(tex->MaxSize > 0);
+   return tex;
+}
 
-      _mesa_GenTextures(1, &tex->TexObj);
-      _mesa_BindTexture(tex->Target, tex->TexObj);
+
+/**
+ * Return pointer to temp_texture info for _mesa_meta_bitmap().
+ * We use a separate texture for bitmaps to reduce texture
+ * allocation/deallocation.
+ */
+static struct temp_texture *
+get_bitmap_temp_texture(GLcontext *ctx)
+{
+   struct temp_texture *tex = &ctx->Meta->Bitmap.Tex;
+
+   if (!tex->TexObj) {
+      init_temp_texture(ctx, tex);
    }
 
    return tex;
@@ -717,6 +856,9 @@ alloc_texture(struct temp_texture *tex,
 {
    GLboolean newTex = GL_FALSE;
 
+   ASSERT(width <= tex->MaxSize);
+   ASSERT(height <= tex->MaxSize);
+
    if (width > tex->Width ||
        height > tex->Height ||
        intFormat != tex->IntFormat) {
@@ -724,13 +866,13 @@ alloc_texture(struct temp_texture *tex,
 
       if (tex->NPOT) {
          /* use non-power of two size */
-         tex->Width = width;
-         tex->Height = height;
+         tex->Width = MAX2(tex->MinSize, width);
+         tex->Height = MAX2(tex->MinSize, height);
       }
       else {
          /* find power of two size */
          GLsizei w, h;
-         w = h = 16;
+         w = h = tex->MinSize;
          while (w < width)
             w *= 2;
          while (h < height)
@@ -803,7 +945,8 @@ setup_copypix_texture(struct temp_texture *tex,
  * Setup/load texture for glDrawPixels.
  */
 static void
-setup_drawpix_texture(struct temp_texture *tex,
+setup_drawpix_texture(GLcontext *ctx,
+		      struct temp_texture *tex,
                       GLboolean newTex,
                       GLenum texIntFormat,
                       GLsizei width, GLsizei height,
@@ -824,9 +967,17 @@ setup_drawpix_texture(struct temp_texture *tex,
                           tex->Width, tex->Height, 0, format, type, pixels);
       }
       else {
+	 struct gl_buffer_object *save_unpack_obj = NULL;
+
+	 _mesa_reference_buffer_object(ctx, &save_unpack_obj,
+				       ctx->Unpack.BufferObj);
+	 _mesa_BindBufferARB(GL_PIXEL_UNPACK_BUFFER_ARB, 0);
          /* create empty texture */
          _mesa_TexImage2D(tex->Target, 0, tex->IntFormat,
                           tex->Width, tex->Height, 0, format, type, NULL);
+	 if (save_unpack_obj != NULL)
+	    _mesa_BindBufferARB(GL_PIXEL_UNPACK_BUFFER_ARB,
+				save_unpack_obj->Name);
          /* load image */
          _mesa_TexSubImage2D(tex->Target, 0,
                              0, 0, width, height, format, type, pixels);
@@ -874,14 +1025,143 @@ init_blit_depth_pixels(GLcontext *ctx)
 
 
 /**
+ * Try to do a glBiltFramebuffer using no-copy texturing.
+ * We can do this when the src renderbuffer is actually a texture.
+ * But if the src buffer == dst buffer we cannot do this.
+ *
+ * \return new buffer mask indicating the buffers left to blit using the
+ *         normal path.
+ */
+static GLbitfield
+blitframebuffer_texture(GLcontext *ctx,
+                        GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
+                        GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
+                        GLbitfield mask, GLenum filter)
+{
+   if (mask & GL_COLOR_BUFFER_BIT) {
+      const struct gl_framebuffer *drawFb = ctx->DrawBuffer;
+      const struct gl_framebuffer *readFb = ctx->ReadBuffer;
+      const struct gl_renderbuffer_attachment *drawAtt =
+         &drawFb->Attachment[drawFb->_ColorDrawBufferIndexes[0]];
+      const struct gl_renderbuffer_attachment *readAtt =
+         &readFb->Attachment[readFb->_ColorReadBufferIndex];
+
+      if (readAtt && readAtt->Texture) {
+         const struct gl_texture_object *texObj = readAtt->Texture;
+         const GLuint srcLevel = readAtt->TextureLevel;
+         const GLenum minFilterSave = texObj->MinFilter;
+         const GLenum magFilterSave = texObj->MagFilter;
+         const GLint baseLevelSave = texObj->BaseLevel;
+         const GLint maxLevelSave = texObj->MaxLevel;
+         const GLenum wrapSSave = texObj->WrapS;
+         const GLenum wrapTSave = texObj->WrapT;
+         const GLenum target = texObj->Target;
+
+         if (drawAtt->Texture == readAtt->Texture) {
+            /* Can't use same texture as both the source and dest.  We need
+             * to handle overlapping blits and besides, some hw may not
+             * support this.
+             */
+            return mask;
+         }
+
+         if (target != GL_TEXTURE_2D && target != GL_TEXTURE_RECTANGLE_ARB) {
+            /* Can't handle other texture types at this time */
+            return mask;
+         }
+
+         /*
+         printf("Blit from texture!\n");
+         printf("  srcAtt %p  dstAtt %p\n", readAtt, drawAtt);
+         printf("  srcTex %p  dstText %p\n", texObj, drawAtt->Texture);
+         */
+
+         /* Prepare src texture state */
+         _mesa_BindTexture(target, texObj->Name);
+         _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, filter);
+         _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, filter);
+         _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, srcLevel);
+         _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, srcLevel);
+         _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
+         _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
+         _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
+         _mesa_set_enable(ctx, target, GL_TRUE);
+
+         /* Prepare vertex data (the VBO was previously created and bound) */
+         {
+            struct vertex {
+               GLfloat x, y, s, t;
+            };
+            struct vertex verts[4];
+            GLfloat s0, t0, s1, t1;
+
+            if (target == GL_TEXTURE_2D) {
+               const struct gl_texture_image *texImage
+                   = _mesa_select_tex_image(ctx, texObj, target, srcLevel);
+               s0 = srcX0 / (float) texImage->Width;
+               s1 = srcX1 / (float) texImage->Width;
+               t0 = srcY0 / (float) texImage->Height;
+               t1 = srcY1 / (float) texImage->Height;
+            }
+            else {
+               assert(target == GL_TEXTURE_RECTANGLE_ARB);
+               s0 = srcX0;
+               s1 = srcX1;
+               t0 = srcY0;
+               t1 = srcY1;
+            }
+
+            verts[0].x = (GLfloat) dstX0;
+            verts[0].y = (GLfloat) dstY0;
+            verts[1].x = (GLfloat) dstX1;
+            verts[1].y = (GLfloat) dstY0;
+            verts[2].x = (GLfloat) dstX1;
+            verts[2].y = (GLfloat) dstY1;
+            verts[3].x = (GLfloat) dstX0;
+            verts[3].y = (GLfloat) dstY1;
+
+            verts[0].s = s0;
+            verts[0].t = t0;
+            verts[1].s = s1;
+            verts[1].t = t0;
+            verts[2].s = s1;
+            verts[2].t = t1;
+            verts[3].s = s0;
+            verts[3].t = t1;
+
+            _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts);
+         }
+
+         _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
+
+         /* Restore texture object state, the texture binding will
+          * be restored by _mesa_meta_end().
+          */
+         _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, minFilterSave);
+         _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, magFilterSave);
+         _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, baseLevelSave);
+         _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, maxLevelSave);
+         _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, wrapSSave);
+         _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, wrapTSave);
+
+         /* Done with color buffer */
+         mask &= ~GL_COLOR_BUFFER_BIT;
+      }
+   }
+
+   return mask;
+}
+
+
+/**
  * Meta implementation of ctx->Driver.BlitFramebuffer() in terms
  * of texture mapping and polygon rendering.
  */
 void
-_mesa_meta_blit_framebuffer(GLcontext *ctx,
-                            GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
-                            GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
-                            GLbitfield mask, GLenum filter)
+_mesa_meta_BlitFramebuffer(GLcontext *ctx,
+                           GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
+                           GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
+                           GLbitfield mask, GLenum filter)
 {
    struct blit_state *blit = &ctx->Meta->Blit;
    struct temp_texture *tex = get_temp_texture(ctx);
@@ -892,7 +1172,10 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx,
    const GLint srcH = abs(srcY1 - srcY0);
    const GLboolean srcFlipX = srcX1 < srcX0;
    const GLboolean srcFlipY = srcY1 < srcY0;
-   GLfloat verts[4][4]; /* four verts of X,Y,S,T */
+   struct vertex {
+      GLfloat x, y, s, t;
+   };
+   struct vertex verts[4];
    GLboolean newTex;
 
    if (srcW > maxTexSize || srcH > maxTexSize) {
@@ -931,10 +1214,8 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx,
                           NULL, GL_DYNAMIC_DRAW_ARB);
 
       /* setup vertex arrays */
-      _mesa_VertexPointer(2, GL_FLOAT, sizeof(verts[0]),
-                          (void *) (0 * sizeof(GLfloat)));
-      _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]),
-                            (void *) (2 * sizeof(GLfloat)));
+      _mesa_VertexPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(x));
+      _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s));
       _mesa_EnableClientState(GL_VERTEX_ARRAY);
       _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY);
    }
@@ -943,33 +1224,45 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx,
       _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, blit->VBO);
    }
 
+   /* Try faster, direct texture approach first */
+   mask = blitframebuffer_texture(ctx, srcX0, srcY0, srcX1, srcY1,
+                                  dstX0, dstY0, dstX1, dstY1, mask, filter);
+   if (mask == 0x0) {
+      _mesa_meta_end(ctx);
+      return;
+   }
+
+   /* Continue with "normal" approach which involves copying the src rect
+    * into a temporary texture and is "blitted" by drawing a textured quad.
+    */
+
    newTex = alloc_texture(tex, srcW, srcH, GL_RGBA);
 
    /* vertex positions/texcoords (after texture allocation!) */
    {
-      verts[0][0] = (GLfloat) dstX0;
-      verts[0][1] = (GLfloat) dstY0;
-      verts[1][0] = (GLfloat) dstX1;
-      verts[1][1] = (GLfloat) dstY0;
-      verts[2][0] = (GLfloat) dstX1;
-      verts[2][1] = (GLfloat) dstY1;
-      verts[3][0] = (GLfloat) dstX0;
-      verts[3][1] = (GLfloat) dstY1;
-
-      verts[0][2] = 0.0F;
-      verts[0][3] = 0.0F;
-      verts[1][2] = tex->Sright;
-      verts[1][3] = 0.0F;
-      verts[2][2] = tex->Sright;
-      verts[2][3] = tex->Ttop;
-      verts[3][2] = 0.0F;
-      verts[3][3] = tex->Ttop;
+      verts[0].x = (GLfloat) dstX0;
+      verts[0].y = (GLfloat) dstY0;
+      verts[1].x = (GLfloat) dstX1;
+      verts[1].y = (GLfloat) dstY0;
+      verts[2].x = (GLfloat) dstX1;
+      verts[2].y = (GLfloat) dstY1;
+      verts[3].x = (GLfloat) dstX0;
+      verts[3].y = (GLfloat) dstY1;
+
+      verts[0].s = 0.0F;
+      verts[0].t = 0.0F;
+      verts[1].s = tex->Sright;
+      verts[1].t = 0.0F;
+      verts[2].s = tex->Sright;
+      verts[2].t = tex->Ttop;
+      verts[3].s = 0.0F;
+      verts[3].t = tex->Ttop;
 
       /* upload new vertex data */
       _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts);
    }
 
-   _mesa_Enable(tex->Target);
+   _mesa_set_enable(ctx, tex->Target, GL_TRUE);
 
    if (mask & GL_COLOR_BUFFER_BIT) {
       setup_copypix_texture(tex, newTex, srcX, srcY, srcW, srcH,
@@ -990,7 +1283,7 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx,
          _mesa_ReadPixels(srcX, srcY, srcW, srcH,
                           GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, tmp);
 
-         setup_drawpix_texture(tex, newTex, GL_DEPTH_COMPONENT, srcW, srcH,
+         setup_drawpix_texture(ctx, tex, newTex, GL_DEPTH_COMPONENT, srcW, srcH,
                                GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, tmp);
 
          _mesa_BindProgram(GL_FRAGMENT_PROGRAM_ARB, blit->DepthFP);
@@ -1011,7 +1304,7 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx,
       /* XXX can't easily do stencil */
    }
 
-   _mesa_Disable(tex->Target);
+   _mesa_set_enable(ctx, tex->Target, GL_FALSE);
 
    _mesa_meta_end(ctx);
 
@@ -1026,10 +1319,13 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx,
  * Meta implementation of ctx->Driver.Clear() in terms of polygon rendering.
  */
 void
-_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers)
+_mesa_meta_Clear(GLcontext *ctx, GLbitfield buffers)
 {
    struct clear_state *clear = &ctx->Meta->Clear;
-   GLfloat verts[4][7]; /* four verts of X,Y,Z,R,G,B,A */
+   struct vertex {
+      GLfloat x, y, z, r, g, b, a;
+   };
+   struct vertex verts[4];
    /* save all state but scissor, pixel pack/unpack */
    GLbitfield metaSave = META_ALL - META_SCISSOR - META_PIXEL_STORE;
 
@@ -1050,14 +1346,10 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers)
       /* create vertex array buffer */
       _mesa_GenBuffersARB(1, &clear->VBO);
       _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, clear->VBO);
-      _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts),
-                          NULL, GL_DYNAMIC_DRAW_ARB);
 
       /* setup vertex arrays */
-      _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]),
-                          (void *) (0 * sizeof(GLfloat)));
-      _mesa_ColorPointer(4, GL_FLOAT, sizeof(verts[0]),
-                         (void *) (3 * sizeof(GLfloat)));
+      _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x));
+      _mesa_ColorPointer(4, GL_FLOAT, sizeof(struct vertex), OFFSET(r));
       _mesa_EnableClientState(GL_VERTEX_ARRAY);
       _mesa_EnableClientState(GL_COLOR_ARRAY);
    }
@@ -1107,26 +1399,30 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers)
       const GLfloat z = 1.0 - 2.0 * ctx->Depth.Clear;
       GLuint i;
 
-      verts[0][0] = x0;
-      verts[0][1] = y0;
-      verts[0][2] = z;
-      verts[1][0] = x1;
-      verts[1][1] = y0;
-      verts[1][2] = z;
-      verts[2][0] = x1;
-      verts[2][1] = y1;
-      verts[2][2] = z;
-      verts[3][0] = x0;
-      verts[3][1] = y1;
-      verts[3][2] = z;
+      verts[0].x = x0;
+      verts[0].y = y0;
+      verts[0].z = z;
+      verts[1].x = x1;
+      verts[1].y = y0;
+      verts[1].z = z;
+      verts[2].x = x1;
+      verts[2].y = y1;
+      verts[2].z = z;
+      verts[3].x = x0;
+      verts[3].y = y1;
+      verts[3].z = z;
 
       /* vertex colors */
       for (i = 0; i < 4; i++) {
-         COPY_4FV(&verts[i][3], ctx->Color.ClearColor);
+         verts[i].r = ctx->Color.ClearColor[0];
+         verts[i].g = ctx->Color.ClearColor[1];
+         verts[i].b = ctx->Color.ClearColor[2];
+         verts[i].a = ctx->Color.ClearColor[3];
       }
 
       /* upload new vertex data */
-      _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts);
+      _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), verts,
+			  GL_DYNAMIC_DRAW_ARB);
    }
 
    /* draw quad */
@@ -1141,13 +1437,16 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers)
  * of texture mapping and polygon rendering.
  */
 void
-_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY,
-                       GLsizei width, GLsizei height,
-                       GLint dstX, GLint dstY, GLenum type)
+_mesa_meta_CopyPixels(GLcontext *ctx, GLint srcX, GLint srcY,
+                      GLsizei width, GLsizei height,
+                      GLint dstX, GLint dstY, GLenum type)
 {
    struct copypix_state *copypix = &ctx->Meta->CopyPix;
    struct temp_texture *tex = get_temp_texture(ctx);
-   GLfloat verts[4][5]; /* four verts of X,Y,Z,S,T */
+   struct vertex {
+      GLfloat x, y, z, s, t;
+   };
+   struct vertex verts[4];
    GLboolean newTex;
    GLenum intFormat = GL_RGBA;
 
@@ -1185,10 +1484,8 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY,
                           NULL, GL_DYNAMIC_DRAW_ARB);
 
       /* setup vertex arrays */
-      _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]),
-                          (void *) (0 * sizeof(GLfloat)));
-      _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]),
-                            (void *) (3 * sizeof(GLfloat)));
+      _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x));
+      _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s));
       _mesa_EnableClientState(GL_VERTEX_ARRAY);
       _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY);
    }
@@ -1207,26 +1504,26 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY,
       const GLfloat dstY1 = dstY + height * ctx->Pixel.ZoomY;
       const GLfloat z = ctx->Current.RasterPos[2];
 
-      verts[0][0] = dstX0;
-      verts[0][1] = dstY0;
-      verts[0][2] = z;
-      verts[0][3] = 0.0F;
-      verts[0][4] = 0.0F;
-      verts[1][0] = dstX1;
-      verts[1][1] = dstY0;
-      verts[1][2] = z;
-      verts[1][3] = tex->Sright;
-      verts[1][4] = 0.0F;
-      verts[2][0] = dstX1;
-      verts[2][1] = dstY1;
-      verts[2][2] = z;
-      verts[2][3] = tex->Sright;
-      verts[2][4] = tex->Ttop;
-      verts[3][0] = dstX0;
-      verts[3][1] = dstY1;
-      verts[3][2] = z;
-      verts[3][3] = 0.0F;
-      verts[3][4] = tex->Ttop;
+      verts[0].x = dstX0;
+      verts[0].y = dstY0;
+      verts[0].z = z;
+      verts[0].s = 0.0F;
+      verts[0].t = 0.0F;
+      verts[1].x = dstX1;
+      verts[1].y = dstY0;
+      verts[1].z = z;
+      verts[1].s = tex->Sright;
+      verts[1].t = 0.0F;
+      verts[2].x = dstX1;
+      verts[2].y = dstY1;
+      verts[2].z = z;
+      verts[2].s = tex->Sright;
+      verts[2].t = tex->Ttop;
+      verts[3].x = dstX0;
+      verts[3].y = dstY1;
+      verts[3].z = z;
+      verts[3].s = 0.0F;
+      verts[3].t = tex->Ttop;
 
       /* upload new vertex data */
       _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts);
@@ -1236,12 +1533,12 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY,
    setup_copypix_texture(tex, newTex, srcX, srcY, width, height,
                          GL_RGBA, GL_NEAREST);
 
-   _mesa_Enable(tex->Target);
+   _mesa_set_enable(ctx, tex->Target, GL_TRUE);
 
    /* draw textured quad */
    _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
 
-   _mesa_Disable(tex->Target);
+   _mesa_set_enable(ctx, tex->Target, GL_FALSE);
 
    _mesa_meta_end(ctx);
 }
@@ -1279,9 +1576,8 @@ tiled_draw_pixels(GLcontext *ctx,
 
          tileUnpack.SkipRows = unpack->SkipRows + j;
 
-         _mesa_meta_draw_pixels(ctx, tileX, tileY,
-                                tileWidth, tileHeight,
-                                format, type, &tileUnpack, pixels);
+         _mesa_meta_DrawPixels(ctx, tileX, tileY, tileWidth, tileHeight,
+                               format, type, &tileUnpack, pixels);
       }
    }
 }
@@ -1390,17 +1686,20 @@ init_draw_depth_pixels(GLcontext *ctx)
  * of texture mapping and polygon rendering.
  */
 void
-_mesa_meta_draw_pixels(GLcontext *ctx,
-		       GLint x, GLint y, GLsizei width, GLsizei height,
-		       GLenum format, GLenum type,
-		       const struct gl_pixelstore_attrib *unpack,
-		       const GLvoid *pixels)
+_mesa_meta_DrawPixels(GLcontext *ctx,
+                      GLint x, GLint y, GLsizei width, GLsizei height,
+                      GLenum format, GLenum type,
+                      const struct gl_pixelstore_attrib *unpack,
+                      const GLvoid *pixels)
 {
    struct drawpix_state *drawpix = &ctx->Meta->DrawPix;
    struct temp_texture *tex = get_temp_texture(ctx);
    const struct gl_pixelstore_attrib unpackSave = ctx->Unpack;
    const GLuint origStencilMask = ctx->Stencil.WriteMask[0];
-   GLfloat verts[4][5]; /* four verts of X,Y,Z,S,T */
+   struct vertex {
+      GLfloat x, y, z, s, t;
+   };
+   struct vertex verts[4];
    GLenum texIntFormat;
    GLboolean fallback, newTex;
    GLbitfield metaExtraSave = 0x0;
@@ -1494,26 +1793,26 @@ _mesa_meta_draw_pixels(GLcontext *ctx,
       const GLfloat y1 = y + height * ctx->Pixel.ZoomY;
       const GLfloat z = ctx->Current.RasterPos[2];
 
-      verts[0][0] = x0;
-      verts[0][1] = y0;
-      verts[0][2] = z;
-      verts[0][3] = 0.0F;
-      verts[0][4] = 0.0F;
-      verts[1][0] = x1;
-      verts[1][1] = y0;
-      verts[1][2] = z;
-      verts[1][3] = tex->Sright;
-      verts[1][4] = 0.0F;
-      verts[2][0] = x1;
-      verts[2][1] = y1;
-      verts[2][2] = z;
-      verts[2][3] = tex->Sright;
-      verts[2][4] = tex->Ttop;
-      verts[3][0] = x0;
-      verts[3][1] = y1;
-      verts[3][2] = z;
-      verts[3][3] = 0.0F;
-      verts[3][4] = tex->Ttop;
+      verts[0].x = x0;
+      verts[0].y = y0;
+      verts[0].z = z;
+      verts[0].s = 0.0F;
+      verts[0].t = 0.0F;
+      verts[1].x = x1;
+      verts[1].y = y0;
+      verts[1].z = z;
+      verts[1].s = tex->Sright;
+      verts[1].t = 0.0F;
+      verts[2].x = x1;
+      verts[2].y = y1;
+      verts[2].z = z;
+      verts[2].s = tex->Sright;
+      verts[2].t = tex->Ttop;
+      verts[3].x = x0;
+      verts[3].y = y1;
+      verts[3].z = z;
+      verts[3].s = 0.0F;
+      verts[3].t = tex->Ttop;
    }
 
    if (drawpix->ArrayObj == 0) {
@@ -1529,17 +1828,15 @@ _mesa_meta_draw_pixels(GLcontext *ctx,
                        verts, GL_DYNAMIC_DRAW_ARB);
 
    /* setup vertex arrays */
-   _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]),
-                       (void *) (0 * sizeof(GLfloat)));
-   _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]),
-                         (void *) (3 * sizeof(GLfloat)));
+   _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x));
+   _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s));
    _mesa_EnableClientState(GL_VERTEX_ARRAY);
    _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY);
 
    /* set given unpack params */
    ctx->Unpack = *unpack;
 
-   _mesa_Enable(tex->Target);
+   _mesa_set_enable(ctx, tex->Target, GL_TRUE);
 
    if (_mesa_is_stencil_format(format)) {
       /* Drawing stencil */
@@ -1548,7 +1845,7 @@ _mesa_meta_draw_pixels(GLcontext *ctx,
       if (!drawpix->StencilFP)
          init_draw_stencil_pixels(ctx);
 
-      setup_drawpix_texture(tex, newTex, texIntFormat, width, height,
+      setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height,
                             GL_ALPHA, type, pixels);
 
       _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE);
@@ -1591,19 +1888,19 @@ _mesa_meta_draw_pixels(GLcontext *ctx,
       _mesa_ProgramLocalParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, 0,
                                         ctx->Current.RasterColor);
 
-      setup_drawpix_texture(tex, newTex, texIntFormat, width, height,
+      setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height,
                             format, type, pixels);
 
       _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
    }
    else {
       /* Drawing RGBA */
-      setup_drawpix_texture(tex, newTex, texIntFormat, width, height,
+      setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height,
                             format, type, pixels);
       _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
    }
 
-   _mesa_Disable(tex->Target);
+   _mesa_set_enable(ctx, tex->Target, GL_FALSE);
 
    _mesa_DeleteBuffersARB(1, &vbo);
 
@@ -1612,3 +1909,889 @@ _mesa_meta_draw_pixels(GLcontext *ctx,
 
    _mesa_meta_end(ctx);
 }
+
+
+/**
+ * Do glBitmap with a alpha texture quad.  Use the alpha test to
+ * cull the 'off' bits.  If alpha test is already enabled, fall back
+ * to swrast (should be a rare case).
+ * A bitmap cache as in the gallium/mesa state tracker would
+ * improve performance a lot.
+ */
+void
+_mesa_meta_Bitmap(GLcontext *ctx,
+                  GLint x, GLint y, GLsizei width, GLsizei height,
+                  const struct gl_pixelstore_attrib *unpack,
+                  const GLubyte *bitmap1)
+{
+   struct bitmap_state *bitmap = &ctx->Meta->Bitmap;
+   struct temp_texture *tex = get_bitmap_temp_texture(ctx);
+   const GLenum texIntFormat = GL_ALPHA;
+   const struct gl_pixelstore_attrib unpackSave = *unpack;
+   struct vertex {
+      GLfloat x, y, z, s, t, r, g, b, a;
+   };
+   struct vertex verts[4];
+   GLboolean newTex;
+   GLubyte *bitmap8;
+
+   /*
+    * Check if swrast fallback is needed.
+    */
+   if (ctx->_ImageTransferState ||
+       ctx->Color.AlphaEnabled ||
+       ctx->Fog.Enabled ||
+       ctx->Texture._EnabledUnits ||
+       width > tex->MaxSize ||
+       height > tex->MaxSize) {
+      _swrast_Bitmap(ctx, x, y, width, height, unpack, bitmap1);
+      return;
+   }
+
+   /* Most GL state applies to glBitmap (like blending, stencil, etc),
+    * but a there's a few things we need to override:
+    */
+   _mesa_meta_begin(ctx, (META_ALPHA_TEST |
+                          META_PIXEL_STORE |
+                          META_RASTERIZATION |
+                          META_SHADER |
+                          META_TEXTURE |
+                          META_TRANSFORM |
+                          META_VERTEX |
+                          META_VIEWPORT));
+
+   if (bitmap->ArrayObj == 0) {
+      /* one-time setup */
+
+      /* create vertex array object */
+      _mesa_GenVertexArraysAPPLE(1, &bitmap->ArrayObj);
+      _mesa_BindVertexArrayAPPLE(bitmap->ArrayObj);
+
+      /* create vertex array buffer */
+      _mesa_GenBuffersARB(1, &bitmap->VBO);
+      _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, bitmap->VBO);
+      _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts),
+                          NULL, GL_DYNAMIC_DRAW_ARB);
+
+      /* setup vertex arrays */
+      _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x));
+      _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s));
+      _mesa_ColorPointer(4, GL_FLOAT, sizeof(struct vertex), OFFSET(r));
+      _mesa_EnableClientState(GL_VERTEX_ARRAY);
+      _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY);
+      _mesa_EnableClientState(GL_COLOR_ARRAY);
+   }
+   else {
+      _mesa_BindVertexArray(bitmap->ArrayObj);
+      _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, bitmap->VBO);
+   }
+
+   newTex = alloc_texture(tex, width, height, texIntFormat);
+
+   /* vertex positions, texcoords, colors (after texture allocation!) */
+   {
+      const GLfloat x0 = (GLfloat) x;
+      const GLfloat y0 = (GLfloat) y;
+      const GLfloat x1 = (GLfloat) (x + width);
+      const GLfloat y1 = (GLfloat) (y + height);
+      const GLfloat z = ctx->Current.RasterPos[2];
+      GLuint i;
+
+      verts[0].x = x0;
+      verts[0].y = y0;
+      verts[0].z = z;
+      verts[0].s = 0.0F;
+      verts[0].t = 0.0F;
+      verts[1].x = x1;
+      verts[1].y = y0;
+      verts[1].z = z;
+      verts[1].s = tex->Sright;
+      verts[1].t = 0.0F;
+      verts[2].x = x1;
+      verts[2].y = y1;
+      verts[2].z = z;
+      verts[2].s = tex->Sright;
+      verts[2].t = tex->Ttop;
+      verts[3].x = x0;
+      verts[3].y = y1;
+      verts[3].z = z;
+      verts[3].s = 0.0F;
+      verts[3].t = tex->Ttop;
+
+      for (i = 0; i < 4; i++) {
+         verts[i].r = ctx->Current.RasterColor[0];
+         verts[i].g = ctx->Current.RasterColor[1];
+         verts[i].b = ctx->Current.RasterColor[2];
+         verts[i].a = ctx->Current.RasterColor[3];
+      }
+
+      /* upload new vertex data */
+      _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts);
+   }
+
+   bitmap1 = _mesa_map_pbo_source(ctx, &unpackSave, bitmap1);
+   if (!bitmap1)
+      return;
+
+   bitmap8 = (GLubyte *) _mesa_calloc(width * height);
+   if (bitmap8) {
+      _mesa_expand_bitmap(width, height, &unpackSave, bitmap1,
+                          bitmap8, width, 0xff);
+
+      _mesa_set_enable(ctx, tex->Target, GL_TRUE);
+
+      _mesa_set_enable(ctx, GL_ALPHA_TEST, GL_TRUE);
+      _mesa_AlphaFunc(GL_GREATER, 0.0);
+
+      setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height,
+                            GL_ALPHA, GL_UNSIGNED_BYTE, bitmap8);
+
+      _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
+
+      _mesa_set_enable(ctx, tex->Target, GL_FALSE);
+
+      _mesa_free(bitmap8);
+   }
+
+   _mesa_unmap_pbo_source(ctx, &unpackSave);
+
+   _mesa_meta_end(ctx);
+}
+
+
+/**
+ * Check if the call to _mesa_meta_GenerateMipmap() will require a
+ * software fallback.  The fallback path will require that the texture
+ * images are mapped.
+ * \return GL_TRUE if a fallback is needed, GL_FALSE otherwise
+ */
+GLboolean
+_mesa_meta_check_generate_mipmap_fallback(GLcontext *ctx, GLenum target,
+                                          struct gl_texture_object *texObj)
+{
+   const GLuint fboSave = ctx->DrawBuffer->Name;
+   struct gen_mipmap_state *mipmap = &ctx->Meta->Mipmap;
+   struct gl_texture_image *baseImage;
+   GLuint srcLevel;
+   GLenum status;
+
+   /* check for fallbacks */
+   if (!ctx->Extensions.EXT_framebuffer_object ||
+       target == GL_TEXTURE_3D) {
+      return GL_TRUE;
+   }
+
+   srcLevel = texObj->BaseLevel;
+   baseImage = _mesa_select_tex_image(ctx, texObj, target, srcLevel);
+   if (!baseImage || baseImage->IsCompressed) {
+      return GL_TRUE;
+   }
+
+   /*
+    * Test that we can actually render in the texture's format.
+    */
+   if (!mipmap->FBO)
+      _mesa_GenFramebuffersEXT(1, &mipmap->FBO);
+   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, mipmap->FBO);
+
+   if (target == GL_TEXTURE_1D) {
+      _mesa_FramebufferTexture1DEXT(GL_FRAMEBUFFER_EXT,
+                                    GL_COLOR_ATTACHMENT0_EXT,
+                                    target, texObj->Name, srcLevel);
+   }
+   else if (target == GL_TEXTURE_3D) {
+      GLint zoffset = 0;
+      _mesa_FramebufferTexture3DEXT(GL_FRAMEBUFFER_EXT,
+                                    GL_COLOR_ATTACHMENT0_EXT,
+                                    target, texObj->Name, srcLevel, zoffset);
+   }
+   else {
+      /* 2D / cube */
+      _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT,
+                                    GL_COLOR_ATTACHMENT0_EXT,
+                                    target, texObj->Name, srcLevel);
+   }
+
+   status = _mesa_CheckFramebufferStatusEXT(GL_FRAMEBUFFER_EXT);
+
+   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fboSave);
+
+   if (status != GL_FRAMEBUFFER_COMPLETE_EXT) {
+      return GL_TRUE;
+   }
+
+   return GL_FALSE;
+}
+
+
+/**
+ * Called via ctx->Driver.GenerateMipmap()
+ * Note: texture borders and 3D texture support not yet complete.
+ */
+void
+_mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target,
+                          struct gl_texture_object *texObj)
+{
+   struct gen_mipmap_state *mipmap = &ctx->Meta->Mipmap;
+   struct vertex {
+      GLfloat x, y, s, t, r;
+   };
+   struct vertex verts[4];
+   const GLuint baseLevel = texObj->BaseLevel;
+   const GLuint maxLevel = texObj->MaxLevel;
+   const GLenum minFilterSave = texObj->MinFilter;
+   const GLenum magFilterSave = texObj->MagFilter;
+   const GLint baseLevelSave = texObj->BaseLevel;
+   const GLint maxLevelSave = texObj->MaxLevel;
+   const GLboolean genMipmapSave = texObj->GenerateMipmap;
+   const GLenum wrapSSave = texObj->WrapS;
+   const GLenum wrapTSave = texObj->WrapT;
+   const GLenum wrapRSave = texObj->WrapR;
+   const GLuint fboSave = ctx->DrawBuffer->Name;
+   GLenum faceTarget;
+   GLuint dstLevel;
+   GLuint border = 0;
+
+   if (_mesa_meta_check_generate_mipmap_fallback(ctx, target, texObj)) {
+      _mesa_generate_mipmap(ctx, target, texObj);
+      return;
+   }
+
+   if (target >= GL_TEXTURE_CUBE_MAP_POSITIVE_X &&
+       target <= GL_TEXTURE_CUBE_MAP_NEGATIVE_Z) {
+      faceTarget = target;
+      target = GL_TEXTURE_CUBE_MAP;
+   }
+   else {
+      faceTarget = target;
+   }
+
+   _mesa_meta_begin(ctx, META_ALL);
+
+   if (mipmap->ArrayObj == 0) {
+      /* one-time setup */
+
+      /* create vertex array object */
+      _mesa_GenVertexArraysAPPLE(1, &mipmap->ArrayObj);
+      _mesa_BindVertexArrayAPPLE(mipmap->ArrayObj);
+
+      /* create vertex array buffer */
+      _mesa_GenBuffersARB(1, &mipmap->VBO);
+      _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, mipmap->VBO);
+      _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts),
+                          NULL, GL_DYNAMIC_DRAW_ARB);
+
+      /* setup vertex arrays */
+      _mesa_VertexPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(x));
+      _mesa_TexCoordPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(s));
+      _mesa_EnableClientState(GL_VERTEX_ARRAY);
+      _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY);
+   }
+   else {
+      _mesa_BindVertexArray(mipmap->ArrayObj);
+      _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, mipmap->VBO);
+   }
+
+   if (!mipmap->FBO) {
+      _mesa_GenFramebuffersEXT(1, &mipmap->FBO);
+   }
+   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, mipmap->FBO);
+
+   _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+   _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
+   _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, GL_FALSE);
+   _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
+   _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
+   _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE);
+
+   _mesa_set_enable(ctx, target, GL_TRUE);
+
+   /* setup texcoords once (XXX what about border?) */
+   switch (faceTarget) {
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_2D:
+      verts[0].s = 0.0F;
+      verts[0].t = 0.0F;
+      verts[0].r = 0.0F;
+      verts[1].s = 1.0F;
+      verts[1].t = 0.0F;
+      verts[1].r = 0.0F;
+      verts[2].s = 1.0F;
+      verts[2].t = 1.0F;
+      verts[2].r = 0.0F;
+      verts[3].s = 0.0F;
+      verts[3].t = 1.0F;
+      verts[3].r = 0.0F;
+      break;
+   case GL_TEXTURE_3D:
+      abort();
+      break;
+   default:
+      /* cube face */
+      {
+         static const GLfloat st[4][2] = {
+            {0.0f, 0.0f}, {1.0f, 0.0f}, {1.0f, 1.0f}, {0.0f, 1.0f}
+         };
+         GLuint i;
+
+         /* loop over quad verts */
+         for (i = 0; i < 4; i++) {
+            /* Compute sc = +/-scale and tc = +/-scale.
+             * Not +/-1 to avoid cube face selection ambiguity near the edges,
+             * though that can still sometimes happen with this scale factor...
+             */
+            const GLfloat scale = 0.9999f;
+            const GLfloat sc = (2.0f * st[i][0] - 1.0f) * scale;
+            const GLfloat tc = (2.0f * st[i][1] - 1.0f) * scale;
+
+            switch (faceTarget) {
+            case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
+               verts[i].s = 1.0f;
+               verts[i].t = -tc;
+               verts[i].r = -sc;
+               break;
+            case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
+               verts[i].s = -1.0f;
+               verts[i].t = -tc;
+               verts[i].r = sc;
+               break;
+            case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
+               verts[i].s = sc;
+               verts[i].t = 1.0f;
+               verts[i].r = tc;
+               break;
+            case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
+               verts[i].s = sc;
+               verts[i].t = -1.0f;
+               verts[i].r = -tc;
+               break;
+            case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
+               verts[i].s = sc;
+               verts[i].t = -tc;
+               verts[i].r = 1.0f;
+               break;
+            case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
+               verts[i].s = -sc;
+               verts[i].t = -tc;
+               verts[i].r = -1.0f;
+               break;
+            default:
+               assert(0);
+            }
+         }
+      }
+   }
+
+   _mesa_set_enable(ctx, target, GL_TRUE);
+
+   /* texture is already locked, unlock now */
+   _mesa_unlock_texture(ctx, texObj);
+
+   for (dstLevel = baseLevel + 1; dstLevel <= maxLevel; dstLevel++) {
+      const struct gl_texture_image *srcImage;
+      const GLuint srcLevel = dstLevel - 1;
+      GLsizei srcWidth, srcHeight, srcDepth;
+      GLsizei dstWidth, dstHeight, dstDepth;
+      GLenum status;
+
+      srcImage = _mesa_select_tex_image(ctx, texObj, faceTarget, srcLevel);
+      assert(srcImage->Border == 0); /* XXX we can fix this */
+
+      /* src size w/out border */
+      srcWidth = srcImage->Width - 2 * border;
+      srcHeight = srcImage->Height - 2 * border;
+      srcDepth = srcImage->Depth - 2 * border;
+
+      /* new dst size w/ border */
+      dstWidth = MAX2(1, srcWidth / 2) + 2 * border;
+      dstHeight = MAX2(1, srcHeight / 2) + 2 * border;
+      dstDepth = MAX2(1, srcDepth / 2) + 2 * border;
+
+      if (dstWidth == srcImage->Width &&
+          dstHeight == srcImage->Height &&
+          dstDepth == srcImage->Depth) {
+         /* all done */
+         break;
+      }
+
+      /* Create empty dest image */
+      if (target == GL_TEXTURE_1D) {
+         _mesa_TexImage1D(target, dstLevel, srcImage->InternalFormat,
+                          dstWidth, border,
+                          GL_RGBA, GL_UNSIGNED_BYTE, NULL);
+      }
+      else if (target == GL_TEXTURE_3D) {
+         _mesa_TexImage3D(target, dstLevel, srcImage->InternalFormat,
+                          dstWidth, dstHeight, dstDepth, border,
+                          GL_RGBA, GL_UNSIGNED_BYTE, NULL);
+      }
+      else {
+         /* 2D or cube */
+         _mesa_TexImage2D(faceTarget, dstLevel, srcImage->InternalFormat,
+                          dstWidth, dstHeight, border,
+                          GL_RGBA, GL_UNSIGNED_BYTE, NULL);
+
+         if (target == GL_TEXTURE_CUBE_MAP) {
+            /* If texturing from a cube, we need to make sure all src faces
+             * have been defined (even if we're not sampling from them.)
+             * Otherwise the texture object will be 'incomplete' and
+             * texturing from it will not be allowed.
+             */
+            GLuint face;
+            for (face = 0; face < 6; face++) {
+               if (!texObj->Image[face][srcLevel] ||
+                   texObj->Image[face][srcLevel]->Width != srcWidth) {
+                  _mesa_TexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face,
+                                   srcLevel, srcImage->InternalFormat,
+                                   srcWidth, srcHeight, border,
+                                   GL_RGBA, GL_UNSIGNED_BYTE, NULL);
+               }
+            }
+         }
+      }
+
+      /* setup vertex positions */
+      {
+         verts[0].x = 0.0F;
+         verts[0].y = 0.0F;
+         verts[1].x = (GLfloat) dstWidth;
+         verts[1].y = 0.0F;
+         verts[2].x = (GLfloat) dstWidth;
+         verts[2].y = (GLfloat) dstHeight;
+         verts[3].x = 0.0F;
+         verts[3].y = (GLfloat) dstHeight;
+
+         /* upload new vertex data */
+         _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts);
+      }
+
+      /* limit sampling to src level */
+      _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, srcLevel);
+      _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, srcLevel);
+
+      /* Set to draw into the current dstLevel */
+      if (target == GL_TEXTURE_1D) {
+         _mesa_FramebufferTexture1DEXT(GL_FRAMEBUFFER_EXT,
+                                       GL_COLOR_ATTACHMENT0_EXT,
+                                       target,
+                                       texObj->Name,
+                                       dstLevel);
+      }
+      else if (target == GL_TEXTURE_3D) {
+         GLint zoffset = 0; /* XXX unfinished */
+         _mesa_FramebufferTexture3DEXT(GL_FRAMEBUFFER_EXT,
+                                       GL_COLOR_ATTACHMENT0_EXT,
+                                       target,
+                                       texObj->Name,
+                                       dstLevel, zoffset);
+      }
+      else {
+         /* 2D / cube */
+         _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT,
+                                       GL_COLOR_ATTACHMENT0_EXT,
+                                       faceTarget,
+                                       texObj->Name,
+                                       dstLevel);
+      }
+
+      _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT);
+
+      /* sanity check */
+      status = _mesa_CheckFramebufferStatusEXT(GL_FRAMEBUFFER_EXT);
+      if (status != GL_FRAMEBUFFER_COMPLETE_EXT) {
+         abort();
+         break;
+      }
+
+      _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
+   }
+
+   _mesa_lock_texture(ctx, texObj); /* relock */
+
+   _mesa_meta_end(ctx);
+
+   _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, minFilterSave);
+   _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, magFilterSave);
+   _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, baseLevelSave);
+   _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, maxLevelSave);
+   _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, genMipmapSave);
+   _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, wrapSSave);
+   _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, wrapTSave);
+   _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, wrapRSave);
+
+   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fboSave);
+}
+
+
+/**
+ * Determine the GL data type to use for the temporary image read with
+ * ReadPixels() and passed to Tex[Sub]Image().
+ */
+static GLenum
+get_temp_image_type(GLcontext *ctx, GLenum baseFormat)
+{
+   switch (baseFormat) {
+   case GL_RGBA:
+   case GL_RGB:
+   case GL_ALPHA:
+   case GL_LUMINANCE:
+   case GL_LUMINANCE_ALPHA:
+   case GL_INTENSITY:
+      if (ctx->DrawBuffer->Visual.redBits <= 8)
+         return GL_UNSIGNED_BYTE;
+      else if (ctx->DrawBuffer->Visual.redBits <= 8)
+         return GL_UNSIGNED_SHORT;
+      else
+         return GL_FLOAT;
+   case GL_DEPTH_COMPONENT:
+      return GL_UNSIGNED_INT;
+   case GL_DEPTH_STENCIL:
+      return GL_UNSIGNED_INT_24_8;
+   default:
+      _mesa_problem(ctx, "Unexpected format in get_temp_image_type()");
+      return 0;
+   }
+}
+
+
+/**
+ * Helper for _mesa_meta_CopyTexImage1/2D() functions.
+ * Have to be careful with locking and meta state for pixel transfer.
+ */
+static void
+copy_tex_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level,
+               GLenum internalFormat, GLint x, GLint y,
+               GLsizei width, GLsizei height, GLint border)
+{
+   struct gl_texture_unit *texUnit;
+   struct gl_texture_object *texObj;
+   struct gl_texture_image *texImage;
+   GLsizei postConvWidth = width, postConvHeight = height;
+   GLenum format, type;
+   GLint bpp;
+   void *buf;
+
+   texUnit = _mesa_get_current_tex_unit(ctx);
+   texObj = _mesa_select_tex_object(ctx, texUnit, target);
+   texImage = _mesa_get_tex_image(ctx, texObj, target, level);
+
+   format = _mesa_base_tex_format(ctx, internalFormat);
+   type = get_temp_image_type(ctx, format);
+   bpp = _mesa_bytes_per_pixel(format, type);
+   if (bpp <= 0) {
+      _mesa_problem(ctx, "Bad bpp in meta copy_tex_image()");
+      return;
+   }
+
+   /*
+    * Alloc image buffer (XXX could use a PBO)
+    */
+   buf = _mesa_malloc(width * height * bpp);
+   if (!buf) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage%uD", dims);
+      return;
+   }
+
+   if (texImage->TexFormat == &_mesa_null_texformat)
+	texImage->TexFormat = ctx->Driver.ChooseTextureFormat(ctx,
+							      internalFormat,
+							      format,
+							      type);
+
+   _mesa_unlock_texture(ctx, texObj); /* need to unlock first */
+
+   /*
+    * Read image from framebuffer (disable pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER);
+   ctx->Driver.ReadPixels(ctx, x, y, width, height,
+			  format, type, &ctx->Pack, buf);
+   _mesa_meta_end(ctx);
+
+   /*
+    * Prepare for new texture image size/data
+    */
+#if FEATURE_convolve
+   if (_mesa_is_color_format(internalFormat)) {
+      _mesa_adjust_image_for_convolution(ctx, 2,
+                                         &postConvWidth, &postConvHeight);
+   }
+#endif
+
+   if (texImage->Data) {
+      ctx->Driver.FreeTexImageData(ctx, texImage);
+   }
+
+   _mesa_init_teximage_fields(ctx, target, texImage,
+                              postConvWidth, postConvHeight, 1,
+                              border, internalFormat);
+
+   /*
+    * Store texture data (with pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE);
+
+   _mesa_update_state(ctx); /* to update pixel transfer state */
+
+   if (target == GL_TEXTURE_1D) {
+      ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
+                             width, border, format, type,
+                             buf, &ctx->Unpack, texObj, texImage);
+   }
+   else {
+      ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
+                             width, height, border, format, type,
+                             buf, &ctx->Unpack, texObj, texImage);
+   }
+   _mesa_meta_end(ctx);
+
+   _mesa_lock_texture(ctx, texObj); /* re-lock */
+
+   _mesa_free(buf);
+}
+
+
+void
+_mesa_meta_CopyTexImage1D(GLcontext *ctx, GLenum target, GLint level,
+                          GLenum internalFormat, GLint x, GLint y,
+                          GLsizei width, GLint border)
+{
+   copy_tex_image(ctx, 1, target, level, internalFormat, x, y,
+                  width, 1, border);
+}
+
+
+void
+_mesa_meta_CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level,
+                          GLenum internalFormat, GLint x, GLint y,
+                          GLsizei width, GLsizei height, GLint border)
+{
+   copy_tex_image(ctx, 2, target, level, internalFormat, x, y,
+                  width, height, border);
+}
+
+
+
+/**
+ * Helper for _mesa_meta_CopyTexSubImage1/2/3D() functions.
+ * Have to be careful with locking and meta state for pixel transfer.
+ */
+static void
+copy_tex_sub_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level,
+                   GLint xoffset, GLint yoffset, GLint zoffset,
+                   GLint x, GLint y,
+                   GLsizei width, GLsizei height)
+{
+   struct gl_texture_unit *texUnit;
+   struct gl_texture_object *texObj;
+   struct gl_texture_image *texImage;
+   GLenum format, type;
+   GLint bpp;
+   void *buf;
+
+   texUnit = _mesa_get_current_tex_unit(ctx);
+   texObj = _mesa_select_tex_object(ctx, texUnit, target);
+   texImage = _mesa_select_tex_image(ctx, texObj, target, level);
+
+   format = texImage->TexFormat->BaseFormat;
+   type = get_temp_image_type(ctx, format);
+   bpp = _mesa_bytes_per_pixel(format, type);
+   if (bpp <= 0) {
+      _mesa_problem(ctx, "Bad bpp in meta copy_tex_sub_image()");
+      return;
+   }
+
+   /*
+    * Alloc image buffer (XXX could use a PBO)
+    */
+   buf = _mesa_malloc(width * height * bpp);
+   if (!buf) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage%uD", dims);
+      return;
+   }
+
+   _mesa_unlock_texture(ctx, texObj); /* need to unlock first */
+
+   /*
+    * Read image from framebuffer (disable pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER);
+   ctx->Driver.ReadPixels(ctx, x, y, width, height,
+			  format, type, &ctx->Pack, buf);
+   _mesa_meta_end(ctx);
+
+   _mesa_update_state(ctx); /* to update pixel transfer state */
+
+   /*
+    * Store texture data (with pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE);
+   if (target == GL_TEXTURE_1D) {
+      ctx->Driver.TexSubImage1D(ctx, target, level, xoffset,
+                                width, format, type, buf,
+                                &ctx->Unpack, texObj, texImage);
+   }
+   else if (target == GL_TEXTURE_3D) {
+      ctx->Driver.TexSubImage3D(ctx, target, level, xoffset, yoffset, zoffset,
+                                width, height, 1, format, type, buf,
+                                &ctx->Unpack, texObj, texImage);
+   }
+   else {
+      ctx->Driver.TexSubImage2D(ctx, target, level, xoffset, yoffset,
+                                width, height, format, type, buf,
+                                &ctx->Unpack, texObj, texImage);
+   }
+   _mesa_meta_end(ctx);
+
+   _mesa_lock_texture(ctx, texObj); /* re-lock */
+
+   _mesa_free(buf);
+}
+
+
+void
+_mesa_meta_CopyTexSubImage1D(GLcontext *ctx, GLenum target, GLint level,
+                             GLint xoffset,
+                             GLint x, GLint y, GLsizei width)
+{
+   copy_tex_sub_image(ctx, 1, target, level, xoffset, 0, 0,
+                      x, y, width, 1);
+}
+
+
+void
+_mesa_meta_CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
+                             GLint xoffset, GLint yoffset,
+                             GLint x, GLint y,
+                             GLsizei width, GLsizei height)
+{
+   copy_tex_sub_image(ctx, 2, target, level, xoffset, yoffset, 0,
+                      x, y, width, height);
+}
+
+
+void
+_mesa_meta_CopyTexSubImage3D(GLcontext *ctx, GLenum target, GLint level,
+                             GLint xoffset, GLint yoffset, GLint zoffset,
+                             GLint x, GLint y,
+                             GLsizei width, GLsizei height)
+{
+   copy_tex_sub_image(ctx, 3, target, level, xoffset, yoffset, zoffset,
+                      x, y, width, height);
+}
+
+
+void
+_mesa_meta_CopyColorTable(GLcontext *ctx,
+                          GLenum target, GLenum internalformat,
+                          GLint x, GLint y, GLsizei width)
+{
+   GLfloat *buf;
+
+   buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat));
+   if (!buf) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyColorTable");
+      return;
+   }
+
+   /*
+    * Read image from framebuffer (disable pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER);
+   ctx->Driver.ReadPixels(ctx, x, y, width, 1,
+                          GL_RGBA, GL_FLOAT, &ctx->Pack, buf);
+
+   _mesa_ColorTable(target, internalformat, width, GL_RGBA, GL_FLOAT, buf);
+
+   _mesa_meta_end(ctx);
+
+   _mesa_free(buf);
+}
+
+
+void
+_mesa_meta_CopyColorSubTable(GLcontext *ctx,GLenum target, GLsizei start,
+                             GLint x, GLint y, GLsizei width)
+{
+   GLfloat *buf;
+
+   buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat));
+   if (!buf) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyColorSubTable");
+      return;
+   }
+
+   /*
+    * Read image from framebuffer (disable pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER);
+   ctx->Driver.ReadPixels(ctx, x, y, width, 1,
+                          GL_RGBA, GL_FLOAT, &ctx->Pack, buf);
+
+   _mesa_ColorSubTable(target, start, width, GL_RGBA, GL_FLOAT, buf);
+
+   _mesa_meta_end(ctx);
+
+   _mesa_free(buf);
+}
+
+
+void
+_mesa_meta_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target,
+                                   GLenum internalFormat,
+                                   GLint x, GLint y, GLsizei width)
+{
+   GLfloat *buf;
+
+   buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat));
+   if (!buf) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyConvolutionFilter2D");
+      return;
+   }
+
+   /*
+    * Read image from framebuffer (disable pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER);
+   _mesa_update_state(ctx);
+   ctx->Driver.ReadPixels(ctx, x, y, width, 1,
+                          GL_RGBA, GL_FLOAT, &ctx->Pack, buf);
+
+   _mesa_ConvolutionFilter1D(target, internalFormat, width,
+                             GL_RGBA, GL_FLOAT, buf);
+
+   _mesa_meta_end(ctx);
+
+   _mesa_free(buf);
+}
+
+
+void
+_mesa_meta_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target,
+                                   GLenum internalFormat, GLint x, GLint y,
+                                   GLsizei width, GLsizei height)
+{
+   GLfloat *buf;
+
+   buf = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat));
+   if (!buf) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyConvolutionFilter2D");
+      return;
+   }
+
+   /*
+    * Read image from framebuffer (disable pixel transfer ops)
+    */
+   _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER);
+   _mesa_update_state(ctx);
+
+   ctx->Driver.ReadPixels(ctx, x, y, width, height,
+                          GL_RGBA, GL_FLOAT, &ctx->Pack, buf);
+
+   _mesa_ConvolutionFilter2D(target, internalFormat, width, height,
+                             GL_RGBA, GL_FLOAT, buf);
+
+   _mesa_meta_end(ctx);
+
+   _mesa_free(buf);
+}
diff --git a/src/mesa/drivers/common/meta.h b/src/mesa/drivers/common/meta.h
index b03b64c48a..6225b94189 100644
--- a/src/mesa/drivers/common/meta.h
+++ b/src/mesa/drivers/common/meta.h
@@ -27,29 +27,6 @@
 #define META_H
 
 
-/**
- * Flags passed to _mesa_meta_begin().
- * XXX these flags may evolve...
- */
-/*@{*/
-#define META_ALPHA_TEST      0x1
-#define META_BLEND           0x2  /**< includes logicop */
-#define META_COLOR_MASK      0x4
-#define META_DEPTH_TEST      0x8
-#define META_FOG            0x10
-#define META_RASTERIZATION  0x20
-#define META_SCISSOR        0x40
-#define META_SHADER         0x80
-#define META_STENCIL_TEST  0x100
-#define META_TRANSFORM     0x200 /**< modelview, projection */
-#define META_TEXTURE       0x400
-#define META_VERTEX        0x800
-#define META_VIEWPORT     0x1000
-#define META_PIXEL_STORE  0x2000
-#define META_ALL            ~0x0
-/*@}*/
-
-
 extern void
 _mesa_meta_init(GLcontext *ctx);
 
@@ -57,25 +34,85 @@ extern void
 _mesa_meta_free(GLcontext *ctx);
 
 extern void
-_mesa_meta_blit_framebuffer(GLcontext *ctx,
-                            GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
-                            GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
-                            GLbitfield mask, GLenum filter);
+_mesa_meta_BlitFramebuffer(GLcontext *ctx,
+                           GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
+                           GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
+                           GLbitfield mask, GLenum filter);
+
+extern void
+_mesa_meta_Clear(GLcontext *ctx, GLbitfield buffers);
+
+extern void
+_mesa_meta_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
+                      GLsizei width, GLsizei height,
+                      GLint dstx, GLint dsty, GLenum type);
+
+extern void
+_mesa_meta_DrawPixels(GLcontext *ctx,
+                      GLint x, GLint y, GLsizei width, GLsizei height,
+                      GLenum format, GLenum type,
+                      const struct gl_pixelstore_attrib *unpack,
+                      const GLvoid *pixels);
+
+extern void
+_mesa_meta_Bitmap(GLcontext *ctx,
+                  GLint x, GLint y, GLsizei width, GLsizei height,
+                  const struct gl_pixelstore_attrib *unpack,
+                  const GLubyte *bitmap);
+
+extern GLboolean
+_mesa_meta_check_generate_mipmap_fallback(GLcontext *ctx, GLenum target,
+                                          struct gl_texture_object *texObj);
+
+extern void
+_mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target,
+                          struct gl_texture_object *texObj);
+
+extern void
+_mesa_meta_CopyTexImage1D(GLcontext *ctx, GLenum target, GLint level,
+                          GLenum internalFormat, GLint x, GLint y,
+                          GLsizei width, GLint border);
+
+extern void
+_mesa_meta_CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level,
+                          GLenum internalFormat, GLint x, GLint y,
+                          GLsizei width, GLsizei height, GLint border);
+
+extern void
+_mesa_meta_CopyTexSubImage1D(GLcontext *ctx, GLenum target, GLint level,
+                             GLint xoffset,
+                             GLint x, GLint y, GLsizei width);
+
+extern void
+_mesa_meta_CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
+                             GLint xoffset, GLint yoffset,
+                             GLint x, GLint y,
+                             GLsizei width, GLsizei height);
+
+extern void
+_mesa_meta_CopyTexSubImage3D(GLcontext *ctx, GLenum target, GLint level,
+                             GLint xoffset, GLint yoffset, GLint zoffset,
+                             GLint x, GLint y,
+                             GLsizei width, GLsizei height);
+
+extern void
+_mesa_meta_CopyColorTable(GLcontext *ctx,
+                          GLenum target, GLenum internalformat,
+                          GLint x, GLint y, GLsizei width);
 
 extern void
-_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers);
+_mesa_meta_CopyColorSubTable(GLcontext *ctx,GLenum target, GLsizei start,
+                             GLint x, GLint y, GLsizei width);
 
 extern void
-_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
-                       GLsizei width, GLsizei height,
-                       GLint dstx, GLint dsty, GLenum type);
+_mesa_meta_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target,
+                                   GLenum internalFormat,
+                                   GLint x, GLint y, GLsizei width);
 
 extern void
-_mesa_meta_draw_pixels(GLcontext *ctx,
-		       GLint x, GLint y, GLsizei width, GLsizei height,
-		       GLenum format, GLenum type,
-		       const struct gl_pixelstore_attrib *unpack,
-		       const GLvoid *pixels);
+_mesa_meta_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target,
+                                   GLenum internalFormat, GLint x, GLint y,
+                                   GLsizei width, GLsizei height);
 
 
 #endif /* META_H */
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template
index 18dbeba24a..1ce9315530 100644
--- a/src/mesa/drivers/dri/Makefile.template
+++ b/src/mesa/drivers/dri/Makefile.template
@@ -60,18 +60,36 @@ SHARED_INCLUDES = \
 
 ##### TARGETS #####
 
-default: symlinks depend $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
+default: symlinks subdirs depend $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
 
 
-$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(PIPE_DRIVERS) $(WINOBJ) Makefile $(TOP)/src/mesa/drivers/dri/Makefile.template
+$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(EXTRA_MODULES) $(WINOBJ) Makefile \
+		$(TOP)/src/mesa/drivers/dri/Makefile.template
 	$(MKLIB) -o $@ -noprefix -linker '$(CC)' -ldflags '$(LDFLAGS)' \
-		$(OBJECTS) $(PIPE_DRIVERS) $(MESA_MODULES) $(WINOBJ) $(DRI_LIB_DEPS)
+		$(OBJECTS) $(MESA_MODULES) $(EXTRA_MODULES) $(WINOBJ) \
+		$(DRI_LIB_DEPS)
 
 
 $(TOP)/$(LIB_DIR)/$(LIBNAME): $(LIBNAME)
 	$(INSTALL) $(LIBNAME) $(TOP)/$(LIB_DIR) 
 
 
+# If the Makefile defined SUBDIRS, run make in each
+.PHONY: subdirs
+subdirs:
+	@if test -n "$(SUBDIRS)" ; then \
+		for dir in $(SUBDIRS) ; do \
+			if [ -d $$dir ] ; then \
+				(cd $$dir && $(MAKE)) || exit 1; \
+			fi \
+		done \
+	fi
+
+
+.PHONY: symlinks
+symlinks:
+
+
 depend: $(C_SOURCES) $(ASM_SOURCES) $(SYMLINKS)
 	@ echo "running $(MKDEP)"
 	@ rm -f depend
diff --git a/src/mesa/drivers/dri/common/drirenderbuffer.h b/src/mesa/drivers/dri/common/drirenderbuffer.h
index cf55286b30..5ae28cb53a 100644
--- a/src/mesa/drivers/dri/common/drirenderbuffer.h
+++ b/src/mesa/drivers/dri/common/drirenderbuffer.h
@@ -56,7 +56,7 @@ typedef struct {
     * A handy flag to know if this is the back color buffer.
     * 
     * \note
-    * This is currently only used by s3v and tdfx.
+    * This is currently only used by tdfx.
     */
    GLboolean backBuffer;
 } driRenderbuffer;
diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h
index 40a030ce0d..5e86324eec 100644
--- a/src/mesa/drivers/dri/common/extension_helper.h
+++ b/src/mesa/drivers/dri/common/extension_helper.h
@@ -316,6 +316,13 @@ static const char CombinerOutputNV_names[] =
     "";
 #endif
 
+#if defined(need_GL_NV_vertex_program)
+static const char VertexAttribs3fvNV_names[] =
+    "iip\0" /* Parameter signature */
+    "glVertexAttribs3fvNV\0"
+    "";
+#endif
+
 #if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
 static const char Uniform2fARB_names[] =
     "iff\0" /* Parameter signature */
@@ -575,6 +582,13 @@ static const char MatrixIndexusvARB_names[] =
     "";
 #endif
 
+#if defined(need_GL_ARB_draw_elements_base_vertex)
+static const char DrawElementsBaseVertex_names[] =
+    "iiipi\0" /* Parameter signature */
+    "glDrawElementsBaseVertex\0"
+    "";
+#endif
+
 #if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
 static const char DisableVertexAttribArrayARB_names[] =
     "i\0" /* Parameter signature */
@@ -970,10 +984,10 @@ static const char GlobalAlphaFactordSUN_names[] =
     "";
 #endif
 
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs3fvNV_names[] =
+#if defined(need_GL_NV_register_combiners)
+static const char GetFinalCombinerInputParameterfvNV_names[] =
     "iip\0" /* Parameter signature */
-    "glVertexAttribs3fvNV\0"
+    "glGetFinalCombinerInputParameterfvNV\0"
     "";
 #endif
 
@@ -3918,6 +3932,13 @@ static const char MapBufferARB_names[] =
     "";
 #endif
 
+#if defined(need_GL_ARB_draw_elements_base_vertex)
+static const char MultiDrawElementsBaseVertex_names[] =
+    "ipipip\0" /* Parameter signature */
+    "glMultiDrawElementsBaseVertex\0"
+    "";
+#endif
+
 #if defined(need_GL_EXT_coordinate_frame)
 static const char Binormal3svEXT_names[] =
     "p\0" /* Parameter signature */
@@ -4396,10 +4417,11 @@ static const char SpriteParameterivSGIX_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_provoking_vertex)
+#if defined(need_GL_EXT_provoking_vertex) || defined(need_GL_ARB_provoking_vertex)
 static const char ProvokingVertexEXT_names[] =
     "i\0" /* Parameter signature */
     "glProvokingVertexEXT\0"
+    "glProvokingVertex\0"
     "";
 #endif
 
@@ -4554,10 +4576,10 @@ static const char UniformMatrix3x4fv_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3fvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glBinormal3fvEXT\0"
+#if defined(need_GL_ARB_draw_elements_base_vertex)
+static const char DrawRangeElementsBaseVertex_names[] =
+    "iiiiipi\0" /* Parameter signature */
+    "glDrawRangeElementsBaseVertex\0"
     "";
 #endif
 
@@ -4710,10 +4732,10 @@ static const char GetFragmentLightfvSGIX_names[] =
     "";
 #endif
 
-#if defined(need_GL_NV_register_combiners)
-static const char GetFinalCombinerInputParameterfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFinalCombinerInputParameterfvNV\0"
+#if defined(need_GL_EXT_coordinate_frame)
+static const char Binormal3fvEXT_names[] =
+    "p\0" /* Parameter signature */
+    "glBinormal3fvEXT\0"
     "";
 #endif
 
@@ -5090,6 +5112,15 @@ static const struct dri_extension_function GL_ARB_draw_buffers_functions[] = {
 };
 #endif
 
+#if defined(need_GL_ARB_draw_elements_base_vertex)
+static const struct dri_extension_function GL_ARB_draw_elements_base_vertex_functions[] = {
+    { DrawElementsBaseVertex_names, DrawElementsBaseVertex_remap_index, -1 },
+    { MultiDrawElementsBaseVertex_names, MultiDrawElementsBaseVertex_remap_index, -1 },
+    { DrawRangeElementsBaseVertex_names, DrawRangeElementsBaseVertex_remap_index, -1 },
+    { NULL, 0, 0 }
+};
+#endif
+
 #if defined(need_GL_ARB_framebuffer_object)
 static const struct dri_extension_function GL_ARB_framebuffer_object_functions[] = {
     { BlitFramebufferEXT_names, BlitFramebufferEXT_remap_index, -1 },
@@ -5164,6 +5195,13 @@ static const struct dri_extension_function GL_ARB_point_parameters_functions[] =
 };
 #endif
 
+#if defined(need_GL_ARB_provoking_vertex)
+static const struct dri_extension_function GL_ARB_provoking_vertex_functions[] = {
+    { ProvokingVertexEXT_names, ProvokingVertexEXT_remap_index, -1 },
+    { NULL, 0, 0 }
+};
+#endif
+
 #if defined(need_GL_ARB_shader_objects)
 static const struct dri_extension_function GL_ARB_shader_objects_functions[] = {
     { UniformMatrix3fvARB_names, UniformMatrix3fvARB_remap_index, -1 },
@@ -5520,8 +5558,8 @@ static const struct dri_extension_function GL_EXT_coordinate_frame_functions[] =
     { Binormal3dvEXT_names, Binormal3dvEXT_remap_index, -1 },
     { Tangent3iEXT_names, Tangent3iEXT_remap_index, -1 },
     { Tangent3bvEXT_names, Tangent3bvEXT_remap_index, -1 },
-    { Binormal3fvEXT_names, Binormal3fvEXT_remap_index, -1 },
     { Tangent3bEXT_names, Tangent3bEXT_remap_index, -1 },
+    { Binormal3fvEXT_names, Binormal3fvEXT_remap_index, -1 },
     { BinormalPointerEXT_names, BinormalPointerEXT_remap_index, -1 },
     { Tangent3svEXT_names, Tangent3svEXT_remap_index, -1 },
     { Binormal3bEXT_names, Binormal3bEXT_remap_index, -1 },
@@ -5963,6 +6001,7 @@ static const struct dri_extension_function GL_NV_point_sprite_functions[] = {
 static const struct dri_extension_function GL_NV_register_combiners_functions[] = {
     { CombinerOutputNV_names, CombinerOutputNV_remap_index, -1 },
     { CombinerParameterfvNV_names, CombinerParameterfvNV_remap_index, -1 },
+    { GetFinalCombinerInputParameterfvNV_names, GetFinalCombinerInputParameterfvNV_remap_index, -1 },
     { GetCombinerOutputParameterfvNV_names, GetCombinerOutputParameterfvNV_remap_index, -1 },
     { FinalCombinerInputNV_names, FinalCombinerInputNV_remap_index, -1 },
     { GetCombinerInputParameterfvNV_names, GetCombinerInputParameterfvNV_remap_index, -1 },
@@ -5971,7 +6010,6 @@ static const struct dri_extension_function GL_NV_register_combiners_functions[]
     { GetFinalCombinerInputParameterivNV_names, GetFinalCombinerInputParameterivNV_remap_index, -1 },
     { CombinerInputNV_names, CombinerInputNV_remap_index, -1 },
     { CombinerParameterfNV_names, CombinerParameterfNV_remap_index, -1 },
-    { GetFinalCombinerInputParameterfvNV_names, GetFinalCombinerInputParameterfvNV_remap_index, -1 },
     { GetCombinerInputParameterivNV_names, GetCombinerInputParameterivNV_remap_index, -1 },
     { CombinerParameterivNV_names, CombinerParameterivNV_remap_index, -1 },
     { NULL, 0, 0 }
@@ -5998,6 +6036,7 @@ static const struct dri_extension_function GL_NV_vertex_array_range_functions[]
 static const struct dri_extension_function GL_NV_vertex_program_functions[] = {
     { VertexAttrib4ubvNV_names, VertexAttrib4ubvNV_remap_index, -1 },
     { VertexAttrib4svNV_names, VertexAttrib4svNV_remap_index, -1 },
+    { VertexAttribs3fvNV_names, VertexAttribs3fvNV_remap_index, -1 },
     { VertexAttribs1dvNV_names, VertexAttribs1dvNV_remap_index, -1 },
     { VertexAttrib1fvNV_names, VertexAttrib1fvNV_remap_index, -1 },
     { VertexAttrib4fNV_names, VertexAttrib4fNV_remap_index, -1 },
@@ -6006,7 +6045,6 @@ static const struct dri_extension_function GL_NV_vertex_program_functions[] = {
     { VertexAttribs3dvNV_names, VertexAttribs3dvNV_remap_index, -1 },
     { VertexAttribs4fvNV_names, VertexAttribs4fvNV_remap_index, -1 },
     { VertexAttrib2sNV_names, VertexAttrib2sNV_remap_index, -1 },
-    { VertexAttribs3fvNV_names, VertexAttribs3fvNV_remap_index, -1 },
     { ProgramEnvParameter4fvARB_names, ProgramEnvParameter4fvARB_remap_index, -1 },
     { LoadProgramNV_names, LoadProgramNV_remap_index, -1 },
     { VertexAttrib4fvNV_names, VertexAttrib4fvNV_remap_index, -1 },
diff --git a/src/mesa/drivers/dri/fb/Makefile b/src/mesa/drivers/dri/fb/Makefile
index 309f50b95f..cf9b3a8556 100644
--- a/src/mesa/drivers/dri/fb/Makefile
+++ b/src/mesa/drivers/dri/fb/Makefile
@@ -25,5 +25,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
-
diff --git a/src/mesa/drivers/dri/ffb/Makefile b/src/mesa/drivers/dri/ffb/Makefile
index cb73238c03..e9da8f9066 100644
--- a/src/mesa/drivers/dri/ffb/Makefile
+++ b/src/mesa/drivers/dri/ffb/Makefile
@@ -33,4 +33,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/ffb/ffb_tex.c b/src/mesa/drivers/dri/ffb/ffb_tex.c
index 69d30aedba..95058e9069 100644
--- a/src/mesa/drivers/dri/ffb/ffb_tex.c
+++ b/src/mesa/drivers/dri/ffb/ffb_tex.c
@@ -30,24 +30,6 @@
 #include "ffb_tex.h"
 
 /* No texture unit, all software. */
-/* XXX this function isn't needed since _mesa_init_driver_functions()
- * will make all these assignments.
- */
 void ffbDDInitTexFuncs(GLcontext *ctx)
 {
-        /*
-	ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format;
-	ctx->Driver.TexImage1D = _mesa_store_teximage1d;
-	ctx->Driver.TexImage2D = _mesa_store_teximage2d;
-	ctx->Driver.TexImage3D = _mesa_store_teximage3d;
-	ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d;
-	ctx->Driver.TexSubImage2D = _mesa_store_texsubimage2d;
-	ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d;
-	ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d;
-	ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d;
-	ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d;
-	ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d;
-	ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d;
-	ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage;
-	*/
 }
diff --git a/src/mesa/drivers/dri/gamma/Makefile b/src/mesa/drivers/dri/gamma/Makefile
index 250d3ac089..09df1578fc 100644
--- a/src/mesa/drivers/dri/gamma/Makefile
+++ b/src/mesa/drivers/dri/gamma/Makefile
@@ -32,4 +32,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/glcore/Makefile b/src/mesa/drivers/dri/glcore/Makefile
deleted file mode 100644
index ac7e1de928..0000000000
--- a/src/mesa/drivers/dri/glcore/Makefile
+++ /dev/null
@@ -1,84 +0,0 @@
-# src/mesa/drivers/dri/glcore/Makefile
-
-TOP = ../../../../..
-include $(TOP)/configs/current
-
-LIBNAME = glcore_dri.so
-
-DRIVER_SOURCES = glcore_driver.c \
-		 $(TOP)/src/mesa/drivers/common/driverfuncs.c \
-		 ../common/dri_util.c
-
-C_SOURCES = \
-	$(DRIVER_SOURCES) \
-	$(DRI_SOURCES) 
-
-
-# Include directories
-INCLUDE_DIRS = \
-	-I. \
-	-I../common \
-	-I../dri_client \
-	-I../dri_client/imports \
-	-Iserver \
-	-I$(TOP)/include \
-	-I$(DRM_SOURCE_PATH)/shared-core \
-	-I$(TOP)/src/mesa \
-	-I$(TOP)/src/mesa/main \
-	-I$(TOP)/src/mesa/glapi \
-	-I$(TOP)/src/mesa/math \
-	-I$(TOP)/src/mesa/transform \
-	-I$(TOP)/src/mesa/shader \
-	-I$(TOP)/src/mesa/swrast \
-	-I$(TOP)/src/mesa/swrast_setup
-
-# Core Mesa objects
-MESA_MODULES = $(TOP)/src/mesa/libmesa.a
-
-# Libraries that the driver shared lib depends on
-LIB_DEPS = -lm -lpthread -lc
-# LIB_DEPS = -lGL -lm -lpthread -lc
-
-
-ASM_SOURCES = 
-
-OBJECTS = $(C_SOURCES:.c=.o) \
-	  $(ASM_SOURCES:.S=.o) 
-
-
-##### RULES #####
-
-.c.o:
-	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@
-
-.S.o:
-	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES)  $< -o $@
-
-
-##### TARGETS #####
-
-default: depend $(TOP)/$(LIB_DIR)/$(LIBNAME)
-
-
-$(TOP)/$(LIB_DIR)/$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile
-	CC="$(CC)" CXX="$(CXX)" $(TOP)/bin/mklib -o $(LIBNAME) -noprefix -install $(TOP)/$(LIB_DIR) \
-		$(OBJECTS) $(WINLIB) $(LIB_DEPS) $(WINOBJ) $(MESA_MODULES)
-
-
-depend: $(C_SOURCES) $(ASM_SOURCES)
-	rm -f depend
-	touch depend
-	$(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(C_SOURCES) $(ASM_SOURCES) \
-		> /dev/null 
-
-
-# Emacs tags
-tags:
-	etags `find . -name \*.[ch]` `find ../include`
-
-
-clean:
-	-rm -f *.o server/*.o
-
-
-include depend
diff --git a/src/mesa/drivers/dri/i810/Makefile b/src/mesa/drivers/dri/i810/Makefile
index a7825b49b4..3874faee51 100644
--- a/src/mesa/drivers/dri/i810/Makefile
+++ b/src/mesa/drivers/dri/i810/Makefile
@@ -29,4 +29,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile
index 9d049dea8f..37f15aa767 100644
--- a/src/mesa/drivers/dri/i915/Makefile
+++ b/src/mesa/drivers/dri/i915/Makefile
@@ -19,7 +19,6 @@ DRIVER_SOURCES = \
 	intel_batchbuffer.c \
 	intel_clear.c \
 	intel_extensions.c \
-	intel_generatemipmap.c \
 	intel_mipmap_tree.c \
 	intel_tex_layout.c \
 	intel_tex_image.c \
@@ -73,4 +72,3 @@ intel_decode.o: ../intel/intel_decode.c
 
 intel_tex_layout.o: ../intel/intel_tex_layout.c
 
-symlinks:
diff --git a/src/mesa/drivers/dri/i915/i830_texstate.c b/src/mesa/drivers/dri/i915/i830_texstate.c
index 6f998fa6f7..f270a13781 100644
--- a/src/mesa/drivers/dri/i915/i830_texstate.c
+++ b/src/mesa/drivers/dri/i915/i830_texstate.c
@@ -162,9 +162,8 @@ i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
    } else {
       dri_bo_reference(intelObj->mt->region->buffer);
       i830->state.tex_buffer[unit] = intelObj->mt->region->buffer;
-      i830->state.tex_offset[unit] = intel_miptree_image_offset(intelObj->mt,
-								0, intelObj->
-								firstLevel);
+      i830->state.tex_offset[unit] =
+         intel_miptree_image_offset(intelObj->mt, 0, intelObj->firstLevel, 0);
 
       format = translate_texture_format(firstImage->TexFormat->MesaFormat,
 					firstImage->InternalFormat);
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index 3ab7d682ee..7d4c7cfbab 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -40,6 +40,7 @@
 
 #include "utils.h"
 #include "i915_reg.h"
+#include "i915_program.h"
 
 #include "intel_regions.h"
 #include "intel_batchbuffer.h"
@@ -80,6 +81,8 @@ i915InvalidateState(GLcontext * ctx, GLuint new_state)
       i915_update_stencil(ctx);
    if (new_state & (_NEW_LIGHT))
        i915_update_provoking_vertex(ctx);
+   if (new_state & (_NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS))
+       i915_update_program(ctx);
 }
 
 
@@ -139,7 +142,7 @@ i915CreateContext(const __GLcontextModes * mesaVis,
    ctx->Const.MaxTextureUnits = I915_TEX_UNITS;
    ctx->Const.MaxTextureImageUnits = I915_TEX_UNITS;
    ctx->Const.MaxTextureCoordUnits = I915_TEX_UNITS;
-
+   ctx->Const.MaxVarying = I915_TEX_UNITS;
 
    /* Advertise the full hardware capabilities.  The new memory
     * manager should cope much better with overload situations:
diff --git a/src/mesa/drivers/dri/i915/i915_context.h b/src/mesa/drivers/dri/i915/i915_context.h
index 8de4a9d0d3..082d614442 100644
--- a/src/mesa/drivers/dri/i915/i915_context.h
+++ b/src/mesa/drivers/dri/i915/i915_context.h
@@ -121,10 +121,14 @@ enum {
 #define I915_MAX_CONSTANT      32
 #define I915_CONSTANT_SIZE     (2+(4*I915_MAX_CONSTANT))
 
+#define I915_MAX_INSN          (I915_MAX_DECL_INSN + \
+				I915_MAX_TEX_INSN + \
+				I915_MAX_ALU_INSN)
 
-#define I915_PROGRAM_SIZE      192
-
-#define I915_MAX_INSN          (I915_MAX_TEX_INSN+I915_MAX_ALU_INSN)
+/* Maximum size of the program packet, which matches the limits on
+ * decl, tex, and ALU instructions.
+ */
+#define I915_PROGRAM_SIZE      (I915_MAX_INSN * 3 + 1)
 
 /* Hardware version of a parsed fragment program.  "Derived" from the
  * mesa fragment_program struct.
@@ -154,8 +158,9 @@ struct i915_fragment_program
     */
    GLcontext *ctx;
 
-   GLuint declarations[I915_PROGRAM_SIZE];
-   GLuint program[I915_PROGRAM_SIZE];
+   /* declarations contains the packet header. */
+   GLuint declarations[I915_MAX_DECL_INSN * 3 + 1];
+   GLuint program[(I915_MAX_TEX_INSN + I915_MAX_ALU_INSN) * 3];
 
    GLfloat constant[I915_MAX_CONSTANT][4];
    GLuint constant_flags[I915_MAX_CONSTANT];
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 2db10c60e9..d9c61446f5 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -89,7 +89,8 @@ src_vector(struct i915_fragment_program *p,
        */
    case PROGRAM_TEMPORARY:
       if (source->Index >= I915_MAX_TEMPORARY) {
-         i915_program_error(p, "Exceeded max temporary reg");
+         i915_program_error(p, "Exceeded max temporary reg: %d/%d",
+			    source->Index, I915_MAX_TEMPORARY);
          return 0;
       }
       src = UREG(REG_TYPE_R, source->Index);
@@ -121,10 +122,23 @@ src_vector(struct i915_fragment_program *p,
          src = i915_emit_decl(p, REG_TYPE_T,
                               T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0),
                               D0_CHANNEL_ALL);
+	 break;
+
+      case FRAG_ATTRIB_VAR0:
+      case FRAG_ATTRIB_VAR0 + 1:
+      case FRAG_ATTRIB_VAR0 + 2:
+      case FRAG_ATTRIB_VAR0 + 3:
+      case FRAG_ATTRIB_VAR0 + 4:
+      case FRAG_ATTRIB_VAR0 + 5:
+      case FRAG_ATTRIB_VAR0 + 6:
+      case FRAG_ATTRIB_VAR0 + 7:
+         src = i915_emit_decl(p, REG_TYPE_T,
+                              T_TEX0 + (source->Index - FRAG_ATTRIB_VAR0),
+                              D0_CHANNEL_ALL);
          break;
 
       default:
-         i915_program_error(p, "Bad source->Index");
+         i915_program_error(p, "Bad source->Index: %d", source->Index);
          return 0;
       }
       break;
@@ -146,6 +160,7 @@ src_vector(struct i915_fragment_program *p,
    case PROGRAM_CONSTANT:
    case PROGRAM_STATE_VAR:
    case PROGRAM_NAMED_PARAM:
+   case PROGRAM_UNIFORM:
       src =
          i915_emit_param4fv(p,
                             program->Base.Parameters->ParameterValues[source->
@@ -153,7 +168,7 @@ src_vector(struct i915_fragment_program *p,
       break;
 
    default:
-      i915_program_error(p, "Bad source->File");
+      i915_program_error(p, "Bad source->File: %d", source->File);
       return 0;
    }
 
@@ -186,13 +201,14 @@ get_result_vector(struct i915_fragment_program *p,
          p->depth_written = 1;
          return UREG(REG_TYPE_OD, 0);
       default:
-         i915_program_error(p, "Bad inst->DstReg.Index");
+         i915_program_error(p, "Bad inst->DstReg.Index: %d",
+			    inst->DstReg.Index);
          return 0;
       }
    case PROGRAM_TEMPORARY:
       return UREG(REG_TYPE_R, inst->DstReg.Index);
    default:
-      i915_program_error(p, "Bad inst->DstReg.File");
+      i915_program_error(p, "Bad inst->DstReg.File: %d", inst->DstReg.File);
       return 0;
    }
 }
@@ -231,7 +247,7 @@ translate_tex_src_target(struct i915_fragment_program *p, GLubyte bit)
    case TEXTURE_CUBE_INDEX:
       return D0_SAMPLE_TYPE_CUBE;
    default:
-      i915_program_error(p, "TexSrcBit");
+      i915_program_error(p, "TexSrcBit: %d", bit);
       return 0;
    }
 }
@@ -351,7 +367,7 @@ upload_program(struct i915_fragment_program *p)
 
    while (1) {
       GLuint src0, src1, src2, flags;
-      GLuint tmp = 0, consts0 = 0, consts1 = 0;
+      GLuint tmp = 0, dst, consts0 = 0, consts1 = 0;
 
       switch (inst->Opcode) {
       case OPCODE_ABS:
@@ -503,6 +519,10 @@ upload_program(struct i915_fragment_program *p)
          EMIT_1ARG_ARITH(A0_FLR);
          break;
 
+      case OPCODE_TRUNC:
+	 EMIT_1ARG_ARITH(A0_TRC);
+	 break;
+
       case OPCODE_FRC:
          EMIT_1ARG_ARITH(A0_FRC);
          break;
@@ -516,6 +536,22 @@ upload_program(struct i915_fragment_program *p)
                          0, src0, T0_TEXKILL);
          break;
 
+      case OPCODE_KIL_NV:
+	 if (inst->DstReg.CondMask == COND_TR) {
+	    tmp = i915_get_utemp(p);
+
+	    i915_emit_texld(p, get_live_regs(p, inst),
+			    tmp, A0_DEST_CHANNEL_ALL,
+			    0, /* use a dummy dest reg */
+			    swizzle(tmp, ONE, ONE, ONE, ONE), /* always */
+			    T0_TEXKILL);
+	 } else {
+	    p->error = 1;
+	    i915_program_error(p, "Unsupported KIL_NV condition code: %d",
+			       inst->DstReg.CondMask);
+	 }
+	 break;
+
       case OPCODE_LG2:
          src0 = src_vector(p, &inst->SrcReg[0], program);
 
@@ -615,6 +651,20 @@ upload_program(struct i915_fragment_program *p)
          EMIT_2ARG_ARITH(A0_MUL);
          break;
 
+      case OPCODE_NOISE1:
+      case OPCODE_NOISE2:
+      case OPCODE_NOISE3:
+      case OPCODE_NOISE4:
+	 /* Don't implement noise because we just don't have the instructions
+	  * to spare.  We aren't the first vendor to do so.
+	  */
+	 i915_program_error(p, "Stubbed-out noise functions");
+	 i915_emit_arith(p,
+			 A0_MOV,
+			 get_result_vector(p, inst),
+			 get_result_flags(inst), 0,
+			 swizzle(src0, ZERO, ZERO, ZERO, ZERO), 0, 0);
+
       case OPCODE_POW:
          src0 = src_vector(p, &inst->SrcReg[0], program);
          src1 = src_vector(p, &inst->SrcReg[1], program);
@@ -721,9 +771,38 @@ upload_program(struct i915_fragment_program *p)
          }
          break;
 
-      case OPCODE_SGE:
-         EMIT_2ARG_ARITH(A0_SGE);
-         break;
+      case OPCODE_SEQ:
+	 tmp = i915_get_utemp(p);
+	 flags = get_result_flags(inst);
+	 dst = get_result_vector(p, inst);
+
+	 /* dst = src1 >= src2 */
+	 i915_emit_arith(p,
+			 A0_SGE,
+			 dst,
+			 flags, 0,
+			 src_vector(p, &inst->SrcReg[0], program),
+			 src_vector(p, &inst->SrcReg[1], program),
+			 0);
+	 /* tmp = src1 <= src2 */
+	 i915_emit_arith(p,
+			 A0_SGE,
+			 tmp,
+			 flags, 0,
+			 negate(src_vector(p, &inst->SrcReg[0], program),
+				1, 1, 1, 1),
+			 negate(src_vector(p, &inst->SrcReg[1], program),
+				1, 1, 1, 1),
+			 0);
+	 /* dst = tmp && dst */
+	 i915_emit_arith(p,
+			 A0_MUL,
+			 dst,
+			 flags, 0,
+			 dst,
+			 tmp,
+			 0);
+	 break;
 
       case OPCODE_SIN:
          src0 = src_vector(p, &inst->SrcReg[0], program);
@@ -809,10 +888,71 @@ upload_program(struct i915_fragment_program *p)
 
          break;
 
+      case OPCODE_SGE:
+	 EMIT_2ARG_ARITH(A0_SGE);
+	 break;
+
+      case OPCODE_SGT:
+	 i915_emit_arith(p,
+			 A0_SLT,
+			 get_result_vector( p, inst ),
+			 get_result_flags( inst ), 0,
+			 negate(src_vector( p, &inst->SrcReg[0], program),
+				1, 1, 1, 1),
+			 negate(src_vector( p, &inst->SrcReg[1], program),
+				1, 1, 1, 1),
+			 0);
+         break;
+
+      case OPCODE_SLE:
+	 i915_emit_arith(p,
+			 A0_SGE,
+			 get_result_vector( p, inst ),
+			 get_result_flags( inst ), 0,
+			 negate(src_vector( p, &inst->SrcReg[0], program),
+				1, 1, 1, 1),
+			 negate(src_vector( p, &inst->SrcReg[1], program),
+				1, 1, 1, 1),
+			 0);
+         break;
+
       case OPCODE_SLT:
          EMIT_2ARG_ARITH(A0_SLT);
          break;
 
+      case OPCODE_SNE:
+	 tmp = i915_get_utemp(p);
+	 flags = get_result_flags(inst);
+	 dst = get_result_vector(p, inst);
+
+	 /* dst = src1 < src2 */
+	 i915_emit_arith(p,
+			 A0_SLT,
+			 dst,
+			 flags, 0,
+			 src_vector(p, &inst->SrcReg[0], program),
+			 src_vector(p, &inst->SrcReg[1], program),
+			 0);
+	 /* tmp = src1 > src2 */
+	 i915_emit_arith(p,
+			 A0_SLT,
+			 tmp,
+			 flags, 0,
+			 negate(src_vector(p, &inst->SrcReg[0], program),
+				1, 1, 1, 1),
+			 negate(src_vector(p, &inst->SrcReg[1], program),
+				1, 1, 1, 1),
+			 0);
+	 /* dst = tmp || dst */
+	 i915_emit_arith(p,
+			 A0_ADD,
+			 dst,
+			 flags | A0_DEST_SATURATE, 0,
+			 dst,
+			 tmp,
+			 0);
+         break;
+
       case OPCODE_SUB:
          src0 = src_vector(p, &inst->SrcReg[0], program);
          src1 = src_vector(p, &inst->SrcReg[1], program);
@@ -869,8 +1009,39 @@ upload_program(struct i915_fragment_program *p)
       case OPCODE_END:
          return;
 
+      case OPCODE_BGNLOOP:
+      case OPCODE_BGNSUB:
+      case OPCODE_BRA:
+      case OPCODE_BRK:
+      case OPCODE_CAL:
+      case OPCODE_CONT:
+      case OPCODE_DDX:
+      case OPCODE_DDY:
+      case OPCODE_ELSE:
+      case OPCODE_ENDIF:
+      case OPCODE_ENDLOOP:
+      case OPCODE_ENDSUB:
+      case OPCODE_IF:
+      case OPCODE_RET:
+	 p->error = 1;
+	 i915_program_error(p, "Unsupported opcode: %s",
+			    _mesa_opcode_string(inst->Opcode));
+	 return;
+
+      case OPCODE_EXP:
+      case OPCODE_LOG:
+	 /* These opcodes are claimed as GLSL, NV_vp, and ARB_vp in
+	  * prog_instruction.h, but apparently GLSL doesn't ever emit them.
+	  * Instead, it translates to EX2 or LG2.
+	  */
+      case OPCODE_TXD:
+      case OPCODE_TXL:
+	 /* These opcodes are claimed by GLSL in prog_instruction.h, but
+	  * only NV_vp/fp appears to emit them.
+	  */
       default:
-         i915_program_error(p, "bad opcode");
+         i915_program_error(p, "bad opcode: %s",
+			    _mesa_opcode_string(inst->Opcode));
          return;
       }
 
@@ -906,7 +1077,7 @@ check_wpos(struct i915_fragment_program *p)
    p->wpos_tex = -1;
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
-      if (inputs & FRAG_BIT_TEX(i))
+      if (inputs & (FRAG_BIT_TEX(i) | FRAG_BIT_VAR(i)))
          continue;
       else if (inputs & FRAG_BIT_WPOS) {
          p->wpos_tex = i;
@@ -1055,6 +1226,28 @@ i915ProgramStringNotify(GLcontext * ctx,
    _tnl_program_string(ctx, target, prog);
 }
 
+void
+i915_update_program(GLcontext *ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   struct i915_fragment_program *fp =
+      (struct i915_fragment_program *) ctx->FragmentProgram._Current;
+
+   if (i915->current_program != fp) {
+      if (i915->current_program) {
+         i915->current_program->on_hardware = 0;
+         i915->current_program->params_uptodate = 0;
+      }
+
+      i915->current_program = fp;
+   }
+
+   if (!fp->translated)
+      translate_program(fp);
+
+   FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, fp->error);
+}
 
 void
 i915ValidateFragmentProgram(struct i915_context *i915)
@@ -1072,16 +1265,6 @@ i915ValidateFragmentProgram(struct i915_context *i915)
    GLuint s2 = S2_TEXCOORD_NONE;
    int i, offset = 0;
 
-   if (i915->current_program != p) {
-      if (i915->current_program) {
-         i915->current_program->on_hardware = 0;
-         i915->current_program->params_uptodate = 0;
-      }
-
-      i915->current_program = p;
-   }
-
-
    /* Important:
     */
    VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
@@ -1125,6 +1308,14 @@ i915ValidateFragmentProgram(struct i915_context *i915)
 
          EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, EMIT_SZ(sz), 0, sz * 4);
       }
+      else if (inputsRead & FRAG_BIT_VAR(i)) {
+         int sz = VB->AttribPtr[_TNL_ATTRIB_GENERIC0 + i]->size;
+
+         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
+         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
+
+         EMIT_ATTR(_TNL_ATTRIB_GENERIC0 + i, EMIT_SZ(sz), 0, sz * 4);
+      }
       else if (i == p->wpos_tex) {
 
          /* If WPOS is required, duplicate the XYZ position data in an
diff --git a/src/mesa/drivers/dri/i915/i915_program.c b/src/mesa/drivers/dri/i915/i915_program.c
index e87700f8e0..e7908bd48f 100644
--- a/src/mesa/drivers/dri/i915/i915_program.c
+++ b/src/mesa/drivers/dri/i915/i915_program.c
@@ -130,6 +130,7 @@ i915_emit_decl(struct i915_fragment_program *p,
    *(p->decl++) = (D0_DCL | D0_DEST(reg) | d0_flags);
    *(p->decl++) = D1_MBZ;
    *(p->decl++) = D2_MBZ;
+   assert(p->decl <= p->declarations + ARRAY_SIZE(p->declarations));
 
    p->nr_decl_insn++;
    return reg;
@@ -186,6 +187,11 @@ i915_emit_arith(struct i915_fragment_program * p,
       p->utemp_flag = old_utemp_flag;   /* restore */
    }
 
+   if (p->csr >= p->program + ARRAY_SIZE(p->program)) {
+      i915_program_error(p, "Program contains too many instructions");
+      return UREG_BAD;
+   }
+
    *(p->csr++) = (op | A0_DEST(dest) | mask | saturate | A0_SRC0(src0));
    *(p->csr++) = (A1_SRC0(src0) | A1_SRC1(src1));
    *(p->csr++) = (A2_SRC1(src1) | A2_SRC2(src2));
@@ -270,6 +276,11 @@ GLuint i915_emit_texld( struct i915_fragment_program *p,
 	  p->register_phases[GET_UREG_NR(coord)] == p->nr_tex_indirect)
 	 p->nr_tex_indirect++;
 
+      if (p->csr >= p->program + ARRAY_SIZE(p->program)) {
+	 i915_program_error(p, "Program contains too many instructions");
+	 return UREG_BAD;
+      }
+
       *(p->csr++) = (op | 
 		     T0_DEST( dest ) |
 		     T0_SAMPLER( sampler ));
@@ -424,12 +435,21 @@ i915_emit_param4fv(struct i915_fragment_program * p, const GLfloat * values)
    return 0;
 }
 
-
-
+/* Warning the user about program errors seems to be quite valuable, from
+ * our bug reports.  It unfortunately means piglit reporting errors
+ * when we fall back to software due to an unsupportable program, though.
+ */
 void
-i915_program_error(struct i915_fragment_program *p, const char *msg)
+i915_program_error(struct i915_fragment_program *p, const char *fmt, ...)
 {
-   _mesa_problem(NULL, "i915_program_error: %s", msg);
+   va_list args;
+
+   fprintf(stderr, "i915_program_error: ");
+   va_start(args, fmt);
+   vfprintf(stderr, fmt, args);
+   va_end(args);
+
+   fprintf(stderr, "\n");
    p->error = 1;
 }
 
@@ -511,7 +531,8 @@ i915_upload_program(struct i915_context *i915,
    GLuint program_size = p->csr - p->program;
    GLuint decl_size = p->decl - p->declarations;
 
-   FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, p->error);
+   if (p->error)
+      return;
 
    /* Could just go straight to the batchbuffer from here:
     */
diff --git a/src/mesa/drivers/dri/i915/i915_program.h b/src/mesa/drivers/dri/i915/i915_program.h
index 14a3f08801..0d17d04865 100644
--- a/src/mesa/drivers/dri/i915/i915_program.h
+++ b/src/mesa/drivers/dri/i915/i915_program.h
@@ -145,7 +145,7 @@ extern GLuint i915_emit_param4fv(struct i915_fragment_program *p,
                                  const GLfloat * values);
 
 extern void i915_program_error(struct i915_fragment_program *p,
-                               const char *msg);
+                               const char *fmt, ...);
 
 extern void i915_init_program(struct i915_context *i915,
                               struct i915_fragment_program *p);
@@ -155,7 +155,6 @@ extern void i915_upload_program(struct i915_context *i915,
 
 extern void i915_fini_program(struct i915_fragment_program *p);
 
-
-
+extern void i915_update_program(GLcontext *ctx);
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index 32d4b30cf9..b2f82f5655 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -173,9 +173,8 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
    } else {
       dri_bo_reference(intelObj->mt->region->buffer);
       i915->state.tex_buffer[unit] = intelObj->mt->region->buffer;
-      i915->state.tex_offset[unit] =  intel_miptree_image_offset(intelObj->mt,
-								 0, intelObj->
-								 firstLevel);
+      i915->state.tex_offset[unit] =
+         intel_miptree_image_offset(intelObj->mt, 0, intelObj->firstLevel, 0);
 
       format = translate_texture_format(firstImage->TexFormat->MesaFormat, 
 					firstImage->InternalFormat,
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index 9a723d3cd7..9e2523932f 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -54,8 +54,7 @@ i915_render_prevalidate(struct intel_context *intel)
 {
    struct i915_context *i915 = i915_context(&intel->ctx);
 
-   if (!intel->Fallback)
-       i915ValidateFragmentProgram(i915);
+   i915ValidateFragmentProgram(i915);
 }
 
 static void
diff --git a/src/mesa/drivers/dri/i915/intel_generatemipmap.c b/src/mesa/drivers/dri/i915/intel_generatemipmap.c
deleted file mode 120000
index 4c6b37ada0..0000000000
--- a/src/mesa/drivers/dri/i915/intel_generatemipmap.c
+++ /dev/null
@@ -1 +0,0 @@
-../intel/intel_generatemipmap.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c
index bc6b29281a..c3cbba8404 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -1076,7 +1076,9 @@ intelRunPipeline(GLcontext * ctx)
       intel->NewGLState = 0;
    }
 
+   intel_map_vertex_shader_textures(ctx);
    _tnl_run_pipeline(ctx);
+   intel_unmap_vertex_shader_textures(ctx);
 
    _mesa_unlock_context_textures(ctx);
 }
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index 6e9a9a29a3..7a55333e89 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -14,7 +14,6 @@ DRIVER_SOURCES = \
 	intel_decode.c \
 	intel_extensions.c \
 	intel_fbo.c \
-	intel_generatemipmap.c \
 	intel_mipmap_tree.c \
 	intel_regions.c \
 	intel_screen.c \
@@ -101,6 +100,5 @@ DRI_LIB_DEPS += -ldrm_intel
 
 include ../Makefile.template
 
-symlinks:
 intel_decode.o: ../intel/intel_decode.c
 intel_tex_layout.o: ../intel/intel_tex_layout.c
diff --git a/src/mesa/drivers/dri/i965/brw_cc.c b/src/mesa/drivers/dri/i965/brw_cc.c
index c724218cf5..1088a7a607 100644
--- a/src/mesa/drivers/dri/i965/brw_cc.c
+++ b/src/mesa/drivers/dri/i965/brw_cc.c
@@ -39,12 +39,14 @@
 
 static void prepare_cc_vp( struct brw_context *brw )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_cc_viewport ccv;
 
    memset(&ccv, 0, sizeof(ccv));
 
-   ccv.min_depth = 0.0;
-   ccv.max_depth = 1.0;
+   /* _NEW_VIEWPORT */
+   ccv.min_depth = ctx->Viewport.Near;
+   ccv.max_depth = ctx->Viewport.Far;
 
    dri_bo_unreference(brw->cc.vp_bo);
    brw->cc.vp_bo = brw_cache_data( &brw->cache, BRW_CC_VP, &ccv, NULL, 0 );
@@ -52,7 +54,7 @@ static void prepare_cc_vp( struct brw_context *brw )
 
 const struct brw_tracked_state brw_cc_vp = {
    .dirty = {
-      .mesa = 0,
+      .mesa = _NEW_VIEWPORT,
       .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c
index 5762c9577c..234b3744bf 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_state.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_state.c
@@ -43,11 +43,14 @@ struct brw_clip_unit_key {
    unsigned int curbe_offset;
 
    unsigned int nr_urb_entries, urb_size;
+
+   GLboolean depth_clamp;
 };
 
 static void
 clip_unit_populate_key(struct brw_context *brw, struct brw_clip_unit_key *key)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    memset(key, 0, sizeof(*key));
 
    /* CACHE_NEW_CLIP_PROG */
@@ -62,6 +65,9 @@ clip_unit_populate_key(struct brw_context *brw, struct brw_clip_unit_key *key)
    /* BRW_NEW_URB_FENCE */
    key->nr_urb_entries = brw->urb.nr_clip_entries;
    key->urb_size = brw->urb.vsize;
+
+   /* _NEW_TRANSOFORM */
+   key->depth_clamp = ctx->Transform.DepthClamp;
 }
 
 static dri_bo *
@@ -117,7 +123,8 @@ clip_unit_create_from_key(struct brw_context *brw,
    clip.clip5.userclip_enable_flags = 0x7f;
    clip.clip5.userclip_must_clip = 1;
    clip.clip5.guard_band_enable = 0;
-   clip.clip5.viewport_z_clip_enable = 1;
+   if (!key->depth_clamp)
+      clip.clip5.viewport_z_clip_enable = 1;
    clip.clip5.viewport_xy_clip_enable = 1;
    clip.clip5.vertex_position_space = BRW_CLIP_NDCSPACE;
    clip.clip5.api_mode = BRW_CLIP_API_OGL;
@@ -168,7 +175,7 @@ static void upload_clip_unit( struct brw_context *brw )
 
 const struct brw_tracked_state brw_clip_unit = {
    .dirty = {
-      .mesa  = 0,
+      .mesa  = _NEW_TRANSFORM,
       .brw   = (BRW_NEW_CURBE_OFFSETS |
 		BRW_NEW_URB_FENCE),
       .cache = CACHE_NEW_CLIP_PROG
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index b1e7ec8465..01b6a4a168 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -706,10 +706,6 @@ void brw_debug_batch(struct intel_context *intel);
 /*======================================================================
  * brw_tex.c
  */
-void brwUpdateTextureState( struct intel_context *intel );
-void brw_FrameBufferTexInit( struct brw_context *brw,
-			     struct intel_region *region );
-void brw_FrameBufferTexDestroy( struct brw_context *brw );
 void brw_validate_textures( struct brw_context *brw );
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index c53bd47bb5..44bb7bd588 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -25,13 +25,15 @@
  * 
  **************************************************************************/
 
-#include <stdlib.h>
 
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/state.h"
-#include "main/api_validate.h"
 #include "main/enums.h"
+#include "tnl/tnl.h"
+#include "vbo/vbo_context.h"
+#include "swrast/swrast.h"
+#include "swrast_setup/swrast_setup.h"
 
 #include "brw_draw.h"
 #include "brw_defines.h"
@@ -42,11 +44,6 @@
 #include "intel_batchbuffer.h"
 #include "intel_buffer_objects.h"
 
-#include "tnl/tnl.h"
-#include "vbo/vbo_context.h"
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-
 #define FILE_DEBUG_FLAG DEBUG_BATCH
 
 static GLuint prim_to_hw_prim[GL_POLYGON+1] = {
@@ -145,7 +142,7 @@ static void brw_emit_prim(struct brw_context *brw,
       prim_packet.start_vert_location += brw->ib.start_vertex_offset;
    prim_packet.instance_count = 1;
    prim_packet.start_instance_location = 0;
-   prim_packet.base_vert_location = 0;
+   prim_packet.base_vert_location = prim->basevertex;
 
    /* Can't wrap here, since we rely on the validated state. */
    brw->no_batch_wrap = GL_TRUE;
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 4f8ceb37bd..375afadcbe 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -25,9 +25,9 @@
  * 
  **************************************************************************/
 
-#include <stdlib.h>
 
 #include "main/glheader.h"
+#include "main/bufferobj.h"
 #include "main/context.h"
 #include "main/state.h"
 #include "main/api_validate.h"
@@ -385,7 +385,7 @@ static void brw_prepare_vertices(struct brw_context *brw)
 
       input->element_size = get_size(input->glarray->Type) * input->glarray->Size;
 
-      if (input->glarray->BufferObj->Name != 0) {
+      if (_mesa_is_bufferobj(input->glarray->BufferObj)) {
 	 struct intel_buffer_object *intel_buffer =
 	    intel_buffer_object(input->glarray->BufferObj);
 
@@ -625,7 +625,7 @@ static void brw_prepare_indices(struct brw_context *brw)
 
    /* Turn into a proper VBO:
     */
-   if (!bufferobj->Name) {
+   if (!_mesa_is_bufferobj(bufferobj)) {
       brw->ib.start_vertex_offset = 0;
 
       /* Get new bufferobj, offset:
diff --git a/src/mesa/drivers/dri/i965/brw_gs_state.c b/src/mesa/drivers/dri/i965/brw_gs_state.c
index a761c03153..ed9d2ffe60 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_state.c
@@ -93,7 +93,10 @@ gs_unit_create_from_key(struct brw_context *brw, struct brw_gs_unit_key *key)
    gs.thread4.nr_urb_entries = key->nr_urb_entries;
    gs.thread4.urb_entry_allocation_size = key->urb_size - 1;
 
-   gs.thread4.max_threads = 0; /* Hardware requirement */
+   if (key->nr_urb_entries >= 8)
+      gs.thread4.max_threads = 1;
+   else
+      gs.thread4.max_threads = 0;
 
    if (BRW_IS_IGDNG(brw))
       gs.thread4.rendering_enable = 1;
diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index f8e46aacf7..c262e1db8b 100644
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -534,14 +534,9 @@ brw_state_cache_bo_delete(struct brw_cache *cache, dri_bo *bo)
    for (i = 0; i < cache->size; i++) {
       for (prev = &cache->items[i]; *prev;) {
 	 struct brw_cache_item *c = *prev;
-	 int j;
-
-	 for (j = 0; j < c->nr_reloc_bufs; j++) {
-	    if (c->reloc_bufs[j] == bo)
-	       break;
-	 }
 
-	 if (j != c->nr_reloc_bufs) {
+	 if (drm_intel_bo_references(c->bo, bo)) {
+	    int j;
 
 	    *prev = c->next;
 
@@ -551,17 +546,8 @@ brw_state_cache_bo_delete(struct brw_cache *cache, dri_bo *bo)
 	    free((void *)c->key);
 	    free(c);
 	    cache->n_items--;
-
-	    /* Delete up the tree.  Notably we're trying to get from
-	     * a request to delete the surface, to deleting the surface state
-	     * object, to deleting the binding table.  We're slack and restart
-	     * the deletion process when we do this because the other delete
-	     * may kill our *prev.
-	     */
-	    brw_state_cache_bo_delete(cache, c->bo);
-	    prev = &cache->items[i];
 	 } else {
-	    prev = &(*prev)->next;
+	    prev = &c->next;
 	 }
       }
    }
diff --git a/src/mesa/drivers/dri/i965/brw_tex.c b/src/mesa/drivers/dri/i965/brw_tex.c
index 71bff166dd..e911b105b2 100644
--- a/src/mesa/drivers/dri/i965/brw_tex.c
+++ b/src/mesa/drivers/dri/i965/brw_tex.c
@@ -39,38 +39,6 @@
 #include "intel_tex.h"
 #include "brw_context.h"
 
-
-void brw_FrameBufferTexInit( struct brw_context *brw,
-			     struct intel_region *region )
-{
-   struct intel_context *intel = &brw->intel;
-   GLcontext *ctx = &intel->ctx;
-   struct gl_texture_object *obj;
-   struct gl_texture_image *img;
-   
-   intel->frame_buffer_texobj = obj =
-      ctx->Driver.NewTextureObject( ctx, (GLuint) -1, GL_TEXTURE_2D );
-
-   obj->MinFilter = GL_NEAREST;
-   obj->MagFilter = GL_NEAREST;
-
-   img = ctx->Driver.NewTextureImage( ctx );
-
-   _mesa_init_teximage_fields( ctx, GL_TEXTURE_2D, img,
-			       region->pitch, region->height, 1, 0,
-			       region->cpp == 4 ? GL_RGBA : GL_RGB );
-   
-   _mesa_set_tex_image( obj, GL_TEXTURE_2D, 0, img );
-}
-
-void brw_FrameBufferTexDestroy( struct brw_context *brw )
-{
-   if (brw->intel.frame_buffer_texobj != NULL)
-      brw->intel.ctx.Driver.DeleteTexture( &brw->intel.ctx,
-					   brw->intel.frame_buffer_texobj );
-   brw->intel.frame_buffer_texobj = NULL;
-}
-
 /**
  * Finalizes all textures, completing any rendering that needs to be done
  * to prepare them.
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 108e19cdbc..1638ef8111 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -1208,7 +1208,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 		 MIN2(c->nr_outputs + 1 + len_vertext_header, (BRW_MAX_MRF-1)), /* msg len */
 		 0,		/* response len */
 		 eot, 		/* eot */
-		 1, 		/* writes complete */
+		 eot, 		/* writes complete */
 		 0, 		/* urb destination offset */
 		 BRW_URB_SWIZZLE_INTERLEAVE);
 
@@ -1270,9 +1270,27 @@ post_vs_emit( struct brw_vs_compile *c,
 }
 
 static uint32_t
-get_predicate(uint32_t swizzle)
+get_predicate(const struct prog_instruction *inst)
 {
-   switch (swizzle) {
+   if (inst->DstReg.CondMask == COND_TR)
+      return BRW_PREDICATE_NONE;
+
+   /* All of GLSL only produces predicates for COND_NE and one channel per
+    * vector.  Fail badly if someone starts doing something else, as it might
+    * mean infinite looping or something.
+    *
+    * We'd like to support all the condition codes, but our hardware doesn't
+    * quite match the Mesa IR, which is modeled after the NV extensions.  For
+    * those, the instruction may update the condition codes or not, then any
+    * later instruction may use one of those condition codes.  For gen4, the
+    * instruction may update the flags register based on one of the condition
+    * codes output by the instruction, and then further instructions may
+    * predicate on that.  We can probably support this, but it won't
+    * necessarily be easy.
+    */
+   assert(inst->DstReg.CondMask == COND_NE);
+
+   switch (inst->DstReg.CondSwizzle) {
    case SWIZZLE_XXXX:
       return BRW_PREDICATE_ALIGN16_REPLICATE_X;
    case SWIZZLE_YYYY:
@@ -1282,7 +1300,8 @@ get_predicate(uint32_t swizzle)
    case SWIZZLE_WWWW:
       return BRW_PREDICATE_ALIGN16_REPLICATE_W;
    default:
-      _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", swizzle);
+      _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n",
+		    inst->DstReg.CondMask);
       return BRW_PREDICATE_NORMAL;
    }
 }
@@ -1294,6 +1313,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 #define MAX_IF_DEPTH 32
 #define MAX_LOOP_DEPTH 32
    struct brw_compile *p = &c->func;
+   struct brw_context *brw = p->brw;
    const GLuint nr_insns = c->vp->program.Base.NumInstructions;
    GLuint insn, if_depth = 0, loop_depth = 0;
    GLuint end_offset = 0;
@@ -1492,8 +1512,8 @@ void brw_vs_emit(struct brw_vs_compile *c )
       case OPCODE_IF:
 	 assert(if_depth < MAX_IF_DEPTH);
 	 if_inst[if_depth] = brw_IF(p, BRW_EXECUTE_8);
-	 if_inst[if_depth]->header.predicate_control =
-	    get_predicate(inst->DstReg.CondSwizzle);
+	 /* Note that brw_IF smashes the predicate_control field. */
+	 if_inst[if_depth]->header.predicate_control = get_predicate(inst);
 	 if_depth++;
 	 break;
       case OPCODE_ELSE:
@@ -1503,45 +1523,48 @@ void brw_vs_emit(struct brw_vs_compile *c )
          assert(if_depth > 0);
 	 brw_ENDIF(p, if_inst[--if_depth]);
 	 break;			
-#if 0
       case OPCODE_BGNLOOP:
          loop_inst[loop_depth++] = brw_DO(p, BRW_EXECUTE_8);
          break;
       case OPCODE_BRK:
+	 brw_set_predicate_control(p, get_predicate(inst));
          brw_BREAK(p);
-         brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
          break;
       case OPCODE_CONT:
+	 brw_set_predicate_control(p, get_predicate(inst));
          brw_CONT(p);
          brw_set_predicate_control(p, BRW_PREDICATE_NONE);
          break;
       case OPCODE_ENDLOOP: 
          {
             struct brw_instruction *inst0, *inst1;
+	    GLuint br = 1;
+
             loop_depth--;
+
+	    if (BRW_IS_IGDNG(brw))
+	       br = 2;
+
             inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]);
             /* patch all the BREAK/CONT instructions from last BEGINLOOP */
             while (inst0 > loop_inst[loop_depth]) {
                inst0--;
                if (inst0->header.opcode == BRW_OPCODE_BREAK) {
-                  inst0->bits3.if_else.jump_count = inst1 - inst0 + 1;
+                  inst0->bits3.if_else.jump_count = br * (inst1 - inst0 + 1);
                   inst0->bits3.if_else.pop_count = 0;
                }
                else if (inst0->header.opcode == BRW_OPCODE_CONTINUE) {
-                  inst0->bits3.if_else.jump_count = inst1 - inst0;
+                  inst0->bits3.if_else.jump_count = br * (inst1 - inst0);
                   inst0->bits3.if_else.pop_count = 0;
                }
             }
          }
          break;
-#else
-         (void) loop_inst;
-         (void) loop_depth;
-#endif
       case OPCODE_BRA:
-         brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+	 brw_set_predicate_control(p, get_predicate(inst));
          brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-         brw_set_predicate_control_flag_value(p, 0xff);
+	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
          break;
       case OPCODE_CAL:
 	 brw_set_access_mode(p, BRW_ALIGN_1);
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index ac11790151..124fde25fe 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -69,8 +69,6 @@ static void brw_destroy_context( struct intel_context *intel )
 
    _mesa_free(brw->wm.compile_data);
 
-   brw_FrameBufferTexDestroy( brw );
-
    for (i = 0; i < brw->state.nr_color_regions; i++)
       intel_region_release(&brw->state.color_regions[i]);
    brw->state.nr_color_regions = 0;
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index ae98b5492d..872b1f3ecf 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -38,6 +38,8 @@
 #include "brw_context.h"
 #include "brw_eu.h"
 
+#define SATURATE (1<<5)
+
 /* A big lookup table is used to figure out which and how many
  * additional regs will inserted before the main payload in the WM
  * program execution.  These mainly relate to depth and stencil
@@ -203,7 +205,6 @@ struct brw_wm_compile {
    GLuint fp_temp;
    GLuint fp_interp_emitted;
    GLuint fp_fragcolor_emitted;
-   GLuint fp_deriv_emitted;
 
    struct prog_src_register pixel_xy;
    struct prog_src_register delta_xy;
@@ -299,5 +300,10 @@ void brw_wm_lookup_iz( GLuint line_aa,
 GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp);
 void brw_wm_glsl_emit(struct brw_context *brw, struct brw_wm_compile *c);
 
+void emit_ddxy(struct brw_compile *p,
+	       const struct brw_reg *dst,
+	       GLuint mask,
+	       GLboolean is_ddx,
+	       const struct brw_reg *arg0);
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c
index 268f7965c0..bf80a2942a 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c
@@ -34,8 +34,6 @@
 #include "brw_context.h"
 #include "brw_wm.h"
 
-#define SATURATE (1<<5)
-
 /* Not quite sure how correct this is - need to understand horiz
  * vs. vertical strides a little better.
  */
@@ -281,6 +279,79 @@ static void emit_frontfacing( struct brw_compile *p,
    brw_set_predicate_control_flag_value(p, 0xff);
 }
 
+/* For OPCODE_DDX and OPCODE_DDY, per channel of output we've got input
+ * looking like:
+ *
+ * arg0: ss0.tl ss0.tr ss0.bl ss0.br ss1.tl ss1.tr ss1.bl ss1.br
+ *
+ * and we're trying to produce:
+ *
+ *           DDX                     DDY
+ * dst: (ss0.tr - ss0.tl)     (ss0.tl - ss0.bl)
+ *      (ss0.tr - ss0.tl)     (ss0.tr - ss0.br)
+ *      (ss0.br - ss0.bl)     (ss0.tl - ss0.bl)
+ *      (ss0.br - ss0.bl)     (ss0.tr - ss0.br)
+ *      (ss1.tr - ss1.tl)     (ss1.tl - ss1.bl)
+ *      (ss1.tr - ss1.tl)     (ss1.tr - ss1.br)
+ *      (ss1.br - ss1.bl)     (ss1.tl - ss1.bl)
+ *      (ss1.br - ss1.bl)     (ss1.tr - ss1.br)
+ *
+ * and add another set of two more subspans if in 16-pixel dispatch mode.
+ *
+ * For DDX, it ends up being easy: width = 2, horiz=0 gets us the same result
+ * for each pair, and vertstride = 2 jumps us 2 elements after processing a
+ * pair. But for DDY, it's harder, as we want to produce the pairs swizzled
+ * between each other.  We could probably do it like ddx and swizzle the right
+ * order later, but bail for now and just produce
+ * ((ss0.tl - ss0.bl)x4 (ss1.tl - ss1.bl)x4)
+ */
+void emit_ddxy(struct brw_compile *p,
+	       const struct brw_reg *dst,
+	       GLuint mask,
+	       GLboolean is_ddx,
+	       const struct brw_reg *arg0)
+{
+   int i;
+   struct brw_reg src0, src1;
+
+   if (mask & SATURATE)
+      brw_set_saturate(p, 1);
+   for (i = 0; i < 4; i++ ) {
+      if (mask & (1<<i)) {
+	 if (is_ddx) {
+	    src0 = brw_reg(arg0[i].file, arg0[i].nr, 1,
+			   BRW_REGISTER_TYPE_F,
+			   BRW_VERTICAL_STRIDE_2,
+			   BRW_WIDTH_2,
+			   BRW_HORIZONTAL_STRIDE_0,
+			   BRW_SWIZZLE_XYZW, WRITEMASK_XYZW);
+	    src1 = brw_reg(arg0[i].file, arg0[i].nr, 0,
+			   BRW_REGISTER_TYPE_F,
+			   BRW_VERTICAL_STRIDE_2,
+			   BRW_WIDTH_2,
+			   BRW_HORIZONTAL_STRIDE_0,
+			   BRW_SWIZZLE_XYZW, WRITEMASK_XYZW);
+	 } else {
+	    src0 = brw_reg(arg0[i].file, arg0[i].nr, 0,
+			   BRW_REGISTER_TYPE_F,
+			   BRW_VERTICAL_STRIDE_4,
+			   BRW_WIDTH_4,
+			   BRW_HORIZONTAL_STRIDE_0,
+			   BRW_SWIZZLE_XYZW, WRITEMASK_XYZW);
+	    src1 = brw_reg(arg0[i].file, arg0[i].nr, 2,
+			   BRW_REGISTER_TYPE_F,
+			   BRW_VERTICAL_STRIDE_4,
+			   BRW_WIDTH_4,
+			   BRW_HORIZONTAL_STRIDE_0,
+			   BRW_SWIZZLE_XYZW, WRITEMASK_XYZW);
+	 }
+	 brw_ADD(p, dst[i], src0, negate(src1));
+      }
+   }
+   if (mask & SATURATE)
+      brw_set_saturate(p, 0);
+}
+
 static void emit_alu1( struct brw_compile *p, 
 		       struct brw_instruction *(*func)(struct brw_compile *, 
 						       struct brw_reg, 
@@ -908,6 +979,20 @@ static void emit_kil( struct brw_wm_compile *c,
    }
 }
 
+/* KIL_NV kills the pixels that are currently executing, not based on a test
+ * of the arguments.
+ */
+static void emit_kil_nv( struct brw_wm_compile *c )
+{
+   struct brw_compile *p = &c->func;
+   struct brw_reg r0uw = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
+
+   brw_push_insn_state(p);
+   brw_set_mask_control(p, BRW_MASK_DISABLE);
+   brw_NOT(p, c->emit_mask_reg, brw_mask_reg(1)); //IMASK
+   brw_AND(p, r0uw, c->emit_mask_reg, r0uw);
+   brw_pop_insn_state(p);
+}
 
 static void fire_fb_write( struct brw_wm_compile *c,
 			   GLuint base_reg,
@@ -1258,6 +1343,14 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]);
 	 break;
 
+      case OPCODE_DDX:
+	 emit_ddxy(p, dst, dst_flags, GL_TRUE, args[0]);
+	 break;
+
+      case OPCODE_DDY:
+	 emit_ddxy(p, dst, dst_flags, GL_FALSE, args[0]);
+	 break;
+
       case OPCODE_DP3:
 	 emit_dp3(p, dst, dst_flags, args[0], args[1]);
 	 break;
@@ -1387,6 +1480,10 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_kil(c, args[0]);
 	 break;
 
+      case OPCODE_KIL_NV:
+	 emit_kil_nv(c);
+	 break;
+
       default:
 	 _mesa_printf("Unsupported opcode %i (%s) in fragment shader\n",
 		      inst->opcode, inst->opcode < MAX_OPCODE ?
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index 123fe841c3..4e3edfbbff 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -494,38 +494,6 @@ static void emit_interp( struct brw_wm_compile *c,
    c->fp_interp_emitted |= 1<<idx;
 }
 
-static void emit_ddx( struct brw_wm_compile *c,
-        const struct prog_instruction *inst )
-{
-    GLuint idx = inst->SrcReg[0].Index;
-    struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx);
-
-    c->fp_deriv_emitted |= 1<<idx;
-    emit_op(c,
-            OPCODE_DDX,
-            inst->DstReg,
-            0,
-            interp,
-            get_pixel_w(c),
-            src_undef());
-}
-
-static void emit_ddy( struct brw_wm_compile *c,
-        const struct prog_instruction *inst )
-{
-    GLuint idx = inst->SrcReg[0].Index;
-    struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx);
-
-    c->fp_deriv_emitted |= 1<<idx;
-    emit_op(c,
-            OPCODE_DDY,
-            inst->DstReg,
-            0,
-            interp,
-            get_pixel_w(c),
-            src_undef());
-}
-
 /***********************************************************************
  * Hacks to extend the program parameter and constant lists.
  */
@@ -1186,12 +1154,6 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
 	  */
 	 out->DstReg.WriteMask = 0;
 	 break;
-      case OPCODE_DDX:
-	 emit_ddx(c, inst);
-	 break;
-      case OPCODE_DDY:
-         emit_ddy(c, inst);
-	break;
       case OPCODE_END:
 	 emit_fb_write(c);
 	 break;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index 7c210abbce..c9fe1dd8ad 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -22,6 +22,7 @@ static struct brw_reg get_dst_reg(struct brw_wm_compile *c,
 GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp)
 {
     int i;
+
     for (i = 0; i < fp->Base.NumInstructions; i++) {
 	const struct prog_instruction *inst = &fp->Base.Instructions[i];
 	switch (inst->Opcode) {
@@ -31,8 +32,6 @@ GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp)
 	    case OPCODE_CAL:
 	    case OPCODE_BRK:
 	    case OPCODE_RET:
-	    case OPCODE_DDX:
-	    case OPCODE_DDY:
 	    case OPCODE_NOISE1:
 	    case OPCODE_NOISE2:
 	    case OPCODE_NOISE3:
@@ -293,7 +292,7 @@ static void prealloc_reg(struct brw_wm_compile *c)
     int i, j;
     struct brw_reg reg;
     int urb_read_length = 0;
-    GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted | c->fp_deriv_emitted;
+    GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted;
     GLuint reg_index = 0;
 
     memset(c->used_grf, GL_FALSE, sizeof(c->used_grf));
@@ -1474,61 +1473,6 @@ static void emit_sne(struct brw_wm_compile *c,
     emit_sop(c, inst, BRW_CONDITIONAL_NEQ);
 }
 
-static void emit_ddx(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    struct brw_reg interp[4];
-    struct brw_reg dst;
-    struct brw_reg src0, w;
-    GLuint nr, i;
-    src0 = get_src_reg(c, inst, 0, 0);
-    w = get_src_reg(c, inst, 1, 3);
-    nr = src0.nr;
-    interp[0] = brw_vec1_grf(nr, 0);
-    interp[1] = brw_vec1_grf(nr, 4);
-    interp[2] = brw_vec1_grf(nr+1, 0);
-    interp[3] = brw_vec1_grf(nr+1, 4);
-    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
-    for(i = 0; i < 4; i++ ) {
-        if (mask & (1<<i)) {
-            dst = get_dst_reg(c, inst, i);
-            brw_MOV(p, dst, interp[i]);
-            brw_MUL(p, dst, dst, w);
-        }
-    }
-    brw_set_saturate(p, 0);
-}
-
-static void emit_ddy(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    struct brw_reg interp[4];
-    struct brw_reg dst;
-    struct brw_reg src0, w;
-    GLuint nr, i;
-
-    src0 = get_src_reg(c, inst, 0, 0);
-    nr = src0.nr;
-    w = get_src_reg(c, inst, 1, 3);
-    interp[0] = brw_vec1_grf(nr, 0);
-    interp[1] = brw_vec1_grf(nr, 4);
-    interp[2] = brw_vec1_grf(nr+1, 0);
-    interp[3] = brw_vec1_grf(nr+1, 4);
-    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
-    for(i = 0; i < 4; i++ ) {
-        if (mask & (1<<i)) {
-            dst = get_dst_reg(c, inst, i);
-            brw_MOV(p, dst, suboffset(interp[i], 1));
-            brw_MUL(p, dst, dst, w);
-        }
-    }
-    brw_set_saturate(p, 0);
-}
-
 static INLINE struct brw_reg high_words( struct brw_reg reg )
 {
     return stride( suboffset( retype( reg, BRW_REGISTER_TYPE_W ), 1 ),
@@ -2780,6 +2724,21 @@ static void post_wm_emit( struct brw_wm_compile *c )
     brw_resolve_cals(&c->func);
 }
 
+static void
+get_argument_regs(struct brw_wm_compile *c,
+		  const struct prog_instruction *inst,
+		  int index,
+		  struct brw_reg *regs,
+		  int mask)
+{
+    int i;
+
+    for (i = 0; i < 4; i++) {
+	if (mask & (1 << i))
+	    regs[i] = get_src_reg(c, inst, index, i);
+    }
+}
+
 static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 {
 #define MAX_IF_DEPTH 32
@@ -2797,6 +2756,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 
     for (i = 0; i < c->nr_fp_insns; i++) {
         const struct prog_instruction *inst = &c->prog_instructions[i];
+	int dst_flags;
+	struct brw_reg args[3][4], dst[4];
+	int j;
 
         c->cur_inst = i;
 
@@ -2814,6 +2776,10 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 	else
 	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE);
 
+	dst_flags = inst->DstReg.WriteMask;
+	if (inst->SaturateMode == SATURATE_ZERO_ONE)
+	    dst_flags |= SATURATE;
+
 	switch (inst->Opcode) {
 	    case WM_PIXELXY:
 		emit_pixel_xy(c, inst);
@@ -2899,10 +2865,16 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 		emit_min_max(c, inst);
 		break;
 	    case OPCODE_DDX:
-		emit_ddx(c, inst);
-		break;
 	    case OPCODE_DDY:
-                emit_ddy(c, inst);
+		for (j = 0; j < 4; j++) {
+		    if (inst->DstReg.WriteMask & (1 << j))
+			dst[j] = get_dst_reg(c, inst, j);
+		    else
+			dst[j] = brw_null_reg();
+		}
+		get_argument_regs(c, inst, 0, args[0], WRITEMASK_XYZW);
+		emit_ddxy(p, dst, dst_flags, (inst->Opcode == OPCODE_DDX),
+			  args[0]);
                 break;
 	    case OPCODE_SLT:
 		emit_slt(c, inst);
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass1.c b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
index 3436a24717..b449394029 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass1.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
@@ -178,6 +178,11 @@ void brw_wm_pass1( struct brw_wm_compile *c )
 	 read1 = writemask;
 	 break;
 
+      case OPCODE_DDX:
+      case OPCODE_DDY:
+	 read0 = writemask;
+	 break;
+
       case OPCODE_MAD:	
       case OPCODE_CMP:
       case OPCODE_LRP:
@@ -270,6 +275,7 @@ void brw_wm_pass1( struct brw_wm_compile *c )
 
       case OPCODE_DST:
       case WM_FRONTFACING:
+      case OPCODE_KIL_NV:
       default:
 	 break;
       }
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index 39f8c6d522..361f91292b 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -107,6 +107,12 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
    /* as far as we can tell */
    key->computes_depth =
       (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH)) != 0;
+   /* BRW_NEW_DEPTH_BUFFER
+    * Override for NULL depthbuffer case, required by the Pixel Shader Computed
+    * Depth field.
+    */
+   if (brw->state.depth_region == NULL)
+      key->computes_depth = 0;
 
    /* _NEW_COLOR */
    key->uses_kill = fp->UsesKill || ctx->Color.AlphaEnabled;
@@ -300,6 +306,7 @@ const struct brw_tracked_state brw_wm_unit = {
 
       .brw = (BRW_NEW_FRAGMENT_PROGRAM | 
 	      BRW_NEW_CURBE_OFFSETS |
+	      BRW_NEW_DEPTH_BUFFER |
 	      BRW_NEW_NR_WM_SURFACES),
 
       .cache = (CACHE_NEW_WM_PROG |
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index 3dcc592bde..f7cc5153a8 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -545,15 +545,20 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
 		       irb->texformat->MesaFormat);
       }
       key.tiling = region->tiling;
-      key.width = region->width;
-      key.height = region->height;
+      if (brw->intel.intelScreen->driScrnPriv->dri2.enabled) {
+	 key.width = rb->Width;
+	 key.height = rb->Height;
+      } else {
+	 key.width = region->width;
+	 key.height = region->height;
+      }
       key.pitch = region->pitch;
       key.cpp = region->cpp;
       key.draw_offset = region->draw_offset; /* cur 3d or cube face offset */
    } else {
       key.surface_type = BRW_SURFACE_NULL;
       key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-      key.tiling = 0;
+      key.tiling = I915_TILING_X;
       key.width = 1;
       key.height = 1;
       key.cpp = 4;
@@ -655,7 +660,7 @@ brw_wm_get_binding_table(struct brw_context *brw)
 
    if (bind_bo == NULL) {
       GLuint data_size = brw->wm.nr_surfaces * sizeof(GLuint);
-      uint32_t *data = malloc(data_size);
+      uint32_t data[BRW_WM_MAX_SURF];
       int i;
 
       for (i = 0; i < brw->wm.nr_surfaces; i++)
@@ -680,8 +685,6 @@ brw_wm_get_binding_table(struct brw_context *brw)
 			      brw->wm.surf_bo[i]);
 	 }
       }
-
-      free(data);
    }
 
    return bind_bo;
@@ -690,7 +693,6 @@ brw_wm_get_binding_table(struct brw_context *brw)
 static void prepare_wm_surfaces(struct brw_context *brw )
 {
    GLcontext *ctx = &brw->intel.ctx;
-   struct intel_context *intel = &brw->intel;
    GLuint i;
    int old_nr_surfaces;
 
@@ -719,17 +721,8 @@ static void prepare_wm_surfaces(struct brw_context *brw )
 
       /* _NEW_TEXTURE, BRW_NEW_TEXDATA */
       if (texUnit->_ReallyEnabled) {
-         if (texUnit->_Current == intel->frame_buffer_texobj) {
-            /* render to texture */
-            dri_bo_unreference(brw->wm.surf_bo[surf]);
-            brw->wm.surf_bo[surf] = brw->wm.surf_bo[0];
-            dri_bo_reference(brw->wm.surf_bo[surf]);
-            brw->wm.nr_surfaces = surf + 1;
-         } else {
-            /* regular texture */
-            brw_update_texture_surface(ctx, i);
-            brw->wm.nr_surfaces = surf + 1;
-         }
+	 brw_update_texture_surface(ctx, i);
+	 brw->wm.nr_surfaces = surf + 1;
       } else {
          dri_bo_unreference(brw->wm.surf_bo[surf]);
          brw->wm.surf_bo[surf] = NULL;
diff --git a/src/mesa/drivers/dri/i965/intel_generatemipmap.c b/src/mesa/drivers/dri/i965/intel_generatemipmap.c
deleted file mode 120000
index 4c6b37ada0..0000000000
--- a/src/mesa/drivers/dri/i965/intel_generatemipmap.c
+++ /dev/null
@@ -1 +0,0 @@
-../intel/intel_generatemipmap.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
index 51579df09e..d4899aab7f 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
@@ -157,7 +157,7 @@ intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
 #define OUT_BATCH(d) intel_batchbuffer_emit_dword(intel->batch, d)
 
 #define OUT_RELOC(buf, read_domains, write_domain, delta) do {		\
-   assert((delta) >= 0);						\
+   assert((unsigned) (delta) < buf->size);				\
    intel_batchbuffer_emit_reloc(intel->batch, buf,			\
 				read_domains, write_domain, delta);	\
 } while (0)
diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c
index 0c5be4c798..9e114db6c7 100644
--- a/src/mesa/drivers/dri/intel/intel_blit.c
+++ b/src/mesa/drivers/dri/intel/intel_blit.c
@@ -26,9 +26,6 @@
  **************************************************************************/
 
 
-#include <stdio.h>
-#include <errno.h>
-
 #include "main/mtypes.h"
 #include "main/context.h"
 #include "main/enums.h"
@@ -374,8 +371,6 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
       skipBuffers = BUFFER_BIT_STENCIL;
    }
 
-   /* XXX Move this flush/lock into the following conditional? */
-   intelFlush(&intel->ctx);
    LOCK_HARDWARE(intel);
 
    intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
index a0225936c8..ea9d5a6276 100644
--- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
@@ -209,7 +209,8 @@ intel_bufferobj_subdata(GLcontext * ctx,
       memcpy((char *)intel_obj->sys_buffer + offset, data, size);
    else {
       /* Flush any existing batchbuffer that might reference this data. */
-      intelFlush(ctx);
+      if (drm_intel_bo_references(intel->batch->buf, intel_obj->buffer))
+	 intelFlush(ctx);
 
       dri_bo_subdata(intel_obj->buffer, offset, size, data);
    }
@@ -257,10 +258,9 @@ intel_bufferobj_map(GLcontext * ctx,
       return obj->Pointer;
    }
 
-   /* Flush any existing batchbuffer that might have written to this
-    * buffer.
-    */
-   intelFlush(ctx);
+   /* Flush any existing batchbuffer that might reference this data. */
+   if (drm_intel_bo_references(intel->batch->buf, intel_obj->buffer))
+      intelFlush(ctx);
 
    if (intel_obj->region)
       intel_bufferobj_cow(intel, intel_obj);
@@ -330,7 +330,8 @@ intel_bufferobj_map_range(GLcontext * ctx,
     * the batchbuffer so that GEM knows about the buffer access for later
     * syncing.
     */
-   if (!(access & GL_MAP_UNSYNCHRONIZED_BIT))
+   if (!(access & GL_MAP_UNSYNCHRONIZED_BIT) &&
+       drm_intel_bo_references(intel->batch->buf, intel_obj->buffer))
       intelFlush(ctx);
 
    if (intel_obj->buffer == NULL) {
diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c
index 1b0e221789..fb62f0f430 100644
--- a/src/mesa/drivers/dri/intel/intel_clear.c
+++ b/src/mesa/drivers/dri/intel/intel_clear.c
@@ -172,7 +172,7 @@ intelClear(GLcontext *ctx, GLbitfield mask)
 	 DBG("\n");
       }
 
-      _mesa_meta_clear(&intel->ctx, tri_mask);
+      _mesa_meta_Clear(&intel->ctx, tri_mask);
    }
 
    if (swrast_mask) {
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index ddb0550f77..15f4bc6f69 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -588,11 +588,6 @@ intelInitDriverFunctions(struct dd_function_table *functions)
    functions->GetString = intelGetString;
    functions->UpdateState = intelInvalidateState;
 
-   functions->CopyColorTable = _swrast_CopyColorTable;
-   functions->CopyColorSubTable = _swrast_CopyColorSubTable;
-   functions->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
-   functions->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
-
    intelInitTextureFuncs(functions);
    intelInitTextureImageFuncs(functions);
    intelInitTextureSubImageFuncs(functions);
@@ -645,10 +640,6 @@ intelInitContext(struct intel_context *intel,
       intel->maxBatchSize = BATCH_SZ;
 
    intel->bufmgr = intelScreen->bufmgr;
-
-   if (0) /* for debug */
-      drm_intel_bufmgr_set_debug(intel->bufmgr, 1);
-
    intel->ttm = intelScreen->ttm;
    if (intel->ttm) {
       int bo_reuse_mode;
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index d3acf6e4b3..2778cc0136 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -260,9 +260,6 @@ struct intel_context
    intel_line_func draw_line;
    intel_tri_func draw_tri;
 
-   /* These refer to the current drawing buffer:
-    */
-   struct gl_texture_object *frame_buffer_texobj;
    /**
     * Set to true if a single constant cliprect should be used in the
     * batchbuffer.  Otherwise, cliprects must be calculated at batchbuffer
@@ -302,7 +299,6 @@ struct intel_context
 
    GLboolean use_texture_tiling;
    GLboolean use_early_z;
-
    drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
 
    int perf_boxes;
diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c
index 2e61c556d8..2eb08a8f05 100644
--- a/src/mesa/drivers/dri/intel/intel_extensions.c
+++ b/src/mesa/drivers/dri/intel/intel_extensions.c
@@ -31,6 +31,7 @@
 
 
 #define need_GL_ARB_copy_buffer
+#define need_GL_ARB_draw_elements_base_vertex
 #define need_GL_ARB_framebuffer_object
 #define need_GL_ARB_map_buffer_range
 #define need_GL_ARB_occlusion_query
@@ -73,11 +74,15 @@
  */
 static const struct dri_extension card_extensions[] = {
    { "GL_ARB_copy_buffer",                GL_ARB_copy_buffer_functions },
+   { "GL_ARB_draw_elements_base_vertex",  GL_ARB_draw_elements_base_vertex_functions },
    { "GL_ARB_half_float_pixel",           NULL },
    { "GL_ARB_map_buffer_range",           GL_ARB_map_buffer_range_functions },
    { "GL_ARB_multitexture",               NULL },
    { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
    { "GL_ARB_point_sprite",               NULL },
+   { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
+   { "GL_ARB_shading_language_100",       GL_VERSION_2_0_functions },
+   { "GL_ARB_shading_language_120",       GL_VERSION_2_1_functions },
    { "GL_ARB_sync",                       GL_ARB_sync_functions },
    { "GL_ARB_texture_border_clamp",       NULL },
    { "GL_ARB_texture_cube_map",           NULL },
@@ -89,6 +94,7 @@ static const struct dri_extension card_extensions[] = {
    { "GL_ARB_texture_rectangle",          NULL },
    { "GL_ARB_vertex_array_object",        GL_ARB_vertex_array_object_functions},
    { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
+   { "GL_ARB_vertex_shader",              GL_ARB_vertex_shader_functions },
    { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
    { "GL_EXT_blend_color",                GL_EXT_blend_color_functions },
    { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
@@ -139,6 +145,7 @@ static const struct dri_extension i915_extensions[] = {
 
 /** i965-only extensions */
 static const struct dri_extension brw_extensions[] = {
+   { "GL_ARB_depth_clamp",                NULL },
    { "GL_ARB_depth_texture",              NULL },
    { "GL_ARB_fragment_program",           NULL },
    { "GL_ARB_fragment_program_shadow",    NULL },
@@ -147,13 +154,9 @@ static const struct dri_extension brw_extensions[] = {
    { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
    { "GL_ARB_point_sprite", 		  NULL },
    { "GL_ARB_seamless_cube_map",          NULL },
-   { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
-   { "GL_ARB_shading_language_100",       GL_VERSION_2_0_functions },
-   { "GL_ARB_shading_language_120",       GL_VERSION_2_1_functions },
    { "GL_ARB_shadow",                     NULL },
    { "GL_MESA_texture_signed_rgba",       NULL },
    { "GL_ARB_texture_non_power_of_two",   NULL },
-   { "GL_ARB_vertex_shader",              GL_ARB_vertex_shader_functions },
    { "GL_EXT_shadow_funcs",               NULL },
    { "GL_EXT_stencil_two_side",           GL_EXT_stencil_two_side_functions },
    { "GL_EXT_texture_sRGB",		  NULL },
@@ -168,6 +171,7 @@ static const struct dri_extension brw_extensions[] = {
 
 
 static const struct dri_extension arb_oq_extensions[] = {
+   { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
    { NULL, NULL }
 };
 
@@ -179,6 +183,10 @@ static const struct dri_extension ttm_extensions[] = {
    { NULL, NULL }
 };
 
+static const struct dri_extension fragment_shader_extensions[] = {
+   { "GL_ARB_fragment_shader",            NULL },
+   { NULL, NULL }
+};
 
 /**
  * Initializes potential list of extensions if ctx == NULL, or actually enables
@@ -202,6 +210,14 @@ intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging)
       driInitExtensions(ctx, brw_extensions, GL_FALSE);
 
    if (intel == NULL || IS_915(intel->intelScreen->deviceID)
-       || IS_945(intel->intelScreen->deviceID))
+       || IS_945(intel->intelScreen->deviceID)) {
       driInitExtensions(ctx, i915_extensions, GL_FALSE);
+
+      if (intel == NULL || driQueryOptionb(&intel->optionCache, "fragment_shader"))
+	 driInitExtensions(ctx, fragment_shader_extensions, GL_FALSE);
+
+      if (intel == NULL || driQueryOptionb(&intel->optionCache,
+					   "stub_occlusion_query"))
+	 driInitExtensions(ctx, arb_oq_extensions, GL_FALSE);
+   }
 }
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c
index 804c034840..2e61371805 100644
--- a/src/mesa/drivers/dri/intel/intel_fbo.c
+++ b/src/mesa/drivers/dri/intel/intel_fbo.c
@@ -618,15 +618,8 @@ intel_render_texture(GLcontext * ctx,
    }
 
    /* compute offset of the particular 2D image within the texture region */
-   imageOffset = intel_miptree_image_offset(intel_image->mt,
-                                            att->CubeMapFace,
-                                            att->TextureLevel);
-
-   if (att->Texture->Target == GL_TEXTURE_3D) {
-      const GLuint *offsets = intel_miptree_depth_offsets(intel_image->mt,
-                                                          att->TextureLevel);
-      imageOffset += offsets[att->Zoffset];
-   }
+   imageOffset = intel_miptree_image_offset(intel_image->mt, att->CubeMapFace,
+                                            att->TextureLevel, att->Zoffset);
 
    /* store that offset in the region */
    intel_image->mt->region->draw_offset = imageOffset;
@@ -715,5 +708,5 @@ intel_fbo_init(struct intel_context *intel)
    intel->ctx.Driver.FinishRenderTexture = intel_finish_render_texture;
    intel->ctx.Driver.ResizeBuffers = intel_resize_buffers;
    intel->ctx.Driver.ValidateFramebuffer = intel_validate_framebuffer;
-   intel->ctx.Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer;
+   intel->ctx.Driver.BlitFramebuffer = _mesa_meta_BlitFramebuffer;
 }
diff --git a/src/mesa/drivers/dri/intel/intel_generatemipmap.c b/src/mesa/drivers/dri/intel/intel_generatemipmap.c
deleted file mode 100644
index 12059e122c..0000000000
--- a/src/mesa/drivers/dri/intel/intel_generatemipmap.c
+++ /dev/null
@@ -1,304 +0,0 @@
-/*
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- * Copyright © 2009 Intel Corporation
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
- * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
- * IN THE SOFTWARE.
- *
- * Authors:
- *    Eric Anholt <eric@anholt.net>
- *
- */
-
-#include "main/glheader.h"
-#include "main/enums.h"
-#include "main/image.h"
-#include "main/mtypes.h"
-#include "main/macros.h"
-#include "main/bufferobj.h"
-#include "main/teximage.h"
-#include "main/texenv.h"
-#include "main/texobj.h"
-#include "main/texstate.h"
-#include "main/texparam.h"
-#include "main/varray.h"
-#include "main/attrib.h"
-#include "main/enable.h"
-#include "main/buffers.h"
-#include "main/fbobject.h"
-#include "main/framebuffer.h"
-#include "main/renderbuffer.h"
-#include "main/depth.h"
-#include "main/hash.h"
-#include "main/mipmap.h"
-#include "main/blend.h"
-#include "glapi/dispatch.h"
-#include "swrast/swrast.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_batchbuffer.h"
-#include "intel_pixel.h"
-#include "intel_tex.h"
-#include "intel_mipmap_tree.h"
-
-static const char *intel_fp_tex2d =
-      "!!ARBfp1.0\n"
-      "TEX result.color, fragment.texcoord[0], texture[0], 2D;\n"
-      "END\n";
-
-static GLboolean
-intel_generate_mipmap_level(GLcontext *ctx, GLuint tex_name,
-			    int level, int width, int height)
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLfloat vertices[4][2];
-   GLint status;
-
-   /* Set to source from the previous level */
-   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, level - 1);
-   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, level - 1);
-
-   /* Set to draw into the current level */
-   _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT,
-				 GL_COLOR_ATTACHMENT0_EXT,
-				 GL_TEXTURE_2D,
-				 tex_name,
-				 level);
-   /* Choose to render to the color attachment. */
-   _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT);
-
-   status = _mesa_CheckFramebufferStatusEXT (GL_FRAMEBUFFER_EXT);
-   if (status != GL_FRAMEBUFFER_COMPLETE_EXT)
-      return GL_FALSE;
-
-   meta_set_passthrough_transform(&intel->meta);
-
-   /* XXX: Doing it right would involve setting up the transformation to do
-    * 0-1 mapping or something, and not changing the vertex data.
-    */
-   vertices[0][0] = 0;
-   vertices[0][1] = 0;
-   vertices[1][0] = width;
-   vertices[1][1] = 0;
-   vertices[2][0] = width;
-   vertices[2][1] = height;
-   vertices[3][0] = 0;
-   vertices[3][1] = height;
-
-   _mesa_VertexPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &vertices);
-   _mesa_Enable(GL_VERTEX_ARRAY);
-   meta_set_default_texrect(&intel->meta);
-
-   _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
-
-   meta_restore_texcoords(&intel->meta);
-   meta_restore_transform(&intel->meta);
-
-   return GL_TRUE;
-}
-
-static GLboolean
-intel_generate_mipmap_2d(GLcontext *ctx,
-			 GLenum target,
-			 struct gl_texture_object *texObj)
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLint old_active_texture;
-   int level, max_levels, start_level, end_level;
-   GLuint fb_name;
-   GLboolean success = GL_FALSE;
-   struct gl_framebuffer *saved_fbo = NULL;
-   struct gl_buffer_object *saved_array_buffer = NULL;
-   struct gl_buffer_object *saved_element_buffer = NULL;
-
-   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT |
-		    GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT |
-		    GL_DEPTH_BUFFER_BIT);
-   _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
-   old_active_texture = ctx->Texture.CurrentUnit;
-   _mesa_reference_framebuffer(&saved_fbo, ctx->DrawBuffer);
-
-   /* use default array/index buffers */
-   _mesa_reference_buffer_object(ctx, &saved_array_buffer,
-                                 ctx->Array.ArrayBufferObj);
-   _mesa_reference_buffer_object(ctx, &ctx->Array.ArrayBufferObj,
-                                 ctx->Shared->NullBufferObj);   
-   _mesa_reference_buffer_object(ctx, &saved_element_buffer,
-                                 ctx->Array.ElementArrayBufferObj);
-   _mesa_reference_buffer_object(ctx, &ctx->Array.ElementArrayBufferObj,
-                                 ctx->Shared->NullBufferObj);   
-
-   _mesa_Disable(GL_POLYGON_STIPPLE);
-   _mesa_Disable(GL_DEPTH_TEST);
-   _mesa_Disable(GL_STENCIL_TEST);
-   _mesa_ColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
-   _mesa_DepthMask(GL_FALSE);
-
-   /* Bind the given texture to GL_TEXTURE_2D with linear filtering for our
-    * minification.
-    */
-   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
-   _mesa_Enable(GL_TEXTURE_2D);
-   _mesa_BindTexture(GL_TEXTURE_2D, texObj->Name);
-   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER,
-		       GL_LINEAR_MIPMAP_NEAREST);
-   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
-
-   /* Bind the new renderbuffer to the color attachment point. */
-   _mesa_GenFramebuffersEXT(1, &fb_name);
-   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb_name);
-
-   meta_set_fragment_program(&intel->meta, &intel->meta.tex2d_fp,
-			     intel_fp_tex2d);
-   meta_set_passthrough_vertex_program(&intel->meta);
-
-   max_levels = _mesa_max_texture_levels(ctx, texObj->Target);
-   start_level = texObj->BaseLevel;
-   end_level = texObj->MaxLevel;
-
-   /* Loop generating level+1 from level. */
-   for (level = start_level; level < end_level && level < max_levels - 1; level++) {
-      const struct gl_texture_image *srcImage;
-      int width, height;
-
-      srcImage = _mesa_select_tex_image(ctx, texObj, target, level);
-      if (srcImage->Border != 0)
-	 goto fail;
-
-      width = srcImage->Width / 2;
-      if (width < 1)
-	 width = 1;
-      height = srcImage->Height / 2;
-      if (height < 1)
-	 height = 1;
-
-      if (width == srcImage->Width &&
-	  height == srcImage->Height) {
-	 /* Neither _mesa_max_texture_levels nor texObj->MaxLevel are the
-	  * maximum texture level for the object, so break out when we've gone
-	  * over the edge.
-	  */
-	 break;
-      }
-
-      /* Make sure that there's space allocated for the target level.
-       * We could skip this if there's already space allocated and save some
-       * time.
-       */
-      _mesa_TexImage2D(GL_TEXTURE_2D, level + 1, srcImage->InternalFormat,
-		       width, height, 0,
-		       GL_RGBA, GL_UNSIGNED_INT, NULL);
-
-      if (!intel_generate_mipmap_level(ctx, texObj->Name, level + 1,
-				       width, height))
-	 goto fail;
-   }
-
-   success = GL_TRUE;
-
-fail:
-   meta_restore_fragment_program(&intel->meta);
-   meta_restore_vertex_program(&intel->meta);
-
-   /* restore array/index buffers */
-   _mesa_reference_buffer_object(ctx, &ctx->Array.ArrayBufferObj,
-                                 saved_array_buffer);
-   _mesa_reference_buffer_object(ctx, &saved_array_buffer, NULL);
-   _mesa_reference_buffer_object(ctx, &ctx->Array.ElementArrayBufferObj,
-                                 saved_element_buffer);
-   _mesa_reference_buffer_object(ctx, &saved_element_buffer, NULL);
-
-
-   _mesa_DeleteFramebuffersEXT(1, &fb_name);
-   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture);
-   if (saved_fbo)
-      _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, saved_fbo->Name);
-   _mesa_reference_framebuffer(&saved_fbo, NULL);
-   _mesa_PopClientAttrib();
-   _mesa_PopAttrib();
-
-   return success;
-}
-
-
-/**
- * Generate new mipmap data from BASE+1 to BASE+p (the minimally-sized mipmap
- * level).
- *
- * The texture object's miptree must be mapped.
- *
- * It would be really nice if this was just called by Mesa whenever mipmaps
- * needed to be regenerated, rather than us having to remember to do so in
- * each texture image modification path.
- *
- * This function should also include an accelerated path.
- */
-void
-intel_generate_mipmap(GLcontext *ctx, GLenum target,
-                      struct gl_texture_object *texObj)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-   GLuint nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-   int face, i;
-
-   /* HW path */
-   if (target == GL_TEXTURE_2D &&
-       ctx->Extensions.EXT_framebuffer_object &&
-       ctx->Extensions.ARB_fragment_program &&
-       ctx->Extensions.ARB_vertex_program) {
-      GLboolean success;
-
-      /* We'll be accessing this texture using GL entrypoints, which should
-       * be resilient against other access to this texture.
-       */
-      _mesa_unlock_texture(ctx, texObj);
-      success = intel_generate_mipmap_2d(ctx, target, texObj);
-      _mesa_lock_texture(ctx, texObj);
-
-      if (success)
-	 return;
-   }
-
-   /* SW path */
-   intel_tex_map_level_images(intel, intelObj, texObj->BaseLevel);
-   _mesa_generate_mipmap(ctx, target, texObj);
-   intel_tex_unmap_level_images(intel, intelObj, texObj->BaseLevel);
-
-   /* Update the level information in our private data in the new images, since
-    * it didn't get set as part of a normal TexImage path.
-    */
-   for (face = 0; face < nr_faces; face++) {
-      for (i = texObj->BaseLevel + 1; i < texObj->MaxLevel; i++) {
-         struct intel_texture_image *intelImage;
-
-	 intelImage = intel_texture_image(texObj->Image[face][i]);
-	 if (intelImage == NULL)
-	    break;
-
-	 intelImage->level = i;
-	 intelImage->face = face;
-	 /* Unreference the miptree to signal that the new Data is a bare
-	  * pointer from mesa.
-	  */
-	 intel_miptree_release(intel, &intelImage->mt);
-      }
-   }
-}
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
index 4f5101a312..01e9d4add3 100644
--- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
@@ -28,6 +28,7 @@
 #include "intel_context.h"
 #include "intel_mipmap_tree.h"
 #include "intel_regions.h"
+#include "intel_tex_layout.h"
 #include "intel_chipset.h"
 #ifndef I915
 #include "brw_state.h"
@@ -36,6 +37,7 @@
 
 #define FILE_DEBUG_FLAG DEBUG_MIPTREE
 
+
 static GLenum
 target_to_target(GLenum target)
 {
@@ -52,6 +54,7 @@ target_to_target(GLenum target)
    }
 }
 
+
 static struct intel_mipmap_tree *
 intel_miptree_create_internal(struct intel_context *intel,
 			      GLenum target,
@@ -101,6 +104,7 @@ intel_miptree_create_internal(struct intel_context *intel,
    return mt;
 }
 
+
 struct intel_mipmap_tree *
 intel_miptree_create(struct intel_context *intel,
 		     GLenum target,
@@ -153,6 +157,7 @@ intel_miptree_create(struct intel_context *intel,
    return mt;
 }
 
+
 struct intel_mipmap_tree *
 intel_miptree_create_for_region(struct intel_context *intel,
 				GLenum target,
@@ -190,7 +195,8 @@ intel_miptree_create_for_region(struct intel_context *intel,
    intel_region_reference(&mt->region, region);
 
    return mt;
- }
+}
+
 
 /**
  * intel_miptree_pitch_align:
@@ -204,7 +210,6 @@ intel_miptree_create_for_region(struct intel_context *intel,
  * Given @pitch, compute a larger value which accounts for
  * any necessary alignment required by the device
  */
-
 int intel_miptree_pitch_align (struct intel_context *intel,
 			       struct intel_mipmap_tree *mt,
 			       uint32_t tiling,
@@ -250,6 +255,7 @@ int intel_miptree_pitch_align (struct intel_context *intel,
    return pitch;
 }
 
+
 void
 intel_miptree_reference(struct intel_mipmap_tree **dst,
                         struct intel_mipmap_tree *src)
@@ -259,6 +265,7 @@ intel_miptree_reference(struct intel_mipmap_tree **dst,
    DBG("%s %p refcount now %d\n", __FUNCTION__, src, src->refcount);
 }
 
+
 void
 intel_miptree_release(struct intel_context *intel,
                       struct intel_mipmap_tree **mt)
@@ -297,9 +304,8 @@ intel_miptree_release(struct intel_context *intel,
 }
 
 
-
-
-/* Can the image be pulled into a unified mipmap tree.  This mirrors
+/**
+ * Can the image be pulled into a unified mipmap tree?  This mirrors
  * the completeness test in a lot of ways.
  *
  * Not sure whether I want to pass gl_texture_image here.
@@ -386,6 +392,7 @@ intel_miptree_set_image_offset_ex(struct intel_mipmap_tree *mt,
        __FUNCTION__, level, img, x, y, mt->level[level].image_offset[img]);
 }
 
+
 void
 intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
 			       GLuint level, GLuint img,
@@ -395,35 +402,27 @@ intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
 }
 
 
-/* Although we use the image_offset[] array to store relative offsets
- * to cube faces, Mesa doesn't know anything about this and expects
- * each cube face to be treated as a separate image.
- *
- * These functions present that view to mesa:
+/**
+ * Return offset to the start of a 2D slice of a texture (a mipmap level,
+ * cube face, 3D Z slice).
+ * \param mt  the texture object/miptree
+ * \param face  cube map face in [0,5] or zero for non-cube textures
+ * \param level  mipmap level
+ * \param zslice  Z slice of a 3D texture, or zero for non-3D textures
  */
-const GLuint *
-intel_miptree_depth_offsets(struct intel_mipmap_tree *mt, GLuint level)
-{
-   static const GLuint zero = 0;
-
-   if (mt->target != GL_TEXTURE_3D || mt->level[level].nr_images == 1)
-      return &zero;
-   else
-      return mt->level[level].image_offset;
-}
-
-
 GLuint
-intel_miptree_image_offset(struct intel_mipmap_tree *mt,
-			   GLuint face, GLuint level)
+intel_miptree_image_offset(const struct intel_mipmap_tree *mt,
+                           GLuint face, GLuint level, GLuint zslice)
 {
+   GLuint offset = mt->level[level].level_offset;
+
    if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
-      return (mt->level[level].level_offset +
-	      mt->level[level].image_offset[face]);
-   else
-      return mt->level[level].level_offset;
-}
+      offset += mt->level[level].image_offset[face];
+   else if (mt->target == GL_TEXTURE_3D)
+      offset += mt->level[level].image_offset[zslice];
 
+   return offset;
+}
 
 
 /**
@@ -459,9 +458,10 @@ intel_miptree_image_map(struct intel_context * intel,
    }
 
    return (intel_region_map(intel, mt->region) +
-           intel_miptree_image_offset(mt, face, level));
+           intel_miptree_image_offset(mt, face, level, 0));
 }
 
+
 void
 intel_miptree_image_unmap(struct intel_context *intel,
                           struct intel_mipmap_tree *mt)
@@ -471,8 +471,8 @@ intel_miptree_image_unmap(struct intel_context *intel,
 }
 
 
-
-/* Upload data for a particular image.
+/**
+ * Upload data for a particular image.
  */
 void
 intel_miptree_image_data(struct intel_context *intel,
@@ -483,20 +483,20 @@ intel_miptree_image_data(struct intel_context *intel,
 			 GLuint src_row_pitch,
 			 GLuint src_image_pitch)
 {
-   GLuint depth = dst->level[level].depth;
-   GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
-   const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
+   const GLuint depth = dst->level[level].depth;
    GLuint i;
-   GLuint height = 0;
 
    DBG("%s: %d/%d\n", __FUNCTION__, face, level);
    for (i = 0; i < depth; i++) {
-      height = dst->level[level].height;
-      if(dst->compressed)
+      GLuint dst_offset = intel_miptree_image_offset(dst, face, level, i);
+      GLuint height = dst->level[level].height;
+
+      if (dst->compressed)
 	 height = (height + 3) / 4;
+
       intel_region_data(intel,
 			dst->region,
-			dst_offset + dst_depth_offset[i], /* dst_offset */
+			dst_offset,
 			0, 0,                             /* dstx, dsty */
 			src,
 			src_row_pitch,
@@ -507,8 +507,9 @@ intel_miptree_image_data(struct intel_context *intel,
    }
 }
 
-extern void intel_get_texture_alignment_unit(GLenum, GLuint *, GLuint *);
-/* Copy mipmap image between trees
+
+/**
+ * Copy mipmap image between trees
  */
 void
 intel_miptree_image_copy(struct intel_context *intel,
@@ -519,26 +520,26 @@ intel_miptree_image_copy(struct intel_context *intel,
    GLuint width = src->level[level].width;
    GLuint height = src->level[level].height;
    GLuint depth = src->level[level].depth;
-   GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
-   GLuint src_offset = intel_miptree_image_offset(src, face, level);
-   const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
-   const GLuint *src_depth_offset = intel_miptree_depth_offsets(src, level);
    GLuint i;
    GLboolean success;
 
    if (dst->compressed) {
        GLuint align_w, align_h;
 
-       intel_get_texture_alignment_unit(dst->internal_format, &align_w, &align_h);
+       intel_get_texture_alignment_unit(dst->internal_format,
+                                        &align_w, &align_h);
        height = (height + 3) / 4;
        width = ALIGN(width, align_w);
    }
 
    for (i = 0; i < depth; i++) {
+      GLuint dst_offset = intel_miptree_image_offset(dst, face, level, i);
+      GLuint src_offset = intel_miptree_image_offset(src, face, level, i);
+
       success = intel_region_copy(intel,
-				  dst->region, dst_offset + dst_depth_offset[i],
+				  dst->region, dst_offset,
 				  0, 0,
-				  src->region, src_offset + src_depth_offset[i],
+				  src->region, src_offset,
 				  0, 0, width, height, GL_COPY);
       if (!success) {
 	 GLubyte *src_ptr, *dst_ptr;
@@ -546,11 +547,11 @@ intel_miptree_image_copy(struct intel_context *intel,
 	 src_ptr = intel_region_map(intel, src->region);
 	 dst_ptr = intel_region_map(intel, dst->region);
 
-	 _mesa_copy_rect(dst_ptr + dst_offset + dst_depth_offset[i],
+	 _mesa_copy_rect(dst_ptr + dst_offset,
 			 dst->cpp,
 			 dst->pitch,
 			 0, 0, width, height,
-			 src_ptr + src_offset + src_depth_offset[i],
+			 src_ptr + src_offset,
 			 src->pitch,
 			 0, 0);
 	 intel_region_unmap(intel, src->region);
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.h b/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
index c890b2a0d0..c08f9cd8b6 100644
--- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
@@ -177,17 +177,9 @@ void intel_miptree_image_unmap(struct intel_context *intel,
                                struct intel_mipmap_tree *mt);
 
 
-/* Return the linear offset of an image relative to the start of the
- * tree:
- */
-GLuint intel_miptree_image_offset(struct intel_mipmap_tree *mt,
-                                  GLuint face, GLuint level);
-
-/* Return pointers to each 2d slice within an image.  Indexed by depth
- * value.
- */
-const GLuint *intel_miptree_depth_offsets(struct intel_mipmap_tree *mt,
-                                          GLuint level);
+GLuint
+intel_miptree_image_offset(const struct intel_mipmap_tree *mt,
+                           GLuint face, GLuint level, GLuint zslice);
 
 
 void intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c
index a300141655..993e427a99 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel.c
@@ -129,20 +129,6 @@ intel_check_blit_fragment_ops(GLcontext * ctx, GLboolean src_alpha_is_one)
    return GL_TRUE;
 }
 
-
-GLboolean
-intel_check_meta_tex_fragment_ops(GLcontext * ctx)
-{
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
-   /* Some of _ImageTransferState (scale, bias) could be done with
-    * fragment programs on i915.
-    */
-   return !(ctx->_ImageTransferState || ctx->Fog.Enabled ||     /* not done yet */
-            ctx->Texture._EnabledUnits || ctx->FragmentProgram._Enabled);
-}
-
 /* The intel_region struct doesn't really do enough to capture the
  * format of the pixels in the region.  For now this code assumes that
  * the region is a display surface and hence is either ARGB8888 or
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.h b/src/mesa/drivers/dri/intel/intel_pixel.h
index 96a6dd17b2..743b6497c5 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.h
+++ b/src/mesa/drivers/dri/intel/intel_pixel.h
@@ -34,8 +34,6 @@ void intelInitPixelFuncs(struct dd_function_table *functions);
 GLboolean intel_check_blit_fragment_ops(GLcontext * ctx,
 					GLboolean src_alpha_is_one);
 
-GLboolean intel_check_meta_tex_fragment_ops(GLcontext * ctx);
-
 GLboolean intel_check_blit_format(struct intel_region *region,
                                   GLenum format, GLenum type);
 
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index b543a0bbc3..9a0bcc07a5 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -209,7 +209,7 @@ do_blit_bitmap( GLcontext *ctx,
    if (!dst)
        return GL_FALSE;
 
-   if (unpack->BufferObj->Name) {
+   if (_mesa_is_bufferobj(unpack->BufferObj)) {
       bitmap = map_pbo(ctx, width, height, unpack, bitmap);
       if (bitmap == NULL)
 	 return GL_TRUE;	/* even though this is an error, we're done */
@@ -329,7 +329,7 @@ out:
    if (INTEL_DEBUG & DEBUG_SYNC)
       intel_batchbuffer_flush(intel->batch);
 
-   if (unpack->BufferObj->Name) {
+   if (_mesa_is_bufferobj(unpack->BufferObj)) {
       /* done with PBO so unmap it now */
       ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
                               unpack->BufferObj);
@@ -418,7 +418,7 @@ intel_texture_bitmap(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   if (unpack->BufferObj->Name) {
+   if (_mesa_is_bufferobj(unpack->BufferObj)) {
       bitmap = map_pbo(ctx, width, height, unpack, bitmap);
       if (bitmap == NULL)
 	 return GL_TRUE;	/* even though this is an error, we're done */
@@ -428,7 +428,7 @@ intel_texture_bitmap(GLcontext * ctx,
    a8_bitmap = _mesa_calloc(width * height);
    _mesa_expand_bitmap(width, height, unpack, bitmap, a8_bitmap, width, 0xff);
 
-   if (unpack->BufferObj->Name) {
+   if (_mesa_is_bufferobj(unpack->BufferObj)) {
       /* done with PBO so unmap it now */
       ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
                               unpack->BufferObj);
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
index 07ca8f7ddb..f058b3c8e4 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
@@ -240,5 +240,5 @@ intelCopyPixels(GLcontext * ctx,
       return;
 
    /* this will use swrast if needed */
-   _mesa_meta_copy_pixels(ctx, srcx, srcy, width, height, destx, desty, type);
+   _mesa_meta_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type);
 }
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
index 7fbb89fd6a..5ffa847fd4 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
@@ -54,7 +54,7 @@
 #include "intel_fbo.h"
 
 
-/** XXX compare perf of this vs. _mesa_meta_draw_pixels(STENCIL) */
+/** XXX compare perf of this vs. _mesa_meta_DrawPixels(STENCIL) */
 static GLboolean
 intel_stencil_drawpixels(GLcontext * ctx,
 			 GLint x, GLint y,
@@ -265,7 +265,7 @@ intelDrawPixels(GLcontext * ctx,
    /* XXX this function doesn't seem to work reliably even when all
     * the pre-requisite conditions are met.
     * Note that this function is never hit with conform.
-    * Fall back to swrast because even the _mesa_meta_draw_pixels() approach
+    * Fall back to swrast because even the _mesa_meta_DrawPixels() approach
     * isn't working because of an apparent stencil bug.
     */
    if (intel_stencil_drawpixels(ctx, x, y, width, height, format, type,
@@ -280,6 +280,6 @@ intelDrawPixels(GLcontext * ctx,
    }
 #endif
 
-   _mesa_meta_draw_pixels(ctx, x, y, width, height, format, type,
-                          unpack, pixels);
+   _mesa_meta_DrawPixels(ctx, x, y, width, height, format, type,
+                         unpack, pixels);
 }
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c
index e036736323..4707500180 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_read.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c
@@ -180,16 +180,7 @@ do_blit_readpixels(GLcontext * ctx,
    if (!src)
       return GL_FALSE;
 
-   if (pack->BufferObj->Name) {
-      /* XXX This validation should be done by core mesa:
-       */
-      if (!_mesa_validate_pbo_access(2, pack, width, height, 1,
-                                     format, type, pixels)) {
-         _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels");
-         return GL_TRUE;
-      }
-   }
-   else {
+   if (!_mesa_is_bufferobj(pack->BufferObj)) {
       /* PBO only for now:
        */
       if (INTEL_DEBUG & DEBUG_PIXEL)
@@ -225,9 +216,8 @@ do_blit_readpixels(GLcontext * ctx,
       rowLength = -rowLength;
    }
 
-   /* XXX 64-bit cast? */
-   dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height,
-                                             format, type, 0, 0, 0);
+   dst_offset = (GLintptr) _mesa_image_address(2, pack, pixels, width, height,
+					       format, type, 0, 0, 0);
 
 
    /* Although the blits go on the command buffer, need to do this and
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index 1b8c56e68d..24f7fbc992 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -79,6 +79,10 @@ PUBLIC const char __driConfigOptions[] =
 	 DRI_CONF_DESC(en, "Enable early Z in classic mode (unstable, 945-only).")
       DRI_CONF_OPT_END
 
+      DRI_CONF_OPT_BEGIN(fragment_shader, bool, false)
+	 DRI_CONF_DESC(en, "Enable limited ARB_fragment_shader support on 915/945.")
+      DRI_CONF_OPT_END
+
    DRI_CONF_SECTION_END
    DRI_CONF_SECTION_QUALITY
       DRI_CONF_FORCE_S3TC_ENABLE(false)
@@ -88,10 +92,14 @@ PUBLIC const char __driConfigOptions[] =
      DRI_CONF_NO_RAST(false)
      DRI_CONF_ALWAYS_FLUSH_BATCH(false)
      DRI_CONF_ALWAYS_FLUSH_CACHE(false)
+
+      DRI_CONF_OPT_BEGIN(stub_occlusion_query, bool, false)
+	 DRI_CONF_DESC(en, "Enable stub ARB_occlusion_query support on 915/945.")
+      DRI_CONF_OPT_END
    DRI_CONF_SECTION_END
 DRI_CONF_END;
 
-const GLuint __driNConfigOptions = 10;
+const GLuint __driNConfigOptions = 12;
 
 #ifdef USE_NEW_INTERFACE
 static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c
index 8df4990880..dcfcad1d95 100644
--- a/src/mesa/drivers/dri/intel/intel_span.c
+++ b/src/mesa/drivers/dri/intel/intel_span.c
@@ -444,23 +444,30 @@ intel_renderbuffer_unmap(struct intel_context *intel,
  * _ColorReadBuffer, _DepthBuffer or _StencilBuffer fields.
  */
 static void
-intel_map_unmap_buffers(struct intel_context *intel, GLboolean map)
+intel_map_unmap_framebuffer(struct intel_context *intel,
+			    struct gl_framebuffer *fb,
+			    GLboolean map)
 {
-   GLcontext *ctx = &intel->ctx;
-   GLuint i, j;
+   GLuint i;
 
    /* color draw buffers */
-   for (j = 0; j < ctx->DrawBuffer->_NumColorDrawBuffers; j++) {
+   for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
       if (map)
-	 intel_renderbuffer_map(intel, ctx->DrawBuffer->_ColorDrawBuffers[j]);
+         intel_renderbuffer_map(intel, fb->_ColorDrawBuffers[i]);
       else
-	 intel_renderbuffer_unmap(intel, ctx->DrawBuffer->_ColorDrawBuffers[j]);
+         intel_renderbuffer_unmap(intel, fb->_ColorDrawBuffers[i]);
    }
 
+   /* color read buffer */
+   if (map)
+      intel_renderbuffer_map(intel, fb->_ColorReadBuffer);
+   else
+      intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer);
+
    /* check for render to textures */
    for (i = 0; i < BUFFER_COUNT; i++) {
       struct gl_renderbuffer_attachment *att =
-         ctx->DrawBuffer->Attachment + i;
+         fb->Attachment + i;
       struct gl_texture_object *tex = att->Texture;
       if (tex) {
          /* render to texture */
@@ -472,34 +479,23 @@ intel_map_unmap_buffers(struct intel_context *intel, GLboolean map)
       }
    }
 
-   /* color read buffers */
-   if (map)
-      intel_renderbuffer_map(intel, ctx->ReadBuffer->_ColorReadBuffer);
-   else
-      intel_renderbuffer_unmap(intel, ctx->ReadBuffer->_ColorReadBuffer);
-
    /* depth buffer (Note wrapper!) */
-   if (ctx->DrawBuffer->_DepthBuffer) {
+   if (fb->_DepthBuffer) {
       if (map)
-	 intel_renderbuffer_map(intel, ctx->DrawBuffer->_DepthBuffer->Wrapped);
+         intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped);
       else
-	 intel_renderbuffer_unmap(intel,
-				  ctx->DrawBuffer->_DepthBuffer->Wrapped);
+         intel_renderbuffer_unmap(intel, fb->_DepthBuffer->Wrapped);
    }
 
    /* stencil buffer (Note wrapper!) */
-   if (ctx->DrawBuffer->_StencilBuffer) {
+   if (fb->_StencilBuffer) {
       if (map)
-	 intel_renderbuffer_map(intel,
-				ctx->DrawBuffer->_StencilBuffer->Wrapped);
+         intel_renderbuffer_map(intel, fb->_StencilBuffer->Wrapped);
       else
-	 intel_renderbuffer_unmap(intel,
-				  ctx->DrawBuffer->_StencilBuffer->Wrapped);
+         intel_renderbuffer_unmap(intel, fb->_StencilBuffer->Wrapped);
    }
 }
 
-
-
 /**
  * Prepare for software rendering.  Map current read/draw framebuffers'
  * renderbuffes and all currently bound texture objects.
@@ -522,7 +518,9 @@ intelSpanRenderStart(GLcontext * ctx)
       }
    }
 
-   intel_map_unmap_buffers(intel, GL_TRUE);
+   intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_TRUE);
+   if (ctx->ReadBuffer != ctx->DrawBuffer)
+      intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE);
 }
 
 /**
@@ -544,7 +542,9 @@ intelSpanRenderFinish(GLcontext * ctx)
       }
    }
 
-   intel_map_unmap_buffers(intel, GL_FALSE);
+   intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_FALSE);
+   if (ctx->ReadBuffer != ctx->DrawBuffer)
+      intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE);
 
    UNLOCK_HARDWARE(intel);
 }
@@ -558,6 +558,43 @@ intelInitSpanFuncs(GLcontext * ctx)
    swdd->SpanRenderFinish = intelSpanRenderFinish;
 }
 
+void
+intel_map_vertex_shader_textures(GLcontext *ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   int i;
+
+   if (ctx->VertexProgram._Current == NULL)
+      return;
+
+   for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
+      if (ctx->Texture.Unit[i]._ReallyEnabled &&
+	  ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) {
+         struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
+
+         intel_tex_map_images(intel, intel_texture_object(texObj));
+      }
+   }
+}
+
+void
+intel_unmap_vertex_shader_textures(GLcontext *ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   int i;
+
+   if (ctx->VertexProgram._Current == NULL)
+      return;
+
+   for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
+      if (ctx->Texture.Unit[i]._ReallyEnabled &&
+	  ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) {
+         struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
+
+         intel_tex_unmap_images(intel, intel_texture_object(texObj));
+      }
+   }
+}
 
 /**
  * Plug in appropriate span read/write functions for the given renderbuffer.
@@ -711,6 +748,9 @@ intel_set_span_functions(struct intel_context *intel,
 	    intel_YTile_InitStencilPointers_z24_s8(rb);
 	    break;
 	 }
+      } else {
+	 _mesa_problem(NULL,
+		       "Unexpected ActualFormat in intelSetSpanFunctions");
       }
       break;
    default:
diff --git a/src/mesa/drivers/dri/intel/intel_span.h b/src/mesa/drivers/dri/intel/intel_span.h
index acbeb4abe1..bffe109aa5 100644
--- a/src/mesa/drivers/dri/intel/intel_span.h
+++ b/src/mesa/drivers/dri/intel/intel_span.h
@@ -36,5 +36,7 @@ void intel_renderbuffer_map(struct intel_context *intel,
 			    struct gl_renderbuffer *rb);
 void intel_renderbuffer_unmap(struct intel_context *intel,
 			      struct gl_renderbuffer *rb);
+void intel_map_vertex_shader_textures(GLcontext *ctx);
+void intel_unmap_vertex_shader_textures(GLcontext *ctx);
 
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_syncobj.c b/src/mesa/drivers/dri/intel/intel_syncobj.c
index 1286fe929b..0d7889d3c2 100644
--- a/src/mesa/drivers/dri/intel/intel_syncobj.c
+++ b/src/mesa/drivers/dri/intel/intel_syncobj.c
@@ -114,7 +114,7 @@ static void intel_check_sync(GLcontext *ctx, struct gl_sync_object *s)
 {
    struct intel_sync_object *sync = (struct intel_sync_object *)s;
 
-   if (sync->bo && drm_intel_bo_busy(sync->bo)) {
+   if (sync->bo && !drm_intel_bo_busy(sync->bo)) {
       drm_intel_bo_unreference(sync->bo);
       sync->bo = NULL;
       s->StatusFlag = 1;
diff --git a/src/mesa/drivers/dri/intel/intel_tex.c b/src/mesa/drivers/dri/intel/intel_tex.c
index df63f29a42..215a534a5c 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.c
+++ b/src/mesa/drivers/dri/intel/intel_tex.c
@@ -2,6 +2,7 @@
 #include "main/texobj.h"
 #include "main/teximage.h"
 #include "main/mipmap.h"
+#include "drivers/common/meta.h"
 #include "intel_context.h"
 #include "intel_mipmap_tree.h"
 #include "intel_tex.h"
@@ -158,11 +159,58 @@ timed_memcpy(void *dest, const void *src, size_t n)
 }
 #endif /* DO_DEBUG */
 
+
+/**
+ * Called via ctx->Driver.GenerateMipmap()
+ * This is basically a wrapper for _mesa_meta_GenerateMipmap() which checks
+ * if we'll be using software mipmap generation.  In that case, we need to
+ * map/unmap the base level texture image.
+ */
+static void
+intelGenerateMipmap(GLcontext *ctx, GLenum target,
+                    struct gl_texture_object *texObj)
+{
+   if (_mesa_meta_check_generate_mipmap_fallback(ctx, target, texObj)) {
+      /* sw path: need to map texture images */
+      struct intel_context *intel = intel_context(ctx);
+      struct intel_texture_object *intelObj = intel_texture_object(texObj);
+      intel_tex_map_level_images(intel, intelObj, texObj->BaseLevel);
+      _mesa_generate_mipmap(ctx, target, texObj);
+      intel_tex_unmap_level_images(intel, intelObj, texObj->BaseLevel);
+
+      {
+         GLuint nr_faces = (texObj->Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
+         GLuint face, i;
+         /* Update the level information in our private data in the new images,
+          * since it didn't get set as part of a normal TexImage path.
+          */
+         for (face = 0; face < nr_faces; face++) {
+            for (i = texObj->BaseLevel + 1; i < texObj->MaxLevel; i++) {
+               struct intel_texture_image *intelImage =
+                  intel_texture_image(texObj->Image[face][i]);
+               if (!intelImage)
+                  break;
+               intelImage->level = i;
+               intelImage->face = face;
+               /* Unreference the miptree to signal that the new Data is a
+                * bare pointer from mesa.
+                */
+               intel_miptree_release(intel, &intelImage->mt);
+            }
+         }
+      }
+   }
+   else {
+      _mesa_meta_GenerateMipmap(ctx, target, texObj);
+   }
+}
+
+
 void
 intelInitTextureFuncs(struct dd_function_table *functions)
 {
    functions->ChooseTextureFormat = intelChooseTextureFormat;
-   functions->GenerateMipmap = intel_generate_mipmap;
+   functions->GenerateMipmap = intelGenerateMipmap;
 
    functions->NewTextureObject = intelNewTextureObject;
    functions->NewTextureImage = intelNewTextureImage;
diff --git a/src/mesa/drivers/dri/intel/intel_tex.h b/src/mesa/drivers/dri/intel/intel_tex.h
index 471aa2a240..57ed0b1aab 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.h
+++ b/src/mesa/drivers/dri/intel/intel_tex.h
@@ -71,7 +71,4 @@ void intel_tex_unmap_images(struct intel_context *intel,
 
 int intel_compressed_num_bytes(GLuint mesaFormat);
 
-void intel_generate_mipmap(GLcontext *ctx, GLenum target,
-			   struct gl_texture_object *texObj);
-
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index b241c11625..99cd818417 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -30,7 +30,8 @@
 #include "main/image.h"
 #include "main/teximage.h"
 #include "main/mipmap.h"
-#include "swrast/swrast.h"
+
+#include "drivers/common/meta.h"
 
 #include "intel_screen.h"
 #include "intel_context.h"
@@ -91,7 +92,6 @@ do_copy_texsubimage(struct intel_context *intel,
                     GLint x, GLint y, GLsizei width, GLsizei height)
 {
    GLcontext *ctx = &intel->ctx;
-   struct gl_texture_object *texObj = intelImage->base.TexObject;
    const struct intel_region *src =
       get_teximage_source(intel, internalFormat);
 
@@ -117,7 +117,7 @@ do_copy_texsubimage(struct intel_context *intel,
 						 INTEL_WRITE_PART);
       GLuint image_offset = intel_miptree_image_offset(intelImage->mt,
                                                        intelImage->face,
-                                                       intelImage->level);
+                                                       intelImage->level, 0);
       const GLint orig_x = x;
       const GLint orig_y = y;
       GLshort src_pitch;
@@ -171,11 +171,6 @@ do_copy_texsubimage(struct intel_context *intel,
 
    UNLOCK_HARDWARE(intel);
 
-   /* GL_SGIS_generate_mipmap */
-   if (intelImage->level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      intel_generate_mipmap(ctx, target, texObj);
-   }
-
    return GL_TRUE;
 }
 
@@ -222,8 +217,8 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
-   _swrast_copy_teximage1d(ctx, target, level, internalFormat, x, y,
-                           width, border);
+   _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y,
+                             width, border);
 }
 
 
@@ -270,8 +265,8 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
-   _swrast_copy_teximage2d(ctx, target, level, internalFormat, x, y,
-                           width, height, border);
+   _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y,
+                             width, height, border);
 }
 
 
@@ -295,7 +290,7 @@ intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
    if (!do_copy_texsubimage(intel_context(ctx), target,
                             intel_texture_image(texImage),
                             internalFormat, xoffset, 0, x, y, width, 1)) {
-      _swrast_copy_texsubimage1d(ctx, target, level, xoffset, x, y, width);
+      _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width);
    }
 }
 
@@ -321,10 +316,10 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
                             internalFormat,
                             xoffset, yoffset, x, y, width, height)) {
 
-      DBG("%s - fallback to swrast\n", __FUNCTION__);
+      DBG("%s - fallback to _mesa_meta_CopyTexSubImage2D\n", __FUNCTION__);
 
-      _swrast_copy_texsubimage2d(ctx, target, level,
-                                 xoffset, yoffset, x, y, width, height);
+      _mesa_meta_CopyTexSubImage2D(ctx, target, level,
+                                   xoffset, yoffset, x, y, width, height);
    }
 }
 
diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c
index c5f5220837..dcc613449f 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_image.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_image.c
@@ -1,15 +1,11 @@
 
-#include <stdlib.h>
-#include <stdio.h>
-
 #include "main/glheader.h"
 #include "main/macros.h"
 #include "main/mtypes.h"
 #include "main/enums.h"
-#include "main/colortab.h"
+#include "main/bufferobj.h"
 #include "main/convolve.h"
 #include "main/context.h"
-#include "main/simple_list.h"
 #include "main/texcompress.h"
 #include "main/texformat.h"
 #include "main/texgetimage.h"
@@ -205,8 +201,11 @@ try_pbo_upload(struct intel_context *intel,
    struct intel_buffer_object *pbo = intel_buffer_object(unpack->BufferObj);
    GLuint src_offset, src_stride;
    GLuint dst_offset, dst_stride;
+   dri_bo *dst_buffer = intel_region_buffer(intel,
+					    intelImage->mt->region,
+					    INTEL_WRITE_FULL);
 
-   if (unpack->BufferObj->Name == 0 ||
+   if (!_mesa_is_bufferobj(unpack->BufferObj) ||
        intel->ctx._ImageTransferState ||
        unpack->SkipPixels || unpack->SkipRows) {
       DBG("%s: failure 1\n", __FUNCTION__);
@@ -223,11 +222,13 @@ try_pbo_upload(struct intel_context *intel,
 
    dst_offset = intel_miptree_image_offset(intelImage->mt,
                                            intelImage->face,
-                                           intelImage->level);
+                                           intelImage->level,
+                                           0 /* zslice */);
 
    dst_stride = intelImage->mt->pitch;
 
-   intelFlush(&intel->ctx);
+   if (drm_intel_bo_references(intel->batch->buf, dst_buffer))
+      intelFlush(&intel->ctx);
    LOCK_HARDWARE(intel);
    {
       dri_bo *src_buffer = intel_bufferobj_buffer(intel, pbo, INTEL_READ);
@@ -264,7 +265,7 @@ try_pbo_zcopy(struct intel_context *intel,
    GLuint src_offset, src_stride;
    GLuint dst_offset, dst_stride;
 
-   if (unpack->BufferObj->Name == 0 ||
+   if (!_mesa_is_bufferobj(unpack->BufferObj) ||
        intel->ctx._ImageTransferState ||
        unpack->SkipPixels || unpack->SkipRows) {
       DBG("%s: failure 1\n", __FUNCTION__);
@@ -281,8 +282,8 @@ try_pbo_zcopy(struct intel_context *intel,
 
    dst_offset = intel_miptree_image_offset(intelImage->mt,
                                            intelImage->face,
-                                           intelImage->level);
-
+                                           intelImage->level,
+                                           0 /* zslice */);
    dst_stride = intelImage->mt->pitch;
 
    if (src_stride != dst_stride || dst_offset != 0 || src_offset != 0) {
@@ -320,8 +321,6 @@ intelTexImage(GLcontext * ctx,
    DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__,
        _mesa_lookup_enum_by_nr(target), level, width, height, depth, border);
 
-   intelFlush(ctx);
-
    intelImage->face = target_to_face(target);
    intelImage->level = level;
 
@@ -427,7 +426,7 @@ intelTexImage(GLcontext * ctx,
     */
    if (dims <= 2 &&
        intelImage->mt &&
-       unpack->BufferObj->Name != 0 &&
+       _mesa_is_bufferobj(unpack->BufferObj) &&
        check_pbo_format(internalFormat, format,
                         type, intelImage->base.TexFormat)) {
 
@@ -482,13 +481,20 @@ intelTexImage(GLcontext * ctx,
    LOCK_HARDWARE(intel);
 
    if (intelImage->mt) {
-      if (pixels != NULL)
+      if (pixels != NULL) {
+	 /* Flush any queued rendering with the texture before mapping. */
+	 if (drm_intel_bo_references(intel->batch->buf,
+				     intelImage->mt->region->buffer)) {
+	    intelFlush(ctx);
+	 }
          texImage->Data = intel_miptree_image_map(intel,
                                                   intelImage->mt,
                                                   intelImage->face,
                                                   intelImage->level,
                                                   &dstRowStride,
                                                   intelImage->base.ImageOffsets);
+      }
+
       texImage->RowStride = dstRowStride / intelImage->mt->cpp;
    }
    else {
@@ -550,11 +556,6 @@ intelTexImage(GLcontext * ctx,
    }
 
    UNLOCK_HARDWARE(intel);
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      intel_generate_mipmap(ctx, target, texObj);
-   }
 }
 
 
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.h b/src/mesa/drivers/dri/intel/intel_tex_layout.h
index c9de9b5678..a9ac9e7eb4 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.h
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.h
@@ -33,7 +33,7 @@
 #include "main/macros.h"
 
 
-static GLuint minify( GLuint d )
+static INLINE GLuint minify( GLuint d )
 {
    return MAX2(1, d>>1);
 }
diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
index 89037073f8..751ec2c98c 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_subimage.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
@@ -129,11 +129,6 @@ intelTexSubimage(GLcontext * ctx,
    }
 
    UNLOCK_HARDWARE(intel);
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      intel_generate_mipmap(ctx, target, texObj);
-   }
 }
 
 
diff --git a/src/mesa/drivers/dri/mach64/Makefile b/src/mesa/drivers/dri/mach64/Makefile
index 7246d51f5d..a8f463e9fd 100644
--- a/src/mesa/drivers/dri/mach64/Makefile
+++ b/src/mesa/drivers/dri/mach64/Makefile
@@ -30,4 +30,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/mach64/mach64_tex.c b/src/mesa/drivers/dri/mach64/mach64_tex.c
index 225d23179e..5a22c93bd2 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tex.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tex.c
@@ -567,7 +567,6 @@ void mach64InitTextureFuncs( struct dd_function_table *functions )
 
    functions->UpdateTexturePalette	= NULL;
    functions->ActiveTexture		= NULL;
-   functions->PrioritizeTexture		= NULL;
 
    driInitTextureFormats();
 }
diff --git a/src/mesa/drivers/dri/mga/Makefile b/src/mesa/drivers/dri/mga/Makefile
index a871064c62..0cc329fb22 100644
--- a/src/mesa/drivers/dri/mga/Makefile
+++ b/src/mesa/drivers/dri/mga/Makefile
@@ -31,4 +31,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/r128/Makefile b/src/mesa/drivers/dri/r128/Makefile
index 796dfbc516..52c5a38a70 100644
--- a/src/mesa/drivers/dri/r128/Makefile
+++ b/src/mesa/drivers/dri/r128/Makefile
@@ -29,4 +29,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/r200/Makefile b/src/mesa/drivers/dri/r200/Makefile
index 42635bf9d9..776f1e3f3f 100644
--- a/src/mesa/drivers/dri/r200/Makefile
+++ b/src/mesa/drivers/dri/r200/Makefile
@@ -55,8 +55,7 @@ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES)
 
 X86_SOURCES = 
 
-DRIVER_DEFINES = -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R200 \
-				 -Wall
+DRIVER_DEFINES = -DRADEON_R200 -Wall
 
 DRI_LIB_DEPS += $(RADEON_LDFLAGS)
 
@@ -67,4 +66,3 @@ include ../Makefile.template
 
 #INCLUDES += -I../radeon/server
 
-symlinks:
diff --git a/src/mesa/drivers/dri/r300/Makefile b/src/mesa/drivers/dri/r300/Makefile
index 5d8d6f6658..cb0f715fa0 100644
--- a/src/mesa/drivers/dri/r300/Makefile
+++ b/src/mesa/drivers/dri/r300/Makefile
@@ -54,6 +54,7 @@ DRIVER_SOURCES = \
 		 r300_vertprog.c \
 		 r300_fragprog_common.c \
 		 r300_shader.c \
+		 radeon_mesa_to_rc.c \
 		 r300_emit.c \
 		 r300_swtcl.c \
 		 $(RADEON_COMMON_SOURCES) \
@@ -62,23 +63,18 @@ DRIVER_SOURCES = \
 
 C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES)
 
-DRIVER_DEFINES = -DCOMPILE_R300 -DR200_MERGED=0 \
-	-DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R300 \
+DRIVER_DEFINES = -DRADEON_R300
 #	-DRADEON_BO_TRACK \
 	-Wall
 
 DRI_LIB_DEPS += $(RADEON_LDFLAGS)
 
-PIPE_DRIVERS =  compiler/libr300compiler.a
+SUBDIRS = compiler
 
-##### TARGETS #####
+EXTRA_MODULES = compiler/libr300compiler.a
 
-include ../Makefile.template
 
-symlinks:
+##### TARGETS #####
 
-# Mark the archive phony so that we always check for recompilation
-.PHONY : compiler/libr300compiler.a
+include ../Makefile.template
 
-compiler/libr300compiler.a:
-	cd compiler && $(MAKE)
diff --git a/src/mesa/drivers/dri/r300/compiler/Makefile b/src/mesa/drivers/dri/r300/compiler/Makefile
index d973844192..d83888d90a 100644
--- a/src/mesa/drivers/dri/r300/compiler/Makefile
+++ b/src/mesa/drivers/dri/r300/compiler/Makefile
@@ -8,10 +8,17 @@ LIBNAME = r300compiler
 C_SOURCES = \
 		radeon_code.c \
 		radeon_compiler.c \
-		radeon_nqssadce.c \
 		radeon_program.c \
+		radeon_program_print.c \
+		radeon_opcodes.c \
 		radeon_program_alu.c \
 		radeon_program_pair.c \
+		radeon_pair_translate.c \
+		radeon_pair_schedule.c \
+		radeon_pair_regalloc.c \
+		radeon_dataflow.c \
+		radeon_dataflow_deadcode.c \
+		radeon_dataflow_swizzles.c \
 		r3xx_fragprog.c \
 		r300_fragprog.c \
 		r300_fragprog_swizzle.c \
diff --git a/src/mesa/drivers/dri/r300/compiler/SConscript b/src/mesa/drivers/dri/r300/compiler/SConscript
new file mode 100755
index 0000000000..46075a8aee
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/SConscript
@@ -0,0 +1,37 @@
+Import('*')
+
+env = env.Clone()
+env.Append(CPPPATH = '#/include')
+env.Append(CPPPATH = '#/src/mesa')
+
+# temporary fix
+env['CFLAGS'] = str(env['CFLAGS']).replace('-Werror=declaration-after-statement', '')
+
+r300compiler = env.ConvenienceLibrary(
+    target = 'r300compiler',
+    source = [
+        'radeon_code.c',
+        'radeon_compiler.c',
+        'radeon_program.c',
+        'radeon_program_print.c',
+        'radeon_opcodes.c',
+        'radeon_program_alu.c',
+        'radeon_program_pair.c',
+        'radeon_pair_translate.c',
+        'radeon_pair_schedule.c',
+        'radeon_pair_regalloc.c',
+        'radeon_dataflow.c',
+        'radeon_dataflow_deadcode.c',
+        'radeon_dataflow_swizzles.c',
+        'r3xx_fragprog.c',
+        'r300_fragprog.c',
+        'r300_fragprog_swizzle.c',
+        'r300_fragprog_emit.c',
+        'r500_fragprog.c',
+        'r500_fragprog_emit.c',
+        'r3xx_vertprog.c',
+        'r3xx_vertprog_dump.c',
+        'memory_pool.c',
+    ])
+
+Return('r300compiler')
diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog.c b/src/mesa/drivers/dri/r300/compiler/r300_fragprog.c
index 6c9fba4914..aa69b0fc72 100644
--- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog.c
@@ -27,17 +27,17 @@
 
 #include "r300_fragprog.h"
 
-#include "shader/prog_parameter.h"
+#include <stdio.h>
 
 #include "../r300_reg.h"
 
-static struct prog_src_register shadow_ambient(struct radeon_compiler * c, int tmu)
+static struct rc_src_register shadow_ambient(struct radeon_compiler * c, int tmu)
 {
-	struct prog_src_register reg = { 0, };
+	struct rc_src_register reg = { 0, };
 
-	reg.File = PROGRAM_STATE_VAR;
+	reg.File = RC_FILE_CONSTANT;
 	reg.Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_SHADOW_AMBIENT, tmu);
-	reg.Swizzle = SWIZZLE_WWWW;
+	reg.Swizzle = RC_SWIZZLE_WWWW;
 	return reg;
 }
 
@@ -47,7 +47,7 @@ static struct prog_src_register shadow_ambient(struct radeon_compiler * c, int t
  *  - extract operand swizzles
  *  - introduce a temporary register when write masks are needed
  */
-GLboolean r300_transform_TEX(
+int r300_transform_TEX(
 	struct radeon_compiler * c,
 	struct rc_instruction* inst,
 	void* data)
@@ -55,77 +55,77 @@ GLboolean r300_transform_TEX(
 	struct r300_fragment_program_compiler *compiler =
 		(struct r300_fragment_program_compiler*)data;
 
-	if (inst->I.Opcode != OPCODE_TEX &&
-	    inst->I.Opcode != OPCODE_TXB &&
-	    inst->I.Opcode != OPCODE_TXP &&
-	    inst->I.Opcode != OPCODE_KIL)
-		return GL_FALSE;
+	if (inst->U.I.Opcode != RC_OPCODE_TEX &&
+	    inst->U.I.Opcode != RC_OPCODE_TXB &&
+	    inst->U.I.Opcode != RC_OPCODE_TXP &&
+	    inst->U.I.Opcode != RC_OPCODE_KIL)
+		return 0;
 
 	/* ARB_shadow & EXT_shadow_funcs */
-	if (inst->I.Opcode != OPCODE_KIL &&
-	    c->Program.ShadowSamplers & (1 << inst->I.TexSrcUnit)) {
-		GLuint comparefunc = GL_NEVER + compiler->state.unit[inst->I.TexSrcUnit].texture_compare_func;
+	if (inst->U.I.Opcode != RC_OPCODE_KIL &&
+	    c->Program.ShadowSamplers & (1 << inst->U.I.TexSrcUnit)) {
+		rc_compare_func comparefunc = compiler->state.unit[inst->U.I.TexSrcUnit].texture_compare_func;
 
-		if (comparefunc == GL_NEVER || comparefunc == GL_ALWAYS) {
-			inst->I.Opcode = OPCODE_MOV;
+		if (comparefunc == RC_COMPARE_FUNC_NEVER || comparefunc == RC_COMPARE_FUNC_ALWAYS) {
+			inst->U.I.Opcode = RC_OPCODE_MOV;
 
-			if (comparefunc == GL_ALWAYS) {
-				inst->I.SrcReg[0].File = PROGRAM_BUILTIN;
-				inst->I.SrcReg[0].Swizzle = SWIZZLE_1111;
+			if (comparefunc == RC_COMPARE_FUNC_ALWAYS) {
+				inst->U.I.SrcReg[0].File = RC_FILE_NONE;
+				inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_1111;
 			} else {
-				inst->I.SrcReg[0] = shadow_ambient(c, inst->I.TexSrcUnit);
+				inst->U.I.SrcReg[0] = shadow_ambient(c, inst->U.I.TexSrcUnit);
 			}
 
-			return GL_TRUE;
+			return 1;
 		} else {
-			GLuint comparefunc = GL_NEVER + compiler->state.unit[inst->I.TexSrcUnit].texture_compare_func;
-			GLuint depthmode = compiler->state.unit[inst->I.TexSrcUnit].depth_texture_mode;
+			rc_compare_func comparefunc = compiler->state.unit[inst->U.I.TexSrcUnit].texture_compare_func;
+			unsigned int depthmode = compiler->state.unit[inst->U.I.TexSrcUnit].depth_texture_mode;
 			struct rc_instruction * inst_rcp = rc_insert_new_instruction(c, inst);
 			struct rc_instruction * inst_mad = rc_insert_new_instruction(c, inst_rcp);
 			struct rc_instruction * inst_cmp = rc_insert_new_instruction(c, inst_mad);
 			int pass, fail;
 
-			inst_rcp->I.Opcode = OPCODE_RCP;
-			inst_rcp->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst_rcp->I.DstReg.Index = rc_find_free_temporary(c);
-			inst_rcp->I.DstReg.WriteMask = WRITEMASK_W;
-			inst_rcp->I.SrcReg[0] = inst->I.SrcReg[0];
-			inst_rcp->I.SrcReg[0].Swizzle = SWIZZLE_WWWW;
-
-			inst_cmp->I.DstReg = inst->I.DstReg;
-			inst->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst->I.DstReg.Index = rc_find_free_temporary(c);
-			inst->I.DstReg.WriteMask = WRITEMASK_XYZW;
-
-			inst_mad->I.Opcode = OPCODE_MAD;
-			inst_mad->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst_mad->I.DstReg.Index = rc_find_free_temporary(c);
-			inst_mad->I.SrcReg[0] = inst->I.SrcReg[0];
-			inst_mad->I.SrcReg[0].Swizzle = SWIZZLE_ZZZZ;
-			inst_mad->I.SrcReg[1].File = PROGRAM_TEMPORARY;
-			inst_mad->I.SrcReg[1].Index = inst_rcp->I.DstReg.Index;
-			inst_mad->I.SrcReg[1].Swizzle = SWIZZLE_WWWW;
-			inst_mad->I.SrcReg[2].File = PROGRAM_TEMPORARY;
-			inst_mad->I.SrcReg[2].Index = inst->I.DstReg.Index;
+			inst_rcp->U.I.Opcode = RC_OPCODE_RCP;
+			inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst_rcp->U.I.DstReg.Index = rc_find_free_temporary(c);
+			inst_rcp->U.I.DstReg.WriteMask = RC_MASK_W;
+			inst_rcp->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
+			inst_rcp->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_WWWW;
+
+			inst_cmp->U.I.DstReg = inst->U.I.DstReg;
+			inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst->U.I.DstReg.Index = rc_find_free_temporary(c);
+			inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
+
+			inst_mad->U.I.Opcode = RC_OPCODE_MAD;
+			inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst_mad->U.I.DstReg.Index = rc_find_free_temporary(c);
+			inst_mad->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
+			inst_mad->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_ZZZZ;
+			inst_mad->U.I.SrcReg[1].File = RC_FILE_TEMPORARY;
+			inst_mad->U.I.SrcReg[1].Index = inst_rcp->U.I.DstReg.Index;
+			inst_mad->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_WWWW;
+			inst_mad->U.I.SrcReg[2].File = RC_FILE_TEMPORARY;
+			inst_mad->U.I.SrcReg[2].Index = inst->U.I.DstReg.Index;
 			if (depthmode == 0) /* GL_LUMINANCE */
-				inst_mad->I.SrcReg[2].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z);
+				inst_mad->U.I.SrcReg[2].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_Z);
 			else if (depthmode == 2) /* GL_ALPHA */
-				inst_mad->I.SrcReg[2].Swizzle = SWIZZLE_WWWW;
+				inst_mad->U.I.SrcReg[2].Swizzle = RC_SWIZZLE_WWWW;
 
 			/* Recall that SrcReg[0] is tex, SrcReg[2] is r and:
 			 *   r  < tex  <=>      -tex+r < 0
 			 *   r >= tex  <=> not (-tex+r < 0 */
-			if (comparefunc == GL_LESS || comparefunc == GL_GEQUAL)
-				inst_mad->I.SrcReg[2].Negate = inst_mad->I.SrcReg[2].Negate ^ NEGATE_XYZW;
+			if (comparefunc == RC_COMPARE_FUNC_LESS || comparefunc == RC_COMPARE_FUNC_GEQUAL)
+				inst_mad->U.I.SrcReg[2].Negate = inst_mad->U.I.SrcReg[2].Negate ^ RC_MASK_XYZW;
 			else
-				inst_mad->I.SrcReg[0].Negate = inst_mad->I.SrcReg[0].Negate ^ NEGATE_XYZW;
+				inst_mad->U.I.SrcReg[0].Negate = inst_mad->U.I.SrcReg[0].Negate ^ RC_MASK_XYZW;
 
-			inst_cmp->I.Opcode = OPCODE_CMP;
+			inst_cmp->U.I.Opcode = RC_OPCODE_CMP;
 			/* DstReg has been filled out above */
-			inst_cmp->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-			inst_cmp->I.SrcReg[0].Index = inst_mad->I.DstReg.Index;
+			inst_cmp->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+			inst_cmp->U.I.SrcReg[0].Index = inst_mad->U.I.DstReg.Index;
 
-			if (comparefunc == GL_LESS || comparefunc == GL_GREATER) {
+			if (comparefunc == RC_COMPARE_FUNC_LESS || comparefunc == RC_COMPARE_FUNC_GREATER) {
 				pass = 1;
 				fail = 2;
 			} else {
@@ -133,9 +133,9 @@ GLboolean r300_transform_TEX(
 				fail = 1;
 			}
 
-			inst_cmp->I.SrcReg[pass].File = PROGRAM_BUILTIN;
-			inst_cmp->I.SrcReg[pass].Swizzle = SWIZZLE_1111;
-			inst_cmp->I.SrcReg[fail] = shadow_ambient(c, inst->I.TexSrcUnit);
+			inst_cmp->U.I.SrcReg[pass].File = RC_FILE_NONE;
+			inst_cmp->U.I.SrcReg[pass].Swizzle = RC_SWIZZLE_1111;
+			inst_cmp->U.I.SrcReg[fail] = shadow_ambient(c, inst->U.I.TexSrcUnit);
 		}
 	}
 
@@ -143,52 +143,52 @@ GLboolean r300_transform_TEX(
 	 * instead of [0..Width]x[0..Height].
 	 * Add a scaling instruction.
 	 */
-	if (inst->I.Opcode != OPCODE_KIL && inst->I.TexSrcTarget == TEXTURE_RECT_INDEX) {
+	if (inst->U.I.Opcode != RC_OPCODE_KIL && inst->U.I.TexSrcTarget == RC_TEXTURE_RECT) {
 		struct rc_instruction * inst_mul = rc_insert_new_instruction(c, inst->Prev);
 
-		inst_mul->I.Opcode = OPCODE_MUL;
-		inst_mul->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst_mul->I.DstReg.Index = rc_find_free_temporary(c);
-		inst_mul->I.SrcReg[0] = inst->I.SrcReg[0];
-		inst_mul->I.SrcReg[1].File = PROGRAM_STATE_VAR;
-		inst_mul->I.SrcReg[1].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_TEXRECT_FACTOR, inst->I.TexSrcUnit);
+		inst_mul->U.I.Opcode = RC_OPCODE_MUL;
+		inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst_mul->U.I.DstReg.Index = rc_find_free_temporary(c);
+		inst_mul->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
+		inst_mul->U.I.SrcReg[1].File = RC_FILE_CONSTANT;
+		inst_mul->U.I.SrcReg[1].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_TEXRECT_FACTOR, inst->U.I.TexSrcUnit);
 
-		reset_srcreg(&inst->I.SrcReg[0]);
-		inst->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-		inst->I.SrcReg[0].Index = inst_mul->I.DstReg.Index;
+		reset_srcreg(&inst->U.I.SrcReg[0]);
+		inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+		inst->U.I.SrcReg[0].Index = inst_mul->U.I.DstReg.Index;
 	}
 
 	/* Cannot write texture to output registers or with masks */
-	if (inst->I.Opcode != OPCODE_KIL &&
-	    (inst->I.DstReg.File != PROGRAM_TEMPORARY || inst->I.DstReg.WriteMask != WRITEMASK_XYZW)) {
+	if (inst->U.I.Opcode != RC_OPCODE_KIL &&
+	    (inst->U.I.DstReg.File != RC_FILE_TEMPORARY || inst->U.I.DstReg.WriteMask != RC_MASK_XYZW)) {
 		struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst);
 
-		inst_mov->I.Opcode = OPCODE_MOV;
-		inst_mov->I.DstReg = inst->I.DstReg;
-		inst_mov->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-		inst_mov->I.SrcReg[0].Index = rc_find_free_temporary(c);
+		inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+		inst_mov->U.I.DstReg = inst->U.I.DstReg;
+		inst_mov->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+		inst_mov->U.I.SrcReg[0].Index = rc_find_free_temporary(c);
 
-		inst->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst->I.DstReg.Index = inst_mov->I.SrcReg[0].Index;
-		inst->I.DstReg.WriteMask = WRITEMASK_XYZW;
+		inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst->U.I.DstReg.Index = inst_mov->U.I.SrcReg[0].Index;
+		inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
 	}
 
 
 	/* Cannot read texture coordinate from constants file */
-	if (inst->I.SrcReg[0].File != PROGRAM_TEMPORARY && inst->I.SrcReg[0].File != PROGRAM_INPUT) {
+	if (inst->U.I.SrcReg[0].File != RC_FILE_TEMPORARY && inst->U.I.SrcReg[0].File != RC_FILE_INPUT) {
 		struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev);
 
-		inst_mov->I.Opcode = OPCODE_MOV;
-		inst_mov->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst_mov->I.DstReg.Index = rc_find_free_temporary(c);
-		inst_mov->I.SrcReg[0] = inst->I.SrcReg[0];
+		inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+		inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst_mov->U.I.DstReg.Index = rc_find_free_temporary(c);
+		inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
 
-		reset_srcreg(&inst->I.SrcReg[0]);
-		inst->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-		inst->I.SrcReg[0].Index = inst_mov->I.DstReg.Index;
+		reset_srcreg(&inst->U.I.SrcReg[0]);
+		inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+		inst->U.I.SrcReg[0].Index = inst_mov->U.I.DstReg.Index;
 	}
 
-	return GL_TRUE;
+	return 1;
 }
 
 /* just some random things... */
diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog.h b/src/mesa/drivers/dri/r300/compiler/r300_fragprog.h
index 0ac46dbd9c..418df36c93 100644
--- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog.h
+++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog.h
@@ -33,9 +33,6 @@
 #ifndef __R300_FRAGPROG_H_
 #define __R300_FRAGPROG_H_
 
-#include "shader/program.h"
-#include "shader/prog_instruction.h"
-
 #include "radeon_compiler.h"
 #include "radeon_program.h"
 
@@ -44,6 +41,6 @@ extern void r300BuildFragmentProgramHwCode(struct r300_fragment_program_compiler
 
 extern void r300FragmentProgramDump(struct rX00_fragment_program_code *c);
 
-extern GLboolean r300_transform_TEX(struct radeon_compiler * c, struct rc_instruction* inst, void* data);
+extern int r300_transform_TEX(struct radeon_compiler * c, struct rc_instruction* inst, void* data);
 
 #endif
diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c
index c7227bbd15..375838d98e 100644
--- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c
+++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c
@@ -56,7 +56,6 @@ struct r300_emit_state {
 };
 
 #define PROG_CODE \
-	struct r300_emit_state * emit = (struct r300_emit_state*)data; \
 	struct r300_fragment_program_compiler *c = emit->compiler; \
 	struct r300_fragment_program_code *code = &c->code->code.r300
 
@@ -69,64 +68,76 @@ struct r300_emit_state {
 /**
  * Mark a temporary register as used.
  */
-static void use_temporary(struct r300_fragment_program_code *code, GLuint index)
+static void use_temporary(struct r300_fragment_program_code *code, unsigned int index)
 {
 	if (index > code->pixsize)
 		code->pixsize = index;
 }
 
+static unsigned int use_source(struct r300_fragment_program_code* code, struct radeon_pair_instruction_source src)
+{
+	if (src.File == RC_FILE_CONSTANT) {
+		return src.Index | (1 << 5);
+	} else if (src.File == RC_FILE_TEMPORARY) {
+		use_temporary(code, src.Index);
+		return src.Index;
+	}
+
+	return 0;
+}
+
 
-static GLuint translate_rgb_opcode(struct r300_fragment_program_compiler * c, GLuint opcode)
+static unsigned int translate_rgb_opcode(struct r300_fragment_program_compiler * c, rc_opcode opcode)
 {
 	switch(opcode) {
-	case OPCODE_CMP: return R300_ALU_OUTC_CMP;
-	case OPCODE_DP3: return R300_ALU_OUTC_DP3;
-	case OPCODE_DP4: return R300_ALU_OUTC_DP4;
-	case OPCODE_FRC: return R300_ALU_OUTC_FRC;
+	case RC_OPCODE_CMP: return R300_ALU_OUTC_CMP;
+	case RC_OPCODE_DP3: return R300_ALU_OUTC_DP3;
+	case RC_OPCODE_DP4: return R300_ALU_OUTC_DP4;
+	case RC_OPCODE_FRC: return R300_ALU_OUTC_FRC;
 	default:
 		error("translate_rgb_opcode(%i): Unknown opcode", opcode);
 		/* fall through */
-	case OPCODE_NOP:
+	case RC_OPCODE_NOP:
 		/* fall through */
-	case OPCODE_MAD: return R300_ALU_OUTC_MAD;
-	case OPCODE_MAX: return R300_ALU_OUTC_MAX;
-	case OPCODE_MIN: return R300_ALU_OUTC_MIN;
-	case OPCODE_REPL_ALPHA: return R300_ALU_OUTC_REPL_ALPHA;
+	case RC_OPCODE_MAD: return R300_ALU_OUTC_MAD;
+	case RC_OPCODE_MAX: return R300_ALU_OUTC_MAX;
+	case RC_OPCODE_MIN: return R300_ALU_OUTC_MIN;
+	case RC_OPCODE_REPL_ALPHA: return R300_ALU_OUTC_REPL_ALPHA;
 	}
 }
 
-static GLuint translate_alpha_opcode(struct r300_fragment_program_compiler * c, GLuint opcode)
+static unsigned int translate_alpha_opcode(struct r300_fragment_program_compiler * c, rc_opcode opcode)
 {
 	switch(opcode) {
-	case OPCODE_CMP: return R300_ALU_OUTA_CMP;
-	case OPCODE_DP3: return R300_ALU_OUTA_DP4;
-	case OPCODE_DP4: return R300_ALU_OUTA_DP4;
-	case OPCODE_EX2: return R300_ALU_OUTA_EX2;
-	case OPCODE_FRC: return R300_ALU_OUTA_FRC;
-	case OPCODE_LG2: return R300_ALU_OUTA_LG2;
+	case RC_OPCODE_CMP: return R300_ALU_OUTA_CMP;
+	case RC_OPCODE_DP3: return R300_ALU_OUTA_DP4;
+	case RC_OPCODE_DP4: return R300_ALU_OUTA_DP4;
+	case RC_OPCODE_EX2: return R300_ALU_OUTA_EX2;
+	case RC_OPCODE_FRC: return R300_ALU_OUTA_FRC;
+	case RC_OPCODE_LG2: return R300_ALU_OUTA_LG2;
 	default:
 		error("translate_rgb_opcode(%i): Unknown opcode", opcode);
 		/* fall through */
-	case OPCODE_NOP:
+	case RC_OPCODE_NOP:
 		/* fall through */
-	case OPCODE_MAD: return R300_ALU_OUTA_MAD;
-	case OPCODE_MAX: return R300_ALU_OUTA_MAX;
-	case OPCODE_MIN: return R300_ALU_OUTA_MIN;
-	case OPCODE_RCP: return R300_ALU_OUTA_RCP;
-	case OPCODE_RSQ: return R300_ALU_OUTA_RSQ;
+	case RC_OPCODE_MAD: return R300_ALU_OUTA_MAD;
+	case RC_OPCODE_MAX: return R300_ALU_OUTA_MAX;
+	case RC_OPCODE_MIN: return R300_ALU_OUTA_MIN;
+	case RC_OPCODE_RCP: return R300_ALU_OUTA_RCP;
+	case RC_OPCODE_RSQ: return R300_ALU_OUTA_RSQ;
 	}
 }
 
 /**
  * Emit one paired ALU instruction.
  */
-static GLboolean emit_alu(void* data, struct radeon_pair_instruction* inst)
+static int emit_alu(struct r300_emit_state * emit, struct rc_pair_instruction* inst)
 {
 	PROG_CODE;
 
 	if (code->alu.length >= R300_PFS_MAX_ALU_INST) {
 		error("Too many ALU instructions");
-		return GL_FALSE;
+		return 0;
 	}
 
 	int ip = code->alu.length++;
@@ -136,17 +147,13 @@ static GLboolean emit_alu(void* data, struct radeon_pair_instruction* inst)
 	code->alu.inst[ip].alpha_inst = translate_alpha_opcode(c, inst->Alpha.Opcode);
 
 	for(j = 0; j < 3; ++j) {
-		GLuint src = inst->RGB.Src[j].Index | (inst->RGB.Src[j].Constant << 5);
-		if (!inst->RGB.Src[j].Constant)
-			use_temporary(code, inst->RGB.Src[j].Index);
+		unsigned int src = use_source(code, inst->RGB.Src[j]);
 		code->alu.inst[ip].rgb_addr |= src << (6*j);
 
-		src = inst->Alpha.Src[j].Index | (inst->Alpha.Src[j].Constant << 5);
-		if (!inst->Alpha.Src[j].Constant)
-			use_temporary(code, inst->Alpha.Src[j].Index);
+		src = use_source(code, inst->Alpha.Src[j]);
 		code->alu.inst[ip].alpha_addr |= src << (6*j);
 
-		GLuint arg = r300FPTranslateRGBSwizzle(inst->RGB.Arg[j].Source, inst->RGB.Arg[j].Swizzle);
+		unsigned int arg = r300FPTranslateRGBSwizzle(inst->RGB.Arg[j].Source, inst->RGB.Arg[j].Swizzle);
 		arg |= inst->RGB.Arg[j].Abs << 6;
 		arg |= inst->RGB.Arg[j].Negate << 5;
 		code->alu.inst[ip].rgb_inst |= arg << (7*j);
@@ -186,27 +193,27 @@ static GLboolean emit_alu(void* data, struct radeon_pair_instruction* inst)
 	if (inst->Alpha.DepthWriteMask) {
 		code->alu.inst[ip].alpha_addr |= R300_ALU_DSTA_DEPTH;
 		emit->node_flags |= R300_W_OUT;
-		c->code->writes_depth = GL_TRUE;
+		c->code->writes_depth = 1;
 	}
 
-	return GL_TRUE;
+	return 1;
 }
 
 
 /**
  * Finish the current node without advancing to the next one.
  */
-static GLboolean finish_node(struct r300_emit_state * emit)
+static int finish_node(struct r300_emit_state * emit)
 {
 	struct r300_fragment_program_compiler * c = emit->compiler;
 	struct r300_fragment_program_code *code = &emit->compiler->code->code.r300;
 
 	if (code->alu.length == emit->node_first_alu) {
 		/* Generate a single NOP for this node */
-		struct radeon_pair_instruction inst;
-		_mesa_bzero(&inst, sizeof(inst));
+		struct rc_pair_instruction inst;
+		memset(&inst, 0, sizeof(inst));
 		if (!emit_alu(emit, &inst))
-			return GL_FALSE;
+			return 0;
 	}
 
 	unsigned alu_offset = emit->node_first_alu;
@@ -217,7 +224,7 @@ static GLboolean finish_node(struct r300_emit_state * emit)
 	if (code->tex.length == emit->node_first_tex) {
 		if (emit->current_node > 0) {
 			error("Node %i has no TEX instructions", emit->current_node);
-			return GL_FALSE;
+			return 0;
 		}
 
 		tex_end = 0;
@@ -240,7 +247,7 @@ static GLboolean finish_node(struct r300_emit_state * emit)
 			(tex_end << R300_TEX_SIZE_SHIFT) |
 			emit->node_flags;
 
-	return GL_TRUE;
+	return 1;
 }
 
 
@@ -248,79 +255,72 @@ static GLboolean finish_node(struct r300_emit_state * emit)
  * Begin a block of texture instructions.
  * Create the necessary indirection.
  */
-static GLboolean begin_tex(void* data)
+static int begin_tex(struct r300_emit_state * emit)
 {
 	PROG_CODE;
 
 	if (code->alu.length == emit->node_first_alu &&
 	    code->tex.length == emit->node_first_tex) {
-		return GL_TRUE;
+		return 1;
 	}
 
 	if (emit->current_node == 3) {
 		error("Too many texture indirections");
-		return GL_FALSE;
+		return 0;
 	}
 
 	if (!finish_node(emit))
-		return GL_FALSE;
+		return 0;
 
 	emit->current_node++;
 	emit->node_first_tex = code->tex.length;
 	emit->node_first_alu = code->alu.length;
 	emit->node_flags = 0;
-	return GL_TRUE;
+	return 1;
 }
 
 
-static GLboolean emit_tex(void* data, struct radeon_pair_texture_instruction* inst)
+static int emit_tex(struct r300_emit_state * emit, struct rc_instruction * inst)
 {
 	PROG_CODE;
 
 	if (code->tex.length >= R300_PFS_MAX_TEX_INST) {
 		error("Too many TEX instructions");
-		return GL_FALSE;
+		return 0;
 	}
 
-	GLuint unit = inst->TexSrcUnit;
-	GLuint dest = inst->DestIndex;
-	GLuint opcode;
+	unsigned int unit = inst->U.I.TexSrcUnit;
+	unsigned int dest = inst->U.I.DstReg.Index;
+	unsigned int opcode;
 
-	switch(inst->Opcode) {
-	case RADEON_OPCODE_KIL: opcode = R300_TEX_OP_KIL; break;
-	case RADEON_OPCODE_TEX: opcode = R300_TEX_OP_LD; break;
-	case RADEON_OPCODE_TXB: opcode = R300_TEX_OP_TXB; break;
-	case RADEON_OPCODE_TXP: opcode = R300_TEX_OP_TXP; break;
+	switch(inst->U.I.Opcode) {
+	case RC_OPCODE_KIL: opcode = R300_TEX_OP_KIL; break;
+	case RC_OPCODE_TEX: opcode = R300_TEX_OP_LD; break;
+	case RC_OPCODE_TXB: opcode = R300_TEX_OP_TXB; break;
+	case RC_OPCODE_TXP: opcode = R300_TEX_OP_TXP; break;
 	default:
-		error("Unknown texture opcode %i", inst->Opcode);
-		return GL_FALSE;
+		error("Unknown texture opcode %i", inst->U.I.Opcode);
+		return 0;
 	}
 
-	if (inst->Opcode == RADEON_OPCODE_KIL) {
+	if (inst->U.I.Opcode == RC_OPCODE_KIL) {
 		unit = 0;
 		dest = 0;
 	} else {
 		use_temporary(code, dest);
 	}
 
-	use_temporary(code, inst->SrcIndex);
+	use_temporary(code, inst->U.I.SrcReg[0].Index);
 
 	code->tex.inst[code->tex.length++] =
-		(inst->SrcIndex << R300_SRC_ADDR_SHIFT) |
+		(inst->U.I.SrcReg[0].Index << R300_SRC_ADDR_SHIFT) |
 		(dest << R300_DST_ADDR_SHIFT) |
 		(unit << R300_TEX_ID_SHIFT) |
 		(opcode << R300_TEX_INST_SHIFT);
-	return GL_TRUE;
+	return 1;
 }
 
 
-static const struct radeon_pair_handler pair_handler = {
-	.EmitPaired = &emit_alu,
-	.EmitTex = &emit_tex,
-	.BeginTexBlock = &begin_tex,
-	.MaxHwTemps = R300_PFS_NUM_TEMP_REGS
-};
-
 /**
  * Final compilation step: Turn the intermediate radeon_program into
  * machine-readable instructions.
@@ -333,9 +333,26 @@ void r300BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi
 	memset(&emit, 0, sizeof(emit));
 	emit.compiler = compiler;
 
-	_mesa_bzero(code, sizeof(struct r300_fragment_program_code));
+	memset(code, 0, sizeof(struct r300_fragment_program_code));
+
+	for(struct rc_instruction * inst = compiler->Base.Program.Instructions.Next;
+	    inst != &compiler->Base.Program.Instructions && !compiler->Base.Error;
+	    inst = inst->Next) {
+		if (inst->Type == RC_INSTRUCTION_NORMAL) {
+			if (inst->U.I.Opcode == RC_OPCODE_BEGIN_TEX) {
+				begin_tex(&emit);
+				continue;
+			}
+
+			emit_tex(&emit, inst);
+		} else {
+			emit_alu(&emit, &inst->U.P);
+		}
+	}
+
+	if (code->pixsize >= R300_PFS_NUM_TEMP_REGS)
+		rc_error(&compiler->Base, "Too many hardware temporaries used.\n");
 
-	radeonPairProgram(compiler, &pair_handler, &emit);
 	if (compiler->Base.Error)
 		return;
 
diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.c b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.c
index 1b14cc3888..cfa48a59e3 100644
--- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.c
+++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.c
@@ -33,16 +33,17 @@
 
 #include "r300_fragprog_swizzle.h"
 
+#include <stdio.h>
+
 #include "../r300_reg.h"
-#include "radeon_nqssadce.h"
 #include "radeon_compiler.h"
 
-#define MAKE_SWZ3(x, y, z) (MAKE_SWIZZLE4(SWIZZLE_##x, SWIZZLE_##y, SWIZZLE_##z, SWIZZLE_ZERO))
+#define MAKE_SWZ3(x, y, z) (RC_MAKE_SWIZZLE(RC_SWIZZLE_##x, RC_SWIZZLE_##y, RC_SWIZZLE_##z, RC_SWIZZLE_ZERO))
 
 struct swizzle_data {
-	GLuint hash; /**< swizzle value this matches */
-	GLuint base; /**< base value for hw swizzle */
-	GLuint stride; /**< difference in base between arg0/1/2 */
+	unsigned int hash; /**< swizzle value this matches */
+	unsigned int base; /**< base value for hw swizzle */
+	unsigned int stride; /**< difference in base between arg0/1/2 */
 };
 
 static const struct swizzle_data native_swizzles[] = {
@@ -65,15 +66,15 @@ static const int num_native_swizzles = sizeof(native_swizzles)/sizeof(native_swi
  * Find a native RGB swizzle that matches the given swizzle.
  * Returns 0 if none found.
  */
-static const struct swizzle_data* lookup_native_swizzle(GLuint swizzle)
+static const struct swizzle_data* lookup_native_swizzle(unsigned int swizzle)
 {
 	int i, comp;
 
 	for(i = 0; i < num_native_swizzles; ++i) {
 		const struct swizzle_data* sd = &native_swizzles[i];
 		for(comp = 0; comp < 3; ++comp) {
-			GLuint swz = GET_SWZ(swizzle, comp);
-			if (swz == SWIZZLE_NIL)
+			unsigned int swz = GET_SWZ(swizzle, comp);
+			if (swz == RC_SWIZZLE_UNUSED)
 				continue;
 			if (swz != GET_SWZ(sd->hash, comp))
 				break;
@@ -90,71 +91,72 @@ static const struct swizzle_data* lookup_native_swizzle(GLuint swizzle)
  * Check whether the given instruction supports the swizzle and negate
  * combinations in the given source register.
  */
-GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
+static int r300_swizzle_is_native(rc_opcode opcode, struct rc_src_register reg)
 {
 	if (reg.Abs)
-		reg.Negate = NEGATE_NONE;
+		reg.Negate = RC_MASK_NONE;
 
-	if (opcode == OPCODE_KIL ||
-	    opcode == OPCODE_TEX ||
-	    opcode == OPCODE_TXB ||
-	    opcode == OPCODE_TXP) {
+	if (opcode == RC_OPCODE_KIL ||
+	    opcode == RC_OPCODE_TEX ||
+	    opcode == RC_OPCODE_TXB ||
+	    opcode == RC_OPCODE_TXP) {
 		int j;
 
 		if (reg.Abs || reg.Negate)
-			return GL_FALSE;
+			return 0;
 
 		for(j = 0; j < 4; ++j) {
-			GLuint swz = GET_SWZ(reg.Swizzle, j);
-			if (swz == SWIZZLE_NIL)
+			unsigned int swz = GET_SWZ(reg.Swizzle, j);
+			if (swz == RC_SWIZZLE_UNUSED)
 				continue;
 			if (swz != j)
-				return GL_FALSE;
+				return 0;
 		}
 
-		return GL_TRUE;
+		return 1;
 	}
 
-	GLuint relevant = 0;
+	unsigned int relevant = 0;
 	int j;
 
 	for(j = 0; j < 3; ++j)
-		if (GET_SWZ(reg.Swizzle, j) != SWIZZLE_NIL)
+		if (GET_SWZ(reg.Swizzle, j) != RC_SWIZZLE_UNUSED)
 			relevant |= 1 << j;
 
 	if ((reg.Negate & relevant) && ((reg.Negate & relevant) != relevant))
-		return GL_FALSE;
+		return 0;
 
 	if (!lookup_native_swizzle(reg.Swizzle))
-		return GL_FALSE;
+		return 0;
 
-	return GL_TRUE;
+	return 1;
 }
 
 
-/**
- * Generate MOV dst, src using only native swizzles.
- */
-void r300FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst, struct prog_src_register src)
+static void r300_swizzle_split(
+		struct rc_src_register src, unsigned int mask,
+		struct rc_swizzle_split * split)
 {
 	if (src.Abs)
-		src.Negate = NEGATE_NONE;
+		src.Negate = RC_MASK_NONE;
+
+	split->NumPhases = 0;
 
-	while(dst.WriteMask) {
+	while(mask) {
 		const struct swizzle_data *best_swizzle = 0;
-		GLuint best_matchcount = 0;
-		GLuint best_matchmask = 0;
+		unsigned int best_matchcount = 0;
+		unsigned int best_matchmask = 0;
 		int i, comp;
 
 		for(i = 0; i < num_native_swizzles; ++i) {
 			const struct swizzle_data *sd = &native_swizzles[i];
-			GLuint matchcount = 0;
-			GLuint matchmask = 0;
+			unsigned int matchcount = 0;
+			unsigned int matchmask = 0;
 			for(comp = 0; comp < 3; ++comp) {
-				if (!GET_BIT(dst.WriteMask, comp))
+				if (!GET_BIT(mask, comp))
 					continue;
-				GLuint swz = GET_SWZ(src.Swizzle, comp);
-				if (swz == SWIZZLE_NIL)
+				unsigned int swz = GET_SWZ(src.Swizzle, comp);
+				if (swz == RC_SWIZZLE_UNUSED)
 					continue;
 				if (swz == GET_SWZ(sd->hash, comp)) {
 					/* check if the negate bit of current component
@@ -170,34 +172,35 @@ void r300FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst,
 				best_swizzle = sd;
 				best_matchcount = matchcount;
 				best_matchmask = matchmask;
-				if (matchmask == (dst.WriteMask & WRITEMASK_XYZ))
+				if (matchmask == (mask & RC_MASK_XYZ))
 					break;
 			}
 		}
 
-		struct rc_instruction *inst = rc_insert_new_instruction(s->Compiler, s->IP->Prev);
-		inst->I.Opcode = OPCODE_MOV;
-		inst->I.DstReg = dst;
-		inst->I.DstReg.WriteMask &= (best_matchmask | WRITEMASK_W);
-		inst->I.SrcReg[0] = src;
-		inst->I.SrcReg[0].Negate = (best_matchmask & src.Negate) ? NEGATE_XYZW : NEGATE_NONE;
-		/* Note: We rely on NqSSA/DCE to set unused swizzle components to NIL */
+		if (mask & RC_MASK_W)
+			best_matchmask |= RC_MASK_W;
 
-		dst.WriteMask &= ~inst->I.DstReg.WriteMask;
+		split->Phase[split->NumPhases++] = best_matchmask;
+		mask &= ~best_matchmask;
 	}
 }
 
+struct rc_swizzle_caps r300_swizzle_caps = {
+	.IsNative = r300_swizzle_is_native,
+	.Split = r300_swizzle_split
+};
+
 
 /**
  * Translate an RGB (XYZ) swizzle into the hardware code for the given
  * instruction source.
  */
-GLuint r300FPTranslateRGBSwizzle(GLuint src, GLuint swizzle)
+unsigned int r300FPTranslateRGBSwizzle(unsigned int src, unsigned int swizzle)
 {
 	const struct swizzle_data* sd = lookup_native_swizzle(swizzle);
 
 	if (!sd) {
-		_mesa_printf("Not a native swizzle: %08x\n", swizzle);
+		fprintf(stderr, "Not a native swizzle: %08x\n", swizzle);
 		return 0;
 	}
 
@@ -209,15 +212,15 @@ GLuint r300FPTranslateRGBSwizzle(GLuint src, GLuint swizzle)
  * Translate an Alpha (W) swizzle into the hardware code for the given
  * instruction source.
  */
-GLuint r300FPTranslateAlphaSwizzle(GLuint src, GLuint swizzle)
+unsigned int r300FPTranslateAlphaSwizzle(unsigned int src, unsigned int swizzle)
 {
 	if (swizzle < 3)
 		return swizzle + 3*src;
 
 	switch(swizzle) {
-	case SWIZZLE_W: return R300_ALU_ARGA_SRC0A + src;
-	case SWIZZLE_ONE: return R300_ALU_ARGA_ONE;
-	case SWIZZLE_ZERO: return R300_ALU_ARGA_ZERO;
+	case RC_SWIZZLE_W: return R300_ALU_ARGA_SRC0A + src;
+	case RC_SWIZZLE_ONE: return R300_ALU_ARGA_ONE;
+	case RC_SWIZZLE_ZERO: return R300_ALU_ARGA_ZERO;
 	default: return R300_ALU_ARGA_ONE;
 	}
 }
diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.h b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.h
index 231bf4eef5..118476af13 100644
--- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.h
+++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_swizzle.h
@@ -28,15 +28,11 @@
 #ifndef __R300_FRAGPROG_SWIZZLE_H_
 #define __R300_FRAGPROG_SWIZZLE_H_
 
-#include "main/glheader.h"
-#include "shader/prog_instruction.h"
+#include "radeon_swizzle.h"
 
-struct nqssadce_state;
+extern struct rc_swizzle_caps r300_swizzle_caps;
 
-GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg);
-void r300FPBuildSwizzle(struct nqssadce_state*, struct prog_dst_register dst, struct prog_src_register src);
-
-GLuint r300FPTranslateRGBSwizzle(GLuint src, GLuint swizzle);
-GLuint r300FPTranslateAlphaSwizzle(GLuint src, GLuint swizzle);
+unsigned int r300FPTranslateRGBSwizzle(unsigned int src, unsigned int swizzle);
+unsigned int r300FPTranslateAlphaSwizzle(unsigned int src, unsigned int swizzle);
 
 #endif /* __R300_FRAGPROG_SWIZZLE_H_ */
diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c b/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c
index 76c3a7ecfd..5581f25352 100644
--- a/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c
@@ -22,22 +22,21 @@
 
 #include "radeon_compiler.h"
 
-#include "shader/prog_parameter.h"
-#include "shader/prog_print.h"
-#include "shader/prog_statevars.h"
+#include <stdio.h>
 
-#include "radeon_nqssadce.h"
+#include "radeon_dataflow.h"
 #include "radeon_program_alu.h"
 #include "r300_fragprog.h"
 #include "r300_fragprog_swizzle.h"
 #include "r500_fragprog.h"
 
 
-static void nqssadce_init(struct nqssadce_state* s)
+static void dataflow_outputs_mark_use(void * userdata, void * data,
+		void (*callback)(void *, unsigned int, unsigned int))
 {
-	struct r300_fragment_program_compiler * c = s->UserData;
-	s->Outputs[c->OutputColor].Sourced = WRITEMASK_XYZW;
-	s->Outputs[c->OutputDepth].Sourced = WRITEMASK_W;
+	struct r300_fragment_program_compiler * c = userdata;
+	callback(data, c->OutputColor, RC_MASK_XYZW);
+	callback(data, c->OutputDepth, RC_MASK_W);
 }
 
 static void rewrite_depth_out(struct r300_fragment_program_compiler * c)
@@ -45,35 +44,35 @@ static void rewrite_depth_out(struct r300_fragment_program_compiler * c)
 	struct rc_instruction *rci;
 
 	for (rci = c->Base.Program.Instructions.Next; rci != &c->Base.Program.Instructions; rci = rci->Next) {
-		struct prog_instruction * inst = &rci->I;
+		struct rc_sub_instruction * inst = &rci->U.I;
 
-		if (inst->DstReg.File != PROGRAM_OUTPUT || inst->DstReg.Index != c->OutputDepth)
+		if (inst->DstReg.File != RC_FILE_OUTPUT || inst->DstReg.Index != c->OutputDepth)
 			continue;
 
-		if (inst->DstReg.WriteMask & WRITEMASK_Z) {
-			inst->DstReg.WriteMask = WRITEMASK_W;
+		if (inst->DstReg.WriteMask & RC_MASK_Z) {
+			inst->DstReg.WriteMask = RC_MASK_W;
 		} else {
 			inst->DstReg.WriteMask = 0;
 			continue;
 		}
 
 		switch (inst->Opcode) {
-			case OPCODE_FRC:
-			case OPCODE_MOV:
-				inst->SrcReg[0] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[0]);
+			case RC_OPCODE_FRC:
+			case RC_OPCODE_MOV:
+				inst->SrcReg[0] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[0]);
 				break;
-			case OPCODE_ADD:
-			case OPCODE_MAX:
-			case OPCODE_MIN:
-			case OPCODE_MUL:
-				inst->SrcReg[0] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[0]);
-				inst->SrcReg[1] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[1]);
+			case RC_OPCODE_ADD:
+			case RC_OPCODE_MAX:
+			case RC_OPCODE_MIN:
+			case RC_OPCODE_MUL:
+				inst->SrcReg[0] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[0]);
+				inst->SrcReg[1] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[1]);
 				break;
-			case OPCODE_CMP:
-			case OPCODE_MAD:
-				inst->SrcReg[0] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[0]);
-				inst->SrcReg[1] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[1]);
-				inst->SrcReg[2] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[2]);
+			case RC_OPCODE_CMP:
+			case RC_OPCODE_MAD:
+				inst->SrcReg[0] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[0]);
+				inst->SrcReg[1] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[1]);
+				inst->SrcReg[2] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[2]);
 				break;
 			default:
 				// Scalar instructions needn't be reswizzled
@@ -89,11 +88,14 @@ void r3xx_compile_fragment_program(struct r300_fragment_program_compiler* c)
 	if (c->is_r500) {
 		struct radeon_program_transformation transformations[] = {
 			{ &r500_transform_TEX, c },
+			{ &r500_transform_IF, 0 },
 			{ &radeonTransformALU, 0 },
 			{ &radeonTransformDeriv, 0 },
 			{ &radeonTransformTrigScale, 0 }
 		};
-		radeonLocalTransform(&c->Base, 4, transformations);
+		radeonLocalTransform(&c->Base, 5, transformations);
+
+		c->Base.SwizzleCaps = &r500_swizzle_caps;
 	} else {
 		struct radeon_program_transformation transformations[] = {
 			{ &r300_transform_TEX, c },
@@ -101,32 +103,66 @@ void r3xx_compile_fragment_program(struct r300_fragment_program_compiler* c)
 			{ &radeonTransformTrigSimple, 0 }
 		};
 		radeonLocalTransform(&c->Base, 3, transformations);
+
+		c->Base.SwizzleCaps = &r300_swizzle_caps;
 	}
 
 	if (c->Base.Debug) {
-		_mesa_printf("Fragment Program: After native rewrite:\n");
+		fprintf(stderr, "Fragment Program: After native rewrite:\n");
 		rc_print_program(&c->Base.Program);
 		fflush(stderr);
 	}
 
-	if (c->is_r500) {
-		struct radeon_nqssadce_descr nqssadce = {
-			.Init = &nqssadce_init,
-			.IsNativeSwizzle = &r500FPIsNativeSwizzle,
-			.BuildSwizzle = &r500FPBuildSwizzle
-		};
-		radeonNqssaDce(&c->Base, &nqssadce, c);
-	} else {
-		struct radeon_nqssadce_descr nqssadce = {
-			.Init = &nqssadce_init,
-			.IsNativeSwizzle = &r300FPIsNativeSwizzle,
-			.BuildSwizzle = &r300FPBuildSwizzle
-		};
-		radeonNqssaDce(&c->Base, &nqssadce, c);
+	rc_dataflow_deadcode(&c->Base, &dataflow_outputs_mark_use, c);
+	if (c->Base.Error)
+		return;
+
+	if (c->Base.Debug) {
+		fprintf(stderr, "Fragment Program: After deadcode:\n");
+		rc_print_program(&c->Base.Program);
+		fflush(stderr);
+	}
+
+	rc_dataflow_swizzles(&c->Base);
+	if (c->Base.Error)
+		return;
+
+	if (c->Base.Debug) {
+		fprintf(stderr, "Compiler: after dataflow passes:\n");
+		rc_print_program(&c->Base.Program);
+		fflush(stderr);
+	}
+
+	rc_pair_translate(c);
+	if (c->Base.Error)
+		return;
+
+	if (c->Base.Debug) {
+		fprintf(stderr, "Compiler: after pair translate:\n");
+		rc_print_program(&c->Base.Program);
+		fflush(stderr);
 	}
 
+	rc_pair_schedule(c);
+	if (c->Base.Error)
+		return;
+
+	if (c->Base.Debug) {
+		fprintf(stderr, "Compiler: after pair scheduling:\n");
+		rc_print_program(&c->Base.Program);
+		fflush(stderr);
+	}
+
+	if (c->is_r500)
+		rc_pair_regalloc(c, 128);
+	else
+		rc_pair_regalloc(c, R300_PFS_NUM_TEMP_REGS);
+
+	if (c->Base.Error)
+		return;
+
 	if (c->Base.Debug) {
-		_mesa_printf("Compiler: after NqSSA-DCE:\n");
+		fprintf(stderr, "Compiler: after pair register allocation:\n");
 		rc_print_program(&c->Base.Program);
 		fflush(stderr);
 	}
diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c
index dad27fc98e..1b2cb8dde7 100644
--- a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c
@@ -22,13 +22,13 @@
 
 #include "radeon_compiler.h"
 
+#include <stdio.h>
+
 #include "../r300_reg.h"
 
-#include "radeon_nqssadce.h"
-#include "radeon_program.h"
+#include "radeon_dataflow.h"
 #include "radeon_program_alu.h"
-
-#include "shader/prog_print.h"
+#include "radeon_swizzle.h"
 
 
 /*
@@ -42,104 +42,83 @@
 			   t_swizzle(y),	\
 			   t_swizzle(y),	\
 			   t_src_class(vpi->SrcReg[x].File), \
-			   NEGATE_NONE) | (vpi->SrcReg[x].RelAddr << 4))
+			   RC_MASK_NONE) | (vpi->SrcReg[x].RelAddr << 4))
 
 
-static unsigned long t_dst_mask(GLuint mask)
+static unsigned long t_dst_mask(unsigned int mask)
 {
-	/* WRITEMASK_* is equivalent to VSF_FLAG_* */
-	return mask & WRITEMASK_XYZW;
+	/* RC_MASK_* is equivalent to VSF_FLAG_* */
+	return mask & RC_MASK_XYZW;
 }
 
-static unsigned long t_dst_class(gl_register_file file)
+static unsigned long t_dst_class(rc_register_file file)
 {
-
 	switch (file) {
-	case PROGRAM_TEMPORARY:
+	default:
+		fprintf(stderr, "%s: Bad register file %i\n", __FUNCTION__, file);
+		/* fall-through */
+	case RC_FILE_TEMPORARY:
 		return PVS_DST_REG_TEMPORARY;
-	case PROGRAM_OUTPUT:
+	case RC_FILE_OUTPUT:
 		return PVS_DST_REG_OUT;
-	case PROGRAM_ADDRESS:
+	case RC_FILE_ADDRESS:
 		return PVS_DST_REG_A0;
-		/*
-		   case PROGRAM_INPUT:
-		   case PROGRAM_LOCAL_PARAM:
-		   case PROGRAM_ENV_PARAM:
-		   case PROGRAM_NAMED_PARAM:
-		   case PROGRAM_STATE_VAR:
-		   case PROGRAM_WRITE_ONLY:
-		   case PROGRAM_ADDRESS:
-		 */
-	default:
-		fprintf(stderr, "problem in %s", __FUNCTION__);
-		_mesa_exit(-1);
-		return -1;
 	}
 }
 
 static unsigned long t_dst_index(struct r300_vertex_program_code *vp,
-				 struct prog_dst_register *dst)
+				 struct rc_dst_register *dst)
 {
-	if (dst->File == PROGRAM_OUTPUT)
+	if (dst->File == RC_FILE_OUTPUT)
 		return vp->outputs[dst->Index];
 
 	return dst->Index;
 }
 
-static unsigned long t_src_class(gl_register_file file)
+static unsigned long t_src_class(rc_register_file file)
 {
 	switch (file) {
-	case PROGRAM_BUILTIN:
-	case PROGRAM_TEMPORARY:
+	default:
+		fprintf(stderr, "%s: Bad register file %i\n", __FUNCTION__, file);
+		/* fall-through */
+	case RC_FILE_NONE:
+	case RC_FILE_TEMPORARY:
 		return PVS_SRC_REG_TEMPORARY;
-	case PROGRAM_INPUT:
+	case RC_FILE_INPUT:
 		return PVS_SRC_REG_INPUT;
-	case PROGRAM_LOCAL_PARAM:
-	case PROGRAM_ENV_PARAM:
-	case PROGRAM_NAMED_PARAM:
-	case PROGRAM_CONSTANT:
-	case PROGRAM_STATE_VAR:
+	case RC_FILE_CONSTANT:
 		return PVS_SRC_REG_CONSTANT;
-		/*
-		   case PROGRAM_OUTPUT:
-		   case PROGRAM_WRITE_ONLY:
-		   case PROGRAM_ADDRESS:
-		 */
-	default:
-		fprintf(stderr, "problem in %s", __FUNCTION__);
-		_mesa_exit(-1);
-		return -1;
 	}
 }
 
-static GLboolean t_src_conflict(struct prog_src_register a, struct prog_src_register b)
+static int t_src_conflict(struct rc_src_register a, struct rc_src_register b)
 {
 	unsigned long aclass = t_src_class(a.File);
 	unsigned long bclass = t_src_class(b.File);
 
 	if (aclass != bclass)
-		return GL_FALSE;
+		return 0;
 	if (aclass == PVS_SRC_REG_TEMPORARY)
-		return GL_FALSE;
+		return 0;
 
 	if (a.RelAddr || b.RelAddr)
-		return GL_TRUE;
+		return 1;
 	if (a.Index != b.Index)
-		return GL_TRUE;
+		return 1;
 
-	return GL_FALSE;
+	return 0;
 }
 
-static INLINE unsigned long t_swizzle(GLubyte swizzle)
+static inline unsigned long t_swizzle(unsigned int swizzle)
 {
-	/* this is in fact a NOP as the Mesa SWIZZLE_* are all identical to VSF_IN_COMPONENT_* */
+	/* this is in fact a NOP as the Mesa RC_SWIZZLE_* are all identical to VSF_IN_COMPONENT_* */
 	return swizzle;
 }
 
 static unsigned long t_src_index(struct r300_vertex_program_code *vp,
-				 struct prog_src_register *src)
+				 struct rc_src_register *src)
 {
-	if (src->File == PROGRAM_INPUT) {
+	if (src->File == RC_FILE_INPUT) {
 		assert(vp->inputs[src->Index] != -1);
 		return vp->inputs[src->Index];
 	} else {
@@ -155,9 +134,9 @@ static unsigned long t_src_index(struct r300_vertex_program_code *vp,
 /* these two functions should probably be merged... */
 
 static unsigned long t_src(struct r300_vertex_program_code *vp,
-			   struct prog_src_register *src)
+			   struct rc_src_register *src)
 {
-	/* src->Negate uses the NEGATE_ flags from program_instruction.h,
+	/* src->Negate uses the RC_MASK_ flags from program_instruction.h,
 	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
 	 */
 	return PVS_SRC_OPERAND(t_src_index(vp, src),
@@ -170,9 +149,9 @@ static unsigned long t_src(struct r300_vertex_program_code *vp,
 }
 
 static unsigned long t_src_scalar(struct r300_vertex_program_code *vp,
-				  struct prog_src_register *src)
+				  struct rc_src_register *src)
 {
-	/* src->Negate uses the NEGATE_ flags from program_instruction.h,
+	/* src->Negate uses the RC_MASK_ flags from program_instruction.h,
 	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
 	 */
 	return PVS_SRC_OPERAND(t_src_index(vp, src),
@@ -181,79 +160,79 @@ static unsigned long t_src_scalar(struct r300_vertex_program_code *vp,
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_src_class(src->File),
-			       src->Negate ? NEGATE_XYZW : NEGATE_NONE) |
+			       src->Negate ? RC_MASK_XYZW : RC_MASK_NONE) |
 	    (src->RelAddr << 4);
 }
 
-static GLboolean valid_dst(struct r300_vertex_program_code *vp,
-			   struct prog_dst_register *dst)
+static int valid_dst(struct r300_vertex_program_code *vp,
+			   struct rc_dst_register *dst)
 {
-	if (dst->File == PROGRAM_OUTPUT && vp->outputs[dst->Index] == -1) {
-		return GL_FALSE;
-	} else if (dst->File == PROGRAM_ADDRESS) {
+	if (dst->File == RC_FILE_OUTPUT && vp->outputs[dst->Index] == -1) {
+		return 0;
+	} else if (dst->File == RC_FILE_ADDRESS) {
 		assert(dst->Index == 0);
 	}
 
-	return GL_TRUE;
+	return 1;
 }
 
 static void ei_vector1(struct r300_vertex_program_code *vp,
-				GLuint hw_opcode,
-				struct prog_instruction *vpi,
-				GLuint * inst)
+				unsigned int hw_opcode,
+				struct rc_sub_instruction *vpi,
+				unsigned int * inst)
 {
 	inst[0] = PVS_OP_DST_OPERAND(hw_opcode,
-				     GL_FALSE,
-				     GL_FALSE,
+				     0,
+				     0,
 				     t_dst_index(vp, &vpi->DstReg),
 				     t_dst_mask(vpi->DstReg.WriteMask),
 				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &vpi->SrcReg[0]);
-	inst[2] = __CONST(0, SWIZZLE_ZERO);
-	inst[3] = __CONST(0, SWIZZLE_ZERO);
+	inst[2] = __CONST(0, RC_SWIZZLE_ZERO);
+	inst[3] = __CONST(0, RC_SWIZZLE_ZERO);
 }
 
 static void ei_vector2(struct r300_vertex_program_code *vp,
-				GLuint hw_opcode,
-				struct prog_instruction *vpi,
-				GLuint * inst)
+				unsigned int hw_opcode,
+				struct rc_sub_instruction *vpi,
+				unsigned int * inst)
 {
 	inst[0] = PVS_OP_DST_OPERAND(hw_opcode,
-				     GL_FALSE,
-				     GL_FALSE,
+				     0,
+				     0,
 				     t_dst_index(vp, &vpi->DstReg),
 				     t_dst_mask(vpi->DstReg.WriteMask),
 				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &vpi->SrcReg[0]);
 	inst[2] = t_src(vp, &vpi->SrcReg[1]);
-	inst[3] = __CONST(1, SWIZZLE_ZERO);
+	inst[3] = __CONST(1, RC_SWIZZLE_ZERO);
 }
 
 static void ei_math1(struct r300_vertex_program_code *vp,
-				GLuint hw_opcode,
-				struct prog_instruction *vpi,
-				GLuint * inst)
+				unsigned int hw_opcode,
+				struct rc_sub_instruction *vpi,
+				unsigned int * inst)
 {
 	inst[0] = PVS_OP_DST_OPERAND(hw_opcode,
-				     GL_TRUE,
-				     GL_FALSE,
+				     1,
+				     0,
 				     t_dst_index(vp, &vpi->DstReg),
 				     t_dst_mask(vpi->DstReg.WriteMask),
 				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &vpi->SrcReg[0]);
-	inst[2] = __CONST(0, SWIZZLE_ZERO);
-	inst[3] = __CONST(0, SWIZZLE_ZERO);
+	inst[2] = __CONST(0, RC_SWIZZLE_ZERO);
+	inst[3] = __CONST(0, RC_SWIZZLE_ZERO);
 }
 
 static void ei_lit(struct r300_vertex_program_code *vp,
-				      struct prog_instruction *vpi,
-				      GLuint * inst)
+				      struct rc_sub_instruction *vpi,
+				      unsigned int * inst)
 {
 	//LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W}
 
 	inst[0] = PVS_OP_DST_OPERAND(ME_LIGHT_COEFF_DX,
-				     GL_TRUE,
-				     GL_FALSE,
+				     1,
+				     0,
 				     t_dst_index(vp, &vpi->DstReg),
 				     t_dst_mask(vpi->DstReg.WriteMask),
 				     t_dst_class(vpi->DstReg.File));
@@ -263,27 +242,27 @@ static void ei_lit(struct r300_vertex_program_code *vp,
 				  PVS_SRC_SELECT_FORCE_0,	// Z
 				  t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 1)),	// Y
 				  t_src_class(vpi->SrcReg[0].File),
-				  vpi->SrcReg[0].Negate ? NEGATE_XYZW : NEGATE_NONE) |
+				  vpi->SrcReg[0].Negate ? RC_MASK_XYZW : RC_MASK_NONE) |
 	    (vpi->SrcReg[0].RelAddr << 4);
 	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &vpi->SrcReg[0]), t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 1)),	// Y
 				  t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 3)),	// W
 				  PVS_SRC_SELECT_FORCE_0,	// Z
 				  t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 0)),	// X
 				  t_src_class(vpi->SrcReg[0].File),
-				  vpi->SrcReg[0].Negate ? NEGATE_XYZW : NEGATE_NONE) |
+				  vpi->SrcReg[0].Negate ? RC_MASK_XYZW : RC_MASK_NONE) |
 	    (vpi->SrcReg[0].RelAddr << 4);
 	inst[3] = PVS_SRC_OPERAND(t_src_index(vp, &vpi->SrcReg[0]), t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 1)),	// Y
 				  t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 0)),	// X
 				  PVS_SRC_SELECT_FORCE_0,	// Z
 				  t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 3)),	// W
 				  t_src_class(vpi->SrcReg[0].File),
-				  vpi->SrcReg[0].Negate ? NEGATE_XYZW : NEGATE_NONE) |
+				  vpi->SrcReg[0].Negate ? RC_MASK_XYZW : RC_MASK_NONE) |
 	    (vpi->SrcReg[0].RelAddr << 4);
 }
 
 static void ei_mad(struct r300_vertex_program_code *vp,
-				      struct prog_instruction *vpi,
-				      GLuint * inst)
+				      struct rc_sub_instruction *vpi,
+				      unsigned int * inst)
 {
 	/* Remarks about hardware limitations of MAD
 	 * (please preserve this comment, as this information is _NOT_
@@ -311,22 +290,22 @@ static void ei_mad(struct r300_vertex_program_code *vp,
 	 * according to AMD docs, this should improve performance by one clock
 	 * as a nice side bonus.
 	 */
-	if (vpi->SrcReg[0].File == PROGRAM_TEMPORARY &&
-	    vpi->SrcReg[1].File == PROGRAM_TEMPORARY &&
-	    vpi->SrcReg[2].File == PROGRAM_TEMPORARY &&
+	if (vpi->SrcReg[0].File == RC_FILE_TEMPORARY &&
+	    vpi->SrcReg[1].File == RC_FILE_TEMPORARY &&
+	    vpi->SrcReg[2].File == RC_FILE_TEMPORARY &&
 	    vpi->SrcReg[0].Index != vpi->SrcReg[1].Index &&
 	    vpi->SrcReg[0].Index != vpi->SrcReg[2].Index &&
 	    vpi->SrcReg[1].Index != vpi->SrcReg[2].Index) {
 		inst[0] = PVS_OP_DST_OPERAND(PVS_MACRO_OP_2CLK_MADD,
-				GL_FALSE,
-				GL_TRUE,
+				0,
+				1,
 				t_dst_index(vp, &vpi->DstReg),
 				t_dst_mask(vpi->DstReg.WriteMask),
 				t_dst_class(vpi->DstReg.File));
 	} else {
 		inst[0] = PVS_OP_DST_OPERAND(VE_MULTIPLY_ADD,
-				GL_FALSE,
-				GL_FALSE,
+				0,
+				0,
 				t_dst_index(vp, &vpi->DstReg),
 				t_dst_mask(vpi->DstReg.WriteMask),
 				t_dst_class(vpi->DstReg.File));
@@ -337,17 +316,17 @@ static void ei_mad(struct r300_vertex_program_code *vp,
 }
 
 static void ei_pow(struct r300_vertex_program_code *vp,
-				      struct prog_instruction *vpi,
-				      GLuint * inst)
+				      struct rc_sub_instruction *vpi,
+				      unsigned int * inst)
 {
 	inst[0] = PVS_OP_DST_OPERAND(ME_POWER_FUNC_FF,
-				     GL_TRUE,
-				     GL_FALSE,
+				     1,
+				     0,
 				     t_dst_index(vp, &vpi->DstReg),
 				     t_dst_mask(vpi->DstReg.WriteMask),
 				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &vpi->SrcReg[0]);
-	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[2] = __CONST(0, RC_SWIZZLE_ZERO);
 	inst[3] = t_src_scalar(vp, &vpi->SrcReg[1]);
 }
 
@@ -362,8 +341,8 @@ static void translate_vertex_program(struct r300_vertex_program_compiler * compi
 	compiler->SetHwInputOutput(compiler);
 
 	for(rci = compiler->Base.Program.Instructions.Next; rci != &compiler->Base.Program.Instructions; rci = rci->Next) {
-		struct prog_instruction *vpi = &rci->I;
-		GLuint *inst = compiler->code->body.d + compiler->code->length;
+		struct rc_sub_instruction *vpi = &rci->U.I;
+		unsigned int *inst = compiler->code->body.d + compiler->code->length;
 
 		/* Skip instructions writing to non-existing destination */
 		if (!valid_dst(compiler->code, &vpi->DstReg))
@@ -375,26 +354,26 @@ static void translate_vertex_program(struct r300_vertex_program_compiler * compi
 		}
 
 		switch (vpi->Opcode) {
-		case OPCODE_ADD: ei_vector2(compiler->code, VE_ADD, vpi, inst); break;
-		case OPCODE_ARL: ei_vector1(compiler->code, VE_FLT2FIX_DX, vpi, inst); break;
-		case OPCODE_DP4: ei_vector2(compiler->code, VE_DOT_PRODUCT, vpi, inst); break;
-		case OPCODE_DST: ei_vector2(compiler->code, VE_DISTANCE_VECTOR, vpi, inst); break;
-		case OPCODE_EX2: ei_math1(compiler->code, ME_EXP_BASE2_FULL_DX, vpi, inst); break;
-		case OPCODE_EXP: ei_math1(compiler->code, ME_EXP_BASE2_DX, vpi, inst); break;
-		case OPCODE_FRC: ei_vector1(compiler->code, VE_FRACTION, vpi, inst); break;
-		case OPCODE_LG2: ei_math1(compiler->code, ME_LOG_BASE2_FULL_DX, vpi, inst); break;
-		case OPCODE_LIT: ei_lit(compiler->code, vpi, inst); break;
-		case OPCODE_LOG: ei_math1(compiler->code, ME_LOG_BASE2_DX, vpi, inst); break;
-		case OPCODE_MAD: ei_mad(compiler->code, vpi, inst); break;
-		case OPCODE_MAX: ei_vector2(compiler->code, VE_MAXIMUM, vpi, inst); break;
-		case OPCODE_MIN: ei_vector2(compiler->code, VE_MINIMUM, vpi, inst); break;
-		case OPCODE_MOV: ei_vector1(compiler->code, VE_ADD, vpi, inst); break;
-		case OPCODE_MUL: ei_vector2(compiler->code, VE_MULTIPLY, vpi, inst); break;
-		case OPCODE_POW: ei_pow(compiler->code, vpi, inst); break;
-		case OPCODE_RCP: ei_math1(compiler->code, ME_RECIP_DX, vpi, inst); break;
-		case OPCODE_RSQ: ei_math1(compiler->code, ME_RECIP_SQRT_DX, vpi, inst); break;
-		case OPCODE_SGE: ei_vector2(compiler->code, VE_SET_GREATER_THAN_EQUAL, vpi, inst); break;
-		case OPCODE_SLT: ei_vector2(compiler->code, VE_SET_LESS_THAN, vpi, inst); break;
+		case RC_OPCODE_ADD: ei_vector2(compiler->code, VE_ADD, vpi, inst); break;
+		case RC_OPCODE_ARL: ei_vector1(compiler->code, VE_FLT2FIX_DX, vpi, inst); break;
+		case RC_OPCODE_DP4: ei_vector2(compiler->code, VE_DOT_PRODUCT, vpi, inst); break;
+		case RC_OPCODE_DST: ei_vector2(compiler->code, VE_DISTANCE_VECTOR, vpi, inst); break;
+		case RC_OPCODE_EX2: ei_math1(compiler->code, ME_EXP_BASE2_FULL_DX, vpi, inst); break;
+		case RC_OPCODE_EXP: ei_math1(compiler->code, ME_EXP_BASE2_DX, vpi, inst); break;
+		case RC_OPCODE_FRC: ei_vector1(compiler->code, VE_FRACTION, vpi, inst); break;
+		case RC_OPCODE_LG2: ei_math1(compiler->code, ME_LOG_BASE2_FULL_DX, vpi, inst); break;
+		case RC_OPCODE_LIT: ei_lit(compiler->code, vpi, inst); break;
+		case RC_OPCODE_LOG: ei_math1(compiler->code, ME_LOG_BASE2_DX, vpi, inst); break;
+		case RC_OPCODE_MAD: ei_mad(compiler->code, vpi, inst); break;
+		case RC_OPCODE_MAX: ei_vector2(compiler->code, VE_MAXIMUM, vpi, inst); break;
+		case RC_OPCODE_MIN: ei_vector2(compiler->code, VE_MINIMUM, vpi, inst); break;
+		case RC_OPCODE_MOV: ei_vector1(compiler->code, VE_ADD, vpi, inst); break;
+		case RC_OPCODE_MUL: ei_vector2(compiler->code, VE_MULTIPLY, vpi, inst); break;
+		case RC_OPCODE_POW: ei_pow(compiler->code, vpi, inst); break;
+		case RC_OPCODE_RCP: ei_math1(compiler->code, ME_RECIP_DX, vpi, inst); break;
+		case RC_OPCODE_RSQ: ei_math1(compiler->code, ME_RECIP_SQRT_DX, vpi, inst); break;
+		case RC_OPCODE_SGE: ei_vector2(compiler->code, VE_SET_GREATER_THAN_EQUAL, vpi, inst); break;
+		case RC_OPCODE_SLT: ei_vector2(compiler->code, VE_SET_LESS_THAN, vpi, inst); break;
 		default:
 			rc_error(&compiler->Base, "Unknown opcode %i\n", vpi->Opcode);
 			return;
@@ -408,38 +387,37 @@ static void translate_vertex_program(struct r300_vertex_program_compiler * compi
 }
 
 struct temporary_allocation {
-	GLuint Allocated:1;
-	GLuint HwTemp:15;
+	unsigned int Allocated:1;
+	unsigned int HwTemp:15;
 	struct rc_instruction * LastRead;
 };
 
 static void allocate_temporary_registers(struct r300_vertex_program_compiler * compiler)
 {
 	struct rc_instruction *inst;
-	GLuint num_orig_temps = 0;
-	GLboolean hwtemps[VSF_MAX_FRAGMENT_TEMPS];
+	unsigned int num_orig_temps = 0;
+	char hwtemps[VSF_MAX_FRAGMENT_TEMPS];
 	struct temporary_allocation * ta;
-	GLuint i, j;
+	unsigned int i, j;
 
 	compiler->code->num_temporaries = 0;
 	memset(hwtemps, 0, sizeof(hwtemps));
 
 	/* Pass 1: Count original temporaries and allocate structures */
 	for(inst = compiler->Base.Program.Instructions.Next; inst != &compiler->Base.Program.Instructions; inst = inst->Next) {
-		GLuint numsrcs = _mesa_num_inst_src_regs(inst->I.Opcode);
-		GLuint numdsts = _mesa_num_inst_dst_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 
-		for (i = 0; i < numsrcs; ++i) {
-			if (inst->I.SrcReg[i].File == PROGRAM_TEMPORARY) {
-				if (inst->I.SrcReg[i].Index >= num_orig_temps)
-					num_orig_temps = inst->I.SrcReg[i].Index + 1;
+		for (i = 0; i < opcode->NumSrcRegs; ++i) {
+			if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY) {
+				if (inst->U.I.SrcReg[i].Index >= num_orig_temps)
+					num_orig_temps = inst->U.I.SrcReg[i].Index + 1;
 			}
 		}
 
-		if (numdsts) {
-			if (inst->I.DstReg.File == PROGRAM_TEMPORARY) {
-				if (inst->I.DstReg.Index >= num_orig_temps)
-					num_orig_temps = inst->I.DstReg.Index + 1;
+		if (opcode->HasDstReg) {
+			if (inst->U.I.DstReg.File == RC_FILE_TEMPORARY) {
+				if (inst->U.I.DstReg.Index >= num_orig_temps)
+					num_orig_temps = inst->U.I.DstReg.Index + 1;
 			}
 		}
 	}
@@ -450,32 +428,31 @@ static void allocate_temporary_registers(struct r300_vertex_program_compiler * c
 
 	/* Pass 2: Determine original temporary lifetimes */
 	for(inst = compiler->Base.Program.Instructions.Next; inst != &compiler->Base.Program.Instructions; inst = inst->Next) {
-		GLuint numsrcs = _mesa_num_inst_src_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 
-		for (i = 0; i < numsrcs; ++i) {
-			if (inst->I.SrcReg[i].File == PROGRAM_TEMPORARY)
-				ta[inst->I.SrcReg[i].Index].LastRead = inst;
+		for (i = 0; i < opcode->NumSrcRegs; ++i) {
+			if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY)
+				ta[inst->U.I.SrcReg[i].Index].LastRead = inst;
 		}
 	}
 
 	/* Pass 3: Register allocation */
 	for(inst = compiler->Base.Program.Instructions.Next; inst != &compiler->Base.Program.Instructions; inst = inst->Next) {
-		GLuint numsrcs = _mesa_num_inst_src_regs(inst->I.Opcode);
-		GLuint numdsts = _mesa_num_inst_dst_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 
-		for (i = 0; i < numsrcs; ++i) {
-			if (inst->I.SrcReg[i].File == PROGRAM_TEMPORARY) {
-				GLuint orig = inst->I.SrcReg[i].Index;
-				inst->I.SrcReg[i].Index = ta[orig].HwTemp;
+		for (i = 0; i < opcode->NumSrcRegs; ++i) {
+			if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY) {
+				unsigned int orig = inst->U.I.SrcReg[i].Index;
+				inst->U.I.SrcReg[i].Index = ta[orig].HwTemp;
 
 				if (ta[orig].Allocated && inst == ta[orig].LastRead)
-					hwtemps[ta[orig].HwTemp] = GL_FALSE;
+					hwtemps[ta[orig].HwTemp] = 0;
 			}
 		}
 
-		if (numdsts) {
-			if (inst->I.DstReg.File == PROGRAM_TEMPORARY) {
-				GLuint orig = inst->I.DstReg.Index;
+		if (opcode->HasDstReg) {
+			if (inst->U.I.DstReg.File == RC_FILE_TEMPORARY) {
+				unsigned int orig = inst->U.I.DstReg.Index;
 
 				if (!ta[orig].Allocated) {
 					for(j = 0; j < VSF_MAX_FRAGMENT_TEMPS; ++j) {
@@ -485,16 +462,16 @@ static void allocate_temporary_registers(struct r300_vertex_program_compiler * c
 					if (j >= VSF_MAX_FRAGMENT_TEMPS) {
 						fprintf(stderr, "Out of hw temporaries\n");
 					} else {
-						ta[orig].Allocated = GL_TRUE;
+						ta[orig].Allocated = 1;
 						ta[orig].HwTemp = j;
-						hwtemps[j] = GL_TRUE;
+						hwtemps[j] = 1;
 
 						if (j >= compiler->code->num_temporaries)
 							compiler->code->num_temporaries = j + 1;
 					}
 				}
 
-				inst->I.DstReg.Index = ta[orig].HwTemp;
+				inst->U.I.DstReg.Index = ta[orig].HwTemp;
 			}
 		}
 	}
@@ -505,45 +482,45 @@ static void allocate_temporary_registers(struct r300_vertex_program_compiler * c
  * Vertex engine cannot read two inputs or two constants at the same time.
  * Introduce intermediate MOVs to temporary registers to account for this.
  */
-static GLboolean transform_source_conflicts(
+static int transform_source_conflicts(
 	struct radeon_compiler *c,
 	struct rc_instruction* inst,
 	void* unused)
 {
-	GLuint num_operands = _mesa_num_inst_src_regs(inst->I.Opcode);
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 
-	if (num_operands == 3) {
-		if (t_src_conflict(inst->I.SrcReg[1], inst->I.SrcReg[2])
-		    || t_src_conflict(inst->I.SrcReg[0], inst->I.SrcReg[2])) {
+	if (opcode->NumSrcRegs == 3) {
+		if (t_src_conflict(inst->U.I.SrcReg[1], inst->U.I.SrcReg[2])
+		    || t_src_conflict(inst->U.I.SrcReg[0], inst->U.I.SrcReg[2])) {
 			int tmpreg = rc_find_free_temporary(c);
 			struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev);
-			inst_mov->I.Opcode = OPCODE_MOV;
-			inst_mov->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst_mov->I.DstReg.Index = tmpreg;
-			inst_mov->I.SrcReg[0] = inst->I.SrcReg[2];
-
-			reset_srcreg(&inst->I.SrcReg[2]);
-			inst->I.SrcReg[2].File = PROGRAM_TEMPORARY;
-			inst->I.SrcReg[2].Index = tmpreg;
+			inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+			inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst_mov->U.I.DstReg.Index = tmpreg;
+			inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[2];
+
+			reset_srcreg(&inst->U.I.SrcReg[2]);
+			inst->U.I.SrcReg[2].File = RC_FILE_TEMPORARY;
+			inst->U.I.SrcReg[2].Index = tmpreg;
 		}
 	}
 
-	if (num_operands >= 2) {
-		if (t_src_conflict(inst->I.SrcReg[1], inst->I.SrcReg[0])) {
+	if (opcode->NumSrcRegs >= 2) {
+		if (t_src_conflict(inst->U.I.SrcReg[1], inst->U.I.SrcReg[0])) {
 			int tmpreg = rc_find_free_temporary(c);
 			struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev);
-			inst_mov->I.Opcode = OPCODE_MOV;
-			inst_mov->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst_mov->I.DstReg.Index = tmpreg;
-			inst_mov->I.SrcReg[0] = inst->I.SrcReg[1];
-
-			reset_srcreg(&inst->I.SrcReg[1]);
-			inst->I.SrcReg[1].File = PROGRAM_TEMPORARY;
-			inst->I.SrcReg[1].Index = tmpreg;
+			inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+			inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst_mov->U.I.DstReg.Index = tmpreg;
+			inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[1];
+
+			reset_srcreg(&inst->U.I.SrcReg[1]);
+			inst->U.I.SrcReg[1].File = RC_FILE_TEMPORARY;
+			inst->U.I.SrcReg[1].Index = tmpreg;
 		}
 	}
 
-	return GL_TRUE;
+	return 1;
 }
 
 static void addArtificialOutputs(struct r300_vertex_program_compiler * compiler)
@@ -554,44 +531,52 @@ static void addArtificialOutputs(struct r300_vertex_program_compiler * compiler)
 		if ((compiler->RequiredOutputs & (1 << i)) &&
 		    !(compiler->Base.Program.OutputsWritten & (1 << i))) {
 			struct rc_instruction * inst = rc_insert_new_instruction(&compiler->Base, compiler->Base.Program.Instructions.Prev);
-			inst->I.Opcode = OPCODE_MOV;
+			inst->U.I.Opcode = RC_OPCODE_MOV;
 
-			inst->I.DstReg.File = PROGRAM_OUTPUT;
-			inst->I.DstReg.Index = i;
-			inst->I.DstReg.WriteMask = WRITEMASK_XYZW;
+			inst->U.I.DstReg.File = RC_FILE_OUTPUT;
+			inst->U.I.DstReg.Index = i;
+			inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
 
-			inst->I.SrcReg[0].File = PROGRAM_CONSTANT;
-			inst->I.SrcReg[0].Index = 0;
-			inst->I.SrcReg[0].Swizzle = SWIZZLE_XYZW;
+			inst->U.I.SrcReg[0].File = RC_FILE_CONSTANT;
+			inst->U.I.SrcReg[0].Index = 0;
+			inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
 
 			compiler->Base.Program.OutputsWritten |= 1 << i;
 		}
 	}
 }
 
-static void nqssadceInit(struct nqssadce_state* s)
+static void dataflow_outputs_mark_used(void * userdata, void * data,
+		void (*callback)(void *, unsigned int, unsigned int))
 {
-	struct r300_vertex_program_compiler * compiler = s->UserData;
+	struct r300_vertex_program_compiler * c = userdata;
 	int i;
 
-	for(i = 0; i < VERT_RESULT_MAX; ++i) {
-		if (compiler->RequiredOutputs & (1 << i))
-			s->Outputs[i].Sourced = WRITEMASK_XYZW;
+	for(i = 0; i < 32; ++i) {
+		if (c->RequiredOutputs & (1 << i))
+			callback(data, i, RC_MASK_XYZW);
 	}
 }
 
-static GLboolean swizzleIsNative(GLuint opcode, struct prog_src_register reg)
+static int swizzle_is_native(rc_opcode opcode, struct rc_src_register reg)
 {
 	(void) opcode;
 	(void) reg;
 
-	return GL_TRUE;
+	return 1;
 }
 
 
+static struct rc_swizzle_caps r300_vertprog_swizzle_caps = {
+	.IsNative = &swizzle_is_native,
+	.Split = 0 /* should never be called */
+};
+
 
 void r3xx_compile_vertex_program(struct r300_vertex_program_compiler* compiler)
 {
+	compiler->Base.SwizzleCaps = &r300_vertprog_swizzle_caps;
+
 	addArtificialOutputs(compiler);
 
 	{
@@ -624,22 +609,22 @@ void r3xx_compile_vertex_program(struct r300_vertex_program_compiler* compiler)
 		fflush(stderr);
 	}
 
-	{
-		struct radeon_nqssadce_descr nqssadce = {
-			.Init = &nqssadceInit,
-			.IsNativeSwizzle = &swizzleIsNative,
-			.BuildSwizzle = NULL
-		};
-		radeonNqssaDce(&compiler->Base, &nqssadce, compiler);
+	rc_dataflow_deadcode(&compiler->Base, &dataflow_outputs_mark_used, compiler);
+
+	if (compiler->Base.Debug) {
+		fprintf(stderr, "Vertex program after deadcode:\n");
+		rc_print_program(&compiler->Base.Program);
+		fflush(stderr);
+	}
 
-		/* We need this step for reusing temporary registers */
-		allocate_temporary_registers(compiler);
+	rc_dataflow_swizzles(&compiler->Base);
 
-		if (compiler->Base.Debug) {
-			fprintf(stderr, "Vertex program after NQSSADCE:\n");
-			rc_print_program(&compiler->Base.Program);
-			fflush(stderr);
-		}
+	allocate_temporary_registers(compiler);
+
+	if (compiler->Base.Debug) {
+		fprintf(stderr, "Vertex program after dataflow:\n");
+		rc_print_program(&compiler->Base.Program);
+		fflush(stderr);
 	}
 
 	translate_vertex_program(compiler);
diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog_dump.c b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog_dump.c
index 980ef3eaea..66f9b0529f 100644
--- a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog_dump.c
+++ b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog_dump.c
@@ -146,7 +146,7 @@ static void r300_vs_op_dump(uint32_t op)
 static void r300_vs_src_dump(uint32_t src)
 {
 	fprintf(stderr, " reg: %d%s swiz: %s%s/%s%s/%s%s/%s%s\n",
-			(src >> 5) & 0x7f, r300_vs_src_debug[src & 0x3],
+			(src >> 5) & 0xff, r300_vs_src_debug[src & 0x3],
 			src & (1 << 25) ? "-" : " ",
 			r300_vs_swiz_debug[(src >> 13) & 0x7],
 			src & (1 << 26) ? "-" : " ",
diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c
index 7e2faed690..d87acecdab 100644
--- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c
@@ -27,15 +27,17 @@
 
 #include "r500_fragprog.h"
 
+#include <stdio.h>
+
 #include "../r300_reg.h"
 
-static struct prog_src_register shadow_ambient(struct radeon_compiler * c, int tmu)
+static struct rc_src_register shadow_ambient(struct radeon_compiler * c, int tmu)
 {
-	struct prog_src_register reg = { 0, };
+	struct rc_src_register reg = { 0, };
 
-	reg.File = PROGRAM_STATE_VAR;
+	reg.File = RC_FILE_CONSTANT;
 	reg.Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_SHADOW_AMBIENT, tmu);
-	reg.Swizzle = SWIZZLE_WWWW;
+	reg.Swizzle = RC_SWIZZLE_WWWW;
 	return reg;
 }
 
@@ -44,7 +46,7 @@ static struct prog_src_register shadow_ambient(struct radeon_compiler * c, int t
  *  - implement texture compare (shadow extensions)
  *  - extract non-native source / destination operands
  */
-GLboolean r500_transform_TEX(
+int r500_transform_TEX(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void* data)
@@ -52,77 +54,77 @@ GLboolean r500_transform_TEX(
 	struct r300_fragment_program_compiler *compiler =
 		(struct r300_fragment_program_compiler*)data;
 
-	if (inst->I.Opcode != OPCODE_TEX &&
-	    inst->I.Opcode != OPCODE_TXB &&
-	    inst->I.Opcode != OPCODE_TXP &&
-	    inst->I.Opcode != OPCODE_KIL)
-		return GL_FALSE;
+	if (inst->U.I.Opcode != RC_OPCODE_TEX &&
+	    inst->U.I.Opcode != RC_OPCODE_TXB &&
+	    inst->U.I.Opcode != RC_OPCODE_TXP &&
+	    inst->U.I.Opcode != RC_OPCODE_KIL)
+		return 0;
 
 	/* ARB_shadow & EXT_shadow_funcs */
-	if (inst->I.Opcode != OPCODE_KIL &&
-	    c->Program.ShadowSamplers & (1 << inst->I.TexSrcUnit)) {
-		GLuint comparefunc = GL_NEVER + compiler->state.unit[inst->I.TexSrcUnit].texture_compare_func;
+	if (inst->U.I.Opcode != RC_OPCODE_KIL &&
+	    c->Program.ShadowSamplers & (1 << inst->U.I.TexSrcUnit)) {
+		rc_compare_func comparefunc = compiler->state.unit[inst->U.I.TexSrcUnit].texture_compare_func;
 
-		if (comparefunc == GL_NEVER || comparefunc == GL_ALWAYS) {
-			inst->I.Opcode = OPCODE_MOV;
+		if (comparefunc == RC_COMPARE_FUNC_NEVER || comparefunc == RC_COMPARE_FUNC_ALWAYS) {
+			inst->U.I.Opcode = RC_OPCODE_MOV;
 
-			if (comparefunc == GL_ALWAYS) {
-				inst->I.SrcReg[0].File = PROGRAM_BUILTIN;
-				inst->I.SrcReg[0].Swizzle = SWIZZLE_1111;
+			if (comparefunc == RC_COMPARE_FUNC_ALWAYS) {
+				inst->U.I.SrcReg[0].File = RC_FILE_NONE;
+				inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_1111;
 			} else {
-				inst->I.SrcReg[0] = shadow_ambient(c, inst->I.TexSrcUnit);
+				inst->U.I.SrcReg[0] = shadow_ambient(c, inst->U.I.TexSrcUnit);
 			}
 
-			return GL_TRUE;
+			return 1;
 		} else {
-			GLuint comparefunc = GL_NEVER + compiler->state.unit[inst->I.TexSrcUnit].texture_compare_func;
-			GLuint depthmode = compiler->state.unit[inst->I.TexSrcUnit].depth_texture_mode;
+			rc_compare_func comparefunc = compiler->state.unit[inst->U.I.TexSrcUnit].texture_compare_func;
+			unsigned int depthmode = compiler->state.unit[inst->U.I.TexSrcUnit].depth_texture_mode;
 			struct rc_instruction * inst_rcp = rc_insert_new_instruction(c, inst);
 			struct rc_instruction * inst_mad = rc_insert_new_instruction(c, inst_rcp);
 			struct rc_instruction * inst_cmp = rc_insert_new_instruction(c, inst_mad);
 			int pass, fail;
 
-			inst_rcp->I.Opcode = OPCODE_RCP;
-			inst_rcp->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst_rcp->I.DstReg.Index = rc_find_free_temporary(c);
-			inst_rcp->I.DstReg.WriteMask = WRITEMASK_W;
-			inst_rcp->I.SrcReg[0] = inst->I.SrcReg[0];
-			inst_rcp->I.SrcReg[0].Swizzle = SWIZZLE_WWWW;
-
-			inst_cmp->I.DstReg = inst->I.DstReg;
-			inst->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst->I.DstReg.Index = rc_find_free_temporary(c);
-			inst->I.DstReg.WriteMask = WRITEMASK_XYZW;
-
-			inst_mad->I.Opcode = OPCODE_MAD;
-			inst_mad->I.DstReg.File = PROGRAM_TEMPORARY;
-			inst_mad->I.DstReg.Index = rc_find_free_temporary(c);
-			inst_mad->I.SrcReg[0] = inst->I.SrcReg[0];
-			inst_mad->I.SrcReg[0].Swizzle = SWIZZLE_ZZZZ;
-			inst_mad->I.SrcReg[1].File = PROGRAM_TEMPORARY;
-			inst_mad->I.SrcReg[1].Index = inst_rcp->I.DstReg.Index;
-			inst_mad->I.SrcReg[1].Swizzle = SWIZZLE_WWWW;
-			inst_mad->I.SrcReg[2].File = PROGRAM_TEMPORARY;
-			inst_mad->I.SrcReg[2].Index = inst->I.DstReg.Index;
+			inst_rcp->U.I.Opcode = RC_OPCODE_RCP;
+			inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst_rcp->U.I.DstReg.Index = rc_find_free_temporary(c);
+			inst_rcp->U.I.DstReg.WriteMask = RC_MASK_W;
+			inst_rcp->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
+			inst_rcp->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_WWWW;
+
+			inst_cmp->U.I.DstReg = inst->U.I.DstReg;
+			inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst->U.I.DstReg.Index = rc_find_free_temporary(c);
+			inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
+
+			inst_mad->U.I.Opcode = RC_OPCODE_MAD;
+			inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY;
+			inst_mad->U.I.DstReg.Index = rc_find_free_temporary(c);
+			inst_mad->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
+			inst_mad->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_ZZZZ;
+			inst_mad->U.I.SrcReg[1].File = RC_FILE_TEMPORARY;
+			inst_mad->U.I.SrcReg[1].Index = inst_rcp->U.I.DstReg.Index;
+			inst_mad->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_WWWW;
+			inst_mad->U.I.SrcReg[2].File = RC_FILE_TEMPORARY;
+			inst_mad->U.I.SrcReg[2].Index = inst->U.I.DstReg.Index;
 			if (depthmode == 0) /* GL_LUMINANCE */
-				inst_mad->I.SrcReg[2].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z);
+				inst_mad->U.I.SrcReg[2].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_Z);
 			else if (depthmode == 2) /* GL_ALPHA */
-				inst_mad->I.SrcReg[2].Swizzle = SWIZZLE_WWWW;
+				inst_mad->U.I.SrcReg[2].Swizzle = RC_SWIZZLE_WWWW;
 
 			/* Recall that SrcReg[0] is tex, SrcReg[2] is r and:
 			 *   r  < tex  <=>      -tex+r < 0
 			 *   r >= tex  <=> not (-tex+r < 0 */
-			if (comparefunc == GL_LESS || comparefunc == GL_GEQUAL)
-				inst_mad->I.SrcReg[2].Negate = inst_mad->I.SrcReg[2].Negate ^ NEGATE_XYZW;
+			if (comparefunc == RC_COMPARE_FUNC_LESS || comparefunc == RC_COMPARE_FUNC_GEQUAL)
+				inst_mad->U.I.SrcReg[2].Negate = inst_mad->U.I.SrcReg[2].Negate ^ RC_MASK_XYZW;
 			else
-				inst_mad->I.SrcReg[0].Negate = inst_mad->I.SrcReg[0].Negate ^ NEGATE_XYZW;
+				inst_mad->U.I.SrcReg[0].Negate = inst_mad->U.I.SrcReg[0].Negate ^ RC_MASK_XYZW;
 
-			inst_cmp->I.Opcode = OPCODE_CMP;
+			inst_cmp->U.I.Opcode = RC_OPCODE_CMP;
 			/* DstReg has been filled out above */
-			inst_cmp->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-			inst_cmp->I.SrcReg[0].Index = inst_mad->I.DstReg.Index;
+			inst_cmp->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+			inst_cmp->U.I.SrcReg[0].Index = inst_mad->U.I.DstReg.Index;
 
-			if (comparefunc == GL_LESS || comparefunc == GL_GREATER) {
+			if (comparefunc == RC_COMPARE_FUNC_LESS || comparefunc == RC_COMPARE_FUNC_GREATER) {
 				pass = 1;
 				fail = 2;
 			} else {
@@ -130,131 +132,161 @@ GLboolean r500_transform_TEX(
 				fail = 1;
 			}
 
-			inst_cmp->I.SrcReg[pass].File = PROGRAM_BUILTIN;
-			inst_cmp->I.SrcReg[pass].Swizzle = SWIZZLE_1111;
-			inst_cmp->I.SrcReg[fail] = shadow_ambient(c, inst->I.TexSrcUnit);
+			inst_cmp->U.I.SrcReg[pass].File = RC_FILE_NONE;
+			inst_cmp->U.I.SrcReg[pass].Swizzle = RC_SWIZZLE_1111;
+			inst_cmp->U.I.SrcReg[fail] = shadow_ambient(c, inst->U.I.TexSrcUnit);
 		}
 	}
 
 	/* Cannot write texture to output registers */
-	if (inst->I.Opcode != OPCODE_KIL && inst->I.DstReg.File != PROGRAM_TEMPORARY) {
+	if (inst->U.I.Opcode != RC_OPCODE_KIL && inst->U.I.DstReg.File != RC_FILE_TEMPORARY) {
 		struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst);
 
-		inst_mov->I.Opcode = OPCODE_MOV;
-		inst_mov->I.DstReg = inst->I.DstReg;
-		inst_mov->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-		inst_mov->I.SrcReg[0].Index = rc_find_free_temporary(c);
+		inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+		inst_mov->U.I.DstReg = inst->U.I.DstReg;
+		inst_mov->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+		inst_mov->U.I.SrcReg[0].Index = rc_find_free_temporary(c);
 
-		inst->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst->I.DstReg.Index = inst_mov->I.SrcReg[0].Index;
-		inst->I.DstReg.WriteMask = WRITEMASK_XYZW;
+		inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst->U.I.DstReg.Index = inst_mov->U.I.SrcReg[0].Index;
+		inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
 	}
 
 	/* Cannot read texture coordinate from constants file */
-	if (inst->I.SrcReg[0].File != PROGRAM_TEMPORARY && inst->I.SrcReg[0].File != PROGRAM_INPUT) {
+	if (inst->U.I.SrcReg[0].File != RC_FILE_TEMPORARY && inst->U.I.SrcReg[0].File != RC_FILE_INPUT) {
 		struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev);
 
-		inst_mov->I.Opcode = OPCODE_MOV;
-		inst_mov->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst_mov->I.DstReg.Index = rc_find_free_temporary(c);
-		inst_mov->I.SrcReg[0] = inst->I.SrcReg[0];
+		inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+		inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst_mov->U.I.DstReg.Index = rc_find_free_temporary(c);
+		inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
 
-		reset_srcreg(&inst->I.SrcReg[0]);
-		inst->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-		inst->I.SrcReg[0].Index = inst_mov->I.DstReg.Index;
+		reset_srcreg(&inst->U.I.SrcReg[0]);
+		inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+		inst->U.I.SrcReg[0].Index = inst_mov->U.I.DstReg.Index;
 	}
 
-	return GL_TRUE;
+	return 1;
+}
+
+/**
+ * Rewrite IF instructions to use the ALU result special register.
+ */
+int r500_transform_IF(
+	struct radeon_compiler * c,
+	struct rc_instruction * inst,
+	void* data)
+{
+	if (inst->U.I.Opcode != RC_OPCODE_IF)
+		return 0;
+
+	struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev);
+	inst_mov->U.I.Opcode = RC_OPCODE_MOV;
+	inst_mov->U.I.DstReg.WriteMask = 0;
+	inst_mov->U.I.WriteALUResult = RC_ALURESULT_W;
+	inst_mov->U.I.ALUResultCompare = RC_COMPARE_FUNC_NOTEQUAL;
+	inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0];
+	inst_mov->U.I.SrcReg[0].Swizzle = combine_swizzles4(inst_mov->U.I.SrcReg[0].Swizzle,
+			RC_SWIZZLE_UNUSED, RC_SWIZZLE_UNUSED, RC_SWIZZLE_UNUSED, RC_SWIZZLE_X);
+
+	inst->U.I.SrcReg[0].File = RC_FILE_SPECIAL;
+	inst->U.I.SrcReg[0].Index = RC_SPECIAL_ALU_RESULT;
+	inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
+	inst->U.I.SrcReg[0].Negate = 0;
+
+	return 1;
 }
 
-GLboolean r500FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
+static int r500_swizzle_is_native(rc_opcode opcode, struct rc_src_register reg)
 {
-	GLuint relevant;
+	unsigned int relevant;
 	int i;
 
-	if (opcode == OPCODE_TEX ||
-	    opcode == OPCODE_TXB ||
-	    opcode == OPCODE_TXP ||
-	    opcode == OPCODE_KIL) {
+	if (opcode == RC_OPCODE_TEX ||
+	    opcode == RC_OPCODE_TXB ||
+	    opcode == RC_OPCODE_TXP ||
+	    opcode == RC_OPCODE_KIL) {
 		if (reg.Abs)
-			return GL_FALSE;
+			return 0;
 
-		if (opcode == OPCODE_KIL && (reg.Swizzle != SWIZZLE_NOOP || reg.Negate != NEGATE_NONE))
-			return GL_FALSE;
+		if (opcode == RC_OPCODE_KIL && (reg.Swizzle != RC_SWIZZLE_XYZW || reg.Negate != RC_MASK_NONE))
+			return 0;
 
 		if (reg.Negate)
-			reg.Negate ^= NEGATE_XYZW;
+			reg.Negate ^= RC_MASK_XYZW;
 
 		for(i = 0; i < 4; ++i) {
-			GLuint swz = GET_SWZ(reg.Swizzle, i);
-			if (swz == SWIZZLE_NIL) {
+			unsigned int swz = GET_SWZ(reg.Swizzle, i);
+			if (swz == RC_SWIZZLE_UNUSED) {
 				reg.Negate &= ~(1 << i);
 				continue;
 			}
 			if (swz >= 4)
-				return GL_FALSE;
+				return 0;
 		}
 
 		if (reg.Negate)
-			return GL_FALSE;
+			return 0;
 
-		return GL_TRUE;
-	} else if (opcode == OPCODE_DDX || opcode == OPCODE_DDY) {
+		return 1;
+	} else if (opcode == RC_OPCODE_DDX || opcode == RC_OPCODE_DDY) {
 		/* DDX/MDH and DDY/MDV explicitly ignore incoming swizzles;
 		 * if it doesn't fit perfectly into a .xyzw case... */
-		if (reg.Swizzle == SWIZZLE_NOOP && !reg.Abs && !reg.Negate)
-			return GL_TRUE;
+		if (reg.Swizzle == RC_SWIZZLE_XYZW && !reg.Abs && !reg.Negate)
+			return 1;
 
-		return GL_FALSE;
+		return 0;
 	} else {
 		/* ALU instructions support almost everything */
 		if (reg.Abs)
-			return GL_TRUE;
+			return 1;
 
 		relevant = 0;
 		for(i = 0; i < 3; ++i) {
-			GLuint swz = GET_SWZ(reg.Swizzle, i);
-			if (swz != SWIZZLE_NIL && swz != SWIZZLE_ZERO)
+			unsigned int swz = GET_SWZ(reg.Swizzle, i);
+			if (swz != RC_SWIZZLE_UNUSED && swz != RC_SWIZZLE_ZERO)
 				relevant |= 1 << i;
 		}
 		if ((reg.Negate & relevant) && ((reg.Negate & relevant) != relevant))
-			return GL_FALSE;
+			return 0;
 
-		return GL_TRUE;
+		return 1;
 	}
 }
 
 /**
- * Implement a MOV with a potentially non-native swizzle.
+ * Split source register access.
  *
  * The only thing we *cannot* do in an ALU instruction is per-component
- * negation. Therefore, we split the MOV into two instructions when necessary.
+ * negation.
  */
-void r500FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst, struct prog_src_register src)
+static void r500_swizzle_split(struct rc_src_register src, unsigned int usemask,
+		struct rc_swizzle_split * split)
 {
-	GLuint negatebase[2] = { 0, 0 };
+	unsigned int negatebase[2] = { 0, 0 };
 	int i;
 
 	for(i = 0; i < 4; ++i) {
-		GLuint swz = GET_SWZ(src.Swizzle, i);
-		if (swz == SWIZZLE_NIL)
+		unsigned int swz = GET_SWZ(src.Swizzle, i);
+		if (swz == RC_SWIZZLE_UNUSED || !GET_BIT(usemask, i))
 			continue;
 		negatebase[GET_BIT(src.Negate, i)] |= 1 << i;
 	}
 
+	split->NumPhases = 0;
+
 	for(i = 0; i <= 1; ++i) {
 		if (!negatebase[i])
 			continue;
 
-		struct rc_instruction *inst = rc_insert_new_instruction(s->Compiler, s->IP->Prev);
-		inst->I.Opcode = OPCODE_MOV;
-		inst->I.DstReg = dst;
-		inst->I.DstReg.WriteMask = negatebase[i];
-		inst->I.SrcReg[0] = src;
-		inst->I.SrcReg[0].Negate = (i == 0) ? NEGATE_NONE : NEGATE_XYZW;
+		split->Phase[split->NumPhases++] = negatebase[i];
 	}
 }
 
+struct rc_swizzle_caps r500_swizzle_caps = {
+	.IsNative = r500_swizzle_is_native,
+	.Split = r500_swizzle_split
+};
 
 static char *toswiz(int swiz_val) {
   switch(swiz_val) {
diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h
index 9091f65cd2..0918cdf518 100644
--- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h
+++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h
@@ -33,21 +33,21 @@
 #ifndef __R500_FRAGPROG_H_
 #define __R500_FRAGPROG_H_
 
-#include "shader/prog_parameter.h"
-#include "shader/prog_instruction.h"
-
 #include "radeon_compiler.h"
-#include "radeon_nqssadce.h"
+#include "radeon_swizzle.h"
 
 extern void r500BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compiler);
 
 extern void r500FragmentProgramDump(struct rX00_fragment_program_code *c);
 
-extern GLboolean r500FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg);
+extern struct rc_swizzle_caps r500_swizzle_caps;
 
-extern void r500FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst, struct prog_src_register src);
+extern int r500_transform_TEX(
+	struct radeon_compiler * c,
+	struct rc_instruction * inst,
+	void* data);
 
-extern GLboolean r500_transform_TEX(
+extern int r500_transform_IF(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void* data);
diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c
index d694725c9b..b1b14394b6 100644
--- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c
+++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c
@@ -37,10 +37,6 @@
  *
  * \author Corbin Simpson <MostAwesomeDude@gmail.com>
  *
- * \todo Depth write, WPOS/FOGC inputs
- *
- * \todo FogOption
- *
  */
 
 #include "r500_fragprog.h"
@@ -51,7 +47,6 @@
 
 
 #define PROG_CODE \
-	struct r300_fragment_program_compiler *c = (struct r300_fragment_program_compiler*)data; \
 	struct r500_fragment_program_code *code = &c->code->code.r500
 
 #define error(fmt, args...) do {			\
@@ -60,63 +55,80 @@
 	} while(0)
 
 
-static GLuint translate_rgb_op(struct r300_fragment_program_compiler *c, GLuint opcode)
+struct branch_info {
+	int If;
+	int Else;
+	int Endif;
+};
+
+struct emit_state {
+	struct radeon_compiler * C;
+	struct r500_fragment_program_code * Code;
+
+	struct branch_info * Branches;
+	unsigned int CurrentBranchDepth;
+	unsigned int BranchesReserved;
+
+	unsigned int MaxBranchDepth;
+};
+
+static unsigned int translate_rgb_op(struct r300_fragment_program_compiler *c, rc_opcode opcode)
 {
 	switch(opcode) {
-	case OPCODE_CMP: return R500_ALU_RGBA_OP_CMP;
-	case OPCODE_DDX: return R500_ALU_RGBA_OP_MDH;
-	case OPCODE_DDY: return R500_ALU_RGBA_OP_MDV;
-	case OPCODE_DP3: return R500_ALU_RGBA_OP_DP3;
-	case OPCODE_DP4: return R500_ALU_RGBA_OP_DP4;
-	case OPCODE_FRC: return R500_ALU_RGBA_OP_FRC;
+	case RC_OPCODE_CMP: return R500_ALU_RGBA_OP_CMP;
+	case RC_OPCODE_DDX: return R500_ALU_RGBA_OP_MDH;
+	case RC_OPCODE_DDY: return R500_ALU_RGBA_OP_MDV;
+	case RC_OPCODE_DP3: return R500_ALU_RGBA_OP_DP3;
+	case RC_OPCODE_DP4: return R500_ALU_RGBA_OP_DP4;
+	case RC_OPCODE_FRC: return R500_ALU_RGBA_OP_FRC;
 	default:
 		error("translate_rgb_op(%d): unknown opcode\n", opcode);
 		/* fall through */
-	case OPCODE_NOP:
+	case RC_OPCODE_NOP:
 		/* fall through */
-	case OPCODE_MAD: return R500_ALU_RGBA_OP_MAD;
-	case OPCODE_MAX: return R500_ALU_RGBA_OP_MAX;
-	case OPCODE_MIN: return R500_ALU_RGBA_OP_MIN;
-	case OPCODE_REPL_ALPHA: return R500_ALU_RGBA_OP_SOP;
+	case RC_OPCODE_MAD: return R500_ALU_RGBA_OP_MAD;
+	case RC_OPCODE_MAX: return R500_ALU_RGBA_OP_MAX;
+	case RC_OPCODE_MIN: return R500_ALU_RGBA_OP_MIN;
+	case RC_OPCODE_REPL_ALPHA: return R500_ALU_RGBA_OP_SOP;
 	}
 }
 
-static GLuint translate_alpha_op(struct r300_fragment_program_compiler *c, GLuint opcode)
+static unsigned int translate_alpha_op(struct r300_fragment_program_compiler *c, rc_opcode opcode)
 {
 	switch(opcode) {
-	case OPCODE_CMP: return R500_ALPHA_OP_CMP;
-	case OPCODE_COS: return R500_ALPHA_OP_COS;
-	case OPCODE_DDX: return R500_ALPHA_OP_MDH;
-	case OPCODE_DDY: return R500_ALPHA_OP_MDV;
-	case OPCODE_DP3: return R500_ALPHA_OP_DP;
-	case OPCODE_DP4: return R500_ALPHA_OP_DP;
-	case OPCODE_EX2: return R500_ALPHA_OP_EX2;
-	case OPCODE_FRC: return R500_ALPHA_OP_FRC;
-	case OPCODE_LG2: return R500_ALPHA_OP_LN2;
+	case RC_OPCODE_CMP: return R500_ALPHA_OP_CMP;
+	case RC_OPCODE_COS: return R500_ALPHA_OP_COS;
+	case RC_OPCODE_DDX: return R500_ALPHA_OP_MDH;
+	case RC_OPCODE_DDY: return R500_ALPHA_OP_MDV;
+	case RC_OPCODE_DP3: return R500_ALPHA_OP_DP;
+	case RC_OPCODE_DP4: return R500_ALPHA_OP_DP;
+	case RC_OPCODE_EX2: return R500_ALPHA_OP_EX2;
+	case RC_OPCODE_FRC: return R500_ALPHA_OP_FRC;
+	case RC_OPCODE_LG2: return R500_ALPHA_OP_LN2;
 	default:
 		error("translate_alpha_op(%d): unknown opcode\n", opcode);
 		/* fall through */
-	case OPCODE_NOP:
+	case RC_OPCODE_NOP:
 		/* fall through */
-	case OPCODE_MAD: return R500_ALPHA_OP_MAD;
-	case OPCODE_MAX: return R500_ALPHA_OP_MAX;
-	case OPCODE_MIN: return R500_ALPHA_OP_MIN;
-	case OPCODE_RCP: return R500_ALPHA_OP_RCP;
-	case OPCODE_RSQ: return R500_ALPHA_OP_RSQ;
-	case OPCODE_SIN: return R500_ALPHA_OP_SIN;
+	case RC_OPCODE_MAD: return R500_ALPHA_OP_MAD;
+	case RC_OPCODE_MAX: return R500_ALPHA_OP_MAX;
+	case RC_OPCODE_MIN: return R500_ALPHA_OP_MIN;
+	case RC_OPCODE_RCP: return R500_ALPHA_OP_RCP;
+	case RC_OPCODE_RSQ: return R500_ALPHA_OP_RSQ;
+	case RC_OPCODE_SIN: return R500_ALPHA_OP_SIN;
 	}
 }
 
-static GLuint fix_hw_swizzle(GLuint swz)
+static unsigned int fix_hw_swizzle(unsigned int swz)
 {
 	if (swz == 5) swz = 6;
-	if (swz == SWIZZLE_NIL) swz = 4;
+	if (swz == RC_SWIZZLE_UNUSED) swz = 4;
 	return swz;
 }
 
-static GLuint translate_arg_rgb(struct radeon_pair_instruction *inst, int arg)
+static unsigned int translate_arg_rgb(struct rc_pair_instruction *inst, int arg)
 {
-	GLuint t = inst->RGB.Arg[arg].Source;
+	unsigned int t = inst->RGB.Arg[arg].Source;
 	int comp;
 	t |= inst->RGB.Arg[arg].Negate << 11;
 	t |= inst->RGB.Arg[arg].Abs << 12;
@@ -127,39 +139,57 @@ static GLuint translate_arg_rgb(struct radeon_pair_instruction *inst, int arg)
 	return t;
 }
 
-static GLuint translate_arg_alpha(struct radeon_pair_instruction *inst, int i)
+static unsigned int translate_arg_alpha(struct rc_pair_instruction *inst, int i)
 {
-	GLuint t = inst->Alpha.Arg[i].Source;
+	unsigned int t = inst->Alpha.Arg[i].Source;
 	t |= fix_hw_swizzle(inst->Alpha.Arg[i].Swizzle) << 2;
 	t |= inst->Alpha.Arg[i].Negate << 5;
 	t |= inst->Alpha.Arg[i].Abs << 6;
 	return t;
 }
 
-static void use_temporary(struct r500_fragment_program_code* code, GLuint index)
+static uint32_t translate_alu_result_op(struct r300_fragment_program_compiler * c, rc_compare_func func)
+{
+	switch(func) {
+	case RC_COMPARE_FUNC_EQUAL: return R500_INST_ALU_RESULT_OP_EQ;
+	case RC_COMPARE_FUNC_LESS: return R500_INST_ALU_RESULT_OP_LT;
+	case RC_COMPARE_FUNC_GEQUAL: return R500_INST_ALU_RESULT_OP_GE;
+	case RC_COMPARE_FUNC_NOTEQUAL: return R500_INST_ALU_RESULT_OP_NE;
+	default:
+		rc_error(&c->Base, "%s: unsupported compare func %i\n", __FUNCTION__, func);
+		return 0;
+	}
+}
+
+static void use_temporary(struct r500_fragment_program_code* code, unsigned int index)
 {
 	if (index > code->max_temp_idx)
 		code->max_temp_idx = index;
 }
 
-static GLuint use_source(struct r500_fragment_program_code* code, struct radeon_pair_instruction_source src)
+static unsigned int use_source(struct r500_fragment_program_code* code, struct radeon_pair_instruction_source src)
 {
-	if (!src.Constant)
+	if (src.File == RC_FILE_CONSTANT) {
+		return src.Index | 0x100;
+	} else if (src.File == RC_FILE_TEMPORARY) {
 		use_temporary(code, src.Index);
-	return src.Index | src.Constant << 8;
+		return src.Index;
+	}
+
+	return 0;
 }
 
 
 /**
  * Emit a paired ALU instruction.
  */
-static GLboolean emit_paired(void *data, struct radeon_pair_instruction *inst)
+static void emit_paired(struct r300_fragment_program_compiler *c, struct rc_pair_instruction *inst)
 {
 	PROG_CODE;
 
 	if (code->inst_end >= 511) {
 		error("emit_alu: Too many instructions");
-		return GL_FALSE;
+		return;
 	}
 
 	int ip = ++code->inst_end;
@@ -167,17 +197,22 @@ static GLboolean emit_paired(void *data, struct radeon_pair_instruction *inst)
 	code->inst[ip].inst5 = translate_rgb_op(c, inst->RGB.Opcode);
 	code->inst[ip].inst4 = translate_alpha_op(c, inst->Alpha.Opcode);
 
-	if (inst->RGB.OutputWriteMask || inst->Alpha.OutputWriteMask || inst->Alpha.DepthWriteMask)
+	if (inst->RGB.OutputWriteMask || inst->Alpha.OutputWriteMask || inst->Alpha.DepthWriteMask) {
 		code->inst[ip].inst0 = R500_INST_TYPE_OUT;
-	else
+		if (inst->WriteALUResult) {
+			error("%s: cannot write output and ALU result at the same time");
+			return;
+		}
+	} else {
 		code->inst[ip].inst0 = R500_INST_TYPE_ALU;
+	}
 	code->inst[ip].inst0 |= R500_INST_TEX_SEM_WAIT;
 
 	code->inst[ip].inst0 |= (inst->RGB.WriteMask << 11) | (inst->Alpha.WriteMask << 14);
 	code->inst[ip].inst0 |= (inst->RGB.OutputWriteMask << 15) | (inst->Alpha.OutputWriteMask << 18);
 	if (inst->Alpha.DepthWriteMask) {
 		code->inst[ip].inst4 |= R500_ALPHA_W_OMASK;
-		c->code->writes_depth = GL_TRUE;
+		c->code->writes_depth = 1;
 	}
 
 	code->inst[ip].inst4 |= R500_ALPHA_ADDRD(inst->Alpha.DestIndex);
@@ -206,12 +241,21 @@ static GLboolean emit_paired(void *data, struct radeon_pair_instruction *inst)
 	code->inst[ip].inst4 |= translate_arg_alpha(inst, 1) << R500_ALPHA_SEL_B_SHIFT;
 	code->inst[ip].inst5 |= translate_arg_alpha(inst, 2) << R500_ALU_RGBA_ALPHA_SEL_C_SHIFT;
 
-	return GL_TRUE;
+	if (inst->WriteALUResult) {
+		code->inst[ip].inst3 |= R500_ALU_RGB_WMASK;
+
+		if (inst->WriteALUResult == RC_ALURESULT_X)
+			code->inst[ip].inst0 |= R500_INST_ALU_RESULT_SEL_RED;
+		else
+			code->inst[ip].inst0 |= R500_INST_ALU_RESULT_SEL_ALPHA;
+
+		code->inst[ip].inst0 |= translate_alu_result_op(c, inst->ALUResultCompare);
+	}
 }
 
-static GLuint translate_strq_swizzle(GLuint swizzle)
+static unsigned int translate_strq_swizzle(unsigned int swizzle)
 {
-	GLuint swiz = 0;
+	unsigned int swiz = 0;
 	int i;
 	for (i = 0; i < 4; i++)
 		swiz |= (GET_SWZ(swizzle, i) & 0x3) << i*2;
@@ -221,67 +265,193 @@ static GLuint translate_strq_swizzle(GLuint swizzle)
 /**
  * Emit a single TEX instruction
  */
-static GLboolean emit_tex(void *data, struct radeon_pair_texture_instruction *inst)
+static int emit_tex(struct r300_fragment_program_compiler *c, struct rc_sub_instruction *inst)
 {
 	PROG_CODE;
 
 	if (code->inst_end >= 511) {
 		error("emit_tex: Too many instructions");
-		return GL_FALSE;
+		return 0;
 	}
 
 	int ip = ++code->inst_end;
 
 	code->inst[ip].inst0 = R500_INST_TYPE_TEX
-		| (inst->WriteMask << 11)
+		| (inst->DstReg.WriteMask << 11)
 		| R500_INST_TEX_SEM_WAIT;
 	code->inst[ip].inst1 = R500_TEX_ID(inst->TexSrcUnit)
 		| R500_TEX_SEM_ACQUIRE | R500_TEX_IGNORE_UNCOVERED;
 
-	if (inst->TexSrcTarget == TEXTURE_RECT_INDEX)
-	        code->inst[ip].inst1 |= R500_TEX_UNSCALED;
+	if (inst->TexSrcTarget == RC_TEXTURE_RECT)
+		code->inst[ip].inst1 |= R500_TEX_UNSCALED;
 
 	switch (inst->Opcode) {
-	case RADEON_OPCODE_KIL:
+	case RC_OPCODE_KIL:
 		code->inst[ip].inst1 |= R500_TEX_INST_TEXKILL;
 		break;
-	case RADEON_OPCODE_TEX:
+	case RC_OPCODE_TEX:
 		code->inst[ip].inst1 |= R500_TEX_INST_LD;
 		break;
-	case RADEON_OPCODE_TXB:
+	case RC_OPCODE_TXB:
 		code->inst[ip].inst1 |= R500_TEX_INST_LODBIAS;
 		break;
-	case RADEON_OPCODE_TXP:
+	case RC_OPCODE_TXP:
 		code->inst[ip].inst1 |= R500_TEX_INST_PROJ;
 		break;
 	default:
 		error("emit_tex can't handle opcode %x\n", inst->Opcode);
 	}
 
-	code->inst[ip].inst2 = R500_TEX_SRC_ADDR(inst->SrcIndex)
-		| (translate_strq_swizzle(inst->SrcSwizzle) << 8)
-		| R500_TEX_DST_ADDR(inst->DestIndex)
+	use_temporary(code, inst->SrcReg[0].Index);
+	if (inst->Opcode != RC_OPCODE_KIL)
+		use_temporary(code, inst->DstReg.Index);
+
+	code->inst[ip].inst2 = R500_TEX_SRC_ADDR(inst->SrcReg[0].Index)
+		| (translate_strq_swizzle(inst->SrcReg[0].Swizzle) << 8)
+		| R500_TEX_DST_ADDR(inst->DstReg.Index)
 		| R500_TEX_DST_R_SWIZ_R | R500_TEX_DST_G_SWIZ_G
 		| R500_TEX_DST_B_SWIZ_B | R500_TEX_DST_A_SWIZ_A;
 
-	return GL_TRUE;
+	return 1;
 }
 
-static const struct radeon_pair_handler pair_handler = {
-	.EmitPaired = emit_paired,
-	.EmitTex = emit_tex,
-	.MaxHwTemps = 128
-};
+static void grow_branches(struct emit_state * s)
+{
+	unsigned int newreserved = s->BranchesReserved * 2;
+	struct branch_info * newbranches;
+
+	if (!newreserved)
+		newreserved = 4;
+
+	newbranches = memory_pool_malloc(&s->C->Pool, newreserved*sizeof(struct branch_info));
+	memcpy(newbranches, s->Branches, s->CurrentBranchDepth*sizeof(struct branch_info));
+
+	s->Branches = newbranches;
+	s->BranchesReserved = newreserved;
+}
+
+static void emit_flowcontrol(struct emit_state * s, struct rc_instruction * inst)
+{
+	if (s->Code->inst_end >= 511) {
+		rc_error(s->C, "emit_tex: Too many instructions");
+		return;
+	}
+
+	unsigned int newip = ++s->Code->inst_end;
+
+	s->Code->inst[newip].inst0 = R500_INST_TYPE_FC | R500_INST_ALU_WAIT;
+
+	if (inst->U.I.Opcode == RC_OPCODE_IF) {
+		if (s->CurrentBranchDepth >= 32) {
+			rc_error(s->C, "Branch depth exceeds hardware limit");
+			return;
+		}
+
+		if (s->CurrentBranchDepth >= s->BranchesReserved)
+			grow_branches(s);
+
+		struct branch_info * branch = &s->Branches[s->CurrentBranchDepth++];
+		branch->If = newip;
+		branch->Else = -1;
+		branch->Endif = -1;
+
+		if (s->CurrentBranchDepth > s->MaxBranchDepth)
+			s->MaxBranchDepth = s->CurrentBranchDepth;
+
+		/* actual instruction is filled in at ENDIF time */
+	} else if (inst->U.I.Opcode == RC_OPCODE_ELSE) {
+		if (!s->CurrentBranchDepth) {
+			rc_error(s->C, "%s: got ELSE outside a branch", __FUNCTION__);
+			return;
+		}
+
+		struct branch_info * branch = &s->Branches[s->CurrentBranchDepth - 1];
+		branch->Else = newip;
+
+		/* actual instruction is filled in at ENDIF time */
+	} else if (inst->U.I.Opcode == RC_OPCODE_ENDIF) {
+		if (!s->CurrentBranchDepth) {
+			rc_error(s->C, "%s: got ELSE outside a branch", __FUNCTION__);
+			return;
+		}
+
+		struct branch_info * branch = &s->Branches[s->CurrentBranchDepth - 1];
+		branch->Endif = newip;
+
+		s->Code->inst[branch->If].inst2 = R500_FC_OP_JUMP
+			| R500_FC_A_OP_NONE /* no address stack */
+			| R500_FC_JUMP_FUNC(0x0f) /* jump if ALU result is false */
+			| R500_FC_B_OP0_INCR /* increment branch counter if stay */
+		;
+
+		if (branch->Else >= 0) {
+			/* increment branch counter also if jump */
+			s->Code->inst[branch->If].inst2 |= R500_FC_B_OP1_INCR;
+			s->Code->inst[branch->If].inst3 = R500_FC_JUMP_ADDR(branch->Else + 1);
+
+			s->Code->inst[branch->Else].inst2 = R500_FC_OP_JUMP
+				| R500_FC_A_OP_NONE /* no address stack */
+				| R500_FC_B_ELSE /* all active pixels want to jump */
+				| R500_FC_B_OP0_NONE /* no counter op if stay */
+				| R500_FC_B_OP1_DECR /* decrement branch counter if jump */
+				| R500_FC_B_POP_CNT(1)
+			;
+			s->Code->inst[branch->Else].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
+		} else {
+			/* don't touch branch counter on jump */
+			s->Code->inst[branch->If].inst2 |= R500_FC_B_OP1_NONE;
+			s->Code->inst[branch->If].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
+		}
+
+		s->Code->inst[branch->Endif].inst2 = R500_FC_OP_JUMP
+			| R500_FC_A_OP_NONE /* no address stack */
+			| R500_FC_JUMP_ANY /* docs says set this, but I don't understand why */
+			| R500_FC_B_OP0_DECR /* decrement branch counter if stay */
+			| R500_FC_B_OP1_NONE /* no branch counter if stay */
+			| R500_FC_B_POP_CNT(1)
+		;
+		s->Code->inst[branch->Endif].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
+
+		s->CurrentBranchDepth--;
+	} else {
+		rc_error(s->C, "%s: unknown opcode %i\n", __FUNCTION__, inst->U.I.Opcode);
+	}
+}
 
 void r500BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compiler)
 {
+	struct emit_state s;
 	struct r500_fragment_program_code *code = &compiler->code->code.r500;
 
-	_mesa_bzero(code, sizeof(*code));
+	memset(&s, 0, sizeof(s));
+	s.C = &compiler->Base;
+	s.Code = code;
+
+	memset(code, 0, sizeof(*code));
 	code->max_temp_idx = 1;
 	code->inst_end = -1;
 
-	radeonPairProgram(compiler, &pair_handler, compiler);
+	for(struct rc_instruction * inst = compiler->Base.Program.Instructions.Next;
+	    inst != &compiler->Base.Program.Instructions && !compiler->Base.Error;
+	    inst = inst->Next) {
+		if (inst->Type == RC_INSTRUCTION_NORMAL) {
+			const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+
+			if (opcode->IsFlowControl) {
+				emit_flowcontrol(&s, inst);
+			} else if (inst->U.I.Opcode == RC_OPCODE_BEGIN_TEX) {
+				continue;
+			} else {
+				emit_tex(compiler, &inst->U.I);
+			}
+		} else {
+			emit_paired(compiler, &inst->U.P);
+		}
+	}
+
+	if (code->max_temp_idx >= 128)
+		rc_error(&compiler->Base, "Too many hardware temporaries used");
+
 	if (compiler->Base.Error)
 		return;
 
@@ -296,4 +466,11 @@ void r500BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi
 		int ip = ++code->inst_end;
 		code->inst[ip].inst0 = R500_INST_TYPE_OUT | R500_INST_TEX_SEM_WAIT;
 	}
+
+	if (s.MaxBranchDepth >= 4) {
+		if (code->max_temp_idx < 1)
+			code->max_temp_idx = 1;
+
+		code->us_fc_ctrl |= R500_FC_FULL_FC_EN;
+	}
 }
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_code.c b/src/mesa/drivers/dri/r300/compiler/radeon_code.c
index c7923004df..1a3d8bb641 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_code.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_code.c
@@ -25,11 +25,13 @@
  *
  */
 
-#include "main/mtypes.h"
-#include "shader/prog_instruction.h"
-
 #include "radeon_code.h"
 
+#include <stdlib.h>
+#include <string.h>
+
+#include "radeon_program.h"
+
 void rc_constants_init(struct rc_constant_list * c)
 {
 	memset(c, 0, sizeof(*c));
@@ -143,7 +145,7 @@ unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float da
 		if (c->Constants[index].Type == RC_CONSTANT_IMMEDIATE) {
 			for(unsigned comp = 0; comp < c->Constants[index].Size; ++comp) {
 				if (c->Constants[index].u.Immediate[comp] == data) {
-					*swizzle = MAKE_SWIZZLE4(comp, comp, comp, comp);
+					*swizzle = RC_MAKE_SWIZZLE(comp, comp, comp, comp);
 					return index;
 				}
 			}
@@ -156,7 +158,7 @@ unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float da
 	if (free_index >= 0) {
 		unsigned comp = c->Constants[free_index].Size++;
 		c->Constants[free_index].u.Immediate[comp] = data;
-		*swizzle = MAKE_SWIZZLE4(comp, comp, comp, comp);
+		*swizzle = RC_MAKE_SWIZZLE(comp, comp, comp, comp);
 		return free_index;
 	}
 
@@ -164,7 +166,7 @@ unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float da
 	constant.Type = RC_CONSTANT_IMMEDIATE;
 	constant.Size = 1;
 	constant.u.Immediate[0] = data;
-	*swizzle = SWIZZLE_XXXX;
+	*swizzle = RC_SWIZZLE_XXXX;
 
 	return rc_constants_add(c, &constant);
 }
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_code.h b/src/mesa/drivers/dri/r300/compiler/radeon_code.h
index 3e88554ba1..902b7cfa53 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_code.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_code.h
@@ -89,6 +89,23 @@ unsigned rc_constants_add_immediate_vec4(struct rc_constant_list * c, const floa
 unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float data, unsigned * swizzle);
 
 /**
+ * Compare functions.
+ *
+ * \note By design, RC_COMPARE_FUNC_xxx + GL_NEVER gives you
+ * the correct GL compare function.
+ */
+typedef enum {
+	RC_COMPARE_FUNC_NEVER = 0,
+	RC_COMPARE_FUNC_LESS,
+	RC_COMPARE_FUNC_EQUAL,
+	RC_COMPARE_FUNC_LEQUAL,
+	RC_COMPARE_FUNC_GREATER,
+	RC_COMPARE_FUNC_NOTEQUAL,
+	RC_COMPARE_FUNC_GEQUAL,
+	RC_COMPARE_FUNC_ALWAYS
+} rc_compare_func;
+
+/**
  * Stores state that influences the compilation of a fragment program.
  */
 struct r300_fragment_program_external_state {
@@ -105,10 +122,12 @@ struct r300_fragment_program_external_state {
 
 		/**
 		 * If the sampler is used as a shadow sampler,
-		 * this field is (texture_compare_func - GL_NEVER).
-		 * [e.g. if compare function is GL_LEQUAL, this field is 3]
+		 * this field specifies the compare function.
+		 *
+		 * Otherwise, this field is \ref RC_COMPARE_FUNC_NEVER (aka 0).
 		 *
 		 * Otherwise, this field is 0.
+		 * \sa rc_compare_func
 		 */
 		unsigned texture_compare_func : 3;
 	} unit[16];
@@ -163,6 +182,8 @@ struct r500_fragment_program_code {
 	int inst_end; /* Number of instructions - 1; also, last instruction to be executed */
 
 	int max_temp_idx;
+
+	uint32_t us_fc_ctrl;
 };
 
 struct rX00_fragment_program_code {
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c
index da950d5289..c0e7a7f7a0 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c
@@ -23,6 +23,8 @@
 #include "radeon_compiler.h"
 
 #include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
 
 #include "radeon_program.h"
 
@@ -34,7 +36,7 @@ void rc_init(struct radeon_compiler * c)
 	memory_pool_init(&c->Pool);
 	c->Program.Instructions.Prev = &c->Program.Instructions;
 	c->Program.Instructions.Next = &c->Program.Instructions;
-	c->Program.Instructions.I.Opcode = OPCODE_END;
+	c->Program.Instructions.U.I.Opcode = RC_OPCODE_ILLEGAL_OPCODE;
 }
 
 void rc_destroy(struct radeon_compiler * c)
@@ -60,7 +62,7 @@ void rc_error(struct radeon_compiler * c, const char * fmt, ...)
 {
 	va_list ap;
 
-	c->Error = GL_TRUE;
+	c->Error = 1;
 
 	if (!c->ErrorMsg) {
 		/* Only remember the first error */
@@ -91,28 +93,63 @@ void rc_error(struct radeon_compiler * c, const char * fmt, ...)
 	}
 }
 
+int rc_if_fail_helper(struct radeon_compiler * c, const char * file, int line, const char * assertion)
+{
+	rc_error(c, "ICE at %s:%i: assertion failed: %s\n", file, line, assertion);
+	return 1;
+}
+
+/**
+ * Recompute c->Program.InputsRead and c->Program.OutputsWritten
+ * based on which inputs and outputs are actually referenced
+ * in program instructions.
+ */
+void rc_calculate_inputs_outputs(struct radeon_compiler * c)
+{
+	struct rc_instruction *inst;
+
+	c->Program.InputsRead = 0;
+	c->Program.OutputsWritten = 0;
+
+	for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next)
+	{
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+		int i;
+
+		for (i = 0; i < opcode->NumSrcRegs; ++i) {
+			if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT)
+				c->Program.InputsRead |= 1 << inst->U.I.SrcReg[i].Index;
+		}
+
+		if (opcode->HasDstReg) {
+			if (inst->U.I.DstReg.File == RC_FILE_OUTPUT)
+				c->Program.OutputsWritten |= 1 << inst->U.I.DstReg.Index;
+		}
+	}
+}
+
 /**
  * Rewrite the program such that everything that source the given input
  * register will source new_input instead.
  */
-void rc_move_input(struct radeon_compiler * c, unsigned input, struct prog_src_register new_input)
+void rc_move_input(struct radeon_compiler * c, unsigned input, struct rc_src_register new_input)
 {
 	struct rc_instruction * inst;
 
 	c->Program.InputsRead &= ~(1 << input);
 
 	for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) {
-		const unsigned numsrcs = _mesa_num_inst_src_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 		unsigned i;
 
-		for(i = 0; i < numsrcs; ++i) {
-			if (inst->I.SrcReg[i].File == PROGRAM_INPUT && inst->I.SrcReg[i].Index == input) {
-				inst->I.SrcReg[i].File = new_input.File;
-				inst->I.SrcReg[i].Index = new_input.Index;
-				inst->I.SrcReg[i].Swizzle = combine_swizzles(new_input.Swizzle, inst->I.SrcReg[i].Swizzle);
-				if (!inst->I.SrcReg[i].Abs) {
-					inst->I.SrcReg[i].Negate ^= new_input.Negate;
-					inst->I.SrcReg[i].Abs = new_input.Abs;
+		for(i = 0; i < opcode->NumSrcRegs; ++i) {
+			if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT && inst->U.I.SrcReg[i].Index == input) {
+				inst->U.I.SrcReg[i].File = new_input.File;
+				inst->U.I.SrcReg[i].Index = new_input.Index;
+				inst->U.I.SrcReg[i].Swizzle = combine_swizzles(new_input.Swizzle, inst->U.I.SrcReg[i].Swizzle);
+				if (!inst->U.I.SrcReg[i].Abs) {
+					inst->U.I.SrcReg[i].Negate ^= new_input.Negate;
+					inst->U.I.SrcReg[i].Abs = new_input.Abs;
 				}
 
 				c->Program.InputsRead |= 1 << new_input.Index;
@@ -134,12 +171,12 @@ void rc_move_output(struct radeon_compiler * c, unsigned output, unsigned new_ou
 	c->Program.OutputsWritten &= ~(1 << output);
 
 	for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) {
-		const unsigned numdsts = _mesa_num_inst_dst_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 
-		if (numdsts) {
-			if (inst->I.DstReg.File == PROGRAM_OUTPUT && inst->I.DstReg.Index == output) {
-				inst->I.DstReg.Index = new_output;
-				inst->I.DstReg.WriteMask &= writemask;
+		if (opcode->HasDstReg) {
+			if (inst->U.I.DstReg.File == RC_FILE_OUTPUT && inst->U.I.DstReg.Index == output) {
+				inst->U.I.DstReg.Index = new_output;
+				inst->U.I.DstReg.WriteMask &= writemask;
 
 				c->Program.OutputsWritten |= 1 << new_output;
 			}
@@ -157,33 +194,33 @@ void rc_copy_output(struct radeon_compiler * c, unsigned output, unsigned dup_ou
 	struct rc_instruction * inst;
 
 	for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) {
-		const unsigned numdsts = _mesa_num_inst_dst_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 
-		if (numdsts) {
-			if (inst->I.DstReg.File == PROGRAM_OUTPUT && inst->I.DstReg.Index == output) {
-				inst->I.DstReg.File = PROGRAM_TEMPORARY;
-				inst->I.DstReg.Index = tempreg;
+		if (opcode->HasDstReg) {
+			if (inst->U.I.DstReg.File == RC_FILE_OUTPUT && inst->U.I.DstReg.Index == output) {
+				inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+				inst->U.I.DstReg.Index = tempreg;
 			}
 		}
 	}
 
 	inst = rc_insert_new_instruction(c, c->Program.Instructions.Prev);
-	inst->I.Opcode = OPCODE_MOV;
-	inst->I.DstReg.File = PROGRAM_OUTPUT;
-	inst->I.DstReg.Index = output;
+	inst->U.I.Opcode = RC_OPCODE_MOV;
+	inst->U.I.DstReg.File = RC_FILE_OUTPUT;
+	inst->U.I.DstReg.Index = output;
 
-	inst->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-	inst->I.SrcReg[0].Index = tempreg;
-	inst->I.SrcReg[0].Swizzle = SWIZZLE_XYZW;
+	inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+	inst->U.I.SrcReg[0].Index = tempreg;
+	inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
 
 	inst = rc_insert_new_instruction(c, c->Program.Instructions.Prev);
-	inst->I.Opcode = OPCODE_MOV;
-	inst->I.DstReg.File = PROGRAM_OUTPUT;
-	inst->I.DstReg.Index = dup_output;
+	inst->U.I.Opcode = RC_OPCODE_MOV;
+	inst->U.I.DstReg.File = RC_FILE_OUTPUT;
+	inst->U.I.DstReg.Index = dup_output;
 
-	inst->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-	inst->I.SrcReg[0].Index = tempreg;
-	inst->I.SrcReg[0].Swizzle = SWIZZLE_XYZW;
+	inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+	inst->U.I.SrcReg[0].Index = tempreg;
+	inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
 
 	c->Program.OutputsWritten |= 1 << dup_output;
 }
@@ -201,60 +238,60 @@ void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsig
 
 	/* perspective divide */
 	struct rc_instruction * inst_rcp = rc_insert_new_instruction(c, &c->Program.Instructions);
-	inst_rcp->I.Opcode = OPCODE_RCP;
+	inst_rcp->U.I.Opcode = RC_OPCODE_RCP;
 
-	inst_rcp->I.DstReg.File = PROGRAM_TEMPORARY;
-	inst_rcp->I.DstReg.Index = tempregi;
-	inst_rcp->I.DstReg.WriteMask = WRITEMASK_W;
+	inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY;
+	inst_rcp->U.I.DstReg.Index = tempregi;
+	inst_rcp->U.I.DstReg.WriteMask = RC_MASK_W;
 
-	inst_rcp->I.SrcReg[0].File = PROGRAM_INPUT;
-	inst_rcp->I.SrcReg[0].Index = new_input;
-	inst_rcp->I.SrcReg[0].Swizzle = SWIZZLE_WWWW;
+	inst_rcp->U.I.SrcReg[0].File = RC_FILE_INPUT;
+	inst_rcp->U.I.SrcReg[0].Index = new_input;
+	inst_rcp->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_WWWW;
 
 	struct rc_instruction * inst_mul = rc_insert_new_instruction(c, inst_rcp);
-	inst_mul->I.Opcode = OPCODE_MUL;
+	inst_mul->U.I.Opcode = RC_OPCODE_MUL;
 
-	inst_mul->I.DstReg.File = PROGRAM_TEMPORARY;
-	inst_mul->I.DstReg.Index = tempregi;
-	inst_mul->I.DstReg.WriteMask = WRITEMASK_XYZ;
+	inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY;
+	inst_mul->U.I.DstReg.Index = tempregi;
+	inst_mul->U.I.DstReg.WriteMask = RC_MASK_XYZ;
 
-	inst_mul->I.SrcReg[0].File = PROGRAM_INPUT;
-	inst_mul->I.SrcReg[0].Index = new_input;
+	inst_mul->U.I.SrcReg[0].File = RC_FILE_INPUT;
+	inst_mul->U.I.SrcReg[0].Index = new_input;
 
-	inst_mul->I.SrcReg[1].File = PROGRAM_TEMPORARY;
-	inst_mul->I.SrcReg[1].Index = tempregi;
-	inst_mul->I.SrcReg[1].Swizzle = SWIZZLE_WWWW;
+	inst_mul->U.I.SrcReg[1].File = RC_FILE_TEMPORARY;
+	inst_mul->U.I.SrcReg[1].Index = tempregi;
+	inst_mul->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_WWWW;
 
 	/* viewport transformation */
 	struct rc_instruction * inst_mad = rc_insert_new_instruction(c, inst_mul);
-	inst_mad->I.Opcode = OPCODE_MAD;
+	inst_mad->U.I.Opcode = RC_OPCODE_MAD;
 
-	inst_mad->I.DstReg.File = PROGRAM_TEMPORARY;
-	inst_mad->I.DstReg.Index = tempregi;
-	inst_mad->I.DstReg.WriteMask = WRITEMASK_XYZ;
+	inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY;
+	inst_mad->U.I.DstReg.Index = tempregi;
+	inst_mad->U.I.DstReg.WriteMask = RC_MASK_XYZ;
 
-	inst_mad->I.SrcReg[0].File = PROGRAM_TEMPORARY;
-	inst_mad->I.SrcReg[0].Index = tempregi;
-	inst_mad->I.SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+	inst_mad->U.I.SrcReg[0].File = RC_FILE_TEMPORARY;
+	inst_mad->U.I.SrcReg[0].Index = tempregi;
+	inst_mad->U.I.SrcReg[0].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_ZERO);
 
-	inst_mad->I.SrcReg[1].File = PROGRAM_STATE_VAR;
-	inst_mad->I.SrcReg[1].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_WINDOW_DIMENSION, 0);
-	inst_mad->I.SrcReg[1].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+	inst_mad->U.I.SrcReg[1].File = RC_FILE_CONSTANT;
+	inst_mad->U.I.SrcReg[1].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_WINDOW_DIMENSION, 0);
+	inst_mad->U.I.SrcReg[1].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_ZERO);
 
-	inst_mad->I.SrcReg[2].File = PROGRAM_STATE_VAR;
-	inst_mad->I.SrcReg[2].Index = inst_mad->I.SrcReg[1].Index;
-	inst_mad->I.SrcReg[2].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+	inst_mad->U.I.SrcReg[2].File = RC_FILE_CONSTANT;
+	inst_mad->U.I.SrcReg[2].Index = inst_mad->U.I.SrcReg[1].Index;
+	inst_mad->U.I.SrcReg[2].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_ZERO);
 
 	struct rc_instruction * inst;
 	for (inst = inst_mad->Next; inst != &c->Program.Instructions; inst = inst->Next) {
-		const unsigned numsrcs = _mesa_num_inst_src_regs(inst->I.Opcode);
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 		unsigned i;
 
-		for(i = 0; i < numsrcs; i++) {
-			if (inst->I.SrcReg[i].File == PROGRAM_INPUT &&
-			    inst->I.SrcReg[i].Index == wpos) {
-				inst->I.SrcReg[i].File = PROGRAM_TEMPORARY;
-				inst->I.SrcReg[i].Index = tempregi;
+		for(i = 0; i < opcode->NumSrcRegs; i++) {
+			if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT &&
+			    inst->U.I.SrcReg[i].Index == wpos) {
+				inst->U.I.SrcReg[i].File = RC_FILE_TEMPORARY;
+				inst->U.I.SrcReg[i].Index = tempregi;
 			}
 		}
 	}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h
index e63ab8840a..87a732cd90 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h
@@ -23,35 +23,11 @@
 #ifndef RADEON_COMPILER_H
 #define RADEON_COMPILER_H
 
-#include "main/mtypes.h"
-#include "shader/prog_instruction.h"
-
 #include "memory_pool.h"
 #include "radeon_code.h"
+#include "radeon_program.h"
 
-
-struct rc_instruction {
-	struct rc_instruction * Prev;
-	struct rc_instruction * Next;
-	struct prog_instruction I;
-};
-
-struct rc_program {
-	/**
-	 * Instructions.Next points to the first instruction,
-	 * Instructions.Prev points to the last instruction.
-	 */
-	struct rc_instruction Instructions;
-
-	/* Long term, we should probably remove InputsRead & OutputsWritten,
-	 * since updating dependent state can be fragile, and they aren't
-	 * actually used very often. */
-	uint32_t InputsRead;
-	uint32_t OutputsWritten;
-	uint32_t ShadowSamplers; /**< Texture units used for shadow sampling. */
-
-	struct rc_constant_list Constants;
-};
+struct rc_swizzle_caps;
 
 struct radeon_compiler {
 	struct memory_pool Pool;
@@ -59,6 +35,14 @@ struct radeon_compiler {
 	unsigned Debug:1;
 	unsigned Error:1;
 	char * ErrorMsg;
+
+	/**
+	 * Variables used internally, not be touched by callers
+	 * of the compiler
+	 */
+	/*@{*/
+	struct rc_swizzle_caps * SwizzleCaps;
+	/*@}*/
 };
 
 void rc_init(struct radeon_compiler * c);
@@ -67,11 +51,26 @@ void rc_destroy(struct radeon_compiler * c);
 void rc_debug(struct radeon_compiler * c, const char * fmt, ...);
 void rc_error(struct radeon_compiler * c, const char * fmt, ...);
 
-void rc_mesa_to_rc_program(struct radeon_compiler * c, struct gl_program * program);
+int rc_if_fail_helper(struct radeon_compiler * c, const char * file, int line, const char * assertion);
+
+/**
+ * This macro acts like an if-statement that can be used to implement
+ * non-aborting assertions in the compiler.
+ *
+ * It checks whether \p cond is true. If not, an internal compiler error is
+ * flagged and the if-clause is run.
+ *
+ * A typical use-case would be:
+ *
+ *  if (rc_assert(c, condition-that-must-be-true))
+ *  	return;
+ */
+#define rc_assert(c, cond) \
+	(!(cond) && rc_if_fail_helper(c, __FILE__, __LINE__, #cond))
 
 void rc_calculate_inputs_outputs(struct radeon_compiler * c);
 
-void rc_move_input(struct radeon_compiler * c, unsigned input, struct prog_src_register new_input);
+void rc_move_input(struct radeon_compiler * c, unsigned input, struct rc_src_register new_input);
 void rc_move_output(struct radeon_compiler * c, unsigned output, unsigned new_output, unsigned writemask);
 void rc_copy_output(struct radeon_compiler * c, unsigned output, unsigned dup_output);
 void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsigned new_input);
@@ -97,7 +96,7 @@ void r3xx_compile_fragment_program(struct r300_fragment_program_compiler* c);
 struct r300_vertex_program_compiler {
 	struct radeon_compiler Base;
 	struct r300_vertex_program_code *code;
-	GLbitfield RequiredOutputs;
+	uint32_t RequiredOutputs;
 
 	void * UserData;
 	void (*SetHwInputOutput)(struct r300_vertex_program_compiler * c);
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow.c b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow.c
new file mode 100644
index 0000000000..cce9166e64
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow.c
@@ -0,0 +1,162 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_dataflow.h"
+
+#include "radeon_program.h"
+
+
+static void reads_normal(struct rc_instruction * fullinst, rc_read_write_fn cb, void * userdata)
+{
+	struct rc_sub_instruction * inst = &fullinst->U.I;
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode);
+
+	for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) {
+		unsigned int refmask = 0;
+
+		if (inst->SrcReg[src].File == RC_FILE_NONE)
+			return;
+
+		for(unsigned int chan = 0; chan < 4; ++chan)
+			refmask |= 1 << GET_SWZ(inst->SrcReg[src].Swizzle, chan);
+
+		refmask &= RC_MASK_XYZW;
+
+		for(unsigned int chan = 0; chan < 4; ++chan) {
+			if (GET_BIT(refmask, chan)) {
+				cb(userdata, fullinst, inst->SrcReg[src].File, inst->SrcReg[src].Index, chan);
+			}
+		}
+
+		if (refmask && inst->SrcReg[src].RelAddr)
+			cb(userdata, fullinst, RC_FILE_ADDRESS, 0, RC_MASK_X);
+	}
+}
+
+static void reads_pair(struct rc_instruction * fullinst,  rc_read_write_fn cb, void * userdata)
+{
+	struct rc_pair_instruction * inst = &fullinst->U.P;
+	unsigned int refmasks[3] = { 0, 0, 0 };
+
+	if (inst->RGB.Opcode != RC_OPCODE_NOP) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->RGB.Opcode);
+
+		for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) {
+			for(unsigned int chan = 0; chan < 3; ++chan) {
+				unsigned int swz = GET_SWZ(inst->RGB.Arg[arg].Swizzle, chan);
+				if (swz < 4)
+					refmasks[inst->RGB.Arg[arg].Source] |= 1 << swz;
+			}
+		}
+	}
+
+	if (inst->Alpha.Opcode != RC_OPCODE_NOP) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Alpha.Opcode);
+
+		for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) {
+			if (inst->Alpha.Arg[arg].Swizzle < 4)
+				refmasks[inst->Alpha.Arg[arg].Source] |= 1 << inst->Alpha.Arg[arg].Swizzle;
+		}
+	}
+
+	for(unsigned int src = 0; src < 3; ++src) {
+		if (inst->RGB.Src[src].Used) {
+			for(unsigned int chan = 0; chan < 3; ++chan) {
+				if (GET_BIT(refmasks[src], chan))
+					cb(userdata, fullinst, inst->RGB.Src[src].File, inst->RGB.Src[src].Index, chan);
+			}
+		}
+
+		if (inst->Alpha.Src[src].Used) {
+			if (GET_BIT(refmasks[src], 3))
+				cb(userdata, fullinst, inst->Alpha.Src[src].File, inst->Alpha.Src[src].Index, 3);
+		}
+	}
+}
+
+/**
+ * Calls a callback function for all sourced register channels.
+ *
+ * This is conservative, i.e. channels may be called multiple times,
+ * and the writemask of the instruction is not taken into account.
+ */
+void rc_for_all_reads(struct rc_instruction * inst, rc_read_write_fn cb, void * userdata)
+{
+	if (inst->Type == RC_INSTRUCTION_NORMAL) {
+		reads_normal(inst, cb, userdata);
+	} else {
+		reads_pair(inst, cb, userdata);
+	}
+}
+
+
+
+static void writes_normal(struct rc_instruction * fullinst, rc_read_write_fn cb, void * userdata)
+{
+	struct rc_sub_instruction * inst = &fullinst->U.I;
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode);
+
+	if (opcode->HasDstReg) {
+		for(unsigned int chan = 0; chan < 4; ++chan) {
+			if (GET_BIT(inst->DstReg.WriteMask, chan))
+				cb(userdata, fullinst, inst->DstReg.File, inst->DstReg.Index, chan);
+		}
+	}
+
+	if (inst->WriteALUResult)
+		cb(userdata, fullinst, RC_FILE_SPECIAL, RC_SPECIAL_ALU_RESULT, 0);
+}
+
+static void writes_pair(struct rc_instruction * fullinst, rc_read_write_fn cb, void * userdata)
+{
+	struct rc_pair_instruction * inst = &fullinst->U.P;
+
+	for(unsigned int chan = 0; chan < 3; ++chan) {
+		if (GET_BIT(inst->RGB.WriteMask, chan))
+			cb(userdata, fullinst, RC_FILE_TEMPORARY, inst->RGB.DestIndex, chan);
+	}
+
+	if (inst->Alpha.WriteMask)
+		cb(userdata, fullinst, RC_FILE_TEMPORARY, inst->Alpha.DestIndex, 3);
+
+	if (inst->WriteALUResult)
+		cb(userdata, fullinst, RC_FILE_SPECIAL, RC_SPECIAL_ALU_RESULT, 0);
+}
+
+/**
+ * Calls a callback function for all written register channels.
+ *
+ * \warning Does not report output registers for paired instructions!
+ */
+void rc_for_all_writes(struct rc_instruction * inst, rc_read_write_fn cb, void * userdata)
+{
+	if (inst->Type == RC_INSTRUCTION_NORMAL) {
+		writes_normal(inst, cb, userdata);
+	} else {
+		writes_pair(inst, cb, userdata);
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow.h b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow.h
new file mode 100644
index 0000000000..5aa4cb64f3
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef RADEON_DATAFLOW_H
+#define RADEON_DATAFLOW_H
+
+#include "radeon_program_constants.h"
+
+struct radeon_compiler;
+struct rc_instruction;
+struct rc_swizzle_caps;
+
+
+/**
+ * Help analyze the register accesses of instructions.
+ */
+/*@{*/
+typedef void (*rc_read_write_fn)(void * userdata, struct rc_instruction * inst,
+			rc_register_file file, unsigned int index, unsigned int chan);
+void rc_for_all_reads(struct rc_instruction * inst, rc_read_write_fn cb, void * userdata);
+void rc_for_all_writes(struct rc_instruction * inst, rc_read_write_fn cb, void * userdata);
+/*@}*/
+
+
+/**
+ * Compiler passes based on dataflow analysis.
+ */
+/*@{*/
+typedef void (*rc_dataflow_mark_outputs_fn)(void * userdata, void * data,
+			void (*mark_fn)(void * data, unsigned int index, unsigned int mask));
+void rc_dataflow_deadcode(struct radeon_compiler * c, rc_dataflow_mark_outputs_fn dce, void * userdata);
+void rc_dataflow_swizzles(struct radeon_compiler * c);
+/*@}*/
+
+#endif /* RADEON_DATAFLOW_H */
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c
new file mode 100644
index 0000000000..e0c66c4aeb
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c
@@ -0,0 +1,295 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_dataflow.h"
+
+#include "radeon_compiler.h"
+
+
+struct updatemask_state {
+	unsigned char Output[RC_REGISTER_MAX_INDEX];
+	unsigned char Temporary[RC_REGISTER_MAX_INDEX];
+	unsigned char Address;
+	unsigned char Special[RC_NUM_SPECIAL_REGISTERS];
+};
+
+struct instruction_state {
+	unsigned char WriteMask:4;
+	unsigned char WriteALUResult:1;
+	unsigned char SrcReg[3];
+};
+
+struct branchinfo {
+	unsigned int HaveElse:1;
+
+	struct updatemask_state StoreEndif;
+	struct updatemask_state StoreElse;
+};
+
+struct deadcode_state {
+	struct radeon_compiler * C;
+	struct instruction_state * Instructions;
+
+	struct updatemask_state R;
+
+	struct branchinfo * BranchStack;
+	unsigned int BranchStackSize;
+	unsigned int BranchStackReserved;
+};
+
+
+static void or_updatemasks(
+	struct updatemask_state * dst,
+	struct updatemask_state * a,
+	struct updatemask_state * b)
+{
+	for(unsigned int i = 0; i < RC_REGISTER_MAX_INDEX; ++i) {
+		dst->Output[i] = a->Output[i] | b->Output[i];
+		dst->Temporary[i] = a->Temporary[i] | b->Temporary[i];
+	}
+
+	for(unsigned int i = 0; i < RC_NUM_SPECIAL_REGISTERS; ++i)
+		dst->Special[i] = a->Special[i] | b->Special[i];
+
+	dst->Address = a->Address | b->Address;
+}
+
+static void push_branch(struct deadcode_state * s)
+{
+	if (s->BranchStackSize >= s->BranchStackReserved) {
+		unsigned int new_reserve = 2 * s->BranchStackReserved;
+		struct branchinfo * new_stack;
+
+		if (!new_reserve)
+			new_reserve = 4;
+
+		new_stack = memory_pool_malloc(&s->C->Pool, new_reserve * sizeof(struct branchinfo));
+		memcpy(new_stack, s->BranchStack, s->BranchStackSize * sizeof(struct branchinfo));
+
+		s->BranchStack = new_stack;
+		s->BranchStackReserved = new_reserve;
+	}
+
+	struct branchinfo * branch = &s->BranchStack[s->BranchStackSize++];
+	branch->HaveElse = 0;
+	memcpy(&branch->StoreEndif, &s->R, sizeof(s->R));
+}
+
+static unsigned char * get_used_ptr(struct deadcode_state *s, rc_register_file file, unsigned int index)
+{
+	if (file == RC_FILE_OUTPUT || file == RC_FILE_TEMPORARY) {
+		if (index >= RC_REGISTER_MAX_INDEX) {
+			rc_error(s->C, "%s: index %i is out of bounds for file %i\n", __FUNCTION__, index, file);
+			return 0;
+		}
+
+		if (file == RC_FILE_OUTPUT)
+			return &s->R.Output[index];
+		else
+			return &s->R.Temporary[index];
+	} else if (file == RC_FILE_ADDRESS) {
+		return &s->R.Address;
+	} else if (file == RC_FILE_SPECIAL) {
+		if (index >= RC_NUM_SPECIAL_REGISTERS) {
+			rc_error(s->C, "%s: special file index %i out of bounds\n", __FUNCTION__, index);
+			return 0;
+		}
+
+		return &s->R.Special[index];
+	}
+
+	return 0;
+}
+
+static void mark_used(struct deadcode_state * s, rc_register_file file, unsigned int index, unsigned int mask)
+{
+	unsigned char * pused = get_used_ptr(s, file, index);
+	if (pused)
+		*pused |= mask;
+}
+
+static void update_instruction(struct deadcode_state * s, struct rc_instruction * inst)
+{
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+	struct instruction_state * insts = &s->Instructions[inst->IP];
+	unsigned int usedmask = 0;
+
+	if (opcode->HasDstReg) {
+		unsigned char * pused = get_used_ptr(s, inst->U.I.DstReg.File, inst->U.I.DstReg.Index);
+		if (pused) {
+			usedmask = *pused & inst->U.I.DstReg.WriteMask;
+			*pused &= ~usedmask;
+		}
+	}
+
+	insts->WriteMask |= usedmask;
+
+	if (inst->U.I.WriteALUResult) {
+		unsigned char * pused = get_used_ptr(s, RC_FILE_SPECIAL, RC_SPECIAL_ALU_RESULT);
+		if (pused && *pused) {
+			if (inst->U.I.WriteALUResult == RC_ALURESULT_X)
+				usedmask |= RC_MASK_X;
+			else if (inst->U.I.WriteALUResult == RC_ALURESULT_W)
+				usedmask |= RC_MASK_W;
+
+			*pused = 0;
+			insts->WriteALUResult = 1;
+		}
+	}
+
+	unsigned int srcmasks[3];
+	rc_compute_sources_for_writemask(opcode, usedmask, srcmasks);
+
+	for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) {
+		unsigned int refmask = 0;
+		unsigned int newsrcmask = srcmasks[src] & ~insts->SrcReg[src];
+		insts->SrcReg[src] |= newsrcmask;
+
+		for(unsigned int chan = 0; chan < 4; ++chan) {
+			if (GET_BIT(newsrcmask, chan))
+				refmask |= 1 << GET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan);
+		}
+
+		/* get rid of spurious bits from ZERO, ONE, etc. swizzles */
+		refmask &= RC_MASK_XYZW;
+
+		if (!refmask)
+			continue;
+
+		mark_used(s, inst->U.I.SrcReg[src].File, inst->U.I.SrcReg[src].Index, refmask);
+
+		if (inst->U.I.SrcReg[src].RelAddr)
+			mark_used(s, RC_FILE_ADDRESS, 0, RC_MASK_X);
+	}
+}
+
+static void mark_output_use(void * data, unsigned int index, unsigned int mask)
+{
+	struct deadcode_state * s = data;
+
+	mark_used(s, RC_FILE_OUTPUT, index, mask);
+}
+
+void rc_dataflow_deadcode(struct radeon_compiler * c, rc_dataflow_mark_outputs_fn dce, void * userdata)
+{
+	struct deadcode_state s;
+	unsigned int nr_instructions;
+
+	memset(&s, 0, sizeof(s));
+	s.C = c;
+
+	nr_instructions = rc_recompute_ips(c);
+	s.Instructions = memory_pool_malloc(&c->Pool, sizeof(struct instruction_state)*nr_instructions);
+	memset(s.Instructions, 0, sizeof(struct instruction_state)*nr_instructions);
+
+	dce(userdata, &s, &mark_output_use);
+
+	for(struct rc_instruction * inst = c->Program.Instructions.Prev;
+	    inst != &c->Program.Instructions;
+	    inst = inst->Prev) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+
+		if (opcode->IsFlowControl) {
+			if (opcode->Opcode == RC_OPCODE_ENDIF) {
+				push_branch(&s);
+			} else {
+				if (s.BranchStackSize) {
+					struct branchinfo * branch = &s.BranchStack[s.BranchStackSize-1];
+
+					if (opcode->Opcode == RC_OPCODE_IF) {
+						or_updatemasks(&s.R,
+								&s.R,
+								branch->HaveElse ? &branch->StoreElse : &branch->StoreEndif);
+
+						s.BranchStackSize--;
+					} else if (opcode->Opcode == RC_OPCODE_ELSE) {
+						if (branch->HaveElse) {
+							rc_error(c, "%s: Multiple ELSE for one IF/ENDIF\n", __FUNCTION__);
+						} else {
+							memcpy(&branch->StoreElse, &s.R, sizeof(s.R));
+							memcpy(&s.R, &branch->StoreEndif, sizeof(s.R));
+							branch->HaveElse = 1;
+						}
+					} else {
+						rc_error(c, "%s: Unhandled control flow instruction %s\n", __FUNCTION__, opcode->Name);
+					}
+				} else {
+					rc_error(c, "%s: Unexpected control flow instruction\n", __FUNCTION__);
+				}
+			}
+		}
+
+		update_instruction(&s, inst);
+	}
+
+	unsigned int ip = 0;
+	for(struct rc_instruction * inst = c->Program.Instructions.Next;
+	    inst != &c->Program.Instructions;
+	    inst = inst->Next, ++ip) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);\
+		int dead = 1;
+
+		if (!opcode->HasDstReg) {
+			dead = 0;
+		} else {
+			inst->U.I.DstReg.WriteMask = s.Instructions[ip].WriteMask;
+			if (s.Instructions[ip].WriteMask)
+				dead = 0;
+
+			if (s.Instructions[ip].WriteALUResult)
+				dead = 0;
+			else
+				inst->U.I.WriteALUResult = RC_ALURESULT_NONE;
+		}
+
+		if (dead) {
+			struct rc_instruction * todelete = inst;
+			inst = inst->Prev;
+			rc_remove_instruction(todelete);
+			continue;
+		}
+
+		unsigned int srcmasks[3];
+		unsigned int usemask = s.Instructions[ip].WriteMask;
+
+		if (inst->U.I.WriteALUResult == RC_ALURESULT_X)
+			usemask |= RC_MASK_X;
+		else if (inst->U.I.WriteALUResult == RC_ALURESULT_W)
+			usemask |= RC_MASK_W;
+
+		rc_compute_sources_for_writemask(opcode, usemask, srcmasks);
+
+		for(unsigned int src = 0; src < 3; ++src) {
+			for(unsigned int chan = 0; chan < 4; ++chan) {
+				if (!GET_BIT(srcmasks[src], chan))
+					SET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan, RC_SWIZZLE_UNUSED);
+			}
+		}
+	}
+
+	rc_calculate_inputs_outputs(c);
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c
new file mode 100644
index 0000000000..33acbd30f4
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c
@@ -0,0 +1,102 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_dataflow.h"
+
+#include "radeon_compiler.h"
+#include "radeon_swizzle.h"
+
+
+static void rewrite_source(struct radeon_compiler * c,
+		struct rc_instruction * inst, unsigned src)
+{
+	struct rc_swizzle_split split;
+	unsigned int tempreg = rc_find_free_temporary(c);
+	unsigned int usemask;
+
+	usemask = 0;
+	for(unsigned int chan = 0; chan < 4; ++chan) {
+		if (GET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan) != RC_SWIZZLE_UNUSED)
+			usemask |= 1 << chan;
+	}
+
+	c->SwizzleCaps->Split(inst->U.I.SrcReg[src], usemask, &split);
+
+	for(unsigned int phase = 0; phase < split.NumPhases; ++phase) {
+		struct rc_instruction * mov = rc_insert_new_instruction(c, inst->Prev);
+		unsigned int phase_refmask;
+		unsigned int masked_negate;
+
+		mov->U.I.Opcode = RC_OPCODE_MOV;
+		mov->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		mov->U.I.DstReg.Index = tempreg;
+		mov->U.I.DstReg.WriteMask = split.Phase[phase];
+		mov->U.I.SrcReg[0] = inst->U.I.SrcReg[src];
+
+		phase_refmask = 0;
+		for(unsigned int chan = 0; chan < 4; ++chan) {
+			if (!GET_BIT(split.Phase[phase], chan))
+				SET_SWZ(mov->U.I.SrcReg[0].Swizzle, chan, RC_SWIZZLE_UNUSED);
+			else
+				phase_refmask |= 1 << GET_SWZ(mov->U.I.SrcReg[0].Swizzle, chan);
+		}
+
+		phase_refmask &= RC_MASK_XYZW;
+
+		masked_negate = split.Phase[phase] & mov->U.I.SrcReg[0].Negate;
+		if (masked_negate == 0)
+			mov->U.I.SrcReg[0].Negate = 0;
+		else if (masked_negate == split.Phase[phase])
+			mov->U.I.SrcReg[0].Negate = RC_MASK_XYZW;
+
+	}
+
+	inst->U.I.SrcReg[src].File = RC_FILE_TEMPORARY;
+	inst->U.I.SrcReg[src].Index = tempreg;
+	inst->U.I.SrcReg[src].Swizzle = 0;
+	inst->U.I.SrcReg[src].Negate = RC_MASK_NONE;
+	inst->U.I.SrcReg[src].Abs = 0;
+	for(unsigned int chan = 0; chan < 4; ++chan) {
+		SET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan,
+				GET_BIT(usemask, chan) ? chan : RC_SWIZZLE_UNUSED);
+	}
+}
+
+void rc_dataflow_swizzles(struct radeon_compiler * c)
+{
+	struct rc_instruction * inst;
+
+	for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+		unsigned int src;
+
+		for(src = 0; src < opcode->NumSrcRegs; ++src) {
+			if (!c->SwizzleCaps->IsNative(inst->U.I.Opcode, inst->U.I.SrcReg[src]))
+				rewrite_source(c, inst, src);
+		}
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_nqssadce.c b/src/mesa/drivers/dri/r300/compiler/radeon_nqssadce.c
deleted file mode 100644
index aaaa50ad1f..0000000000
--- a/src/mesa/drivers/dri/r300/compiler/radeon_nqssadce.c
+++ /dev/null
@@ -1,294 +0,0 @@
-/*
- * Copyright (C) 2008 Nicolai Haehnle.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/**
- * @file
- *
- * "Not-quite SSA" and Dead-Code Elimination.
- *
- * @note This code uses SWIZZLE_NIL in a source register to indicate that
- * the corresponding component is ignored by the corresponding instruction.
- */
-
-#include "radeon_nqssadce.h"
-
-#include "radeon_compiler.h"
-
-
-/**
- * Return the @ref register_state for the given register (or 0 for untracked
- * registers, i.e. constants).
- */
-static struct register_state *get_reg_state(struct nqssadce_state* s, GLuint file, GLuint index)
-{
-	switch(file) {
-	case PROGRAM_TEMPORARY: return &s->Temps[index];
-	case PROGRAM_OUTPUT: return &s->Outputs[index];
-	case PROGRAM_ADDRESS: return &s->Address;
-	default: return 0;
-	}
-}
-
-
-/**
- * Left multiplication of a register with a swizzle
- *
- * @note Works correctly only for X, Y, Z, W swizzles, not for constant swizzles.
- */
-struct prog_src_register lmul_swizzle(GLuint swizzle, struct prog_src_register srcreg)
-{
-	struct prog_src_register tmp = srcreg;
-	int i;
-	tmp.Swizzle = 0;
-	tmp.Negate = NEGATE_NONE;
-	for(i = 0; i < 4; ++i) {
-		GLuint swz = GET_SWZ(swizzle, i);
-		if (swz < 4) {
-			tmp.Swizzle |= GET_SWZ(srcreg.Swizzle, swz) << (i*3);
-			tmp.Negate |= GET_BIT(srcreg.Negate, swz) << i;
-		} else {
-			tmp.Swizzle |= swz << (i*3);
-		}
-	}
-	return tmp;
-}
-
-
-static void track_used_srcreg(struct nqssadce_state* s,
-	GLint src, GLuint sourced)
-{
-	struct prog_instruction * inst = &s->IP->I;
-	int i;
-	GLuint deswz_source = 0;
-
-	for(i = 0; i < 4; ++i) {
-		if (GET_BIT(sourced, i)) {
-			GLuint swz = GET_SWZ(inst->SrcReg[src].Swizzle, i);
-			deswz_source |= 1 << swz;
-		} else {
-			inst->SrcReg[src].Swizzle &= ~(7 << (3*i));
-			inst->SrcReg[src].Swizzle |= SWIZZLE_NIL << (3*i);
-		}
-	}
-
-	if (!s->Descr->IsNativeSwizzle(inst->Opcode, inst->SrcReg[src])) {
-		struct prog_dst_register dstreg = inst->DstReg;
-		dstreg.File = PROGRAM_TEMPORARY;
-		dstreg.Index = rc_find_free_temporary(s->Compiler);
-		dstreg.WriteMask = sourced;
-
-		s->Descr->BuildSwizzle(s, dstreg, inst->SrcReg[src]);
-
-		inst->SrcReg[src].File = PROGRAM_TEMPORARY;
-		inst->SrcReg[src].Index = dstreg.Index;
-		inst->SrcReg[src].Swizzle = 0;
-		inst->SrcReg[src].Negate = NEGATE_NONE;
-		inst->SrcReg[src].Abs = 0;
-		for(i = 0; i < 4; ++i) {
-			if (GET_BIT(sourced, i))
-				inst->SrcReg[src].Swizzle |= i << (3*i);
-			else
-				inst->SrcReg[src].Swizzle |= SWIZZLE_NIL << (3*i);
-		}
-		deswz_source = sourced;
-	}
-
-	struct register_state *regstate;
-
-	if (inst->SrcReg[src].RelAddr) {
-		regstate = get_reg_state(s, PROGRAM_ADDRESS, 0);
-		if (regstate)
-			regstate->Sourced |= WRITEMASK_X;
-	} else {
-		regstate = get_reg_state(s, inst->SrcReg[src].File, inst->SrcReg[src].Index);
-		if (regstate)
-			regstate->Sourced |= deswz_source & 0xf;
-	}
-}
-
-static void unalias_srcregs(struct rc_instruction *inst, GLuint oldindex, GLuint newindex)
-{
-	int nsrc = _mesa_num_inst_src_regs(inst->I.Opcode);
-	int i;
-	for(i = 0; i < nsrc; ++i)
-		if (inst->I.SrcReg[i].File == PROGRAM_TEMPORARY && inst->I.SrcReg[i].Index == oldindex)
-			inst->I.SrcReg[i].Index = newindex;
-}
-
-static void unalias_temporary(struct nqssadce_state* s, GLuint oldindex)
-{
-	GLuint newindex = rc_find_free_temporary(s->Compiler);
-	struct rc_instruction * inst;
-	for(inst = s->Compiler->Program.Instructions.Next; inst != s->IP; inst = inst->Next) {
-		if (inst->I.DstReg.File == PROGRAM_TEMPORARY && inst->I.DstReg.Index == oldindex)
-			inst->I.DstReg.Index = newindex;
-		unalias_srcregs(inst, oldindex, newindex);
-	}
-	unalias_srcregs(s->IP, oldindex, newindex);
-}
-
-
-/**
- * Handle one instruction.
- */
-static void process_instruction(struct nqssadce_state* s)
-{
-	struct prog_instruction *inst = &s->IP->I;
-	GLuint WriteMask;
-
-	if (inst->Opcode == OPCODE_END)
-		return;
-
-	if (inst->Opcode != OPCODE_KIL) {
-		struct register_state *regstate = get_reg_state(s, inst->DstReg.File, inst->DstReg.Index);
-		if (!regstate) {
-			rc_error(s->Compiler, "NqssaDce: bad destination register (%i[%i])\n",
-				inst->DstReg.File, inst->DstReg.Index);
-			return;
-		}
-
-		inst->DstReg.WriteMask &= regstate->Sourced;
-		regstate->Sourced &= ~inst->DstReg.WriteMask;
-
-		if (inst->DstReg.WriteMask == 0) {
-			struct rc_instruction * inst_remove = s->IP;
-			s->IP = s->IP->Prev;
-			rc_remove_instruction(inst_remove);
-			return;
-		}
-
-		if (inst->DstReg.File == PROGRAM_TEMPORARY && !regstate->Sourced)
-			unalias_temporary(s, inst->DstReg.Index);
-	}
-
-	WriteMask = inst->DstReg.WriteMask;
-
-	switch (inst->Opcode) {
-	case OPCODE_ARL:
-	case OPCODE_DDX:
-	case OPCODE_DDY:
-	case OPCODE_FRC:
-	case OPCODE_MOV:
-		track_used_srcreg(s, 0, WriteMask);
-		break;
-	case OPCODE_ADD:
-	case OPCODE_MAX:
-	case OPCODE_MIN:
-	case OPCODE_MUL:
-	case OPCODE_SGE:
-	case OPCODE_SLT:
-		track_used_srcreg(s, 0, WriteMask);
-		track_used_srcreg(s, 1, WriteMask);
-		break;
-	case OPCODE_CMP:
-	case OPCODE_MAD:
-		track_used_srcreg(s, 0, WriteMask);
-		track_used_srcreg(s, 1, WriteMask);
-		track_used_srcreg(s, 2, WriteMask);
-		break;
-	case OPCODE_COS:
-	case OPCODE_EX2:
-	case OPCODE_LG2:
-	case OPCODE_RCP:
-	case OPCODE_RSQ:
-	case OPCODE_SIN:
-		track_used_srcreg(s, 0, 0x1);
-		break;
-	case OPCODE_DP3:
-		track_used_srcreg(s, 0, 0x7);
-		track_used_srcreg(s, 1, 0x7);
-		break;
-	case OPCODE_DP4:
-		track_used_srcreg(s, 0, 0xf);
-		track_used_srcreg(s, 1, 0xf);
-		break;
-	case OPCODE_KIL:
-	case OPCODE_TEX:
-	case OPCODE_TXB:
-	case OPCODE_TXP:
-		track_used_srcreg(s, 0, 0xf);
-		break;
-	case OPCODE_DST:
-		track_used_srcreg(s, 0, 0x6);
-		track_used_srcreg(s, 1, 0xa);
-		break;
-	case OPCODE_EXP:
-	case OPCODE_LOG:
-	case OPCODE_POW:
-		track_used_srcreg(s, 0, 0x3);
-		break;
-	case OPCODE_LIT:
-		track_used_srcreg(s, 0, 0xb);
-		break;
-	default:
-		rc_error(s->Compiler, "NqssaDce: Unknown opcode %d\n", inst->Opcode);
-		return;
-	}
-
-	s->IP = s->IP->Prev;
-}
-
-void rc_calculate_inputs_outputs(struct radeon_compiler * c)
-{
-	struct rc_instruction *inst;
-
-	c->Program.InputsRead = 0;
-	c->Program.OutputsWritten = 0;
-
-	for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next)
-	{
-		int i;
-		int num_src_regs = _mesa_num_inst_src_regs(inst->I.Opcode);
-
-		for (i = 0; i < num_src_regs; ++i) {
-			if (inst->I.SrcReg[i].File == PROGRAM_INPUT)
-				c->Program.InputsRead |= 1 << inst->I.SrcReg[i].Index;
-		}
-
-		if (_mesa_num_inst_dst_regs(inst->I.Opcode)) {
-			if (inst->I.DstReg.File == PROGRAM_OUTPUT)
-				c->Program.OutputsWritten |= 1 << inst->I.DstReg.Index;
-		}
-	}
-}
-
-void radeonNqssaDce(struct radeon_compiler * c, struct radeon_nqssadce_descr* descr, void * data)
-{
-	struct nqssadce_state s;
-
-	_mesa_bzero(&s, sizeof(s));
-	s.Compiler = c;
-	s.Descr = descr;
-	s.UserData = data;
-	s.Descr->Init(&s);
-	s.IP = c->Program.Instructions.Prev;
-
-	while(s.IP != &c->Program.Instructions && !c->Error)
-		process_instruction(&s);
-
-	rc_calculate_inputs_outputs(c);
-}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_nqssadce.h b/src/mesa/drivers/dri/r300/compiler/radeon_nqssadce.h
deleted file mode 100644
index b3fc77a35a..0000000000
--- a/src/mesa/drivers/dri/r300/compiler/radeon_nqssadce.h
+++ /dev/null
@@ -1,91 +0,0 @@
-/*
- * Copyright (C) 2008 Nicolai Haehnle.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef __RADEON_PROGRAM_NQSSADCE_H_
-#define __RADEON_PROGRAM_NQSSADCE_H_
-
-#include "radeon_program.h"
-
-struct register_state {
-	/**
-	 * Bitmask indicating which components of the register are sourced
-	 * by later instructions.
-	 */
-	GLuint Sourced : 4;
-};
-
-/**
- * Maintain state such as which registers are used, which registers are
- * read from, etc.
- */
-struct nqssadce_state {
-	struct radeon_compiler *Compiler;
-	struct radeon_nqssadce_descr *Descr;
-
-	/**
-	 * All instructions after this instruction pointer have been dealt with.
-	 */
-	struct rc_instruction * IP;
-
-	/**
-	 * Which registers are read by subsequent instructions?
-	 */
-	struct register_state Temps[MAX_PROGRAM_TEMPS];
-	struct register_state Outputs[VERT_RESULT_MAX];
-	struct register_state Address;
-
-	void * UserData;
-};
-
-
-/**
- * This structure contains a description of the hardware in-so-far as
- * it is required for the NqSSA-DCE pass.
- */
-struct radeon_nqssadce_descr {
-	/**
-	 * Fill in which outputs
-	 */
-	void (*Init)(struct nqssadce_state *);
-
-	/**
-	 * Check whether the given swizzle, absolute and negate combination
-	 * can be implemented natively by the hardware for this opcode.
-	 */
-	GLboolean (*IsNativeSwizzle)(GLuint opcode, struct prog_src_register reg);
-
-	/**
-	 * Emit (at the current IP) the instruction MOV dst, src;
-	 * The transformation will work recursively on the emitted instruction(s).
-	 */
-	void (*BuildSwizzle)(struct nqssadce_state*, struct prog_dst_register dst, struct prog_src_register src);
-};
-
-void radeonNqssaDce(struct radeon_compiler * c, struct radeon_nqssadce_descr* descr, void * data);
-struct prog_src_register lmul_swizzle(GLuint swizzle, struct prog_src_register srcreg);
-
-#endif /* __RADEON_PROGRAM_NQSSADCE_H_ */
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_opcodes.c b/src/mesa/drivers/dri/r300/compiler/radeon_opcodes.c
new file mode 100644
index 0000000000..c1c0181fac
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_opcodes.c
@@ -0,0 +1,429 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_opcodes.h"
+
+#include "radeon_program_constants.h"
+
+struct rc_opcode_info rc_opcodes[MAX_RC_OPCODE] = {
+	{
+		.Opcode = RC_OPCODE_NOP,
+		.Name = "NOP"
+	},
+	{
+		.Opcode = RC_OPCODE_ILLEGAL_OPCODE,
+		.Name = "ILLEGAL OPCODE"
+	},
+	{
+		.Opcode = RC_OPCODE_ABS,
+		.Name = "ABS",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_ADD,
+		.Name = "ADD",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_ARL,
+		.Name = "ARL",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_CMP,
+		.Name = "CMP",
+		.NumSrcRegs = 3,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_COS,
+		.Name = "COS",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_DDX,
+		.Name = "DDX",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_DDY,
+		.Name = "DDY",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_DP3,
+		.Name = "DP3",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_DP4,
+		.Name = "DP4",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_DPH,
+		.Name = "DPH",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_DST,
+		.Name = "DST",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_EX2,
+		.Name = "EX2",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_EXP,
+		.Name = "EXP",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_FLR,
+		.Name = "FLR",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_FRC,
+		.Name = "FRC",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_KIL,
+		.Name = "KIL",
+		.NumSrcRegs = 1
+	},
+	{
+		.Opcode = RC_OPCODE_LG2,
+		.Name = "LG2",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_LIT,
+		.Name = "LIT",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_LOG,
+		.Name = "LOG",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_LRP,
+		.Name = "LRP",
+		.NumSrcRegs = 3,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_MAD,
+		.Name = "MAD",
+		.NumSrcRegs = 3,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_MAX,
+		.Name = "MAX",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_MIN,
+		.Name = "MIN",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_MOV,
+		.Name = "MOV",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_MUL,
+		.Name = "MUL",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_POW,
+		.Name = "POW",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_RCP,
+		.Name = "RCP",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_RSQ,
+		.Name = "RSQ",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SCS,
+		.Name = "SCS",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SEQ,
+		.Name = "SEQ",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SFL,
+		.Name = "SFL",
+		.NumSrcRegs = 0,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SGE,
+		.Name = "SGE",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SGT,
+		.Name = "SGT",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SIN,
+		.Name = "SIN",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsStandardScalar = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SLE,
+		.Name = "SLE",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SLT,
+		.Name = "SLT",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SNE,
+		.Name = "SNE",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SUB,
+		.Name = "SUB",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_SWZ,
+		.Name = "SWZ",
+		.NumSrcRegs = 1,
+		.HasDstReg = 1,
+		.IsComponentwise = 1
+	},
+	{
+		.Opcode = RC_OPCODE_XPD,
+		.Name = "XPD",
+		.NumSrcRegs = 2,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_TEX,
+		.Name = "TEX",
+		.HasTexture = 1,
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_TXB,
+		.Name = "TXB",
+		.HasTexture = 1,
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_TXD,
+		.Name = "TXD",
+		.HasTexture = 1,
+		.NumSrcRegs = 3,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_TXL,
+		.Name = "TXL",
+		.HasTexture = 1,
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_TXP,
+		.Name = "TXP",
+		.HasTexture = 1,
+		.NumSrcRegs = 1,
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_IF,
+		.Name = "IF",
+		.IsFlowControl = 1,
+		.NumSrcRegs = 1
+	},
+	{
+		.Opcode = RC_OPCODE_ELSE,
+		.Name = "ELSE",
+		.IsFlowControl = 1,
+		.NumSrcRegs = 0
+	},
+	{
+		.Opcode = RC_OPCODE_ENDIF,
+		.Name = "ENDIF",
+		.IsFlowControl = 1,
+		.NumSrcRegs = 0
+	},
+	{
+		.Opcode = RC_OPCODE_REPL_ALPHA,
+		.Name = "REPL_ALPHA",
+		.HasDstReg = 1
+	},
+	{
+		.Opcode = RC_OPCODE_BEGIN_TEX,
+		.Name = "BEGIN_TEX"
+	}
+};
+
+void rc_compute_sources_for_writemask(
+		const struct rc_opcode_info * opcode,
+		unsigned int writemask,
+		unsigned int *srcmasks)
+{
+	srcmasks[0] = 0;
+	srcmasks[1] = 0;
+	srcmasks[2] = 0;
+
+	if (opcode->Opcode == RC_OPCODE_KIL)
+		srcmasks[0] |= RC_MASK_XYZW;
+	else if (opcode->Opcode == RC_OPCODE_IF)
+		srcmasks[0] |= RC_MASK_X;
+
+	if (!writemask)
+		return;
+
+	if (opcode->IsComponentwise) {
+		for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src)
+			srcmasks[src] |= writemask;
+	} else if (opcode->IsStandardScalar) {
+		for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src)
+			srcmasks[src] |= RC_MASK_X;
+	} else {
+		switch(opcode->Opcode) {
+		case RC_OPCODE_ARL:
+			srcmasks[0] |= RC_MASK_X;
+			break;
+		case RC_OPCODE_DP3:
+			srcmasks[0] |= RC_MASK_XYZ;
+			srcmasks[1] |= RC_MASK_XYZ;
+			break;
+		case RC_OPCODE_DP4:
+			srcmasks[0] |= RC_MASK_XYZW;
+			srcmasks[1] |= RC_MASK_XYZW;
+			break;
+		case RC_OPCODE_TEX:
+		case RC_OPCODE_TXB:
+		case RC_OPCODE_TXP:
+			srcmasks[0] |= RC_MASK_XYZW;
+			break;
+		case RC_OPCODE_DST:
+			srcmasks[0] |= 0x6;
+			srcmasks[1] |= 0xa;
+			break;
+		case RC_OPCODE_EXP:
+		case RC_OPCODE_LOG:
+			srcmasks[0] |= RC_MASK_XY;
+			break;
+		case RC_OPCODE_LIT:
+			srcmasks[0] |= 0xb;
+			break;
+		default:
+			break;
+		}
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_opcodes.h b/src/mesa/drivers/dri/r300/compiler/radeon_opcodes.h
new file mode 100644
index 0000000000..a3c5b86954
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_opcodes.h
@@ -0,0 +1,235 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef RADEON_OPCODES_H
+#define RADEON_OPCODES_H
+
+#include <assert.h>
+
+/**
+ * Opcodes understood by the Radeon compiler.
+ */
+typedef enum {
+	RC_OPCODE_NOP = 0,
+	RC_OPCODE_ILLEGAL_OPCODE,
+
+	/** vec4 instruction: dst.c = abs(src0.c); */
+	RC_OPCODE_ABS,
+
+	/** vec4 instruction: dst.c = src0.c + src1.c; */
+	RC_OPCODE_ADD,
+
+	/** special instruction: load address register
+	 * dst.x = floor(src.x), where dst must be an address register */
+	RC_OPCODE_ARL,
+
+	/** vec4 instruction: dst.c = src0.c < 0.0 ? src1.c : src2.c */
+	RC_OPCODE_CMP,
+
+	/** scalar instruction: dst = cos(src0.x) */
+	RC_OPCODE_COS,
+
+	/** special instruction: take vec4 partial derivative in X direction
+	 * dst.c = d src0.c / dx */
+	RC_OPCODE_DDX,
+
+	/** special instruction: take vec4 partial derivative in Y direction
+	 * dst.c = d src0.c / dy */
+	RC_OPCODE_DDY,
+
+	/** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y + src0.z*src1.z */
+	RC_OPCODE_DP3,
+
+	/** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y + src0.z*src1.z + src0.w*src1.w */
+	RC_OPCODE_DP4,
+
+	/** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y + src0.z*src1.z + src1.w */
+	RC_OPCODE_DPH,
+
+	/** special instruction, see ARB_fragment_program */
+	RC_OPCODE_DST,
+
+	/** scalar instruction: dst = 2**src0.x */
+	RC_OPCODE_EX2,
+
+	/** special instruction, see ARB_vertex_program */
+	RC_OPCODE_EXP,
+
+	/** vec4 instruction: dst.c = floor(src0.c) */
+	RC_OPCODE_FLR,
+
+	/** vec4 instruction: dst.c = src0.c - floor(src0.c) */
+	RC_OPCODE_FRC,
+
+	/** special instruction: stop execution if any component of src0 is negative */
+	RC_OPCODE_KIL,
+
+	/** scalar instruction: dst = log_2(src0.x) */
+	RC_OPCODE_LG2,
+
+	/** special instruction, see ARB_vertex_program */
+	RC_OPCODE_LIT,
+
+	/** special instruction, see ARB_vertex_program */
+	RC_OPCODE_LOG,
+
+	/** vec4 instruction: dst.c = src0.c*src1.c + (1 - src0.c)*src2.c */
+	RC_OPCODE_LRP,
+
+	/** vec4 instruction: dst.c = src0.c*src1.c + src2.c */
+	RC_OPCODE_MAD,
+
+	/** vec4 instruction: dst.c = max(src0.c, src1.c) */
+	RC_OPCODE_MAX,
+
+	/** vec4 instruction: dst.c = min(src0.c, src1.c) */
+	RC_OPCODE_MIN,
+
+	/** vec4 instruction: dst.c = src0.c */
+	RC_OPCODE_MOV,
+
+	/** vec4 instruction: dst.c = src0.c*src1.c */
+	RC_OPCODE_MUL,
+
+	/** scalar instruction: dst = src0.x ** src1.x */
+	RC_OPCODE_POW,
+
+	/** scalar instruction: dst = 1 / src0.x */
+	RC_OPCODE_RCP,
+
+	/** scalar instruction: dst = 1 / sqrt(src0.x) */
+	RC_OPCODE_RSQ,
+
+	/** special instruction, see ARB_fragment_program */
+	RC_OPCODE_SCS,
+
+	/** vec4 instruction: dst.c = (src0.c == src1.c) ? 1.0 : 0.0 */
+	RC_OPCODE_SEQ,
+
+	/** vec4 instruction: dst.c = 0.0 */
+	RC_OPCODE_SFL,
+
+	/** vec4 instruction: dst.c = (src0.c >= src1.c) ? 1.0 : 0.0 */
+	RC_OPCODE_SGE,
+
+	/** vec4 instruction: dst.c = (src0.c > src1.c) ? 1.0 : 0.0 */
+	RC_OPCODE_SGT,
+
+	/** scalar instruction: dst = sin(src0.x) */
+	RC_OPCODE_SIN,
+
+	/** vec4 instruction: dst.c = (src0.c <= src1.c) ? 1.0 : 0.0 */
+	RC_OPCODE_SLE,
+
+	/** vec4 instruction: dst.c = (src0.c < src1.c) ? 1.0 : 0.0 */
+	RC_OPCODE_SLT,
+
+	/** vec4 instruction: dst.c = (src0.c != src1.c) ? 1.0 : 0.0 */
+	RC_OPCODE_SNE,
+
+	/** vec4 instruction: dst.c = src0.c - src1.c */
+	RC_OPCODE_SUB,
+
+	/** vec4 instruction: dst.c = src0.c */
+	RC_OPCODE_SWZ,
+
+	/** special instruction, see ARB_fragment_program */
+	RC_OPCODE_XPD,
+
+	RC_OPCODE_TEX,
+	RC_OPCODE_TXB,
+	RC_OPCODE_TXD,
+	RC_OPCODE_TXL,
+	RC_OPCODE_TXP,
+
+	/** branch instruction:
+	 * If src0.x != 0.0, continue with the next instruction;
+	 * otherwise, jump to matching RC_OPCODE_ELSE or RC_OPCODE_ENDIF.
+	 */
+	RC_OPCODE_IF,
+
+	/** branch instruction: jump to matching RC_OPCODE_ENDIF */
+	RC_OPCODE_ELSE,
+
+	/** branch instruction: has no effect */
+	RC_OPCODE_ENDIF,
+
+	/** special instruction, used in R300-R500 fragment program pair instructions
+	 * indicates that the result of the alpha operation shall be replicated
+	 * across all other channels */
+	RC_OPCODE_REPL_ALPHA,
+
+	/** special instruction, used in R300-R500 fragment programs
+	 * to indicate the start of a block of texture instructions that
+	 * can run simultaneously. */
+	RC_OPCODE_BEGIN_TEX,
+
+	MAX_RC_OPCODE
+} rc_opcode;
+
+
+struct rc_opcode_info {
+	rc_opcode Opcode;
+	const char * Name;
+
+	/** true if the instruction reads from a texture.
+	 *
+	 * \note This is false for the KIL instruction, even though KIL is
+	 * a texture instruction from a hardware point of view. */
+	unsigned int HasTexture:1;
+
+	unsigned int NumSrcRegs:2;
+	unsigned int HasDstReg:1;
+
+	/** true if this instruction affects control flow */
+	unsigned int IsFlowControl:1;
+
+	/** true if this is a vector instruction that operates on components in parallel
+	 * without any cross-component interaction */
+	unsigned int IsComponentwise:1;
+
+	/** true if this instruction sources only its operands X components
+	 * to compute one result which is smeared across all output channels */
+	unsigned int IsStandardScalar:1;
+};
+
+extern struct rc_opcode_info rc_opcodes[MAX_RC_OPCODE];
+
+static inline const struct rc_opcode_info * rc_get_opcode_info(rc_opcode opcode)
+{
+	assert((unsigned int)opcode < MAX_RC_OPCODE);
+	assert(rc_opcodes[opcode].Opcode == opcode);
+
+	return &rc_opcodes[opcode];
+}
+
+void rc_compute_sources_for_writemask(
+		const struct rc_opcode_info * opcode,
+		unsigned int writemask,
+		unsigned int *srcmasks);
+
+#endif /* RADEON_OPCODES_H */
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c b/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c
new file mode 100644
index 0000000000..828d0c8e28
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c
@@ -0,0 +1,350 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_program_pair.h"
+
+#include <stdio.h>
+
+#include "radeon_compiler.h"
+#include "radeon_dataflow.h"
+
+
+#define VERBOSE 0
+
+#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0)
+
+
+struct live_intervals {
+	int Start;
+	int End;
+	struct live_intervals * Next;
+};
+
+struct register_info {
+	struct live_intervals Live;
+
+	unsigned int Used:1;
+	unsigned int Allocated:1;
+	rc_register_file File:3;
+	unsigned int Index:RC_REGISTER_INDEX_BITS;
+};
+
+struct hardware_register {
+	struct live_intervals * Used;
+};
+
+struct regalloc_state {
+	struct radeon_compiler * C;
+
+	struct register_info Input[RC_REGISTER_MAX_INDEX];
+	struct register_info Temporary[RC_REGISTER_MAX_INDEX];
+
+	struct hardware_register * HwTemporary;
+	unsigned int NumHwTemporaries;
+};
+
+static void print_live_intervals(struct live_intervals * src)
+{
+	if (!src) {
+		DBG("(null)");
+		return;
+	}
+
+	while(src) {
+		DBG("(%i,%i)", src->Start, src->End);
+		src = src->Next;
+	}
+}
+
+static void add_live_intervals(struct regalloc_state * s,
+		struct live_intervals ** dst, struct live_intervals * src)
+{
+	struct live_intervals ** dst_backup = dst;
+
+	if (VERBOSE) {
+		DBG("add_live_intervals: ");
+		print_live_intervals(*dst);
+		DBG(" to ");
+		print_live_intervals(src);
+		DBG("\n");
+	}
+
+	while(src) {
+		if (*dst && (*dst)->End < src->Start) {
+			dst = &(*dst)->Next;
+		} else if (!*dst || (*dst)->Start > src->End) {
+			struct live_intervals * li = memory_pool_malloc(&s->C->Pool, sizeof(*li));
+			li->Start = src->Start;
+			li->End = src->End;
+			li->Next = *dst;
+			*dst = li;
+			src = src->Next;
+		} else {
+			if (src->End > (*dst)->End)
+				(*dst)->End = src->End;
+			if (src->Start < (*dst)->Start)
+				(*dst)->Start = src->Start;
+			src = src->Next;
+		}
+	}
+
+	if (VERBOSE) {
+		DBG("    result: ");
+		print_live_intervals(*dst_backup);
+		DBG("\n");
+	}
+}
+
+static int overlap_live_intervals(struct live_intervals * dst, struct live_intervals * src)
+{
+	if (VERBOSE) {
+		DBG("overlap_live_intervals: ");
+		print_live_intervals(dst);
+		DBG(" to ");
+		print_live_intervals(src);
+		DBG("\n");
+	}
+
+	while(src && dst) {
+		if (dst->End <= src->Start) {
+			dst = dst->Next;
+		} else if (dst->End <= src->End) {
+			DBG("    overlap\n");
+			return 1;
+		} else if (dst->Start < src->End) {
+			DBG("    overlap\n");
+			return 1;
+		} else {
+			src = src->Next;
+		}
+	}
+
+	DBG("    no overlap\n");
+
+	return 0;
+}
+
+static int try_add_live_intervals(struct regalloc_state * s,
+		struct live_intervals ** dst, struct live_intervals * src)
+{
+	if (overlap_live_intervals(*dst, src))
+		return 0;
+
+	add_live_intervals(s, dst, src);
+	return 1;
+}
+
+static void scan_callback(void * data, struct rc_instruction * inst,
+		rc_register_file file, unsigned int index, unsigned int chan)
+{
+	struct regalloc_state * s = data;
+	struct register_info * reg;
+
+	if (file == RC_FILE_TEMPORARY)
+		reg = &s->Temporary[index];
+	else if (file == RC_FILE_INPUT)
+		reg = &s->Input[index];
+	else
+		return;
+
+	if (!reg->Used) {
+		reg->Used = 1;
+		if (file == RC_FILE_INPUT)
+			reg->Live.Start = -1;
+		else
+			reg->Live.Start = inst->IP;
+		reg->Live.End = inst->IP;
+	} else {
+		if (inst->IP > reg->Live.End)
+			reg->Live.End = inst->IP;
+	}
+}
+
+static void compute_live_intervals(struct regalloc_state * s)
+{
+	rc_recompute_ips(s->C);
+
+	for(struct rc_instruction * inst = s->C->Program.Instructions.Next;
+	    inst != &s->C->Program.Instructions;
+	    inst = inst->Next) {
+		rc_for_all_reads(inst, scan_callback, s);
+		rc_for_all_writes(inst, scan_callback, s);
+	}
+}
+
+static void rewrite_register(struct regalloc_state * s,
+		rc_register_file * file, unsigned int * index)
+{
+	const struct register_info * reg;
+
+	if (*file == RC_FILE_TEMPORARY)
+		reg = &s->Temporary[*index];
+	else if (*file == RC_FILE_INPUT)
+		reg = &s->Input[*index];
+	else
+		return;
+
+	if (reg->Allocated) {
+		*file = reg->File;
+		*index = reg->Index;
+	}
+}
+
+static void rewrite_normal_instruction(struct regalloc_state * s, struct rc_sub_instruction * inst)
+{
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode);
+
+	if (opcode->HasDstReg) {
+		rc_register_file file = inst->DstReg.File;
+		unsigned int index = inst->DstReg.Index;
+
+		rewrite_register(s, &file, &index);
+
+		inst->DstReg.File = file;
+		inst->DstReg.Index = index;
+	}
+
+	for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) {
+		rc_register_file file = inst->SrcReg[src].File;
+		unsigned int index = inst->SrcReg[src].Index;
+
+		rewrite_register(s, &file, &index);
+
+		inst->SrcReg[src].File = file;
+		inst->SrcReg[src].Index = index;
+	}
+}
+
+static void rewrite_pair_instruction(struct regalloc_state * s, struct rc_pair_instruction * inst)
+{
+	if (inst->RGB.WriteMask) {
+		rc_register_file file = RC_FILE_TEMPORARY;
+		unsigned int index = inst->RGB.DestIndex;
+
+		rewrite_register(s, &file, &index);
+
+		inst->RGB.DestIndex = index;
+	}
+
+	if (inst->Alpha.WriteMask) {
+		rc_register_file file = RC_FILE_TEMPORARY;
+		unsigned int index = inst->Alpha.DestIndex;
+
+		rewrite_register(s, &file, &index);
+
+		inst->Alpha.DestIndex = index;
+	}
+
+	for(unsigned int src = 0; src < 3; ++src) {
+		if (inst->RGB.Src[src].Used) {
+			rc_register_file file = inst->RGB.Src[src].File;
+			unsigned int index = inst->RGB.Src[src].Index;
+
+			rewrite_register(s, &file, &index);
+
+			inst->RGB.Src[src].File = file;
+			inst->RGB.Src[src].Index = index;
+		}
+
+		if (inst->Alpha.Src[src].Used) {
+			rc_register_file file = inst->Alpha.Src[src].File;
+			unsigned int index = inst->Alpha.Src[src].Index;
+
+			rewrite_register(s, &file, &index);
+
+			inst->Alpha.Src[src].File = file;
+			inst->Alpha.Src[src].Index = index;
+		}
+	}
+}
+
+static void do_regalloc(struct regalloc_state * s)
+{
+	/* Simple and stupid greedy register allocation */
+	for(unsigned int index = 0; index < RC_REGISTER_MAX_INDEX; ++index) {
+		struct register_info * reg = &s->Temporary[index];
+
+		if (!reg->Used)
+			continue;
+
+		for(unsigned int hwreg = 0; hwreg < s->NumHwTemporaries; ++hwreg) {
+			if (try_add_live_intervals(s, &s->HwTemporary[hwreg].Used, &reg->Live)) {
+				reg->Allocated = 1;
+				reg->File = RC_FILE_TEMPORARY;
+				reg->Index = hwreg;
+				goto success;
+			}
+		}
+
+		rc_error(s->C, "Ran out of hardware temporaries\n");
+		return;
+
+	success:;
+	}
+
+	/* Rewrite all instructions based on the translation table we built */
+	for(struct rc_instruction * inst = s->C->Program.Instructions.Next;
+	    inst != &s->C->Program.Instructions;
+	    inst = inst->Next) {
+		if (inst->Type == RC_INSTRUCTION_NORMAL)
+			rewrite_normal_instruction(s, &inst->U.I);
+		else
+			rewrite_pair_instruction(s, &inst->U.P);
+	}
+}
+
+static void alloc_input(void * data, unsigned int input, unsigned int hwreg)
+{
+	struct regalloc_state * s = data;
+
+	if (!s->Input[input].Used)
+		return;
+
+	add_live_intervals(s, &s->HwTemporary[hwreg].Used, &s->Input[input].Live);
+
+	s->Input[input].Allocated = 1;
+	s->Input[input].File = RC_FILE_TEMPORARY;
+	s->Input[input].Index = hwreg;
+
+}
+
+void rc_pair_regalloc(struct r300_fragment_program_compiler *c, unsigned maxtemps)
+{
+	struct regalloc_state s;
+
+	memset(&s, 0, sizeof(s));
+	s.C = &c->Base;
+	s.NumHwTemporaries = maxtemps;
+	s.HwTemporary = memory_pool_malloc(&s.C->Pool, maxtemps*sizeof(struct hardware_register));
+	memset(s.HwTemporary, 0, maxtemps*sizeof(struct hardware_register));
+
+	compute_live_intervals(&s);
+
+	c->AllocateHwInputs(c, &alloc_input, &s);
+
+	do_regalloc(&s);
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_pair_schedule.c b/src/mesa/drivers/dri/r300/compiler/radeon_pair_schedule.c
new file mode 100644
index 0000000000..df67aafe02
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_pair_schedule.c
@@ -0,0 +1,501 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_program_pair.h"
+
+#include <stdio.h>
+
+#include "radeon_compiler.h"
+#include "radeon_dataflow.h"
+
+
+#define VERBOSE 0
+
+#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0)
+
+struct schedule_instruction {
+	struct rc_instruction * Instruction;
+
+	/** Next instruction in the linked list of ready instructions. */
+	struct schedule_instruction *NextReady;
+
+	/** Values that this instruction reads and writes */
+	struct reg_value * WriteValues[4];
+	struct reg_value * ReadValues[12];
+	unsigned int NumWriteValues:3;
+	unsigned int NumReadValues:4;
+
+	/**
+	 * Number of (read and write) dependencies that must be resolved before
+	 * this instruction can be scheduled.
+	 */
+	unsigned int NumDependencies:5;
+};
+
+
+/**
+ * Used to keep track of which instructions read a value.
+ */
+struct reg_value_reader {
+	struct schedule_instruction *Reader;
+	struct reg_value_reader *Next;
+};
+
+/**
+ * Used to keep track which values are stored in each component of a
+ * RC_FILE_TEMPORARY.
+ */
+struct reg_value {
+	struct schedule_instruction * Writer;
+
+	/**
+	 * Unordered linked list of instructions that read from this value.
+	 * When this value becomes available, we increase all readers'
+	 * dependency count.
+	 */
+	struct reg_value_reader *Readers;
+
+	/**
+	 * Number of readers of this value. This is decremented each time
+	 * a reader of the value is committed.
+	 * When the reader cound reaches zero, the dependency count
+	 * of the instruction writing \ref Next is decremented.
+	 */
+	unsigned int NumReaders;
+
+	struct reg_value *Next; /**< Pointer to the next value to be written to the same register */
+};
+
+struct register_state {
+	struct reg_value * Values[4];
+};
+
+struct schedule_state {
+	struct radeon_compiler * C;
+	struct schedule_instruction * Current;
+
+	struct register_state Temporary[RC_REGISTER_MAX_INDEX];
+
+	/**
+	 * Linked lists of instructions that can be scheduled right now,
+	 * based on which ALU/TEX resources they require.
+	 */
+	/*@{*/
+	struct schedule_instruction *ReadyFullALU;
+	struct schedule_instruction *ReadyRGB;
+	struct schedule_instruction *ReadyAlpha;
+	struct schedule_instruction *ReadyTEX;
+	/*@}*/
+};
+
+static struct reg_value ** get_reg_valuep(struct schedule_state * s,
+		rc_register_file file, unsigned int index, unsigned int chan)
+{
+	if (file != RC_FILE_TEMPORARY)
+		return 0;
+
+	if (index >= RC_REGISTER_MAX_INDEX) {
+		rc_error(s->C, "%s: index %i out of bounds\n", __FUNCTION__, index);
+		return 0;
+	}
+
+	return &s->Temporary[index].Values[chan];
+}
+
+static struct reg_value * get_reg_value(struct schedule_state * s,
+		rc_register_file file, unsigned int index, unsigned int chan)
+{
+	struct reg_value ** pv = get_reg_valuep(s, file, index, chan);
+	if (!pv)
+		return 0;
+	return *pv;
+}
+
+static void add_inst_to_list(struct schedule_instruction ** list, struct schedule_instruction * inst)
+{
+	inst->NextReady = *list;
+	*list = inst;
+}
+
+static void instruction_ready(struct schedule_state * s, struct schedule_instruction * sinst)
+{
+	DBG("%i is now ready\n", sinst->Instruction->IP);
+
+	if (sinst->Instruction->Type == RC_INSTRUCTION_NORMAL)
+		add_inst_to_list(&s->ReadyTEX, sinst);
+	else if (sinst->Instruction->U.P.Alpha.Opcode == RC_OPCODE_NOP)
+		add_inst_to_list(&s->ReadyRGB, sinst);
+	else if (sinst->Instruction->U.P.RGB.Opcode == RC_OPCODE_NOP)
+		add_inst_to_list(&s->ReadyAlpha, sinst);
+	else
+		add_inst_to_list(&s->ReadyFullALU, sinst);
+}
+
+static void decrease_dependencies(struct schedule_state * s, struct schedule_instruction * sinst)
+{
+	assert(sinst->NumDependencies > 0);
+	sinst->NumDependencies--;
+	if (!sinst->NumDependencies)
+		instruction_ready(s, sinst);
+}
+
+static void commit_instruction(struct schedule_state * s, struct schedule_instruction * sinst)
+{
+	DBG("%i: commit\n", sinst->Instruction->IP);
+
+	for(unsigned int i = 0; i < sinst->NumReadValues; ++i) {
+		struct reg_value * v = sinst->ReadValues[i];
+		assert(v->NumReaders > 0);
+		v->NumReaders--;
+		if (!v->NumReaders) {
+			if (v->Next)
+				decrease_dependencies(s, v->Next->Writer);
+		}
+	}
+
+	for(unsigned int i = 0; i < sinst->NumWriteValues; ++i) {
+		struct reg_value * v = sinst->WriteValues[i];
+		if (v->NumReaders) {
+			for(struct reg_value_reader * r = v->Readers; r; r = r->Next) {
+				decrease_dependencies(s, r->Reader);
+			}
+		} else {
+			/* This happens in instruction sequences of the type
+			 *  OP r.x, ...;
+			 *  OP r.x, r.x, ...;
+			 * See also the subtlety in how instructions that both
+			 * read and write the same register are scanned.
+			 */
+			if (v->Next)
+				decrease_dependencies(s, v->Next->Writer);
+		}
+	}
+}
+
+/**
+ * Emit all ready texture instructions in a single block.
+ *
+ * Emit as a single block to (hopefully) sample many textures in parallel,
+ * and to avoid hardware indirections on R300.
+ */
+static void emit_all_tex(struct schedule_state * s, struct rc_instruction * before)
+{
+	struct schedule_instruction *readytex;
+
+	assert(s->ReadyTEX);
+
+	/* Don't let the ready list change under us! */
+	readytex = s->ReadyTEX;
+	s->ReadyTEX = 0;
+
+	/* Node marker for R300 */
+	struct rc_instruction * inst_begin = rc_insert_new_instruction(s->C, before->Prev);
+	inst_begin->U.I.Opcode = RC_OPCODE_BEGIN_TEX;
+
+	/* Link texture instructions back in */
+	while(readytex) {
+		struct schedule_instruction * tex = readytex;
+		readytex = readytex->NextReady;
+
+		rc_insert_instruction(before->Prev, tex->Instruction);
+		commit_instruction(s, tex);
+	}
+}
+
+
+static int destructive_merge_instructions(
+		struct rc_pair_instruction * rgb,
+		struct rc_pair_instruction * alpha)
+{
+	assert(rgb->Alpha.Opcode == RC_OPCODE_NOP);
+	assert(alpha->RGB.Opcode == RC_OPCODE_NOP);
+
+	/* Copy alpha args into rgb */
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(alpha->Alpha.Opcode);
+
+	for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) {
+		unsigned int srcrgb = 0;
+		unsigned int srcalpha = 0;
+		unsigned int oldsrc = alpha->Alpha.Arg[arg].Source;
+		rc_register_file file = 0;
+		unsigned int index = 0;
+
+		if (alpha->Alpha.Arg[arg].Swizzle < 3) {
+			srcrgb = 1;
+			file = alpha->RGB.Src[oldsrc].File;
+			index = alpha->RGB.Src[oldsrc].Index;
+		} else if (alpha->Alpha.Arg[arg].Swizzle < 4) {
+			srcalpha = 1;
+			file = alpha->Alpha.Src[oldsrc].File;
+			index = alpha->Alpha.Src[oldsrc].Index;
+		}
+
+		int source = rc_pair_alloc_source(rgb, srcrgb, srcalpha, file, index);
+		if (source < 0)
+			return 0;
+
+		rgb->Alpha.Arg[arg].Source = source;
+		rgb->Alpha.Arg[arg].Swizzle = alpha->Alpha.Arg[arg].Swizzle;
+		rgb->Alpha.Arg[arg].Abs = alpha->Alpha.Arg[arg].Abs;
+		rgb->Alpha.Arg[arg].Negate = alpha->Alpha.Arg[arg].Negate;
+	}
+
+	/* Copy alpha opcode into rgb */
+	rgb->Alpha.Opcode = alpha->Alpha.Opcode;
+	rgb->Alpha.DestIndex = alpha->Alpha.DestIndex;
+	rgb->Alpha.WriteMask = alpha->Alpha.WriteMask;
+	rgb->Alpha.OutputWriteMask = alpha->Alpha.OutputWriteMask;
+	rgb->Alpha.DepthWriteMask = alpha->Alpha.DepthWriteMask;
+	rgb->Alpha.Saturate = alpha->Alpha.Saturate;
+
+	/* Merge ALU result writing */
+	if (alpha->WriteALUResult) {
+		if (rgb->WriteALUResult)
+			return 0;
+
+		rgb->WriteALUResult = alpha->WriteALUResult;
+		rgb->ALUResultCompare = alpha->ALUResultCompare;
+	}
+
+	return 1;
+}
+
+/**
+ * Try to merge the given instructions into the rgb instructions.
+ *
+ * Return true on success; on failure, return false, and keep
+ * the instructions untouched.
+ */
+static int merge_instructions(struct rc_pair_instruction * rgb, struct rc_pair_instruction * alpha)
+{
+	struct rc_pair_instruction backup;
+
+	memcpy(&backup, rgb, sizeof(struct rc_pair_instruction));
+
+	if (destructive_merge_instructions(rgb, alpha))
+		return 1;
+
+	memcpy(rgb, &backup, sizeof(struct rc_pair_instruction));
+	return 0;
+}
+
+
+/**
+ * Find a good ALU instruction or pair of ALU instruction and emit it.
+ *
+ * Prefer emitting full ALU instructions, so that when we reach a point
+ * where no full ALU instruction can be emitted, we have more candidates
+ * for RGB/Alpha pairing.
+ */
+static void emit_one_alu(struct schedule_state *s, struct rc_instruction * before)
+{
+	struct schedule_instruction * sinst;
+
+	if (s->ReadyFullALU || !(s->ReadyRGB && s->ReadyAlpha)) {
+		if (s->ReadyFullALU) {
+			sinst = s->ReadyFullALU;
+			s->ReadyFullALU = s->ReadyFullALU->NextReady;
+		} else if (s->ReadyRGB) {
+			sinst = s->ReadyRGB;
+			s->ReadyRGB = s->ReadyRGB->NextReady;
+		} else {
+			sinst = s->ReadyAlpha;
+			s->ReadyAlpha = s->ReadyAlpha->NextReady;
+		}
+
+		rc_insert_instruction(before->Prev, sinst->Instruction);
+		commit_instruction(s, sinst);
+	} else {
+		struct schedule_instruction **prgb;
+		struct schedule_instruction **palpha;
+
+		/* Some pairings might fail because they require too
+		 * many source slots; try all possible pairings if necessary */
+		for(prgb = &s->ReadyRGB; *prgb; prgb = &(*prgb)->NextReady) {
+			for(palpha = &s->ReadyAlpha; *palpha; palpha = &(*palpha)->NextReady) {
+				struct schedule_instruction * psirgb = *prgb;
+				struct schedule_instruction * psialpha = *palpha;
+
+				if (!merge_instructions(&psirgb->Instruction->U.P, &psialpha->Instruction->U.P))
+					continue;
+
+				*prgb = (*prgb)->NextReady;
+				*palpha = (*palpha)->NextReady;
+				rc_insert_instruction(before->Prev, psirgb->Instruction);
+				commit_instruction(s, psirgb);
+				commit_instruction(s, psialpha);
+				goto success;
+			}
+		}
+
+		/* No success in pairing; just take the first RGB instruction */
+		sinst = s->ReadyRGB;
+		s->ReadyRGB = s->ReadyRGB->NextReady;
+
+		rc_insert_instruction(before->Prev, sinst->Instruction);
+		commit_instruction(s, sinst);
+	success: ;
+	}
+}
+
+static void scan_read(void * data, struct rc_instruction * inst,
+		rc_register_file file, unsigned int index, unsigned int chan)
+{
+	struct schedule_state * s = data;
+	struct reg_value * v = get_reg_value(s, file, index, chan);
+
+	if (!v)
+		return;
+
+	if (v->Writer == s->Current) {
+		/* The instruction reads and writes to a register component.
+		 * In this case, we only want to increment dependencies by one.
+		 */
+		return;
+	}
+
+	DBG("%i: read %i[%i] chan %i\n", s->Current->Instruction->IP, file, index, chan);
+
+	struct reg_value_reader * reader = memory_pool_malloc(&s->C->Pool, sizeof(*reader));
+	reader->Reader = s->Current;
+	reader->Next = v->Readers;
+	v->Readers = reader;
+	v->NumReaders++;
+
+	s->Current->NumDependencies++;
+
+	if (s->Current->NumReadValues >= 12) {
+		rc_error(s->C, "%s: NumReadValues overflow\n", __FUNCTION__);
+	} else {
+		s->Current->ReadValues[s->Current->NumReadValues++] = v;
+	}
+}
+
+static void scan_write(void * data, struct rc_instruction * inst,
+		rc_register_file file, unsigned int index, unsigned int chan)
+{
+	struct schedule_state * s = data;
+	struct reg_value ** pv = get_reg_valuep(s, file, index, chan);
+
+	if (!pv)
+		return;
+
+	DBG("%i: write %i[%i] chan %i\n", s->Current->Instruction->IP, file, index, chan);
+
+	struct reg_value * newv = memory_pool_malloc(&s->C->Pool, sizeof(*newv));
+	memset(newv, 0, sizeof(*newv));
+
+	newv->Writer = s->Current;
+
+	if (*pv) {
+		(*pv)->Next = newv;
+		s->Current->NumDependencies++;
+	}
+
+	*pv = newv;
+
+	if (s->Current->NumWriteValues >= 4) {
+		rc_error(s->C, "%s: NumWriteValues overflow\n", __FUNCTION__);
+	} else {
+		s->Current->WriteValues[s->Current->NumWriteValues++] = newv;
+	}
+}
+
+static void schedule_block(struct r300_fragment_program_compiler * c,
+		struct rc_instruction * begin, struct rc_instruction * end)
+{
+	struct schedule_state s;
+
+	memset(&s, 0, sizeof(s));
+	s.C = &c->Base;
+
+	/* Scan instructions for data dependencies */
+	unsigned int ip = 0;
+	for(struct rc_instruction * inst = begin; inst != end; inst = inst->Next) {
+		s.Current = memory_pool_malloc(&c->Base.Pool, sizeof(*s.Current));
+		memset(s.Current, 0, sizeof(struct schedule_instruction));
+
+		s.Current->Instruction = inst;
+		inst->IP = ip++;
+
+		DBG("%i: Scanning\n", inst->IP);
+
+		/* The order of things here is subtle and maybe slightly
+		 * counter-intuitive, to account for the case where an
+		 * instruction writes to the same register as it reads
+		 * from. */
+		rc_for_all_writes(inst, &scan_write, &s);
+		rc_for_all_reads(inst, &scan_read, &s);
+
+		DBG("%i: Has %i dependencies\n", inst->IP, s.Current->NumDependencies);
+
+		if (!s.Current->NumDependencies)
+			instruction_ready(&s, s.Current);
+	}
+
+	/* Temporarily unlink all instructions */
+	begin->Prev->Next = end;
+	end->Prev = begin->Prev;
+
+	/* Schedule instructions back */
+	while(!s.C->Error &&
+	      (s.ReadyTEX || s.ReadyRGB || s.ReadyAlpha || s.ReadyFullALU)) {
+		if (s.ReadyTEX)
+			emit_all_tex(&s, end);
+
+		while(!s.C->Error && (s.ReadyFullALU || s.ReadyRGB || s.ReadyAlpha))
+			emit_one_alu(&s, end);
+	}
+}
+
+static int is_controlflow(struct rc_instruction * inst)
+{
+	if (inst->Type == RC_INSTRUCTION_NORMAL) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+		return opcode->IsFlowControl;
+	}
+	return 0;
+}
+
+void rc_pair_schedule(struct r300_fragment_program_compiler *c)
+{
+	struct rc_instruction * inst = c->Base.Program.Instructions.Next;
+	while(inst != &c->Base.Program.Instructions) {
+		if (is_controlflow(inst)) {
+			inst = inst->Next;
+			continue;
+		}
+
+		struct rc_instruction * first = inst;
+
+		while(inst != &c->Base.Program.Instructions && !is_controlflow(inst))
+			inst = inst->Next;
+
+		DBG("Schedule one block\n");
+		schedule_block(c, first, inst);
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_pair_translate.c b/src/mesa/drivers/dri/r300/compiler/radeon_pair_translate.c
new file mode 100644
index 0000000000..7211768272
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_pair_translate.c
@@ -0,0 +1,253 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_program_pair.h"
+
+#include "radeon_compiler.h"
+
+
+/**
+ * Finally rewrite ADD, MOV, MUL as the appropriate native instruction
+ * and reverse the order of arguments for CMP.
+ */
+static void final_rewrite(struct rc_sub_instruction *inst)
+{
+	struct rc_src_register tmp;
+
+	switch(inst->Opcode) {
+	case RC_OPCODE_ADD:
+		inst->SrcReg[2] = inst->SrcReg[1];
+		inst->SrcReg[1].File = RC_FILE_NONE;
+		inst->SrcReg[1].Swizzle = RC_SWIZZLE_1111;
+		inst->SrcReg[1].Negate = RC_MASK_NONE;
+		inst->Opcode = RC_OPCODE_MAD;
+		break;
+	case RC_OPCODE_CMP:
+		tmp = inst->SrcReg[2];
+		inst->SrcReg[2] = inst->SrcReg[0];
+		inst->SrcReg[0] = tmp;
+		break;
+	case RC_OPCODE_MOV:
+		/* AMD say we should use CMP.
+		 * However, when we transform
+		 *  KIL -r0;
+		 * into
+		 *  CMP tmp, -r0, -r0, 0;
+		 *  KIL tmp;
+		 * we get incorrect behaviour on R500 when r0 == 0.0.
+		 * It appears that the R500 KIL hardware treats -0.0 as less
+		 * than zero.
+		 */
+		inst->SrcReg[1].File = RC_FILE_NONE;
+		inst->SrcReg[1].Swizzle = RC_SWIZZLE_1111;
+		inst->SrcReg[2].File = RC_FILE_NONE;
+		inst->SrcReg[2].Swizzle = RC_SWIZZLE_0000;
+		inst->Opcode = RC_OPCODE_MAD;
+		break;
+	case RC_OPCODE_MUL:
+		inst->SrcReg[2].File = RC_FILE_NONE;
+		inst->SrcReg[2].Swizzle = RC_SWIZZLE_0000;
+		inst->Opcode = RC_OPCODE_MAD;
+		break;
+	default:
+		/* nothing to do */
+		break;
+	}
+}
+
+
+/**
+ * Classify an instruction according to which ALUs etc. it needs
+ */
+static void classify_instruction(struct rc_sub_instruction * inst,
+	int * needrgb, int * needalpha, int * istranscendent)
+{
+	*needrgb = (inst->DstReg.WriteMask & RC_MASK_XYZ) ? 1 : 0;
+	*needalpha = (inst->DstReg.WriteMask & RC_MASK_W) ? 1 : 0;
+	*istranscendent = 0;
+
+	if (inst->WriteALUResult == RC_ALURESULT_X)
+		*needrgb = 1;
+	else if (inst->WriteALUResult == RC_ALURESULT_W)
+		*needalpha = 1;
+
+	switch(inst->Opcode) {
+	case RC_OPCODE_ADD:
+	case RC_OPCODE_CMP:
+	case RC_OPCODE_DDX:
+	case RC_OPCODE_DDY:
+	case RC_OPCODE_FRC:
+	case RC_OPCODE_MAD:
+	case RC_OPCODE_MAX:
+	case RC_OPCODE_MIN:
+	case RC_OPCODE_MOV:
+	case RC_OPCODE_MUL:
+		break;
+	case RC_OPCODE_COS:
+	case RC_OPCODE_EX2:
+	case RC_OPCODE_LG2:
+	case RC_OPCODE_RCP:
+	case RC_OPCODE_RSQ:
+	case RC_OPCODE_SIN:
+		*istranscendent = 1;
+		*needalpha = 1;
+		break;
+	case RC_OPCODE_DP4:
+		*needalpha = 1;
+		/* fall through */
+	case RC_OPCODE_DP3:
+		*needrgb = 1;
+		break;
+	default:
+		break;
+	}
+}
+
+
+/**
+ * Fill the given ALU instruction's opcodes and source operands into the given pair,
+ * if possible.
+ */
+static void set_pair_instruction(struct r300_fragment_program_compiler *c,
+	struct rc_pair_instruction * pair,
+	struct rc_sub_instruction * inst)
+{
+	memset(pair, 0, sizeof(struct rc_pair_instruction));
+
+	int needrgb, needalpha, istranscendent;
+	classify_instruction(inst, &needrgb, &needalpha, &istranscendent);
+
+	if (needrgb) {
+		if (istranscendent)
+			pair->RGB.Opcode = RC_OPCODE_REPL_ALPHA;
+		else
+			pair->RGB.Opcode = inst->Opcode;
+		if (inst->SaturateMode == RC_SATURATE_ZERO_ONE)
+			pair->RGB.Saturate = 1;
+	}
+	if (needalpha) {
+		pair->Alpha.Opcode = inst->Opcode;
+		if (inst->SaturateMode == RC_SATURATE_ZERO_ONE)
+			pair->Alpha.Saturate = 1;
+	}
+
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode);
+	int nargs = opcode->NumSrcRegs;
+	int i;
+
+	/* Special case for DDX/DDY (MDH/MDV). */
+	if (inst->Opcode == RC_OPCODE_DDX || inst->Opcode == RC_OPCODE_DDY) {
+		nargs++;
+	}
+
+	for(i = 0; i < opcode->NumSrcRegs; ++i) {
+		int source;
+		if (needrgb && !istranscendent) {
+			unsigned int srcrgb = 0;
+			unsigned int srcalpha = 0;
+			int j;
+			for(j = 0; j < 3; ++j) {
+				unsigned int swz = GET_SWZ(inst->SrcReg[i].Swizzle, j);
+				if (swz < 3)
+					srcrgb = 1;
+				else if (swz < 4)
+					srcalpha = 1;
+			}
+			source = rc_pair_alloc_source(pair, srcrgb, srcalpha,
+							inst->SrcReg[i].File, inst->SrcReg[i].Index);
+			pair->RGB.Arg[i].Source = source;
+			pair->RGB.Arg[i].Swizzle = inst->SrcReg[i].Swizzle & 0x1ff;
+			pair->RGB.Arg[i].Abs = inst->SrcReg[i].Abs;
+			pair->RGB.Arg[i].Negate = !!(inst->SrcReg[i].Negate & (RC_MASK_X | RC_MASK_Y | RC_MASK_Z));
+		}
+		if (needalpha) {
+			unsigned int srcrgb = 0;
+			unsigned int srcalpha = 0;
+			unsigned int swz = GET_SWZ(inst->SrcReg[i].Swizzle, istranscendent ? 0 : 3);
+			if (swz < 3)
+				srcrgb = 1;
+			else if (swz < 4)
+				srcalpha = 1;
+			source = rc_pair_alloc_source(pair, srcrgb, srcalpha,
+							inst->SrcReg[i].File, inst->SrcReg[i].Index);
+			pair->Alpha.Arg[i].Source = source;
+			pair->Alpha.Arg[i].Swizzle = swz;
+			pair->Alpha.Arg[i].Abs = inst->SrcReg[i].Abs;
+			pair->Alpha.Arg[i].Negate = !!(inst->SrcReg[i].Negate & RC_MASK_W);
+		}
+	}
+
+	/* Destination handling */
+	if (inst->DstReg.File == RC_FILE_OUTPUT) {
+		if (inst->DstReg.Index == c->OutputColor) {
+			pair->RGB.OutputWriteMask |= inst->DstReg.WriteMask & RC_MASK_XYZ;
+			pair->Alpha.OutputWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
+		} else if (inst->DstReg.Index == c->OutputDepth) {
+			pair->Alpha.DepthWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
+		}
+	} else {
+		if (needrgb) {
+			pair->RGB.DestIndex = inst->DstReg.Index;
+			pair->RGB.WriteMask |= inst->DstReg.WriteMask & RC_MASK_XYZ;
+		}
+		if (needalpha) {
+			pair->Alpha.DestIndex = inst->DstReg.Index;
+			pair->Alpha.WriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
+		}
+	}
+
+	if (inst->WriteALUResult) {
+		pair->WriteALUResult = inst->WriteALUResult;
+		pair->ALUResultCompare = inst->ALUResultCompare;
+	}
+}
+
+
+/**
+ * Translate all ALU instructions into corresponding pair instructions,
+ * performing no other changes.
+ */
+void rc_pair_translate(struct r300_fragment_program_compiler *c)
+{
+	for(struct rc_instruction * inst = c->Base.Program.Instructions.Next;
+	    inst != &c->Base.Program.Instructions;
+	    inst = inst->Next) {
+		if (inst->Type != RC_INSTRUCTION_NORMAL)
+			continue;
+
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+
+		if (opcode->HasTexture || opcode->IsFlowControl || opcode->Opcode == RC_OPCODE_KIL)
+			continue;
+
+		struct rc_sub_instruction copy = inst->U.I;
+
+		final_rewrite(&copy);
+		inst->Type = RC_INSTRUCTION_PAIR;
+		set_pair_instruction(c, &inst->U.P, &copy);
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.c b/src/mesa/drivers/dri/r300/compiler/radeon_program.c
index b636f90a96..0dbc5380bb 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.c
@@ -27,9 +27,9 @@
 
 #include "radeon_program.h"
 
+#include <stdio.h>
+
 #include "radeon_compiler.h"
-#include "shader/prog_parameter.h"
-#include "shader/prog_print.h"
 
 
 /**
@@ -69,37 +69,57 @@ void radeonLocalTransform(
 	}
 }
 
+/**
+ * Left multiplication of a register with a swizzle
+ */
+struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register srcreg)
+{
+	struct rc_src_register tmp = srcreg;
+	int i;
+	tmp.Swizzle = 0;
+	tmp.Negate = 0;
+	for(i = 0; i < 4; ++i) {
+		rc_swizzle swz = GET_SWZ(swizzle, i);
+		if (swz < 4) {
+			tmp.Swizzle |= GET_SWZ(srcreg.Swizzle, swz) << (i*3);
+			tmp.Negate |= GET_BIT(srcreg.Negate, swz) << i;
+		} else {
+			tmp.Swizzle |= swz << (i*3);
+		}
+	}
+	return tmp;
+}
 
-GLint rc_find_free_temporary(struct radeon_compiler * c)
+unsigned int rc_find_free_temporary(struct radeon_compiler * c)
 {
-	GLboolean used[MAX_PROGRAM_TEMPS];
-	GLuint i;
+	char used[RC_REGISTER_MAX_INDEX];
+	unsigned int i;
 
 	memset(used, 0, sizeof(used));
 
 	for (struct rc_instruction * rcinst = c->Program.Instructions.Next; rcinst != &c->Program.Instructions; rcinst = rcinst->Next) {
-		const struct prog_instruction *inst = &rcinst->I;
-		const GLuint nsrc = _mesa_num_inst_src_regs(inst->Opcode);
-		const GLuint ndst = _mesa_num_inst_dst_regs(inst->Opcode);
-		GLuint k;
-
-		for (k = 0; k < nsrc; k++) {
-			if (inst->SrcReg[k].File == PROGRAM_TEMPORARY)
-				used[inst->SrcReg[k].Index] = GL_TRUE;
+		const struct rc_sub_instruction *inst = &rcinst->U.I;
+		const struct rc_opcode_info *opcode = rc_get_opcode_info(inst->Opcode);
+		unsigned int k;
+
+		for (k = 0; k < opcode->NumSrcRegs; k++) {
+			if (inst->SrcReg[k].File == RC_FILE_TEMPORARY)
+				used[inst->SrcReg[k].Index] = 1;
 		}
 
-		if (ndst) {
-			if (inst->DstReg.File == PROGRAM_TEMPORARY)
-				used[inst->DstReg.Index] = GL_TRUE;
+		if (opcode->HasDstReg) {
+			if (inst->DstReg.File == RC_FILE_TEMPORARY)
+				used[inst->DstReg.Index] = 1;
 		}
 	}
 
-	for (i = 0; i < MAX_PROGRAM_TEMPS; i++) {
+	for (i = 0; i < RC_REGISTER_MAX_INDEX; i++) {
 		if (!used[i])
 			return i;
 	}
 
-	return -1;
+	rc_error(c, "Ran out of temporary registers\n");
+	return 0;
 }
 
 
@@ -107,24 +127,31 @@ struct rc_instruction *rc_alloc_instruction(struct radeon_compiler * c)
 {
 	struct rc_instruction * inst = memory_pool_malloc(&c->Pool, sizeof(struct rc_instruction));
 
-	inst->Prev = 0;
-	inst->Next = 0;
+	memset(inst, 0, sizeof(struct rc_instruction));
 
-	_mesa_init_instructions(&inst->I, 1);
+	inst->U.I.Opcode = RC_OPCODE_ILLEGAL_OPCODE;
+	inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
+	inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
+	inst->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_XYZW;
+	inst->U.I.SrcReg[2].Swizzle = RC_SWIZZLE_XYZW;
 
 	return inst;
 }
 
-
-struct rc_instruction *rc_insert_new_instruction(struct radeon_compiler * c, struct rc_instruction * after)
+void rc_insert_instruction(struct rc_instruction * after, struct rc_instruction * inst)
 {
-	struct rc_instruction * inst = rc_alloc_instruction(c);
-
 	inst->Prev = after;
 	inst->Next = after->Next;
 
 	inst->Prev->Next = inst;
 	inst->Next->Prev = inst;
+}
+
+struct rc_instruction *rc_insert_new_instruction(struct radeon_compiler * c, struct rc_instruction * after)
+{
+	struct rc_instruction * inst = rc_alloc_instruction(c);
+
+	rc_insert_instruction(after, inst);
 
 	return inst;
 }
@@ -135,76 +162,20 @@ void rc_remove_instruction(struct rc_instruction * inst)
 	inst->Next->Prev = inst->Prev;
 }
 
-
-void rc_mesa_to_rc_program(struct radeon_compiler * c, struct gl_program * program)
-{
-	struct prog_instruction *source;
-	unsigned int i;
-
-	for(source = program->Instructions; source->Opcode != OPCODE_END; ++source) {
-		struct rc_instruction * dest = rc_insert_new_instruction(c, c->Program.Instructions.Prev);
-		dest->I = *source;
-	}
-
-	c->Program.ShadowSamplers = program->ShadowSamplers;
-	c->Program.InputsRead = program->InputsRead;
-	c->Program.OutputsWritten = program->OutputsWritten;
-
-	int isNVProgram = 0;
-
-	if (program->Target == GL_VERTEX_PROGRAM_ARB) {
-		struct gl_vertex_program * vp = (struct gl_vertex_program *) program;
-		isNVProgram = vp->IsNVProgram;
-	}
-
-	if (isNVProgram) {
-		/* NV_vertex_program has a fixed-sized constant environment.
-		 * This could be handled more efficiently for programs that
-		 * do not use relative addressing.
-		 */
-		for(i = 0; i < 96; ++i) {
-			struct rc_constant constant;
-
-			constant.Type = RC_CONSTANT_EXTERNAL;
-			constant.Size = 4;
-			constant.u.External = i;
-
-			rc_constants_add(&c->Program.Constants, &constant);
-		}
-	} else {
-		for(i = 0; i < program->Parameters->NumParameters; ++i) {
-			struct rc_constant constant;
-
-			constant.Type = RC_CONSTANT_EXTERNAL;
-			constant.Size = 4;
-			constant.u.External = i;
-
-			rc_constants_add(&c->Program.Constants, &constant);
-		}
-	}
-}
-
-
 /**
- * Print program to stderr, default options.
+ * Return the number of instructions in the program.
  */
-void rc_print_program(const struct rc_program *prog)
+unsigned int rc_recompute_ips(struct radeon_compiler * c)
 {
-	GLuint indent = 0;
-	GLuint linenum = 1;
-	struct rc_instruction *inst;
-
-	fprintf(stderr, "# Radeon Compiler Program\n");
+	unsigned int ip = 0;
 
-	for(inst = prog->Instructions.Next; inst != &prog->Instructions; inst = inst->Next) {
-		fprintf(stderr, "%3d: ", linenum);
+	for(struct rc_instruction * inst = c->Program.Instructions.Next;
+	    inst != &c->Program.Instructions;
+	    inst = inst->Next) {
+		inst->IP = ip++;
+	}
 
-		/* Massive hack: We rely on the fact that the printers do not actually
-		 * use the gl_program argument (last argument) in debug mode */
-		indent = _mesa_fprint_instruction_opt(
-				stderr, &inst->I,
-				indent, PROG_PRINT_DEBUG, 0);
+	c->Program.Instructions.IP = 0xcafedead;
 
-		linenum++;
-	}
+	return ip;
 }
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.h b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
index 561958608c..33db3ea0ff 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
@@ -28,37 +28,144 @@
 #ifndef __RADEON_PROGRAM_H_
 #define __RADEON_PROGRAM_H_
 
-#include "main/glheader.h"
-#include "main/macros.h"
-#include "main/enums.h"
-#include "shader/program.h"
-#include "shader/prog_instruction.h"
+#include <stdint.h>
+#include <string.h>
+
+#include "radeon_opcodes.h"
+#include "radeon_code.h"
+#include "radeon_program_constants.h"
+#include "radeon_program_pair.h"
 
 struct radeon_compiler;
-struct rc_instruction;
-struct rc_program;
 
-enum {
-	PROGRAM_BUILTIN = PROGRAM_FILE_MAX /**< not a real register, but a special swizzle constant */
+struct rc_src_register {
+	rc_register_file File:3;
+
+	/** Negative values may be used for relative addressing. */
+	signed int Index:(RC_REGISTER_INDEX_BITS+1);
+	unsigned int RelAddr:1;
+
+	unsigned int Swizzle:12;
+
+	/** Take the component-wise absolute value */
+	unsigned int Abs:1;
+
+	/** Post-Abs negation. */
+	unsigned int Negate:4;
+};
+
+struct rc_dst_register {
+	rc_register_file File:3;
+
+	/** Negative values may be used for relative addressing. */
+	signed int Index:(RC_REGISTER_INDEX_BITS+1);
+	unsigned int RelAddr:1;
+
+	unsigned int WriteMask:4;
+};
+
+/**
+ * Instructions are maintained by the compiler in a doubly linked list
+ * of these structures.
+ *
+ * This instruction format is intended to be expanded for hardware-specific
+ * trickery. At different stages of compilation, a different set of
+ * instruction types may be valid.
+ */
+struct rc_sub_instruction {
+	struct rc_src_register SrcReg[3];
+	struct rc_dst_register DstReg;
+
+	/**
+	 * Opcode of this instruction, according to \ref rc_opcode enums.
+	 */
+	rc_opcode Opcode:8;
+
+	/**
+	 * Saturate each value of the result to the range [0,1] or [-1,1],
+	 * according to \ref rc_saturate_mode enums.
+	 */
+	rc_saturate_mode SaturateMode:2;
+
+	/**
+	 * Writing to the special register RC_SPECIAL_ALU_RESULT
+	 */
+	/*@{*/
+	rc_write_aluresult WriteALUResult:2;
+	rc_compare_func ALUResultCompare:3;
+	/*@}*/
+
+	/**
+	 * \name Extra fields for TEX, TXB, TXD, TXL, TXP instructions.
+	 */
+	/*@{*/
+	/** Source texture unit. */
+	unsigned int TexSrcUnit:5;
+
+	/** Source texture target, one of the \ref rc_texture_target enums */
+	rc_texture_target TexSrcTarget:3;
+
+	/** True if tex instruction should do shadow comparison */
+	unsigned int TexShadow:1;
+	/*@}*/
+};
+
+typedef enum {
+	RC_INSTRUCTION_NORMAL = 0,
+	RC_INSTRUCTION_PAIR
+} rc_instruction_type;
+
+struct rc_instruction {
+	struct rc_instruction * Prev;
+	struct rc_instruction * Next;
+
+	rc_instruction_type Type;
+	union {
+		struct rc_sub_instruction I;
+		struct rc_pair_instruction P;
+	} U;
+
+	/**
+	 * Warning: IPs are not stable. If you want to use them,
+	 * you need to recompute them at the beginning of each pass
+	 * using \ref rc_recompute_ips
+	 */
+	unsigned int IP;
+};
+
+struct rc_program {
+	/**
+	 * Instructions.Next points to the first instruction,
+	 * Instructions.Prev points to the last instruction.
+	 */
+	struct rc_instruction Instructions;
+
+	/* Long term, we should probably remove InputsRead & OutputsWritten,
+	 * since updating dependent state can be fragile, and they aren't
+	 * actually used very often. */
+	uint32_t InputsRead;
+	uint32_t OutputsWritten;
+	uint32_t ShadowSamplers; /**< Texture units used for shadow sampling. */
+
+	struct rc_constant_list Constants;
 };
 
 enum {
-	OPCODE_REPL_ALPHA = MAX_OPCODE /**< used in paired instructions */
+	OPCODE_REPL_ALPHA = MAX_RC_OPCODE /**< used in paired instructions */
 };
 
-#define SWIZZLE_0000 MAKE_SWIZZLE4(SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO)
-#define SWIZZLE_1111 MAKE_SWIZZLE4(SWIZZLE_ONE, SWIZZLE_ONE, SWIZZLE_ONE, SWIZZLE_ONE)
 
-static inline GLuint get_swz(GLuint swz, GLuint idx)
+static inline rc_swizzle get_swz(unsigned int swz, rc_swizzle idx)
 {
 	if (idx & 0x4)
 		return idx;
 	return GET_SWZ(swz, idx);
 }
 
-static inline GLuint combine_swizzles4(GLuint src, GLuint swz_x, GLuint swz_y, GLuint swz_z, GLuint swz_w)
+static inline unsigned int combine_swizzles4(unsigned int src,
+		rc_swizzle swz_x, rc_swizzle swz_y, rc_swizzle swz_z, rc_swizzle swz_w)
 {
-	GLuint ret = 0;
+	unsigned int ret = 0;
 
 	ret |= get_swz(src, swz_x);
 	ret |= get_swz(src, swz_y) << 3;
@@ -68,22 +175,24 @@ static inline GLuint combine_swizzles4(GLuint src, GLuint swz_x, GLuint swz_y, G
 	return ret;
 }
 
-static inline GLuint combine_swizzles(GLuint src, GLuint swz)
+static inline unsigned int combine_swizzles(unsigned int src, unsigned int swz)
 {
-	GLuint ret = 0;
+	unsigned int ret = 0;
 
-	ret |= get_swz(src, GET_SWZ(swz, SWIZZLE_X));
-	ret |= get_swz(src, GET_SWZ(swz, SWIZZLE_Y)) << 3;
-	ret |= get_swz(src, GET_SWZ(swz, SWIZZLE_Z)) << 6;
-	ret |= get_swz(src, GET_SWZ(swz, SWIZZLE_W)) << 9;
+	ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_X));
+	ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_Y)) << 3;
+	ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_Z)) << 6;
+	ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_W)) << 9;
 
 	return ret;
 }
 
-static INLINE void reset_srcreg(struct prog_src_register* reg)
+struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register srcreg);
+
+static inline void reset_srcreg(struct rc_src_register* reg)
 {
-	_mesa_bzero(reg, sizeof(*reg));
-	reg->Swizzle = SWIZZLE_NOOP;
+	memset(reg, 0, sizeof(reg));
+	reg->Swizzle = RC_SWIZZLE_XYZW;
 }
 
 
@@ -92,13 +201,13 @@ static INLINE void reset_srcreg(struct prog_src_register* reg)
  *
  * The function will be called once for each instruction.
  * It has to either emit the appropriate transformed code for the instruction
- * and return GL_TRUE, or return GL_FALSE if it doesn't understand the
+ * and return true, or return false if it doesn't understand the
  * instruction.
  *
  * The function gets passed the userData as last parameter.
  */
 struct radeon_program_transformation {
-	GLboolean (*function)(
+	int (*function)(
 		struct radeon_compiler*,
 		struct rc_instruction*,
 		void*);
@@ -110,12 +219,15 @@ void radeonLocalTransform(
 	int num_transformations,
 	struct radeon_program_transformation* transformations);
 
-GLint rc_find_free_temporary(struct radeon_compiler * c);
+unsigned int rc_find_free_temporary(struct radeon_compiler * c);
 
 struct rc_instruction *rc_alloc_instruction(struct radeon_compiler * c);
 struct rc_instruction *rc_insert_new_instruction(struct radeon_compiler * c, struct rc_instruction * after);
+void rc_insert_instruction(struct rc_instruction * after, struct rc_instruction * inst);
 void rc_remove_instruction(struct rc_instruction * inst);
 
+unsigned int rc_recompute_ips(struct radeon_compiler * c);
+
 void rc_print_program(const struct rc_program *prog);
 
 #endif
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
index f23ce301ca..0326d25233 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
@@ -40,175 +40,164 @@
 
 static struct rc_instruction *emit1(
 	struct radeon_compiler * c, struct rc_instruction * after,
-	gl_inst_opcode Opcode, GLuint Saturate, struct prog_dst_register DstReg,
-	struct prog_src_register SrcReg)
+	rc_opcode Opcode, rc_saturate_mode Saturate, struct rc_dst_register DstReg,
+	struct rc_src_register SrcReg)
 {
 	struct rc_instruction *fpi = rc_insert_new_instruction(c, after);
 
-	fpi->I.Opcode = Opcode;
-	fpi->I.SaturateMode = Saturate;
-	fpi->I.DstReg = DstReg;
-	fpi->I.SrcReg[0] = SrcReg;
+	fpi->U.I.Opcode = Opcode;
+	fpi->U.I.SaturateMode = Saturate;
+	fpi->U.I.DstReg = DstReg;
+	fpi->U.I.SrcReg[0] = SrcReg;
 	return fpi;
 }
 
 static struct rc_instruction *emit2(
 	struct radeon_compiler * c, struct rc_instruction * after,
-	gl_inst_opcode Opcode, GLuint Saturate, struct prog_dst_register DstReg,
-	struct prog_src_register SrcReg0, struct prog_src_register SrcReg1)
+	rc_opcode Opcode, rc_saturate_mode Saturate, struct rc_dst_register DstReg,
+	struct rc_src_register SrcReg0, struct rc_src_register SrcReg1)
 {
 	struct rc_instruction *fpi = rc_insert_new_instruction(c, after);
 
-	fpi->I.Opcode = Opcode;
-	fpi->I.SaturateMode = Saturate;
-	fpi->I.DstReg = DstReg;
-	fpi->I.SrcReg[0] = SrcReg0;
-	fpi->I.SrcReg[1] = SrcReg1;
+	fpi->U.I.Opcode = Opcode;
+	fpi->U.I.SaturateMode = Saturate;
+	fpi->U.I.DstReg = DstReg;
+	fpi->U.I.SrcReg[0] = SrcReg0;
+	fpi->U.I.SrcReg[1] = SrcReg1;
 	return fpi;
 }
 
 static struct rc_instruction *emit3(
 	struct radeon_compiler * c, struct rc_instruction * after,
-	gl_inst_opcode Opcode, GLuint Saturate, struct prog_dst_register DstReg,
-	struct prog_src_register SrcReg0, struct prog_src_register SrcReg1,
-	struct prog_src_register SrcReg2)
+	rc_opcode Opcode, rc_saturate_mode Saturate, struct rc_dst_register DstReg,
+	struct rc_src_register SrcReg0, struct rc_src_register SrcReg1,
+	struct rc_src_register SrcReg2)
 {
 	struct rc_instruction *fpi = rc_insert_new_instruction(c, after);
 
-	fpi->I.Opcode = Opcode;
-	fpi->I.SaturateMode = Saturate;
-	fpi->I.DstReg = DstReg;
-	fpi->I.SrcReg[0] = SrcReg0;
-	fpi->I.SrcReg[1] = SrcReg1;
-	fpi->I.SrcReg[2] = SrcReg2;
+	fpi->U.I.Opcode = Opcode;
+	fpi->U.I.SaturateMode = Saturate;
+	fpi->U.I.DstReg = DstReg;
+	fpi->U.I.SrcReg[0] = SrcReg0;
+	fpi->U.I.SrcReg[1] = SrcReg1;
+	fpi->U.I.SrcReg[2] = SrcReg2;
 	return fpi;
 }
 
-static struct prog_dst_register dstreg(int file, int index)
+static struct rc_dst_register dstreg(int file, int index)
 {
-	struct prog_dst_register dst;
+	struct rc_dst_register dst;
 	dst.File = file;
 	dst.Index = index;
-	dst.WriteMask = WRITEMASK_XYZW;
-	dst.CondMask = COND_TR;
+	dst.WriteMask = RC_MASK_XYZW;
 	dst.RelAddr = 0;
-	dst.CondSwizzle = SWIZZLE_NOOP;
-	dst.CondSrc = 0;
-	dst.pad = 0;
 	return dst;
 }
 
-static struct prog_dst_register dstregtmpmask(int index, int mask)
+static struct rc_dst_register dstregtmpmask(int index, int mask)
 {
-	struct prog_dst_register dst = {0};
-	dst.File = PROGRAM_TEMPORARY;
+	struct rc_dst_register dst = {0};
+	dst.File = RC_FILE_TEMPORARY;
 	dst.Index = index;
 	dst.WriteMask = mask;
 	dst.RelAddr = 0;
-	dst.CondMask = COND_TR;
-	dst.CondSwizzle = SWIZZLE_NOOP;
-	dst.CondSrc = 0;
-	dst.pad = 0;
 	return dst;
 }
 
-static const struct prog_src_register builtin_zero = {
-	.File = PROGRAM_BUILTIN,
+static const struct rc_src_register builtin_zero = {
+	.File = RC_FILE_NONE,
 	.Index = 0,
-	.Swizzle = SWIZZLE_0000
+	.Swizzle = RC_SWIZZLE_0000
 };
-static const struct prog_src_register builtin_one = {
-	.File = PROGRAM_BUILTIN,
+static const struct rc_src_register builtin_one = {
+	.File = RC_FILE_NONE,
 	.Index = 0,
-	.Swizzle = SWIZZLE_1111
+	.Swizzle = RC_SWIZZLE_1111
 };
-static const struct prog_src_register srcreg_undefined = {
-	.File = PROGRAM_UNDEFINED,
+static const struct rc_src_register srcreg_undefined = {
+	.File = RC_FILE_NONE,
 	.Index = 0,
-	.Swizzle = SWIZZLE_NOOP
+	.Swizzle = RC_SWIZZLE_XYZW
 };
 
-static struct prog_src_register srcreg(int file, int index)
+static struct rc_src_register srcreg(int file, int index)
 {
-	struct prog_src_register src = srcreg_undefined;
+	struct rc_src_register src = srcreg_undefined;
 	src.File = file;
 	src.Index = index;
 	return src;
 }
 
-static struct prog_src_register srcregswz(int file, int index, int swz)
+static struct rc_src_register srcregswz(int file, int index, int swz)
 {
-	struct prog_src_register src = srcreg_undefined;
+	struct rc_src_register src = srcreg_undefined;
 	src.File = file;
 	src.Index = index;
 	src.Swizzle = swz;
 	return src;
 }
 
-static struct prog_src_register absolute(struct prog_src_register reg)
+static struct rc_src_register absolute(struct rc_src_register reg)
 {
-	struct prog_src_register newreg = reg;
+	struct rc_src_register newreg = reg;
 	newreg.Abs = 1;
-	newreg.Negate = NEGATE_NONE;
+	newreg.Negate = RC_MASK_NONE;
 	return newreg;
 }
 
-static struct prog_src_register negate(struct prog_src_register reg)
+static struct rc_src_register negate(struct rc_src_register reg)
 {
-	struct prog_src_register newreg = reg;
-	newreg.Negate = newreg.Negate ^ NEGATE_XYZW;
+	struct rc_src_register newreg = reg;
+	newreg.Negate = newreg.Negate ^ RC_MASK_XYZW;
 	return newreg;
 }
 
-static struct prog_src_register swizzle(struct prog_src_register reg, GLuint x, GLuint y, GLuint z, GLuint w)
+static struct rc_src_register swizzle(struct rc_src_register reg,
+		rc_swizzle x, rc_swizzle y, rc_swizzle z, rc_swizzle w)
 {
-	struct prog_src_register swizzled = reg;
-	swizzled.Swizzle = MAKE_SWIZZLE4(
-		x >= 4 ? x : GET_SWZ(reg.Swizzle, x),
-		y >= 4 ? y : GET_SWZ(reg.Swizzle, y),
-		z >= 4 ? z : GET_SWZ(reg.Swizzle, z),
-		w >= 4 ? w : GET_SWZ(reg.Swizzle, w));
+	struct rc_src_register swizzled = reg;
+	swizzled.Swizzle = combine_swizzles4(reg.Swizzle, x, y, z, w);
 	return swizzled;
 }
 
-static struct prog_src_register scalar(struct prog_src_register reg)
+static struct rc_src_register scalar(struct rc_src_register reg)
 {
-	return swizzle(reg, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X);
+	return swizzle(reg, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X);
 }
 
 static void transform_ABS(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	struct prog_src_register src = inst->I.SrcReg[0];
+	struct rc_src_register src = inst->U.I.SrcReg[0];
 	src.Abs = 1;
-	src.Negate = NEGATE_NONE;
-	emit1(c, inst->Prev, OPCODE_MOV, inst->I.SaturateMode, inst->I.DstReg, src);
+	src.Negate = RC_MASK_NONE;
+	emit1(c, inst->Prev, RC_OPCODE_MOV, inst->U.I.SaturateMode, inst->U.I.DstReg, src);
 	rc_remove_instruction(inst);
 }
 
 static void transform_DP3(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	struct prog_src_register src0 = inst->I.SrcReg[0];
-	struct prog_src_register src1 = inst->I.SrcReg[1];
-	src0.Negate &= ~NEGATE_W;
+	struct rc_src_register src0 = inst->U.I.SrcReg[0];
+	struct rc_src_register src1 = inst->U.I.SrcReg[1];
+	src0.Negate &= ~RC_MASK_W;
 	src0.Swizzle &= ~(7 << (3 * 3));
-	src0.Swizzle |= SWIZZLE_ZERO << (3 * 3);
-	src1.Negate &= ~NEGATE_W;
+	src0.Swizzle |= RC_SWIZZLE_ZERO << (3 * 3);
+	src1.Negate &= ~RC_MASK_W;
 	src1.Swizzle &= ~(7 << (3 * 3));
-	src1.Swizzle |= SWIZZLE_ZERO << (3 * 3);
-	emit2(c, inst->Prev, OPCODE_DP4, inst->I.SaturateMode, inst->I.DstReg, src0, src1);
+	src1.Swizzle |= RC_SWIZZLE_ZERO << (3 * 3);
+	emit2(c, inst->Prev, RC_OPCODE_DP4, inst->U.I.SaturateMode, inst->U.I.DstReg, src0, src1);
 	rc_remove_instruction(inst);
 }
 
 static void transform_DPH(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	struct prog_src_register src0 = inst->I.SrcReg[0];
-	src0.Negate &= ~NEGATE_W;
+	struct rc_src_register src0 = inst->U.I.SrcReg[0];
+	src0.Negate &= ~RC_MASK_W;
 	src0.Swizzle &= ~(7 << (3 * 3));
-	src0.Swizzle |= SWIZZLE_ONE << (3 * 3);
-	emit2(c, inst->Prev, OPCODE_DP4, inst->I.SaturateMode, inst->I.DstReg, src0, inst->I.SrcReg[1]);
+	src0.Swizzle |= RC_SWIZZLE_ONE << (3 * 3);
+	emit2(c, inst->Prev, RC_OPCODE_DP4, inst->U.I.SaturateMode, inst->U.I.DstReg, src0, inst->U.I.SrcReg[1]);
 	rc_remove_instruction(inst);
 }
 
@@ -219,9 +208,9 @@ static void transform_DPH(struct radeon_compiler* c,
 static void transform_DST(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	emit2(c, inst->Prev, OPCODE_MUL, inst->I.SaturateMode, inst->I.DstReg,
-		swizzle(inst->I.SrcReg[0], SWIZZLE_ONE, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ONE),
-		swizzle(inst->I.SrcReg[1], SWIZZLE_ONE, SWIZZLE_Y, SWIZZLE_ONE, SWIZZLE_W));
+	emit2(c, inst->Prev, RC_OPCODE_MUL, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_ONE, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_ONE),
+		swizzle(inst->U.I.SrcReg[1], RC_SWIZZLE_ONE, RC_SWIZZLE_Y, RC_SWIZZLE_ONE, RC_SWIZZLE_W));
 	rc_remove_instruction(inst);
 }
 
@@ -229,9 +218,9 @@ static void transform_FLR(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
 	int tempreg = rc_find_free_temporary(c);
-	emit1(c, inst->Prev, OPCODE_FRC, 0, dstreg(PROGRAM_TEMPORARY, tempreg), inst->I.SrcReg[0]);
-	emit2(c, inst->Prev, OPCODE_ADD, inst->I.SaturateMode, inst->I.DstReg,
-		inst->I.SrcReg[0], negate(srcreg(PROGRAM_TEMPORARY, tempreg)));
+	emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstreg(RC_FILE_TEMPORARY, tempreg), inst->U.I.SrcReg[0]);
+	emit2(c, inst->Prev, RC_OPCODE_ADD, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		inst->U.I.SrcReg[0], negate(srcreg(RC_FILE_TEMPORARY, tempreg)));
 	rc_remove_instruction(inst);
 }
 
@@ -256,64 +245,64 @@ static void transform_FLR(struct radeon_compiler* c,
 static void transform_LIT(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	GLuint constant;
-	GLuint constant_swizzle;
-	GLuint temp;
-	struct prog_src_register srctemp;
+	unsigned int constant;
+	unsigned int constant_swizzle;
+	unsigned int temp;
+	struct rc_src_register srctemp;
 
 	constant = rc_constants_add_immediate_scalar(&c->Program.Constants, -127.999999, &constant_swizzle);
 
-	if (inst->I.DstReg.WriteMask != WRITEMASK_XYZW || inst->I.DstReg.File != PROGRAM_TEMPORARY) {
+	if (inst->U.I.DstReg.WriteMask != RC_MASK_XYZW || inst->U.I.DstReg.File != RC_FILE_TEMPORARY) {
 		struct rc_instruction * inst_mov;
 
 		inst_mov = emit1(c, inst,
-			OPCODE_MOV, 0, inst->I.DstReg,
-			srcreg(PROGRAM_TEMPORARY, rc_find_free_temporary(c)));
+			RC_OPCODE_MOV, 0, inst->U.I.DstReg,
+			srcreg(RC_FILE_TEMPORARY, rc_find_free_temporary(c)));
 
-		inst->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst->I.DstReg.Index = inst_mov->I.SrcReg[0].Index;
-		inst->I.DstReg.WriteMask = WRITEMASK_XYZW;
+		inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst->U.I.DstReg.Index = inst_mov->U.I.SrcReg[0].Index;
+		inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
 	}
 
-	temp = inst->I.DstReg.Index;
-	srctemp = srcreg(PROGRAM_TEMPORARY, temp);
+	temp = inst->U.I.DstReg.Index;
+	srctemp = srcreg(RC_FILE_TEMPORARY, temp);
 
 	// tmp.x = max(0.0, Src.x);
 	// tmp.y = max(0.0, Src.y);
 	// tmp.w = clamp(Src.z, -128+eps, 128-eps);
-	emit2(c, inst->Prev, OPCODE_MAX, 0,
-		dstregtmpmask(temp, WRITEMASK_XYW),
-		inst->I.SrcReg[0],
-		swizzle(srcreg(PROGRAM_CONSTANT, constant),
-			SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO, constant_swizzle&3));
-	emit2(c, inst->Prev, OPCODE_MIN, 0,
-		dstregtmpmask(temp, WRITEMASK_Z),
-		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-		negate(srcregswz(PROGRAM_CONSTANT, constant, constant_swizzle)));
+	emit2(c, inst->Prev, RC_OPCODE_MAX, 0,
+		dstregtmpmask(temp, RC_MASK_XYW),
+		inst->U.I.SrcReg[0],
+		swizzle(srcreg(RC_FILE_CONSTANT, constant),
+			RC_SWIZZLE_ZERO, RC_SWIZZLE_ZERO, RC_SWIZZLE_ZERO, constant_swizzle&3));
+	emit2(c, inst->Prev, RC_OPCODE_MIN, 0,
+		dstregtmpmask(temp, RC_MASK_Z),
+		swizzle(srctemp, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+		negate(srcregswz(RC_FILE_CONSTANT, constant, constant_swizzle)));
 
 	// tmp.w = Pow(tmp.y, tmp.w)
-	emit1(c, inst->Prev, OPCODE_LG2, 0,
-		dstregtmpmask(temp, WRITEMASK_W),
-		swizzle(srctemp, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y));
-	emit2(c, inst->Prev, OPCODE_MUL, 0,
-		dstregtmpmask(temp, WRITEMASK_W),
-		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-		swizzle(srctemp, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z));
-	emit1(c, inst->Prev, OPCODE_EX2, 0,
-		dstregtmpmask(temp, WRITEMASK_W),
-		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W));
+	emit1(c, inst->Prev, RC_OPCODE_LG2, 0,
+		dstregtmpmask(temp, RC_MASK_W),
+		swizzle(srctemp, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y));
+	emit2(c, inst->Prev, RC_OPCODE_MUL, 0,
+		dstregtmpmask(temp, RC_MASK_W),
+		swizzle(srctemp, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+		swizzle(srctemp, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z));
+	emit1(c, inst->Prev, RC_OPCODE_EX2, 0,
+		dstregtmpmask(temp, RC_MASK_W),
+		swizzle(srctemp, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W));
 
 	// tmp.z = (tmp.x > 0) ? tmp.w : 0.0
-	emit3(c, inst->Prev, OPCODE_CMP, inst->I.SaturateMode,
-		dstregtmpmask(temp, WRITEMASK_Z),
-		negate(swizzle(srctemp, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)),
-		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode,
+		dstregtmpmask(temp, RC_MASK_Z),
+		negate(swizzle(srctemp, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
+		swizzle(srctemp, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
 		builtin_zero);
 
 	// tmp.x, tmp.y, tmp.w = 1.0, tmp.x, 1.0
-	emit1(c, inst->Prev, OPCODE_MOV, inst->I.SaturateMode,
-		dstregtmpmask(temp, WRITEMASK_XYW),
-		swizzle(srctemp, SWIZZLE_ONE, SWIZZLE_X, SWIZZLE_ONE, SWIZZLE_ONE));
+	emit1(c, inst->Prev, RC_OPCODE_MOV, inst->U.I.SaturateMode,
+		dstregtmpmask(temp, RC_MASK_XYW),
+		swizzle(srctemp, RC_SWIZZLE_ONE, RC_SWIZZLE_X, RC_SWIZZLE_ONE, RC_SWIZZLE_ONE));
 
 	rc_remove_instruction(inst);
 }
@@ -323,12 +312,12 @@ static void transform_LRP(struct radeon_compiler* c,
 {
 	int tempreg = rc_find_free_temporary(c);
 
-	emit2(c, inst->Prev, OPCODE_ADD, 0,
-		dstreg(PROGRAM_TEMPORARY, tempreg),
-		inst->I.SrcReg[1], negate(inst->I.SrcReg[2]));
-	emit3(c, inst->Prev, OPCODE_MAD, inst->I.SaturateMode,
-		inst->I.DstReg,
-		inst->I.SrcReg[0], srcreg(PROGRAM_TEMPORARY, tempreg), inst->I.SrcReg[2]);
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0,
+		dstreg(RC_FILE_TEMPORARY, tempreg),
+		inst->U.I.SrcReg[1], negate(inst->U.I.SrcReg[2]));
+	emit3(c, inst->Prev, RC_OPCODE_MAD, inst->U.I.SaturateMode,
+		inst->U.I.DstReg,
+		inst->U.I.SrcReg[0], srcreg(RC_FILE_TEMPORARY, tempreg), inst->U.I.SrcReg[2]);
 
 	rc_remove_instruction(inst);
 }
@@ -337,14 +326,14 @@ static void transform_POW(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
 	int tempreg = rc_find_free_temporary(c);
-	struct prog_dst_register tempdst = dstreg(PROGRAM_TEMPORARY, tempreg);
-	struct prog_src_register tempsrc = srcreg(PROGRAM_TEMPORARY, tempreg);
-	tempdst.WriteMask = WRITEMASK_W;
-	tempsrc.Swizzle = SWIZZLE_WWWW;
+	struct rc_dst_register tempdst = dstreg(RC_FILE_TEMPORARY, tempreg);
+	struct rc_src_register tempsrc = srcreg(RC_FILE_TEMPORARY, tempreg);
+	tempdst.WriteMask = RC_MASK_W;
+	tempsrc.Swizzle = RC_SWIZZLE_WWWW;
 
-	emit1(c, inst->Prev, OPCODE_LG2, 0, tempdst, scalar(inst->I.SrcReg[0]));
-	emit2(c, inst->Prev, OPCODE_MUL, 0, tempdst, tempsrc, scalar(inst->I.SrcReg[1]));
-	emit1(c, inst->Prev, OPCODE_EX2, inst->I.SaturateMode, inst->I.DstReg, tempsrc);
+	emit1(c, inst->Prev, RC_OPCODE_LG2, 0, tempdst, scalar(inst->U.I.SrcReg[0]));
+	emit2(c, inst->Prev, RC_OPCODE_MUL, 0, tempdst, tempsrc, scalar(inst->U.I.SrcReg[1]));
+	emit1(c, inst->Prev, RC_OPCODE_EX2, inst->U.I.SaturateMode, inst->U.I.DstReg, tempsrc);
 
 	rc_remove_instruction(inst);
 }
@@ -352,7 +341,26 @@ static void transform_POW(struct radeon_compiler* c,
 static void transform_RSQ(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	inst->I.SrcReg[0] = absolute(inst->I.SrcReg[0]);
+	inst->U.I.SrcReg[0] = absolute(inst->U.I.SrcReg[0]);
+}
+
+static void transform_SEQ(struct radeon_compiler* c,
+	struct rc_instruction* inst)
+{
+	int tempreg = rc_find_free_temporary(c);
+
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dstreg(RC_FILE_TEMPORARY, tempreg), inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1]));
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		negate(absolute(srcreg(RC_FILE_TEMPORARY, tempreg))), builtin_zero, builtin_one);
+
+	rc_remove_instruction(inst);
+}
+
+static void transform_SFL(struct radeon_compiler* c,
+	struct rc_instruction* inst)
+{
+	emit1(c, inst->Prev, RC_OPCODE_MOV, inst->U.I.SaturateMode, inst->U.I.DstReg, builtin_zero);
+	rc_remove_instruction(inst);
 }
 
 static void transform_SGE(struct radeon_compiler* c,
@@ -360,9 +368,33 @@ static void transform_SGE(struct radeon_compiler* c,
 {
 	int tempreg = rc_find_free_temporary(c);
 
-	emit2(c, inst->Prev, OPCODE_ADD, 0, dstreg(PROGRAM_TEMPORARY, tempreg), inst->I.SrcReg[0], negate(inst->I.SrcReg[1]));
-	emit3(c, inst->Prev, OPCODE_CMP, inst->I.SaturateMode, inst->I.DstReg,
-		srcreg(PROGRAM_TEMPORARY, tempreg), builtin_zero, builtin_one);
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dstreg(RC_FILE_TEMPORARY, tempreg), inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1]));
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		srcreg(RC_FILE_TEMPORARY, tempreg), builtin_zero, builtin_one);
+
+	rc_remove_instruction(inst);
+}
+
+static void transform_SGT(struct radeon_compiler* c,
+	struct rc_instruction* inst)
+{
+	int tempreg = rc_find_free_temporary(c);
+
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dstreg(RC_FILE_TEMPORARY, tempreg), negate(inst->U.I.SrcReg[0]), inst->U.I.SrcReg[1]);
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		srcreg(RC_FILE_TEMPORARY, tempreg), builtin_one, builtin_zero);
+
+	rc_remove_instruction(inst);
+}
+
+static void transform_SLE(struct radeon_compiler* c,
+	struct rc_instruction* inst)
+{
+	int tempreg = rc_find_free_temporary(c);
+
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dstreg(RC_FILE_TEMPORARY, tempreg), negate(inst->U.I.SrcReg[0]), inst->U.I.SrcReg[1]);
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		srcreg(RC_FILE_TEMPORARY, tempreg), builtin_zero, builtin_one);
 
 	rc_remove_instruction(inst);
 }
@@ -372,9 +404,21 @@ static void transform_SLT(struct radeon_compiler* c,
 {
 	int tempreg = rc_find_free_temporary(c);
 
-	emit2(c, inst->Prev, OPCODE_ADD, 0, dstreg(PROGRAM_TEMPORARY, tempreg), inst->I.SrcReg[0], negate(inst->I.SrcReg[1]));
-	emit3(c, inst->Prev, OPCODE_CMP, inst->I.SaturateMode, inst->I.DstReg,
-		srcreg(PROGRAM_TEMPORARY, tempreg), builtin_one, builtin_zero);
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dstreg(RC_FILE_TEMPORARY, tempreg), inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1]));
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		srcreg(RC_FILE_TEMPORARY, tempreg), builtin_one, builtin_zero);
+
+	rc_remove_instruction(inst);
+}
+
+static void transform_SNE(struct radeon_compiler* c,
+	struct rc_instruction* inst)
+{
+	int tempreg = rc_find_free_temporary(c);
+
+	emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dstreg(RC_FILE_TEMPORARY, tempreg), inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1]));
+	emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		negate(absolute(srcreg(RC_FILE_TEMPORARY, tempreg))), builtin_one, builtin_zero);
 
 	rc_remove_instruction(inst);
 }
@@ -382,14 +426,14 @@ static void transform_SLT(struct radeon_compiler* c,
 static void transform_SUB(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	inst->I.Opcode = OPCODE_ADD;
-	inst->I.SrcReg[1] = negate(inst->I.SrcReg[1]);
+	inst->U.I.Opcode = RC_OPCODE_ADD;
+	inst->U.I.SrcReg[1] = negate(inst->U.I.SrcReg[1]);
 }
 
 static void transform_SWZ(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
-	inst->I.Opcode = OPCODE_MOV;
+	inst->U.I.Opcode = RC_OPCODE_MOV;
 }
 
 static void transform_XPD(struct radeon_compiler* c,
@@ -397,13 +441,13 @@ static void transform_XPD(struct radeon_compiler* c,
 {
 	int tempreg = rc_find_free_temporary(c);
 
-	emit2(c, inst->Prev, OPCODE_MUL, 0, dstreg(PROGRAM_TEMPORARY, tempreg),
-		swizzle(inst->I.SrcReg[0], SWIZZLE_Z, SWIZZLE_X, SWIZZLE_Y, SWIZZLE_W),
-		swizzle(inst->I.SrcReg[1], SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_X, SWIZZLE_W));
-	emit3(c, inst->Prev, OPCODE_MAD, inst->I.SaturateMode, inst->I.DstReg,
-		swizzle(inst->I.SrcReg[0], SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_X, SWIZZLE_W),
-		swizzle(inst->I.SrcReg[1], SWIZZLE_Z, SWIZZLE_X, SWIZZLE_Y, SWIZZLE_W),
-		negate(srcreg(PROGRAM_TEMPORARY, tempreg)));
+	emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dstreg(RC_FILE_TEMPORARY, tempreg),
+		swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_W),
+		swizzle(inst->U.I.SrcReg[1], RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_W));
+	emit3(c, inst->Prev, RC_OPCODE_MAD, inst->U.I.SaturateMode, inst->U.I.DstReg,
+		swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_W),
+		swizzle(inst->U.I.SrcReg[1], RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_W),
+		negate(srcreg(RC_FILE_TEMPORARY, tempreg)));
 
 	rc_remove_instruction(inst);
 }
@@ -414,7 +458,7 @@ static void transform_XPD(struct radeon_compiler* c,
  * no userData necessary.
  *
  * Eliminates the following ALU instructions:
- *  ABS, DPH, DST, FLR, LIT, LRP, POW, SGE, SLT, SUB, SWZ, XPD
+ *  ABS, DPH, DST, FLR, LIT, LRP, POW, SEQ, SFL, SGE, SGT, SLE, SLT, SNE, SUB, SWZ, XPD
  * using:
  *  MOV, ADD, MUL, MAD, FRC, DP3, LG2, EX2, CMP
  *
@@ -423,27 +467,32 @@ static void transform_XPD(struct radeon_compiler* c,
  *
  * @note should be applicable to R300 and R500 fragment programs.
  */
-GLboolean radeonTransformALU(
+int radeonTransformALU(
 	struct radeon_compiler * c,
 	struct rc_instruction* inst,
 	void* unused)
 {
-	switch(inst->I.Opcode) {
-	case OPCODE_ABS: transform_ABS(c, inst); return GL_TRUE;
-	case OPCODE_DPH: transform_DPH(c, inst); return GL_TRUE;
-	case OPCODE_DST: transform_DST(c, inst); return GL_TRUE;
-	case OPCODE_FLR: transform_FLR(c, inst); return GL_TRUE;
-	case OPCODE_LIT: transform_LIT(c, inst); return GL_TRUE;
-	case OPCODE_LRP: transform_LRP(c, inst); return GL_TRUE;
-	case OPCODE_POW: transform_POW(c, inst); return GL_TRUE;
-	case OPCODE_RSQ: transform_RSQ(c, inst); return GL_TRUE;
-	case OPCODE_SGE: transform_SGE(c, inst); return GL_TRUE;
-	case OPCODE_SLT: transform_SLT(c, inst); return GL_TRUE;
-	case OPCODE_SUB: transform_SUB(c, inst); return GL_TRUE;
-	case OPCODE_SWZ: transform_SWZ(c, inst); return GL_TRUE;
-	case OPCODE_XPD: transform_XPD(c, inst); return GL_TRUE;
+	switch(inst->U.I.Opcode) {
+	case RC_OPCODE_ABS: transform_ABS(c, inst); return 1;
+	case RC_OPCODE_DPH: transform_DPH(c, inst); return 1;
+	case RC_OPCODE_DST: transform_DST(c, inst); return 1;
+	case RC_OPCODE_FLR: transform_FLR(c, inst); return 1;
+	case RC_OPCODE_LIT: transform_LIT(c, inst); return 1;
+	case RC_OPCODE_LRP: transform_LRP(c, inst); return 1;
+	case RC_OPCODE_POW: transform_POW(c, inst); return 1;
+	case RC_OPCODE_RSQ: transform_RSQ(c, inst); return 1;
+	case RC_OPCODE_SEQ: transform_SEQ(c, inst); return 1;
+	case RC_OPCODE_SFL: transform_SFL(c, inst); return 1;
+	case RC_OPCODE_SGE: transform_SGE(c, inst); return 1;
+	case RC_OPCODE_SGT: transform_SGT(c, inst); return 1;
+	case RC_OPCODE_SLE: transform_SLE(c, inst); return 1;
+	case RC_OPCODE_SLT: transform_SLT(c, inst); return 1;
+	case RC_OPCODE_SNE: transform_SNE(c, inst); return 1;
+	case RC_OPCODE_SUB: transform_SUB(c, inst); return 1;
+	case RC_OPCODE_SWZ: transform_SWZ(c, inst); return 1;
+	case RC_OPCODE_XPD: transform_XPD(c, inst); return 1;
 	default:
-		return GL_FALSE;
+		return 0;
 	}
 }
 
@@ -452,37 +501,37 @@ static void transform_r300_vertex_ABS(struct radeon_compiler* c,
 	struct rc_instruction* inst)
 {
 	/* Note: r500 can take absolute values, but r300 cannot. */
-	inst->I.Opcode = OPCODE_MAX;
-	inst->I.SrcReg[1] = inst->I.SrcReg[0];
-	inst->I.SrcReg[1].Negate ^= NEGATE_XYZW;
+	inst->U.I.Opcode = RC_OPCODE_MAX;
+	inst->U.I.SrcReg[1] = inst->U.I.SrcReg[0];
+	inst->U.I.SrcReg[1].Negate ^= RC_MASK_XYZW;
 }
 
 /**
  * For use with radeonLocalTransform, this transforms non-native ALU
  * instructions of the r300 up to r500 vertex engine.
  */
-GLboolean r300_transform_vertex_alu(
+int r300_transform_vertex_alu(
 	struct radeon_compiler * c,
 	struct rc_instruction* inst,
 	void* unused)
 {
-	switch(inst->I.Opcode) {
-	case OPCODE_ABS: transform_r300_vertex_ABS(c, inst); return GL_TRUE;
-	case OPCODE_DP3: transform_DP3(c, inst); return GL_TRUE;
-	case OPCODE_DPH: transform_DPH(c, inst); return GL_TRUE;
-	case OPCODE_FLR: transform_FLR(c, inst); return GL_TRUE;
-	case OPCODE_LRP: transform_LRP(c, inst); return GL_TRUE;
-	case OPCODE_SUB: transform_SUB(c, inst); return GL_TRUE;
-	case OPCODE_SWZ: transform_SWZ(c, inst); return GL_TRUE;
-	case OPCODE_XPD: transform_XPD(c, inst); return GL_TRUE;
+	switch(inst->U.I.Opcode) {
+	case RC_OPCODE_ABS: transform_r300_vertex_ABS(c, inst); return 1;
+	case RC_OPCODE_DP3: transform_DP3(c, inst); return 1;
+	case RC_OPCODE_DPH: transform_DPH(c, inst); return 1;
+	case RC_OPCODE_FLR: transform_FLR(c, inst); return 1;
+	case RC_OPCODE_LRP: transform_LRP(c, inst); return 1;
+	case RC_OPCODE_SUB: transform_SUB(c, inst); return 1;
+	case RC_OPCODE_SWZ: transform_SWZ(c, inst); return 1;
+	case RC_OPCODE_XPD: transform_XPD(c, inst); return 1;
 	default:
-		return GL_FALSE;
+		return 0;
 	}
 }
 
-static void sincos_constants(struct radeon_compiler* c, GLuint *constants)
+static void sincos_constants(struct radeon_compiler* c, unsigned int *constants)
 {
-	static const GLfloat SinCosConsts[2][4] = {
+	static const float SinCosConsts[2][4] = {
 		{
 			1.273239545,		// 4/PI
 			-0.405284735,		// -4/(PI*PI)
@@ -512,25 +561,25 @@ static void sincos_constants(struct radeon_compiler* c, GLuint *constants)
  */
 static void sin_approx(
 	struct radeon_compiler* c, struct rc_instruction * before,
-	struct prog_dst_register dst, struct prog_src_register src, const GLuint* constants)
-{
-	GLuint tempreg = rc_find_free_temporary(c);
-
-	emit2(c, before->Prev, OPCODE_MUL, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
-		swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
-		srcreg(PROGRAM_CONSTANT, constants[0]));
-	emit3(c, before->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_X),
-		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
-		absolute(swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)),
-		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X));
-	emit3(c, before->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_Y),
-		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
-		absolute(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)),
-		negate(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)));
-	emit3(c, before->Prev, OPCODE_MAD, 0, dst,
-		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
-		swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X));
+	struct rc_dst_register dst, struct rc_src_register src, const unsigned int* constants)
+{
+	unsigned int tempreg = rc_find_free_temporary(c);
+
+	emit2(c, before, RC_OPCODE_MUL, 0, dstregtmpmask(tempreg, RC_MASK_XY),
+		swizzle(src, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
+		srcreg(RC_FILE_CONSTANT, constants[0]));
+	emit3(c, before, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_X),
+		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
+		absolute(swizzle(src, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
+		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
+	emit3(c, before, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_Y),
+		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
+		absolute(swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
+		negate(swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)));
+	emit3(c, before, RC_OPCODE_MAD, 0, dst,
+		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
+		swizzle(srcreg(RC_FILE_CONSTANT, constants[0]), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
 }
 
 /**
@@ -538,81 +587,81 @@ static void sin_approx(
  * using only the basic instructions
  *  MOV, ADD, MUL, MAD, FRC
  */
-GLboolean radeonTransformTrigSimple(struct radeon_compiler* c,
+int radeonTransformTrigSimple(struct radeon_compiler* c,
 	struct rc_instruction* inst,
 	void* unused)
 {
-	if (inst->I.Opcode != OPCODE_COS &&
-	    inst->I.Opcode != OPCODE_SIN &&
-	    inst->I.Opcode != OPCODE_SCS)
-		return GL_FALSE;
+	if (inst->U.I.Opcode != RC_OPCODE_COS &&
+	    inst->U.I.Opcode != RC_OPCODE_SIN &&
+	    inst->U.I.Opcode != RC_OPCODE_SCS)
+		return 0;
 
-	GLuint constants[2];
-	GLuint tempreg = rc_find_free_temporary(c);
+	unsigned int constants[2];
+	unsigned int tempreg = rc_find_free_temporary(c);
 
 	sincos_constants(c, constants);
 
-	if (inst->I.Opcode == OPCODE_COS) {
+	if (inst->U.I.Opcode == RC_OPCODE_COS) {
 		// MAD tmp.x, src, 1/(2*PI), 0.75
 		// FRC tmp.x, tmp.x
 		// MAD tmp.z, tmp.x, 2*PI, -PI
-		emit3(c, inst->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
-			swizzle(inst->I.SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X));
-		emit1(c, inst->Prev, OPCODE_FRC, 0, dstregtmpmask(tempreg, WRITEMASK_W),
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W));
-		emit3(c, inst->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-			negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)));
-
-		sin_approx(c, inst, inst->I.DstReg,
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+		emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W),
+			swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
+		emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(tempreg, RC_MASK_W),
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W));
+		emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W),
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+			negate(swizzle(srcreg(RC_FILE_CONSTANT, constants[0]), RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z)));
+
+		sin_approx(c, inst, inst->U.I.DstReg,
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
 			constants);
-	} else if (inst->I.Opcode == OPCODE_SIN) {
-		emit3(c, inst->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
-			swizzle(inst->I.SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y));
-		emit1(c, inst->Prev, OPCODE_FRC, 0, dstregtmpmask(tempreg, WRITEMASK_W),
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W));
-		emit3(c, inst->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-			negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)));
-
-		sin_approx(c, inst, inst->I.DstReg,
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+	} else if (inst->U.I.Opcode == RC_OPCODE_SIN) {
+		emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W),
+			swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y));
+		emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(tempreg, RC_MASK_W),
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W));
+		emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W),
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+			negate(swizzle(srcreg(RC_FILE_CONSTANT, constants[0]), RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z)));
+
+		sin_approx(c, inst, inst->U.I.DstReg,
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
 			constants);
 	} else {
-		emit3(c, inst->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
-			swizzle(inst->I.SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W));
-		emit1(c, inst->Prev, OPCODE_FRC, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
-			srcreg(PROGRAM_TEMPORARY, tempreg));
-		emit3(c, inst->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
-			srcreg(PROGRAM_TEMPORARY, tempreg),
-			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
-			negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)));
-
-		struct prog_dst_register dst = inst->I.DstReg;
-
-		dst.WriteMask = inst->I.DstReg.WriteMask & WRITEMASK_X;
+		emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_XY),
+			swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_W));
+		emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(tempreg, RC_MASK_XY),
+			srcreg(RC_FILE_TEMPORARY, tempreg));
+		emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_XY),
+			srcreg(RC_FILE_TEMPORARY, tempreg),
+			swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
+			negate(swizzle(srcreg(RC_FILE_CONSTANT, constants[0]), RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z, RC_SWIZZLE_Z)));
+
+		struct rc_dst_register dst = inst->U.I.DstReg;
+
+		dst.WriteMask = inst->U.I.DstReg.WriteMask & RC_MASK_X;
 		sin_approx(c, inst, dst,
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
 			constants);
 
-		dst.WriteMask = inst->I.DstReg.WriteMask & WRITEMASK_Y;
+		dst.WriteMask = inst->U.I.DstReg.WriteMask & RC_MASK_Y;
 		sin_approx(c, inst, dst,
-			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
+			swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
 			constants);
 	}
 
 	rc_remove_instruction(inst);
 
-	return GL_TRUE;
+	return 1;
 }
 
 
@@ -624,53 +673,53 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c,
  *
  * @warning This transformation implicitly changes the semantics of SIN and COS!
  */
-GLboolean radeonTransformTrigScale(struct radeon_compiler* c,
+int radeonTransformTrigScale(struct radeon_compiler* c,
 	struct rc_instruction* inst,
 	void* unused)
 {
-	if (inst->I.Opcode != OPCODE_COS &&
-	    inst->I.Opcode != OPCODE_SIN &&
-	    inst->I.Opcode != OPCODE_SCS)
-		return GL_FALSE;
+	if (inst->U.I.Opcode != RC_OPCODE_COS &&
+	    inst->U.I.Opcode != RC_OPCODE_SIN &&
+	    inst->U.I.Opcode != RC_OPCODE_SCS)
+		return 0;
 
-	static const GLfloat RCP_2PI = 0.15915494309189535;
-	GLuint temp;
-	GLuint constant;
-	GLuint constant_swizzle;
+	static const float RCP_2PI = 0.15915494309189535;
+	unsigned int temp;
+	unsigned int constant;
+	unsigned int constant_swizzle;
 
 	temp = rc_find_free_temporary(c);
 	constant = rc_constants_add_immediate_scalar(&c->Program.Constants, RCP_2PI, &constant_swizzle);
 
-	emit2(c, inst->Prev, OPCODE_MUL, 0, dstregtmpmask(temp, WRITEMASK_W),
-		swizzle(inst->I.SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
-		srcregswz(PROGRAM_CONSTANT, constant, constant_swizzle));
-	emit1(c, inst->Prev, OPCODE_FRC, 0, dstregtmpmask(temp, WRITEMASK_W),
-		srcreg(PROGRAM_TEMPORARY, temp));
-
-	if (inst->I.Opcode == OPCODE_COS) {
-		emit1(c, inst->Prev, OPCODE_COS, inst->I.SaturateMode, inst->I.DstReg,
-			srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
-	} else if (inst->I.Opcode == OPCODE_SIN) {
-		emit1(c, inst->Prev, OPCODE_SIN, inst->I.SaturateMode,
-			inst->I.DstReg, srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
-	} else if (inst->I.Opcode == OPCODE_SCS) {
-		struct prog_dst_register moddst = inst->I.DstReg;
-
-		if (inst->I.DstReg.WriteMask & WRITEMASK_X) {
-			moddst.WriteMask = WRITEMASK_X;
-			emit1(c, inst->Prev, OPCODE_COS, inst->I.SaturateMode, moddst,
-				srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
+	emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dstregtmpmask(temp, RC_MASK_W),
+		swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
+		srcregswz(RC_FILE_CONSTANT, constant, constant_swizzle));
+	emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(temp, RC_MASK_W),
+		srcreg(RC_FILE_TEMPORARY, temp));
+
+	if (inst->U.I.Opcode == RC_OPCODE_COS) {
+		emit1(c, inst->Prev, RC_OPCODE_COS, inst->U.I.SaturateMode, inst->U.I.DstReg,
+			srcregswz(RC_FILE_TEMPORARY, temp, RC_SWIZZLE_WWWW));
+	} else if (inst->U.I.Opcode == RC_OPCODE_SIN) {
+		emit1(c, inst->Prev, RC_OPCODE_SIN, inst->U.I.SaturateMode,
+			inst->U.I.DstReg, srcregswz(RC_FILE_TEMPORARY, temp, RC_SWIZZLE_WWWW));
+	} else if (inst->U.I.Opcode == RC_OPCODE_SCS) {
+		struct rc_dst_register moddst = inst->U.I.DstReg;
+
+		if (inst->U.I.DstReg.WriteMask & RC_MASK_X) {
+			moddst.WriteMask = RC_MASK_X;
+			emit1(c, inst->Prev, RC_OPCODE_COS, inst->U.I.SaturateMode, moddst,
+				srcregswz(RC_FILE_TEMPORARY, temp, RC_SWIZZLE_WWWW));
 		}
-		if (inst->I.DstReg.WriteMask & WRITEMASK_Y) {
-			moddst.WriteMask = WRITEMASK_Y;
-			emit1(c, inst->Prev, OPCODE_SIN, inst->I.SaturateMode, moddst,
-				srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
+		if (inst->U.I.DstReg.WriteMask & RC_MASK_Y) {
+			moddst.WriteMask = RC_MASK_Y;
+			emit1(c, inst->Prev, RC_OPCODE_SIN, inst->U.I.SaturateMode, moddst,
+				srcregswz(RC_FILE_TEMPORARY, temp, RC_SWIZZLE_WWWW));
 		}
 	}
 
 	rc_remove_instruction(inst);
 
-	return GL_TRUE;
+	return 1;
 }
 
 /**
@@ -681,15 +730,15 @@ GLboolean radeonTransformTrigScale(struct radeon_compiler* c,
  * @warning This explicitly changes the form of DDX and DDY!
  */
 
-GLboolean radeonTransformDeriv(struct radeon_compiler* c,
+int radeonTransformDeriv(struct radeon_compiler* c,
 	struct rc_instruction* inst,
 	void* unused)
 {
-	if (inst->I.Opcode != OPCODE_DDX && inst->I.Opcode != OPCODE_DDY)
-		return GL_FALSE;
+	if (inst->U.I.Opcode != RC_OPCODE_DDX && inst->U.I.Opcode != RC_OPCODE_DDY)
+		return 0;
 
-	inst->I.SrcReg[1].Swizzle = MAKE_SWIZZLE4(SWIZZLE_ONE, SWIZZLE_ONE, SWIZZLE_ONE, SWIZZLE_ONE);
-	inst->I.SrcReg[1].Negate = NEGATE_XYZW;
+	inst->U.I.SrcReg[1].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_ONE, RC_SWIZZLE_ONE, RC_SWIZZLE_ONE, RC_SWIZZLE_ONE);
+	inst->U.I.SrcReg[1].Negate = RC_MASK_XYZW;
 
-	return GL_TRUE;
+	return 1;
 }
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.h b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.h
index 147efec6fc..7cb5f84b7f 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.h
@@ -30,27 +30,27 @@
 
 #include "radeon_program.h"
 
-GLboolean radeonTransformALU(
+int radeonTransformALU(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void*);
 
-GLboolean r300_transform_vertex_alu(
+int r300_transform_vertex_alu(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void*);
 
-GLboolean radeonTransformTrigSimple(
+int radeonTransformTrigSimple(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void*);
 
-GLboolean radeonTransformTrigScale(
+int radeonTransformTrigScale(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void*);
 
-GLboolean radeonTransformDeriv(
+int radeonTransformDeriv(
 	struct radeon_compiler * c,
 	struct rc_instruction * inst,
 	void*);
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_constants.h b/src/mesa/drivers/dri/r300/compiler/radeon_program_constants.h
new file mode 100644
index 0000000000..7c0d6720b1
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_constants.h
@@ -0,0 +1,147 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef RADEON_PROGRAM_CONSTANTS_H
+#define RADEON_PROGRAM_CONSTANTS_H
+
+typedef enum {
+	RC_SATURATE_NONE = 0,
+	RC_SATURATE_ZERO_ONE,
+	RC_SATURATE_MINUS_PLUS_ONE
+} rc_saturate_mode;
+
+typedef enum {
+	RC_TEXTURE_2D_ARRAY,
+	RC_TEXTURE_1D_ARRAY,
+	RC_TEXTURE_CUBE,
+	RC_TEXTURE_3D,
+	RC_TEXTURE_RECT,
+	RC_TEXTURE_2D,
+	RC_TEXTURE_1D
+} rc_texture_target;
+
+typedef enum {
+	/**
+	 * Used to indicate unused register descriptions and
+	 * source register that use a constant swizzle.
+	 */
+	RC_FILE_NONE = 0,
+	RC_FILE_TEMPORARY,
+
+	/**
+	 * Input register.
+	 *
+	 * \note The compiler attaches no implicit semantics to input registers.
+	 * Fragment/vertex program specific semantics must be defined explicitly
+	 * using the appropriate compiler interfaces.
+	 */
+	RC_FILE_INPUT,
+
+	/**
+	 * Output register.
+	 *
+	 * \note The compiler attaches no implicit semantics to input registers.
+	 * Fragment/vertex program specific semantics must be defined explicitly
+	 * using the appropriate compiler interfaces.
+	 */
+	RC_FILE_OUTPUT,
+	RC_FILE_ADDRESS,
+
+	/**
+	 * Indicates a constant from the \ref rc_constant_list .
+	 */
+	RC_FILE_CONSTANT,
+
+	/**
+	 * Indicates a special register, see RC_SPECIAL_xxx.
+	 */
+	RC_FILE_SPECIAL
+} rc_register_file;
+
+enum {
+	/** R500 fragment program ALU result "register" */
+	RC_SPECIAL_ALU_RESULT = 0,
+
+	/** Must be last */
+	RC_NUM_SPECIAL_REGISTERS
+};
+
+#define RC_REGISTER_INDEX_BITS 10
+#define RC_REGISTER_MAX_INDEX (1 << RC_REGISTER_INDEX_BITS)
+
+typedef enum {
+	RC_SWIZZLE_X = 0,
+	RC_SWIZZLE_Y,
+	RC_SWIZZLE_Z,
+	RC_SWIZZLE_W,
+	RC_SWIZZLE_ZERO,
+	RC_SWIZZLE_ONE,
+	RC_SWIZZLE_HALF,
+	RC_SWIZZLE_UNUSED
+} rc_swizzle;
+
+#define RC_MAKE_SWIZZLE(a,b,c,d) (((a)<<0) | ((b)<<3) | ((c)<<6) | ((d)<<9))
+#define RC_MAKE_SWIZZLE_SMEAR(a) RC_MAKE_SWIZZLE((a),(a),(a),(a))
+#define GET_SWZ(swz, idx)      (((swz) >> ((idx)*3)) & 0x7)
+#define GET_BIT(msk, idx)      (((msk) >> (idx)) & 0x1)
+#define SET_SWZ(swz, idx, newv) \
+	do { \
+		(swz) = ((swz) & ~(7 << ((idx)*3))) | ((newv) << ((idx)*3)); \
+	} while(0)
+
+#define RC_SWIZZLE_XYZW RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_W)
+#define RC_SWIZZLE_XXXX RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_X)
+#define RC_SWIZZLE_YYYY RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_Y)
+#define RC_SWIZZLE_ZZZZ RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_Z)
+#define RC_SWIZZLE_WWWW RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_W)
+#define RC_SWIZZLE_0000 RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_ZERO)
+#define RC_SWIZZLE_1111 RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_ONE)
+
+/**
+ * \name Bitmasks for components of vectors.
+ *
+ * Used for write masks, negation masks, etc.
+ */
+/*@{*/
+#define RC_MASK_NONE 0
+#define RC_MASK_X 1
+#define RC_MASK_Y 2
+#define RC_MASK_Z 4
+#define RC_MASK_W 8
+#define RC_MASK_XY (RC_MASK_X|RC_MASK_Y)
+#define RC_MASK_XYZ (RC_MASK_X|RC_MASK_Y|RC_MASK_Z)
+#define RC_MASK_XYW (RC_MASK_X|RC_MASK_Y|RC_MASK_W)
+#define RC_MASK_XYZW (RC_MASK_X|RC_MASK_Y|RC_MASK_Z|RC_MASK_W)
+/*@}*/
+
+typedef enum {
+	RC_ALURESULT_NONE = 0,
+	RC_ALURESULT_X,
+	RC_ALURESULT_W
+} rc_write_aluresult;
+
+#endif /* RADEON_PROGRAM_CONSTANTS_H */
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.c
index 4c26db5d24..ee839596aa 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2008 Nicolai Haehnle.
+ * Copyright (C) 2008-2009 Nicolai Haehnle.
  *
  * All Rights Reserved.
  *
@@ -25,584 +25,29 @@
  *
  */
 
-/**
- * @file
- *
- * Perform temporary register allocation and attempt to pair off instructions
- * in RGB and Alpha pairs. Also attempts to optimize the TEX instruction
- * vs. ALU instruction scheduling.
- */
-
 #include "radeon_program_pair.h"
 
-#include "memory_pool.h"
-#include "radeon_compiler.h"
-#include "shader/prog_print.h"
-
-#define error(fmt, args...) do { \
-	rc_error(&s->Compiler->Base, "%s::%s(): " fmt "\n",	\
-		__FILE__, __FUNCTION__, ##args);	\
-} while(0)
-
-struct pair_state_instruction {
-	struct prog_instruction Instruction;
-	GLuint IP; /**< Position of this instruction in original program */
-
-	GLuint IsTex:1; /**< Is a texture instruction */
-	GLuint NeedRGB:1; /**< Needs the RGB ALU */
-	GLuint NeedAlpha:1; /**< Needs the Alpha ALU */
-	GLuint IsTranscendent:1; /**< Is a special transcendent instruction */
-
-	/**
-	 * Number of (read and write) dependencies that must be resolved before
-	 * this instruction can be scheduled.
-	 */
-	GLuint NumDependencies:5;
-
-	/**
-	 * Next instruction in the linked list of ready instructions.
-	 */
-	struct pair_state_instruction *NextReady;
-
-	/**
-	 * Values that this instruction writes
-	 */
-	struct reg_value *Values[4];
-};
-
-
-/**
- * Used to keep track of which instructions read a value.
- */
-struct reg_value_reader {
-	struct pair_state_instruction *Reader;
-	struct reg_value_reader *Next;
-};
-
-/**
- * Used to keep track which values are stored in each component of a
- * PROGRAM_TEMPORARY.
- */
-struct reg_value {
-	struct pair_state_instruction *Writer;
-	struct reg_value *Next; /**< Pointer to the next value to be written to the same PROGRAM_TEMPORARY component */
-
-	/**
-	 * Unordered linked list of instructions that read from this value.
-	 */
-	struct reg_value_reader *Readers;
-
-	/**
-	 * Number of readers of this value. This is calculated during @ref scan_instructions
-	 * and continually decremented during code emission.
-	 * When this count reaches zero, the instruction that writes the @ref Next value
-	 * can be scheduled.
-	 */
-	GLuint NumReaders;
-};
-
-/**
- * Used to translate a PROGRAM_INPUT or PROGRAM_TEMPORARY Mesa register
- * to the proper hardware temporary.
- */
-struct pair_register_translation {
-	GLuint Allocated:1;
-	GLuint HwIndex:8;
-	GLuint RefCount:23; /**< # of times this occurs in an unscheduled instruction SrcReg or DstReg */
-
-	/**
-	 * Notes the value that is currently contained in each component
-	 * (only used for PROGRAM_TEMPORARY registers).
-	 */
-	struct reg_value *Value[4];
-};
-
-struct pair_state {
-	struct r300_fragment_program_compiler * Compiler;
-	const struct radeon_pair_handler *Handler;
-	GLboolean Verbose;
-	void *UserData;
-
-	/**
-	 * Translate Mesa registers to hardware registers
-	 */
-	struct pair_register_translation Inputs[FRAG_ATTRIB_MAX];
-	struct pair_register_translation Temps[MAX_PROGRAM_TEMPS];
-
-	struct {
-		GLuint RefCount; /**< # of times this occurs in an unscheduled SrcReg or DstReg */
-	} HwTemps[128];
-
-	/**
-	 * Linked list of instructions that can be scheduled right now,
-	 * based on which ALU/TEX resources they require.
-	 */
-	struct pair_state_instruction *ReadyFullALU;
-	struct pair_state_instruction *ReadyRGB;
-	struct pair_state_instruction *ReadyAlpha;
-	struct pair_state_instruction *ReadyTEX;
-};
-
-
-static struct pair_register_translation *get_register(struct pair_state *s, GLuint file, GLuint index)
-{
-	switch(file) {
-	case PROGRAM_TEMPORARY: return &s->Temps[index];
-	case PROGRAM_INPUT: return &s->Inputs[index];
-	default: return 0;
-	}
-}
-
-static void alloc_hw_reg(struct pair_state *s, GLuint file, GLuint index, GLuint hwindex)
-{
-	struct pair_register_translation *t = get_register(s, file, index);
-	ASSERT(!s->HwTemps[hwindex].RefCount);
-	ASSERT(!t->Allocated);
-	s->HwTemps[hwindex].RefCount = t->RefCount;
-	t->Allocated = 1;
-	t->HwIndex = hwindex;
-}
-
-static GLuint get_hw_reg(struct pair_state *s, GLuint file, GLuint index)
-{
-	GLuint hwindex;
-
-	struct pair_register_translation *t = get_register(s, file, index);
-	if (!t) {
-		error("get_hw_reg: %i[%i]\n", file, index);
-		return 0;
-	}
-
-	if (t->Allocated)
-		return t->HwIndex;
-
-	for(hwindex = 0; hwindex < s->Handler->MaxHwTemps; ++hwindex)
-		if (!s->HwTemps[hwindex].RefCount)
-			break;
-
-	if (hwindex >= s->Handler->MaxHwTemps) {
-		error("Ran out of hardware temporaries");
-		return 0;
-	}
-
-	alloc_hw_reg(s, file, index, hwindex);
-	return hwindex;
-}
-
-
-static void deref_hw_reg(struct pair_state *s, GLuint hwindex)
-{
-	if (!s->HwTemps[hwindex].RefCount) {
-		error("Hwindex %i refcount error", hwindex);
-		return;
-	}
-
-	s->HwTemps[hwindex].RefCount--;
-}
-
-static void add_pairinst_to_list(struct pair_state_instruction **list, struct pair_state_instruction *pairinst)
-{
-	pairinst->NextReady = *list;
-	*list = pairinst;
-}
-
-/**
- * The given instruction has become ready. Link it into the ready
- * instructions.
- */
-static void instruction_ready(struct pair_state *s, struct pair_state_instruction *pairinst)
-{
-	if (s->Verbose)
-		_mesa_printf("instruction_ready(%i)\n", pairinst->IP);
-
-	if (pairinst->IsTex)
-		add_pairinst_to_list(&s->ReadyTEX, pairinst);
-	else if (!pairinst->NeedAlpha)
-		add_pairinst_to_list(&s->ReadyRGB, pairinst);
-	else if (!pairinst->NeedRGB)
-		add_pairinst_to_list(&s->ReadyAlpha, pairinst);
-	else
-		add_pairinst_to_list(&s->ReadyFullALU, pairinst);
-}
-
-
-/**
- * Finally rewrite ADD, MOV, MUL as the appropriate native instruction
- * and reverse the order of arguments for CMP.
- */
-static void final_rewrite(struct pair_state *s, struct prog_instruction *inst)
-{
-	struct prog_src_register tmp;
-
-	switch(inst->Opcode) {
-	case OPCODE_ADD:
-		inst->SrcReg[2] = inst->SrcReg[1];
-		inst->SrcReg[1].File = PROGRAM_BUILTIN;
-		inst->SrcReg[1].Swizzle = SWIZZLE_1111;
-		inst->SrcReg[1].Negate = NEGATE_NONE;
-		inst->Opcode = OPCODE_MAD;
-		break;
-	case OPCODE_CMP:
-		tmp = inst->SrcReg[2];
-		inst->SrcReg[2] = inst->SrcReg[0];
-		inst->SrcReg[0] = tmp;
-		break;
-	case OPCODE_MOV:
-		/* AMD say we should use CMP.
-		 * However, when we transform
-		 *  KIL -r0;
-		 * into
-		 *  CMP tmp, -r0, -r0, 0;
-		 *  KIL tmp;
-		 * we get incorrect behaviour on R500 when r0 == 0.0.
-		 * It appears that the R500 KIL hardware treats -0.0 as less
-		 * than zero.
-		 */
-		inst->SrcReg[1].File = PROGRAM_BUILTIN;
-		inst->SrcReg[1].Swizzle = SWIZZLE_1111;
-		inst->SrcReg[2].File = PROGRAM_BUILTIN;
-		inst->SrcReg[2].Swizzle = SWIZZLE_0000;
-		inst->Opcode = OPCODE_MAD;
-		break;
-	case OPCODE_MUL:
-		inst->SrcReg[2].File = PROGRAM_BUILTIN;
-		inst->SrcReg[2].Swizzle = SWIZZLE_0000;
-		inst->Opcode = OPCODE_MAD;
-		break;
-	default:
-		/* nothing to do */
-		break;
-	}
-}
-
-
-/**
- * Classify an instruction according to which ALUs etc. it needs
- */
-static void classify_instruction(struct pair_state *s,
-	struct pair_state_instruction *psi)
-{
-	psi->NeedRGB = (psi->Instruction.DstReg.WriteMask & WRITEMASK_XYZ) ? 1 : 0;
-	psi->NeedAlpha = (psi->Instruction.DstReg.WriteMask & WRITEMASK_W) ? 1 : 0;
-
-	switch(psi->Instruction.Opcode) {
-	case OPCODE_ADD:
-	case OPCODE_CMP:
-	case OPCODE_DDX:
-	case OPCODE_DDY:
-	case OPCODE_FRC:
-	case OPCODE_MAD:
-	case OPCODE_MAX:
-	case OPCODE_MIN:
-	case OPCODE_MOV:
-	case OPCODE_MUL:
-		break;
-	case OPCODE_COS:
-	case OPCODE_EX2:
-	case OPCODE_LG2:
-	case OPCODE_RCP:
-	case OPCODE_RSQ:
-	case OPCODE_SIN:
-		psi->IsTranscendent = 1;
-		psi->NeedAlpha = 1;
-		break;
-	case OPCODE_DP4:
-		psi->NeedAlpha = 1;
-		/* fall through */
-	case OPCODE_DP3:
-		psi->NeedRGB = 1;
-		break;
-	case OPCODE_KIL:
-	case OPCODE_TEX:
-	case OPCODE_TXB:
-	case OPCODE_TXP:
-	case OPCODE_END:
-		psi->IsTex = 1;
-		break;
-	default:
-		error("Unknown opcode %d\n", psi->Instruction.Opcode);
-		break;
-	}
-}
-
-
-/**
- * Count which (input, temporary) register is read and written how often,
- * and scan the instruction stream to find dependencies.
- */
-static void scan_instructions(struct pair_state *s)
-{
-	struct rc_instruction *source;
-	GLuint ip;
-
-	for(source = s->Compiler->Base.Program.Instructions.Next, ip = 0;
-	    source != &s->Compiler->Base.Program.Instructions;
-	    source = source->Next, ++ip) {
-		struct pair_state_instruction *pairinst = memory_pool_malloc(&s->Compiler->Base.Pool, sizeof(*pairinst));
-		memset(pairinst, 0, sizeof(struct pair_state_instruction));
-
-		pairinst->Instruction = source->I;
-		pairinst->IP = ip;
-		final_rewrite(s, &pairinst->Instruction);
-		classify_instruction(s, pairinst);
-
-		int nsrc = _mesa_num_inst_src_regs(pairinst->Instruction.Opcode);
-		int j;
-		for(j = 0; j < nsrc; j++) {
-			struct pair_register_translation *t =
-				get_register(s, pairinst->Instruction.SrcReg[j].File, pairinst->Instruction.SrcReg[j].Index);
-			if (!t)
-				continue;
-
-			t->RefCount++;
-
-			if (pairinst->Instruction.SrcReg[j].File == PROGRAM_TEMPORARY) {
-				int i;
-				for(i = 0; i < 4; ++i) {
-					GLuint swz = GET_SWZ(pairinst->Instruction.SrcReg[j].Swizzle, i);
-					if (swz >= 4)
-						continue; /* constant or NIL swizzle */
-					if (!t->Value[swz])
-						continue; /* this is an undefined read */
-
-					/* Do not add a dependency if this instruction
-					 * also rewrites the value. The code below adds
-					 * a dependency for the DstReg, which is a superset
-					 * of the SrcReg dependency. */
-					if (pairinst->Instruction.DstReg.File == PROGRAM_TEMPORARY &&
-					    pairinst->Instruction.DstReg.Index == pairinst->Instruction.SrcReg[j].Index &&
-					    GET_BIT(pairinst->Instruction.DstReg.WriteMask, swz))
-						continue;
-
-					struct reg_value_reader* r = memory_pool_malloc(&s->Compiler->Base.Pool, sizeof(*r));
-					pairinst->NumDependencies++;
-					t->Value[swz]->NumReaders++;
-					r->Reader = pairinst;
-					r->Next = t->Value[swz]->Readers;
-					t->Value[swz]->Readers = r;
-				}
-			}
-		}
-
-		int ndst = _mesa_num_inst_dst_regs(pairinst->Instruction.Opcode);
-		if (ndst) {
-			struct pair_register_translation *t =
-				get_register(s, pairinst->Instruction.DstReg.File, pairinst->Instruction.DstReg.Index);
-			if (t) {
-				t->RefCount++;
-
-				if (pairinst->Instruction.DstReg.File == PROGRAM_TEMPORARY) {
-					int j;
-					for(j = 0; j < 4; ++j) {
-						if (!GET_BIT(pairinst->Instruction.DstReg.WriteMask, j))
-							continue;
-
-						struct reg_value* v = memory_pool_malloc(&s->Compiler->Base.Pool, sizeof(*v));
-						memset(v, 0, sizeof(struct reg_value));
-						v->Writer = pairinst;
-						if (t->Value[j]) {
-							pairinst->NumDependencies++;
-							t->Value[j]->Next = v;
-						}
-						t->Value[j] = v;
-						pairinst->Values[j] = v;
-					}
-				}
-			}
-		}
-
-		if (s->Verbose)
-			_mesa_printf("scan(%i): NumDeps = %i\n", ip, pairinst->NumDependencies);
-
-		if (!pairinst->NumDependencies)
-			instruction_ready(s, pairinst);
-	}
-
-	/* Clear the PROGRAM_TEMPORARY state */
-	int i, j;
-	for(i = 0; i < MAX_PROGRAM_TEMPS; ++i) {
-		for(j = 0; j < 4; ++j)
-			s->Temps[i].Value[j] = 0;
-	}
-}
-
-
-static void decrement_dependencies(struct pair_state *s, struct pair_state_instruction *pairinst)
-{
-	ASSERT(pairinst->NumDependencies > 0);
-	if (!--pairinst->NumDependencies)
-		instruction_ready(s, pairinst);
-}
 
 /**
- * Update the dependency tracking state based on what the instruction
- * at the given IP does.
+ * Return the source slot where we installed the given register access,
+ * or -1 if no slot was free anymore.
  */
-static void commit_instruction(struct pair_state *s, struct pair_state_instruction *pairinst)
-{
-	struct prog_instruction *inst = &pairinst->Instruction;
-
-	if (s->Verbose)
-		_mesa_printf("commit_instruction(%i)\n", pairinst->IP);
-
-	if (inst->DstReg.File == PROGRAM_TEMPORARY) {
-		struct pair_register_translation *t = &s->Temps[inst->DstReg.Index];
-		deref_hw_reg(s, t->HwIndex);
-
-		int i;
-		for(i = 0; i < 4; ++i) {
-			if (!GET_BIT(inst->DstReg.WriteMask, i))
-				continue;
-
-			t->Value[i] = pairinst->Values[i];
-			if (t->Value[i]->NumReaders) {
-				struct reg_value_reader *r;
-				for(r = pairinst->Values[i]->Readers; r; r = r->Next)
-					decrement_dependencies(s, r->Reader);
-			} else if (t->Value[i]->Next) {
-				/* This happens when the only reader writes
-				 * the register at the same time */
-				decrement_dependencies(s, t->Value[i]->Next->Writer);
-			}
-		}
-	}
-
-	int nsrc = _mesa_num_inst_src_regs(inst->Opcode);
-	int i;
-	for(i = 0; i < nsrc; i++) {
-		struct pair_register_translation *t = get_register(s, inst->SrcReg[i].File, inst->SrcReg[i].Index);
-		if (!t)
-			continue;
-
-		deref_hw_reg(s, get_hw_reg(s, inst->SrcReg[i].File, inst->SrcReg[i].Index));
-
-		if (inst->SrcReg[i].File != PROGRAM_TEMPORARY)
-			continue;
-
-		int j;
-		for(j = 0; j < 4; ++j) {
-			GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, j);
-			if (swz >= 4)
-				continue;
-			if (!t->Value[swz])
-				continue;
-
-			/* Do not free a dependency if this instruction
-			 * also rewrites the value. See scan_instructions. */
-			if (inst->DstReg.File == PROGRAM_TEMPORARY &&
-			    inst->DstReg.Index == inst->SrcReg[i].Index &&
-			    GET_BIT(inst->DstReg.WriteMask, swz))
-				continue;
-
-			if (!--t->Value[swz]->NumReaders) {
-				if (t->Value[swz]->Next)
-					decrement_dependencies(s, t->Value[swz]->Next->Writer);
-			}
-		}
-	}
-}
-
-
-/**
- * Emit all ready texture instructions in a single block.
- *
- * Emit as a single block to (hopefully) sample many textures in parallel,
- * and to avoid hardware indirections on R300.
- *
- * In R500, we don't really know when the result of a texture instruction
- * arrives. So allocate all destinations first, to make sure they do not
- * arrive early and overwrite a texture coordinate we're going to use later
- * in the block.
- */
-static void emit_all_tex(struct pair_state *s)
-{
-	struct pair_state_instruction *readytex;
-	struct pair_state_instruction *pairinst;
-
-	ASSERT(s->ReadyTEX);
-
-	// Don't let the ready list change under us!
-	readytex = s->ReadyTEX;
-	s->ReadyTEX = 0;
-
-	// Allocate destination hardware registers in one block to avoid conflicts.
-	for(pairinst = readytex; pairinst; pairinst = pairinst->NextReady) {
-		struct prog_instruction *inst = &pairinst->Instruction;
-		if (inst->Opcode != OPCODE_KIL)
-			get_hw_reg(s, inst->DstReg.File, inst->DstReg.Index);
-	}
-
-	if (s->Compiler->Base.Debug)
-		_mesa_printf(" BEGIN_TEX\n");
-
-	if (s->Handler->BeginTexBlock)
-		s->Compiler->Base.Error = s->Compiler->Base.Error || !s->Handler->BeginTexBlock(s->UserData);
-
-	for(pairinst = readytex; pairinst; pairinst = pairinst->NextReady) {
-		struct prog_instruction *inst = &pairinst->Instruction;
-		commit_instruction(s, pairinst);
-
-		if (inst->Opcode != OPCODE_KIL)
-			inst->DstReg.Index = get_hw_reg(s, inst->DstReg.File, inst->DstReg.Index);
-		inst->SrcReg[0].Index = get_hw_reg(s, inst->SrcReg[0].File, inst->SrcReg[0].Index);
-
-		if (s->Compiler->Base.Debug) {
-			_mesa_printf("   ");
-			_mesa_print_instruction(inst);
-			fflush(stderr);
-		}
-
-		struct radeon_pair_texture_instruction rpti;
-
-		switch(inst->Opcode) {
-		case OPCODE_TEX: rpti.Opcode = RADEON_OPCODE_TEX; break;
-		case OPCODE_TXB: rpti.Opcode = RADEON_OPCODE_TXB; break;
-		case OPCODE_TXP: rpti.Opcode = RADEON_OPCODE_TXP; break;
-		default:
-		case OPCODE_KIL: rpti.Opcode = RADEON_OPCODE_KIL; break;
-		}
-
-		rpti.DestIndex = inst->DstReg.Index;
-		rpti.WriteMask = inst->DstReg.WriteMask;
-		rpti.TexSrcUnit = inst->TexSrcUnit;
-		rpti.TexSrcTarget = inst->TexSrcTarget;
-		rpti.SrcIndex = inst->SrcReg[0].Index;
-		rpti.SrcSwizzle = inst->SrcReg[0].Swizzle;
-
-		s->Compiler->Base.Error = s->Compiler->Base.Error || !s->Handler->EmitTex(s->UserData, &rpti);
-	}
-
-	if (s->Compiler->Base.Debug)
-		_mesa_printf(" END_TEX\n");
-}
-
-
-static int alloc_pair_source(struct pair_state *s, struct radeon_pair_instruction *pair,
-	struct prog_src_register src, GLboolean rgb, GLboolean alpha)
+int rc_pair_alloc_source(struct rc_pair_instruction *pair,
+	unsigned int rgb, unsigned int alpha,
+	rc_register_file file, unsigned int index)
 {
 	int candidate = -1;
 	int candidate_quality = -1;
 	int i;
 
-	if (!rgb && !alpha)
+	if ((!rgb && !alpha) || file == RC_FILE_NONE)
 		return 0;
 
-	GLuint constant;
-	GLuint index;
-
-	if (src.File == PROGRAM_TEMPORARY || src.File == PROGRAM_INPUT) {
-		constant = 0;
-		index = get_hw_reg(s, src.File, src.Index);
-	} else {
-		constant = 1;
-		index = src.Index;
-	}
-
 	for(i = 0; i < 3; ++i) {
 		int q = 0;
 		if (rgb) {
 			if (pair->RGB.Src[i].Used) {
-				if (pair->RGB.Src[i].Constant != constant ||
+				if (pair->RGB.Src[i].File != file ||
 				    pair->RGB.Src[i].Index != index)
 					continue;
 				q++;
@@ -610,7 +55,7 @@ static int alloc_pair_source(struct pair_state *s, struct radeon_pair_instructio
 		}
 		if (alpha) {
 			if (pair->Alpha.Src[i].Used) {
-				if (pair->Alpha.Src[i].Constant != constant ||
+				if (pair->Alpha.Src[i].File != file ||
 				    pair->Alpha.Src[i].Index != index)
 					continue;
 				q++;
@@ -625,334 +70,15 @@ static int alloc_pair_source(struct pair_state *s, struct radeon_pair_instructio
 	if (candidate >= 0) {
 		if (rgb) {
 			pair->RGB.Src[candidate].Used = 1;
-			pair->RGB.Src[candidate].Constant = constant;
+			pair->RGB.Src[candidate].File = file;
 			pair->RGB.Src[candidate].Index = index;
 		}
 		if (alpha) {
 			pair->Alpha.Src[candidate].Used = 1;
-			pair->Alpha.Src[candidate].Constant = constant;
+			pair->Alpha.Src[candidate].File = file;
 			pair->Alpha.Src[candidate].Index = index;
 		}
 	}
 
 	return candidate;
 }
-
-/**
- * Fill the given ALU instruction's opcodes and source operands into the given pair,
- * if possible.
- */
-static GLboolean fill_instruction_into_pair(
-	struct pair_state *s,
-	struct radeon_pair_instruction *pair,
-	struct pair_state_instruction *pairinst)
-{
-	struct prog_instruction *inst = &pairinst->Instruction;
-
-	ASSERT(!pairinst->NeedRGB || pair->RGB.Opcode == OPCODE_NOP);
-	ASSERT(!pairinst->NeedAlpha || pair->Alpha.Opcode == OPCODE_NOP);
-
-	if (pairinst->NeedRGB) {
-		if (pairinst->IsTranscendent)
-			pair->RGB.Opcode = OPCODE_REPL_ALPHA;
-		else
-			pair->RGB.Opcode = inst->Opcode;
-		if (inst->SaturateMode == SATURATE_ZERO_ONE)
-			pair->RGB.Saturate = 1;
-	}
-	if (pairinst->NeedAlpha) {
-		pair->Alpha.Opcode = inst->Opcode;
-		if (inst->SaturateMode == SATURATE_ZERO_ONE)
-			pair->Alpha.Saturate = 1;
-	}
-
-	int nargs = _mesa_num_inst_src_regs(inst->Opcode);
-	int i;
-
-	/* Special case for DDX/DDY (MDH/MDV). */
-	if (inst->Opcode == OPCODE_DDX || inst->Opcode == OPCODE_DDY) {
-		if (pair->RGB.Src[0].Used || pair->Alpha.Src[0].Used)
-			return GL_FALSE;
-		else
-			nargs++;
-	}
-
-	for(i = 0; i < nargs; ++i) {
-		int source;
-		if (pairinst->NeedRGB && !pairinst->IsTranscendent) {
-			GLboolean srcrgb = GL_FALSE;
-			GLboolean srcalpha = GL_FALSE;
-			int j;
-			for(j = 0; j < 3; ++j) {
-				GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, j);
-				if (swz < 3)
-					srcrgb = GL_TRUE;
-				else if (swz < 4)
-					srcalpha = GL_TRUE;
-			}
-			source = alloc_pair_source(s, pair, inst->SrcReg[i], srcrgb, srcalpha);
-			if (source < 0)
-				return GL_FALSE;
-			pair->RGB.Arg[i].Source = source;
-			pair->RGB.Arg[i].Swizzle = inst->SrcReg[i].Swizzle & 0x1ff;
-			pair->RGB.Arg[i].Abs = inst->SrcReg[i].Abs;
-			pair->RGB.Arg[i].Negate = !!(inst->SrcReg[i].Negate & (NEGATE_X | NEGATE_Y | NEGATE_Z));
-		}
-		if (pairinst->NeedAlpha) {
-			GLboolean srcrgb = GL_FALSE;
-			GLboolean srcalpha = GL_FALSE;
-			GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, pairinst->IsTranscendent ? 0 : 3);
-			if (swz < 3)
-				srcrgb = GL_TRUE;
-			else if (swz < 4)
-				srcalpha = GL_TRUE;
-			source = alloc_pair_source(s, pair, inst->SrcReg[i], srcrgb, srcalpha);
-			if (source < 0)
-				return GL_FALSE;
-			pair->Alpha.Arg[i].Source = source;
-			pair->Alpha.Arg[i].Swizzle = swz;
-			pair->Alpha.Arg[i].Abs = inst->SrcReg[i].Abs;
-			pair->Alpha.Arg[i].Negate = !!(inst->SrcReg[i].Negate & NEGATE_W);
-		}
-	}
-
-	return GL_TRUE;
-}
-
-
-/**
- * Fill in the destination register information.
- *
- * This is split from filling in source registers because we want
- * to avoid allocating hardware temporaries for destinations until
- * we are absolutely certain that we're going to emit a certain
- * instruction pairing.
- */
-static void fill_dest_into_pair(
-	struct pair_state *s,
-	struct radeon_pair_instruction *pair,
-	struct pair_state_instruction *pairinst)
-{
-	struct prog_instruction *inst = &pairinst->Instruction;
-
-	if (inst->DstReg.File == PROGRAM_OUTPUT) {
-		if (inst->DstReg.Index == s->Compiler->OutputColor) {
-			pair->RGB.OutputWriteMask |= inst->DstReg.WriteMask & WRITEMASK_XYZ;
-			pair->Alpha.OutputWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
-		} else if (inst->DstReg.Index == s->Compiler->OutputDepth) {
-			pair->Alpha.DepthWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
-		}
-	} else {
-		GLuint hwindex = get_hw_reg(s, inst->DstReg.File, inst->DstReg.Index);
-		if (pairinst->NeedRGB) {
-			pair->RGB.DestIndex = hwindex;
-			pair->RGB.WriteMask |= inst->DstReg.WriteMask & WRITEMASK_XYZ;
-		}
-		if (pairinst->NeedAlpha) {
-			pair->Alpha.DestIndex = hwindex;
-			pair->Alpha.WriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
-		}
-	}
-}
-
-
-/**
- * Find a good ALU instruction or pair of ALU instruction and emit it.
- *
- * Prefer emitting full ALU instructions, so that when we reach a point
- * where no full ALU instruction can be emitted, we have more candidates
- * for RGB/Alpha pairing.
- */
-static void emit_alu(struct pair_state *s)
-{
-	struct radeon_pair_instruction pair;
-	struct pair_state_instruction *psi;
-
-	if (s->ReadyFullALU || !(s->ReadyRGB && s->ReadyAlpha)) {
-		if (s->ReadyFullALU) {
-			psi = s->ReadyFullALU;
-			s->ReadyFullALU = s->ReadyFullALU->NextReady;
-		} else if (s->ReadyRGB) {
-			psi = s->ReadyRGB;
-			s->ReadyRGB = s->ReadyRGB->NextReady;
-		} else {
-			psi = s->ReadyAlpha;
-			s->ReadyAlpha = s->ReadyAlpha->NextReady;
-		}
-
-		_mesa_bzero(&pair, sizeof(pair));
-		fill_instruction_into_pair(s, &pair, psi);
-		fill_dest_into_pair(s, &pair, psi);
-		commit_instruction(s, psi);
-	} else {
-		struct pair_state_instruction **prgb;
-		struct pair_state_instruction **palpha;
-
-		/* Some pairings might fail because they require too
-		 * many source slots; try all possible pairings if necessary */
-		for(prgb = &s->ReadyRGB; *prgb; prgb = &(*prgb)->NextReady) {
-			for(palpha = &s->ReadyAlpha; *palpha; palpha = &(*palpha)->NextReady) {
-				struct pair_state_instruction * psirgb = *prgb;
-				struct pair_state_instruction * psialpha = *palpha;
-				_mesa_bzero(&pair, sizeof(pair));
-				fill_instruction_into_pair(s, &pair, psirgb);
-				if (!fill_instruction_into_pair(s, &pair, psialpha))
-					continue;
-				*prgb = (*prgb)->NextReady;
-				*palpha = (*palpha)->NextReady;
-				fill_dest_into_pair(s, &pair, psirgb);
-				fill_dest_into_pair(s, &pair, psialpha);
-				commit_instruction(s, psirgb);
-				commit_instruction(s, psialpha);
-				goto success;
-			}
-		}
-
-		/* No success in pairing; just take the first RGB instruction */
-		psi = s->ReadyRGB;
-		s->ReadyRGB = s->ReadyRGB->NextReady;
-
-		_mesa_bzero(&pair, sizeof(pair));
-		fill_instruction_into_pair(s, &pair, psi);
-		fill_dest_into_pair(s, &pair, psi);
-		commit_instruction(s, psi);
-	success: ;
-	}
-
-	if (s->Compiler->Base.Debug)
-		radeonPrintPairInstruction(&pair);
-
-	s->Compiler->Base.Error = s->Compiler->Base.Error || !s->Handler->EmitPaired(s->UserData, &pair);
-}
-
-/* Callback function for assigning input registers to hardware registers */
-static void alloc_helper(void * data, unsigned input, unsigned hwreg)
-{
-	struct pair_state * s = data;
-	alloc_hw_reg(s, PROGRAM_INPUT, input, hwreg);
-}
-
-void radeonPairProgram(
-	struct r300_fragment_program_compiler * compiler,
-	const struct radeon_pair_handler* handler, void *userdata)
-{
-	struct pair_state s;
-
-	_mesa_bzero(&s, sizeof(s));
-	s.Compiler = compiler;
-	s.Handler = handler;
-	s.UserData = userdata;
-	s.Verbose = GL_FALSE && s.Compiler->Base.Debug;
-
-	if (s.Compiler->Base.Debug)
-		_mesa_printf("Emit paired program\n");
-
-	scan_instructions(&s);
-	s.Compiler->AllocateHwInputs(s.Compiler, &alloc_helper, &s);
-
-	while(!s.Compiler->Base.Error &&
-	      (s.ReadyTEX || s.ReadyRGB || s.ReadyAlpha || s.ReadyFullALU)) {
-		if (s.ReadyTEX)
-			emit_all_tex(&s);
-
-		while(s.ReadyFullALU || s.ReadyRGB || s.ReadyAlpha)
-			emit_alu(&s);
-	}
-
-	if (s.Compiler->Base.Debug)
-		_mesa_printf(" END\n");
-}
-
-
-static void print_pair_src(int i, struct radeon_pair_instruction_source* src)
-{
-	_mesa_printf("  Src%i = %s[%i]", i, src->Constant ? "CNST" : "TEMP", src->Index);
-}
-
-static const char* opcode_string(GLuint opcode)
-{
-	if (opcode == OPCODE_REPL_ALPHA)
-		return "SOP";
-	else
-		return _mesa_opcode_string(opcode);
-}
-
-static int num_pairinst_args(GLuint opcode)
-{
-	if (opcode == OPCODE_REPL_ALPHA)
-		return 0;
-	else
-		return _mesa_num_inst_src_regs(opcode);
-}
-
-static char swizzle_char(GLuint swz)
-{
-	switch(swz) {
-	case SWIZZLE_X: return 'x';
-	case SWIZZLE_Y: return 'y';
-	case SWIZZLE_Z: return 'z';
-	case SWIZZLE_W: return 'w';
-	case SWIZZLE_ZERO: return '0';
-	case SWIZZLE_ONE: return '1';
-	case SWIZZLE_NIL: return '_';
-	default: return '?';
-	}
-}
-
-void radeonPrintPairInstruction(struct radeon_pair_instruction *inst)
-{
-	int nargs;
-	int i;
-
-	_mesa_printf("       RGB:  ");
-	for(i = 0; i < 3; ++i) {
-		if (inst->RGB.Src[i].Used)
-			print_pair_src(i, inst->RGB.Src + i);
-	}
-	_mesa_printf("\n");
-	_mesa_printf("       Alpha:");
-	for(i = 0; i < 3; ++i) {
-		if (inst->Alpha.Src[i].Used)
-			print_pair_src(i, inst->Alpha.Src + i);
-	}
-	_mesa_printf("\n");
-
-	_mesa_printf("  %s%s", opcode_string(inst->RGB.Opcode), inst->RGB.Saturate ? "_SAT" : "");
-	if (inst->RGB.WriteMask)
-		_mesa_printf(" TEMP[%i].%s%s%s", inst->RGB.DestIndex,
-			(inst->RGB.WriteMask & 1) ? "x" : "",
-			(inst->RGB.WriteMask & 2) ? "y" : "",
-			(inst->RGB.WriteMask & 4) ? "z" : "");
-	if (inst->RGB.OutputWriteMask)
-		_mesa_printf(" COLOR.%s%s%s",
-			(inst->RGB.OutputWriteMask & 1) ? "x" : "",
-			(inst->RGB.OutputWriteMask & 2) ? "y" : "",
-			(inst->RGB.OutputWriteMask & 4) ? "z" : "");
-	nargs = num_pairinst_args(inst->RGB.Opcode);
-	for(i = 0; i < nargs; ++i) {
-		const char* abs = inst->RGB.Arg[i].Abs ? "|" : "";
-		const char* neg = inst->RGB.Arg[i].Negate ? "-" : "";
-		_mesa_printf(", %s%sSrc%i.%c%c%c%s", neg, abs, inst->RGB.Arg[i].Source,
-			swizzle_char(GET_SWZ(inst->RGB.Arg[i].Swizzle, 0)),
-			swizzle_char(GET_SWZ(inst->RGB.Arg[i].Swizzle, 1)),
-			swizzle_char(GET_SWZ(inst->RGB.Arg[i].Swizzle, 2)),
-			abs);
-	}
-	_mesa_printf("\n");
-
-	_mesa_printf("  %s%s", opcode_string(inst->Alpha.Opcode), inst->Alpha.Saturate ? "_SAT" : "");
-	if (inst->Alpha.WriteMask)
-		_mesa_printf(" TEMP[%i].w", inst->Alpha.DestIndex);
-	if (inst->Alpha.OutputWriteMask)
-		_mesa_printf(" COLOR.w");
-	if (inst->Alpha.DepthWriteMask)
-		_mesa_printf(" DEPTH.w");
-	nargs = num_pairinst_args(inst->Alpha.Opcode);
-	for(i = 0; i < nargs; ++i) {
-		const char* abs = inst->Alpha.Arg[i].Abs ? "|" : "";
-		const char* neg = inst->Alpha.Arg[i].Negate ? "-" : "";
-		_mesa_printf(", %s%sSrc%i.%c%s", neg, abs, inst->Alpha.Arg[i].Source,
-			swizzle_char(inst->Alpha.Arg[i].Swizzle), abs);
-	}
-	_mesa_printf("\n");
-}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.h b/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.h
index ff76178551..1600598428 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.h
@@ -28,116 +28,97 @@
 #ifndef __RADEON_PROGRAM_PAIR_H_
 #define __RADEON_PROGRAM_PAIR_H_
 
-#include "radeon_program.h"
+#include "radeon_code.h"
+#include "radeon_opcodes.h"
+#include "radeon_program_constants.h"
 
 struct r300_fragment_program_compiler;
 
 
 /**
- * Represents a paired instruction, as found in R300 and R500
+ * \file
+ * Represents a paired ALU instruction, as found in R300 and R500
  * fragment programs.
+ *
+ * Note that this representation is taking some liberties as far
+ * as register files are concerned, to allow separate register
+ * allocation.
+ *
+ * Also note that there are some subtleties in that the semantics
+ * of certain opcodes are implicitly changed in this representation;
+ * see \ref rc_pair_translate
  */
+
+
 struct radeon_pair_instruction_source {
-	GLuint Index:8;
-	GLuint Constant:1;
-	GLuint Used:1;
+	unsigned int Used:1;
+	rc_register_file File:3;
+	unsigned int Index:RC_REGISTER_INDEX_BITS;
 };
 
 struct radeon_pair_instruction_rgb {
-	GLuint Opcode:8;
-	GLuint DestIndex:8;
-	GLuint WriteMask:3;
-	GLuint OutputWriteMask:3;
-	GLuint Saturate:1;
+	rc_opcode Opcode:8;
+	unsigned int DestIndex:RC_REGISTER_INDEX_BITS;
+	unsigned int WriteMask:3;
+	unsigned int OutputWriteMask:3;
+	unsigned int Saturate:1;
 
 	struct radeon_pair_instruction_source Src[3];
 
 	struct {
-		GLuint Source:2;
-		GLuint Swizzle:9;
-		GLuint Abs:1;
-		GLuint Negate:1;
+		unsigned int Source:2;
+		unsigned int Swizzle:9;
+		unsigned int Abs:1;
+		unsigned int Negate:1;
 	} Arg[3];
 };
 
 struct radeon_pair_instruction_alpha {
-	GLuint Opcode:8;
-	GLuint DestIndex:8;
-	GLuint WriteMask:1;
-	GLuint OutputWriteMask:1;
-	GLuint DepthWriteMask:1;
-	GLuint Saturate:1;
+	rc_opcode Opcode:8;
+	unsigned int DestIndex:RC_REGISTER_INDEX_BITS;
+	unsigned int WriteMask:1;
+	unsigned int OutputWriteMask:1;
+	unsigned int DepthWriteMask:1;
+	unsigned int Saturate:1;
 
 	struct radeon_pair_instruction_source Src[3];
 
 	struct {
-		GLuint Source:2;
-		GLuint Swizzle:3;
-		GLuint Abs:1;
-		GLuint Negate:1;
+		unsigned int Source:2;
+		unsigned int Swizzle:3;
+		unsigned int Abs:1;
+		unsigned int Negate:1;
 	} Arg[3];
 };
 
-struct radeon_pair_instruction {
+struct rc_pair_instruction {
 	struct radeon_pair_instruction_rgb RGB;
 	struct radeon_pair_instruction_alpha Alpha;
-};
-
 
-enum {
-	RADEON_OPCODE_TEX = 0,
-	RADEON_OPCODE_TXB,
-	RADEON_OPCODE_TXP,
-	RADEON_OPCODE_KIL
+	rc_write_aluresult WriteALUResult:2;
+	rc_compare_func ALUResultCompare:3;
 };
 
-struct radeon_pair_texture_instruction {
-	GLuint Opcode:2; /**< one of RADEON_OPCODE_xxx */
-
-	GLuint DestIndex:8;
-	GLuint WriteMask:4;
 
-	GLuint TexSrcUnit:5;
-	GLuint TexSrcTarget:3;
-
-	GLuint SrcIndex:8;
-	GLuint SrcSwizzle:12;
-};
+/**
+ * General helper functions for dealing with the paired instruction format.
+ */
+/*@{*/
+int rc_pair_alloc_source(struct rc_pair_instruction *pair,
+	unsigned int rgb, unsigned int alpha,
+	rc_register_file file, unsigned int index);
+/*@}*/
 
 
 /**
- *
+ * Compiler passes that operate with the paired format.
  */
-struct radeon_pair_handler {
-	/**
-	 * Write a paired instruction to the hardware.
-	 *
-	 * @return GL_FALSE on error.
-	 */
-	GLboolean (*EmitPaired)(void*, struct radeon_pair_instruction*);
-
-	/**
-	 * Write a texture instruction to the hardware.
-	 * Register indices have already been rewritten to the allocated
-	 * hardware register numbers.
-	 *
-	 * @return GL_FALSE on error.
-	 */
-	GLboolean (*EmitTex)(void*, struct radeon_pair_texture_instruction*);
-
-	/**
-	 * Called before a block of contiguous, independent texture
-	 * instructions is emitted.
-	 */
-	GLboolean (*BeginTexBlock)(void*);
-
-	unsigned MaxHwTemps;
-};
-
-void radeonPairProgram(
-	struct r300_fragment_program_compiler * compiler,
-	const struct radeon_pair_handler*, void *userdata);
+/*@{*/
+struct radeon_pair_handler;
 
-void radeonPrintPairInstruction(struct radeon_pair_instruction *inst);
+void rc_pair_translate(struct r300_fragment_program_compiler *c);
+void rc_pair_schedule(struct r300_fragment_program_compiler *c);
+void rc_pair_regalloc(struct r300_fragment_program_compiler *c, unsigned maxtemps);
+/*@}*/
 
 #endif /* __RADEON_PROGRAM_PAIR_H_ */
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_print.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_print.c
new file mode 100644
index 0000000000..d863b82d53
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_print.c
@@ -0,0 +1,300 @@
+/*
+ * Copyright 2009 Nicolai Hähnle <nhaehnle@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE. */
+
+#include "radeon_program.h"
+
+#include <stdio.h>
+
+static const char * textarget_to_string(rc_texture_target target)
+{
+	switch(target) {
+	case RC_TEXTURE_2D_ARRAY: return "2D_ARRAY";
+	case RC_TEXTURE_1D_ARRAY: return "1D_ARRAY";
+	case RC_TEXTURE_CUBE: return "CUBE";
+	case RC_TEXTURE_3D: return "3D";
+	case RC_TEXTURE_RECT: return "RECT";
+	case RC_TEXTURE_2D: return "2D";
+	case RC_TEXTURE_1D: return "1D";
+	default: return "BAD_TEXTURE_TARGET";
+	}
+}
+
+static void rc_print_comparefunc(FILE * f, const char * lhs, rc_compare_func func, const char * rhs)
+{
+	if (func == RC_COMPARE_FUNC_NEVER) {
+		fprintf(f, "false");
+	} else if (func == RC_COMPARE_FUNC_ALWAYS) {
+		fprintf(f, "true");
+	} else {
+		const char * op;
+		switch(func) {
+		case RC_COMPARE_FUNC_LESS: op = "<"; break;
+		case RC_COMPARE_FUNC_EQUAL: op = "=="; break;
+		case RC_COMPARE_FUNC_LEQUAL: op = "<="; break;
+		case RC_COMPARE_FUNC_GREATER: op = ">"; break;
+		case RC_COMPARE_FUNC_NOTEQUAL: op = "!="; break;
+		case RC_COMPARE_FUNC_GEQUAL: op = ">="; break;
+		default: op = "???"; break;
+		}
+		fprintf(f, "%s %s %s", lhs, op, rhs);
+	}
+}
+
+static void rc_print_register(FILE * f, rc_register_file file, int index, unsigned int reladdr)
+{
+	if (file == RC_FILE_NONE) {
+		fprintf(f, "none");
+	} else if (file == RC_FILE_SPECIAL) {
+		switch(index) {
+		case RC_SPECIAL_ALU_RESULT: fprintf(f, "aluresult"); break;
+		default: fprintf(f, "special[%i]", index); break;
+		}
+	} else {
+		const char * filename;
+		switch(file) {
+		case RC_FILE_TEMPORARY: filename = "temp"; break;
+		case RC_FILE_INPUT: filename = "input"; break;
+		case RC_FILE_OUTPUT: filename = "output"; break;
+		case RC_FILE_ADDRESS: filename = "addr"; break;
+		case RC_FILE_CONSTANT: filename = "const"; break;
+		default: filename = "BAD FILE"; break;
+		}
+		fprintf(f, "%s[%i%s]", filename, index, reladdr ? " + addr[0]" : "");
+	}
+}
+
+static void rc_print_mask(FILE * f, unsigned int mask)
+{
+	if (mask & RC_MASK_X) fprintf(f, "x");
+	if (mask & RC_MASK_Y) fprintf(f, "y");
+	if (mask & RC_MASK_Z) fprintf(f, "z");
+	if (mask & RC_MASK_W) fprintf(f, "w");
+}
+
+static void rc_print_dst_register(FILE * f, struct rc_dst_register dst)
+{
+	rc_print_register(f, dst.File, dst.Index, dst.RelAddr);
+	if (dst.WriteMask != RC_MASK_XYZW) {
+		fprintf(f, ".");
+		rc_print_mask(f, dst.WriteMask);
+	}
+}
+
+static char rc_swizzle_char(unsigned int swz)
+{
+	switch(swz) {
+	case RC_SWIZZLE_X: return 'x';
+	case RC_SWIZZLE_Y: return 'y';
+	case RC_SWIZZLE_Z: return 'z';
+	case RC_SWIZZLE_W: return 'w';
+	case RC_SWIZZLE_ZERO: return '0';
+	case RC_SWIZZLE_ONE: return '1';
+	case RC_SWIZZLE_HALF: return 'H';
+	case RC_SWIZZLE_UNUSED: return '_';
+	}
+	return '?';
+}
+
+static void rc_print_swizzle(FILE * f, unsigned int swizzle, unsigned int negate)
+{
+	unsigned int comp;
+	for(comp = 0; comp < 4; ++comp) {
+		rc_swizzle swz = GET_SWZ(swizzle, comp);
+		if (GET_BIT(negate, comp))
+			fprintf(f, "-");
+		fprintf(f, "%c", rc_swizzle_char(swz));
+	}
+}
+
+static void rc_print_src_register(FILE * f, struct rc_src_register src)
+{
+	int trivial_negate = (src.Negate == RC_MASK_NONE || src.Negate == RC_MASK_XYZW);
+
+	if (src.Negate == RC_MASK_XYZW)
+		fprintf(f, "-");
+	if (src.Abs)
+		fprintf(f, "|");
+
+	rc_print_register(f, src.File, src.Index, src.RelAddr);
+
+	if (src.Abs && !trivial_negate)
+		fprintf(f, "|");
+
+	if (src.Swizzle != RC_SWIZZLE_XYZW || !trivial_negate) {
+		fprintf(f, ".");
+		rc_print_swizzle(f, src.Swizzle, trivial_negate ? 0 : src.Negate);
+	}
+
+	if (src.Abs && trivial_negate)
+		fprintf(f, "|");
+}
+
+static void rc_print_normal_instruction(FILE * f, struct rc_instruction * inst)
+{
+	const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
+	unsigned int reg;
+
+	fprintf(f, "%s", opcode->Name);
+
+	switch(inst->U.I.SaturateMode) {
+	case RC_SATURATE_NONE: break;
+	case RC_SATURATE_ZERO_ONE: fprintf(f, "_SAT"); break;
+	case RC_SATURATE_MINUS_PLUS_ONE: fprintf(f, "_SAT2"); break;
+	default: fprintf(f, "_BAD_SAT"); break;
+	}
+
+	if (opcode->HasDstReg) {
+		fprintf(f, " ");
+		rc_print_dst_register(f, inst->U.I.DstReg);
+		if (opcode->NumSrcRegs)
+			fprintf(f, ",");
+	}
+
+	for(reg = 0; reg < opcode->NumSrcRegs; ++reg) {
+		if (reg > 0)
+			fprintf(f, ",");
+		fprintf(f, " ");
+		rc_print_src_register(f, inst->U.I.SrcReg[reg]);
+	}
+
+	if (opcode->HasTexture) {
+		fprintf(f, ", %s%s[%u]",
+			textarget_to_string(inst->U.I.TexSrcTarget),
+			inst->U.I.TexShadow ? "SHADOW" : "",
+			inst->U.I.TexSrcUnit);
+	}
+
+	fprintf(f, ";");
+
+	if (inst->U.I.WriteALUResult) {
+		fprintf(f, " [aluresult = (");
+		rc_print_comparefunc(f,
+			(inst->U.I.WriteALUResult == RC_ALURESULT_X) ? "x" : "w",
+			inst->U.I.ALUResultCompare, "0");
+		fprintf(f, ")]");
+	}
+
+	fprintf(f, "\n");
+}
+
+static void rc_print_pair_instruction(FILE * f, struct rc_instruction * fullinst)
+{
+	struct rc_pair_instruction * inst = &fullinst->U.P;
+	int printedsrc = 0;
+
+	for(unsigned int src = 0; src < 3; ++src) {
+		if (inst->RGB.Src[src].Used) {
+			if (printedsrc)
+				fprintf(f, ", ");
+			fprintf(f, "src%i.xyz = ", src);
+			rc_print_register(f, inst->RGB.Src[src].File, inst->RGB.Src[src].Index, 0);
+			printedsrc = 1;
+		}
+		if (inst->Alpha.Src[src].Used) {
+			if (printedsrc)
+				fprintf(f, ", ");
+			fprintf(f, "src%i.w = ", src);
+			rc_print_register(f, inst->Alpha.Src[src].File, inst->Alpha.Src[src].Index, 0);
+			printedsrc = 1;
+		}
+	}
+	fprintf(f, "\n");
+
+	if (inst->RGB.Opcode != RC_OPCODE_NOP) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->RGB.Opcode);
+
+		fprintf(f, "     %s%s", opcode->Name, inst->RGB.Saturate ? "_SAT" : "");
+		if (inst->RGB.WriteMask)
+			fprintf(f, " temp[%i].%s%s%s", inst->RGB.DestIndex,
+				(inst->RGB.WriteMask & 1) ? "x" : "",
+				(inst->RGB.WriteMask & 2) ? "y" : "",
+				(inst->RGB.WriteMask & 4) ? "z" : "");
+		if (inst->RGB.OutputWriteMask)
+			fprintf(f, " color.%s%s%s",
+				(inst->RGB.OutputWriteMask & 1) ? "x" : "",
+				(inst->RGB.OutputWriteMask & 2) ? "y" : "",
+				(inst->RGB.OutputWriteMask & 4) ? "z" : "");
+		if (inst->WriteALUResult == RC_ALURESULT_X)
+			fprintf(f, " aluresult");
+
+		for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) {
+			const char* abs = inst->RGB.Arg[arg].Abs ? "|" : "";
+			const char* neg = inst->RGB.Arg[arg].Negate ? "-" : "";
+			fprintf(f, ", %s%ssrc%i.%c%c%c%s", neg, abs, inst->RGB.Arg[arg].Source,
+				rc_swizzle_char(GET_SWZ(inst->RGB.Arg[arg].Swizzle, 0)),
+				rc_swizzle_char(GET_SWZ(inst->RGB.Arg[arg].Swizzle, 1)),
+				rc_swizzle_char(GET_SWZ(inst->RGB.Arg[arg].Swizzle, 2)),
+				abs);
+		}
+		fprintf(f, "\n");
+	}
+
+	if (inst->Alpha.Opcode != RC_OPCODE_NOP) {
+		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Alpha.Opcode);
+
+		fprintf(f, "     %s%s", opcode->Name, inst->Alpha.Saturate ? "_SAT" : "");
+		if (inst->Alpha.WriteMask)
+			fprintf(f, " temp[%i].w", inst->Alpha.DestIndex);
+		if (inst->Alpha.OutputWriteMask)
+			fprintf(f, " color.w");
+		if (inst->Alpha.DepthWriteMask)
+			fprintf(f, " depth.w");
+		if (inst->WriteALUResult == RC_ALURESULT_W)
+			fprintf(f, " aluresult");
+
+		for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) {
+			const char* abs = inst->Alpha.Arg[arg].Abs ? "|" : "";
+			const char* neg = inst->Alpha.Arg[arg].Negate ? "-" : "";
+			fprintf(f, ", %s%ssrc%i.%c%s", neg, abs, inst->Alpha.Arg[arg].Source,
+				rc_swizzle_char(inst->Alpha.Arg[arg].Swizzle), abs);
+		}
+		fprintf(f, "\n");
+	}
+
+	if (inst->WriteALUResult) {
+		fprintf(f, "      [aluresult = (");
+		rc_print_comparefunc(f, "result", inst->ALUResultCompare, "0");
+		fprintf(f, ")]\n");
+	}
+}
+
+/**
+ * Print program to stderr, default options.
+ */
+void rc_print_program(const struct rc_program *prog)
+{
+	unsigned int linenum = 0;
+	struct rc_instruction *inst;
+
+	fprintf(stderr, "# Radeon Compiler Program\n");
+
+	for(inst = prog->Instructions.Next; inst != &prog->Instructions; inst = inst->Next) {
+		fprintf(stderr, "%3d: ", linenum);
+
+		if (inst->Type == RC_INSTRUCTION_PAIR)
+			rc_print_pair_instruction(stderr, inst);
+		else
+			rc_print_normal_instruction(stderr, inst);
+
+		linenum++;
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_swizzle.h b/src/mesa/drivers/dri/r300/compiler/radeon_swizzle.h
new file mode 100644
index 0000000000..c81d5f7a5e
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_swizzle.h
@@ -0,0 +1,57 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef RADEON_SWIZZLE_H
+#define RADEON_SWIZZLE_H
+
+#include "radeon_program.h"
+
+struct rc_swizzle_split {
+	unsigned char NumPhases;
+	unsigned char Phase[4];
+};
+
+/**
+ * Describe the swizzling capability of target hardware.
+ */
+struct rc_swizzle_caps {
+	/**
+	 * Check whether the given swizzle, absolute and negate combination
+	 * can be implemented natively by the hardware for this opcode.
+	 *
+	 * \return 1 if the swizzle is native for the given opcode
+	 */
+	int (*IsNative)(rc_opcode opcode, struct rc_src_register reg);
+
+	/**
+	 * Determine how to split access to the masked channels of the
+	 * given source register to obtain ALU-native swizzles.
+	 */
+	void (*Split)(struct rc_src_register reg, unsigned int mask, struct rc_swizzle_split * split);
+};
+
+#endif /* RADEON_SWIZZLE_H */
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
index 0fe32a5443..da5b7ba642 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
@@ -684,11 +684,7 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	r300->hw.rb3d_dither_ctl.cmd[0] = cmdpacket0(r300->radeon.radeonScreen, R300_RB3D_DITHER_CTL, 9);
 	ALLOC_STATE(rb3d_aaresolve_ctl, always, 2, 0);
 	r300->hw.rb3d_aaresolve_ctl.cmd[0] = cmdpacket0(r300->radeon.radeonScreen, R300_RB3D_AARESOLVE_CTL, 1);
-	if ((r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) ||
-	      ( !r300->radeon.radeonScreen->kernel_mm && (
-	    (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS400) ||
-	    (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV410) ||
-	    (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R420) ) ) ) {
+	if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV350) {
 		ALLOC_STATE(rb3d_discard_src_pixel_lte_threshold, always, 3, 0);
 	} else {
 		ALLOC_STATE(rb3d_discard_src_pixel_lte_threshold, never, 3, 0);
@@ -697,6 +693,14 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	ALLOC_STATE(zs, always, R300_ZS_CMDSIZE, 0);
 	r300->hw.zs.cmd[R300_ZS_CMD_0] =
 	    cmdpacket0(r300->radeon.radeonScreen, R300_ZB_CNTL, 3);
+	if (is_r500) {
+		if (r300->radeon.radeonScreen->kernel_mm)
+			ALLOC_STATE(zsb, always, R300_ZSB_CMDSIZE, 0);
+		else
+			ALLOC_STATE(zsb, never, R300_ZSB_CMDSIZE, 0);
+		r300->hw.zsb.cmd[R300_ZSB_CMD_0] =
+			cmdpacket0(r300->radeon.radeonScreen, R500_ZB_STENCILREFMASK_BF, 1);
+	}
 
 	ALLOC_STATE(zstencil_format, always, 5, 0);
 	r300->hw.zstencil_format.cmd[0] =
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 9df3897e65..2c2b16aa98 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -84,6 +84,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define need_GL_EXT_framebuffer_object
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_gpu_program_parameters
+#define need_GL_EXT_provoking_vertex
 #define need_GL_EXT_secondary_color
 #define need_GL_EXT_stencil_two_side
 #define need_GL_ATI_separate_stencil
@@ -116,6 +117,7 @@ const struct dri_extension card_extensions[] = {
   {"GL_EXT_packed_depth_stencil",	NULL},
   {"GL_EXT_fog_coord",			GL_EXT_fog_coord_functions },
   {"GL_EXT_gpu_program_parameters",     GL_EXT_gpu_program_parameters_functions},
+  {"GL_EXT_provoking_vertex",           GL_EXT_provoking_vertex_functions },
   {"GL_EXT_secondary_color", 		GL_EXT_secondary_color_functions},
   {"GL_EXT_shadow_funcs",		NULL},
   {"GL_EXT_stencil_two_side",		GL_EXT_stencil_two_side_functions},
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 1dadcc0a69..518d5cdbf4 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -234,6 +234,10 @@ typedef struct r300_context *r300ContextPtr;
 #define R300_ZS_CNTL_2		3
 #define R300_ZS_CMDSIZE		4
 
+#define R300_ZSB_CMD_0		0
+#define R300_ZSB_CNTL_0		1
+#define R300_ZSB_CMDSIZE	2
+
 #define R300_ZB_CMD_0		0
 #define R300_ZB_OFFSET		1
 #define R300_ZB_PITCH		2
@@ -343,6 +347,7 @@ struct r300_hw_state {
 	struct radeon_state_atom rb3d_aaresolve_ctl;	/* (4E88) */
 	struct radeon_state_atom rb3d_discard_src_pixel_lte_threshold;	/* (4E88) I saw it only written on RV350 hardware..  */
 	struct radeon_state_atom zs;	/* zstencil control (4F00) */
+	struct radeon_state_atom zsb;	/* zstencil bf */
 	struct radeon_state_atom zstencil_format;
 	struct radeon_state_atom zb;	/* z buffer (4F20) */
 	struct radeon_state_atom zb_depthclearvalue;	/* (4F28) */
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_common.c b/src/mesa/drivers/dri/r300/r300_fragprog_common.c
index 70c9252894..267ee81a7a 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog_common.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_common.c
@@ -44,6 +44,7 @@
 
 #include "compiler/radeon_compiler.h"
 
+#include "radeon_mesa_to_rc.h"
 #include "r300_state.h"
 
 
@@ -131,7 +132,7 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler,
  */
 static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r300_fragment_program * fp)
 {
-	struct prog_src_register src;
+	struct rc_src_register src;
 	int i;
 
 	fp->fog_attr = FRAG_ATTRIB_MAX;
@@ -155,7 +156,7 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r
 	}
 
 	memset(&src, 0, sizeof(src));
-	src.File = PROGRAM_INPUT;
+	src.File = RC_FILE_INPUT;
 	src.Index = fp->fog_attr;
 	src.Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ONE);
 	rc_move_input(&compiler->Base, FRAG_ATTRIB_FOGC, src);
@@ -232,7 +233,7 @@ static void translate_fragment_program(GLcontext *ctx, struct r300_fragment_prog
 		fflush(stderr);
 	}
 
-	rc_mesa_to_rc_program(&compiler.Base, &cont->Base.Base);
+	radeon_mesa_to_rc_program(&compiler.Base, &cont->Base.Base);
 
 	insert_WPOS_trailer(&compiler, fp);
 
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index 39b4b61a10..623da60333 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -1022,15 +1022,13 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 	R300_GA_COLOR_CONTROL_RGB0_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA0_SHADING_GOURAUD | \
 	R300_GA_COLOR_CONTROL_RGB1_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA1_SHADING_GOURAUD | \
 	R300_GA_COLOR_CONTROL_RGB2_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA2_SHADING_GOURAUD | \
-	R300_GA_COLOR_CONTROL_RGB3_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD | \
-	R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST )
+	R300_GA_COLOR_CONTROL_RGB3_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD)
 /** TODO: might be candidate for removal, the GOURAUD stuff also looks buggy to me */
 #	define R300_RE_SHADE_MODEL_FLAT     ( \
 	R300_GA_COLOR_CONTROL_RGB0_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA0_SHADING_FLAT | \
 	R300_GA_COLOR_CONTROL_RGB1_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA1_SHADING_GOURAUD | \
 	R300_GA_COLOR_CONTROL_RGB2_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA2_SHADING_FLAT | \
-	R300_GA_COLOR_CONTROL_RGB3_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD | \
-	R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST )
+	R300_GA_COLOR_CONTROL_RGB3_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD)
 
 /* Specifies red & green components of fill color -- S312 format -- Backwards comp. */
 #define R300_GA_SOLID_RG                         0x427c
@@ -2315,6 +2313,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_Z_WRITE_ENABLE		 (1 << 2)
 #	define R300_Z_SIGNED_COMPARE		 (1 << 3)
 #	define R300_STENCIL_FRONT_BACK		 (1 << 4)
+#	define R400_ZSIGNED_MAGNITUDE		 (1 << 5)
+#	define R500_STENCIL_REFMASK_FRONT_BACK	 (1 << 6)
 
 #define R300_ZB_ZSTENCILCNTL                   0x4f04
 	/* functions */
@@ -3002,6 +3002,8 @@ enum {
 #   define R500_INST_RGB_CLAMP				(1 << 19)
 #   define R500_INST_ALPHA_CLAMP			(1 << 20)
 #   define R500_INST_ALU_RESULT_SEL			(1 << 21)
+#   define R500_INST_ALU_RESULT_SEL_RED		(0 << 21)
+#   define R500_INST_ALU_RESULT_SEL_ALPHA		(1 << 21)
 #   define R500_INST_ALPHA_PRED_INV			(1 << 22)
 #   define R500_INST_ALU_RESULT_OP_EQ			(0 << 23)
 #   define R500_INST_ALU_RESULT_OP_LT			(1 << 23)
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index 9301543d38..3060f49aaf 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -590,7 +590,9 @@ static void r300SetDepthState(GLcontext * ctx)
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
 	R300_STATECHANGE(r300, zs);
-	r300->hw.zs.cmd[R300_ZS_CNTL_0] &= R300_STENCIL_ENABLE|R300_STENCIL_FRONT_BACK;
+	r300->hw.zs.cmd[R300_ZS_CNTL_0] &= (R300_STENCIL_ENABLE |
+					    R300_STENCIL_FRONT_BACK |
+					    R500_STENCIL_REFMASK_FRONT_BACK);
 	r300->hw.zs.cmd[R300_ZS_CNTL_1] &= ~(R300_ZS_MASK << R300_Z_FUNC_SHIFT);
 
 	if (ctx->Depth.Test) {
@@ -604,11 +606,16 @@ static void r300SetDepthState(GLcontext * ctx)
 
 static void r300CatchStencilFallback(GLcontext *ctx)
 {
+	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	const unsigned back = ctx->Stencil._BackFace;
 
-	if (ctx->Stencil._Enabled && (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[back]
-		|| ctx->Stencil.ValueMask[0] != ctx->Stencil.ValueMask[back]
-		|| ctx->Stencil.WriteMask[0] != ctx->Stencil.WriteMask[back])) {
+	if (rmesa->radeon.radeonScreen->kernel_mm &&
+	    (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)) {
+		r300SwitchFallback(ctx, R300_FALLBACK_STENCIL_TWOSIDE, GL_FALSE);
+	} else if (ctx->Stencil._Enabled &&
+		   (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[back]
+		    || ctx->Stencil.ValueMask[0] != ctx->Stencil.ValueMask[back]
+		    || ctx->Stencil.WriteMask[0] != ctx->Stencil.WriteMask[back])) {
 		r300SwitchFallback(ctx, R300_FALLBACK_STENCIL_TWOSIDE, GL_TRUE);
 	} else {
 		r300SwitchFallback(ctx, R300_FALLBACK_STENCIL_TWOSIDE, GL_FALSE);
@@ -915,11 +922,24 @@ static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face,
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 	    (flag << R300_S_BACK_FUNC_SHIFT);
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_2] |= refmask;
+
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+		rmesa->hw.zs.cmd[R300_ZS_CNTL_0] |= R500_STENCIL_REFMASK_FRONT_BACK;
+		R300_STATECHANGE(rmesa, zsb);
+		refmask = ((ctx->Stencil.Ref[back] & 0xff) << R300_STENCILREF_SHIFT)
+			| ((ctx->Stencil.ValueMask[back] & 0xff) << R300_STENCILMASK_SHIFT);
+
+		rmesa->hw.zsb.cmd[R300_ZSB_CNTL_0] &=
+			~((R300_STENCILREF_MASK << R300_STENCILREF_SHIFT) |
+			  (R300_STENCILREF_MASK << R300_STENCILMASK_SHIFT));
+		rmesa->hw.zsb.cmd[R300_ZSB_CNTL_0] |= refmask;
+	}
 }
 
 static void r300StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
+	const unsigned back = ctx->Stencil._BackFace;
 
 	r300CatchStencilFallback(ctx);
 
@@ -931,6 +951,13 @@ static void r300StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 	    (ctx->Stencil.
 	     WriteMask[0] & R300_STENCILREF_MASK) <<
 	     R300_STENCILWRITEMASK_SHIFT;
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+		R300_STATECHANGE(rmesa, zsb);
+		rmesa->hw.zsb.cmd[R300_ZSB_CNTL_0] |=
+			(ctx->Stencil.
+			 WriteMask[back] & R300_STENCILREF_MASK) <<
+			R300_STENCILWRITEMASK_SHIFT;
+	}
 }
 
 static void r300StencilOpSeparate(GLcontext * ctx, GLenum face,
@@ -2253,6 +2280,14 @@ static void r300InvalidateState(GLcontext * ctx, GLuint new_state)
 		R300_STATECHANGE(r300, zb);
 	}
 
+	if (new_state & (_NEW_LIGHT)) {
+		R300_STATECHANGE(r300, shade2);
+		if (ctx->Light.ProvokingVertex == GL_LAST_VERTEX_CONVENTION)
+			r300->hw.shade2.cmd[1] |= R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST;
+		else
+			r300->hw.shade2.cmd[1] &= ~R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST;
+	}
+
 	r300->radeon.NewGLState |= new_state;
 }
 
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index ed8b788108..c2f96af2c1 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -41,7 +41,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/tnl.h"
 
 #include "compiler/radeon_compiler.h"
-#include "compiler/radeon_nqssadce.h"
+#include "radeon_mesa_to_rc.h"
 #include "r300_context.h"
 #include "r300_fragprog_common.h"
 #include "r300_state.h"
@@ -211,20 +211,20 @@ static void initialize_NV_registers(struct radeon_compiler * compiler)
 
 	for(reg = 0; reg < 12; ++reg) {
 		inst = rc_insert_new_instruction(compiler, &compiler->Program.Instructions);
-		inst->I.Opcode = OPCODE_MOV;
-		inst->I.DstReg.File = PROGRAM_TEMPORARY;
-		inst->I.DstReg.Index = reg;
-		inst->I.SrcReg[0].File = PROGRAM_BUILTIN;
-		inst->I.SrcReg[0].Swizzle = SWIZZLE_0000;
+		inst->U.I.Opcode = RC_OPCODE_MOV;
+		inst->U.I.DstReg.File = RC_FILE_TEMPORARY;
+		inst->U.I.DstReg.Index = reg;
+		inst->U.I.SrcReg[0].File = RC_FILE_NONE;
+		inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_0000;
 	}
 
 	inst = rc_insert_new_instruction(compiler, &compiler->Program.Instructions);
-	inst->I.Opcode = OPCODE_ARL;
-	inst->I.DstReg.File = PROGRAM_ADDRESS;
-	inst->I.DstReg.Index = 0;
-	inst->I.DstReg.WriteMask = WRITEMASK_X;
-	inst->I.SrcReg[0].File = PROGRAM_BUILTIN;
-	inst->I.SrcReg[0].Swizzle = SWIZZLE_0000;
+	inst->U.I.Opcode = RC_OPCODE_ARL;
+	inst->U.I.DstReg.File = RC_FILE_ADDRESS;
+	inst->U.I.DstReg.Index = 0;
+	inst->U.I.DstReg.WriteMask = WRITEMASK_X;
+	inst->U.I.SrcReg[0].File = RC_FILE_NONE;
+	inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_0000;
 }
 
 static struct r300_vertex_program *build_program(GLcontext *ctx,
@@ -255,7 +255,7 @@ static struct r300_vertex_program *build_program(GLcontext *ctx,
 		_mesa_insert_mvp_code(ctx, vp->Base);
 	}
 
-	rc_mesa_to_rc_program(&compiler.Base, &vp->Base->Base);
+	radeon_mesa_to_rc_program(&compiler.Base, &vp->Base->Base);
 
 	if (mesa_vp->IsNVProgram)
 		initialize_NV_registers(&compiler.Base);
diff --git a/src/mesa/drivers/dri/r300/radeon_context.h b/src/mesa/drivers/dri/r300/radeon_context.h
index 250570f6b8..da4812d323 100644
--- a/src/mesa/drivers/dri/r300/radeon_context.h
+++ b/src/mesa/drivers/dri/r300/radeon_context.h
@@ -51,26 +51,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "radeon_screen.h"
 
-#if R200_MERGED
-extern void radeonFallback(GLcontext * ctx, GLuint bit, GLboolean mode);
-
-#define FALLBACK( radeon, bit, mode ) do {			\
-   if ( 0 ) fprintf( stderr, "FALLBACK in %s: #%d=%d\n",	\
-		     __FUNCTION__, bit, mode );			\
-   radeonFallback( (radeon)->glCtx, bit, mode );		\
-} while (0)
-#else
 #define FALLBACK( radeon, bit, mode ) fprintf(stderr, "%s:%s\n", __LINE__, __FILE__);
-#endif
 
 /* TCL fallbacks */
 extern void radeonTclFallback(GLcontext * ctx, GLuint bit, GLboolean mode);
 
-#if R200_MERGED
-#define TCL_FALLBACK( ctx, bit, mode )	radeonTclFallback( ctx, bit, mode )
-#else
 #define TCL_FALLBACK( ctx, bit, mode )	;
-#endif
 
 
 #endif				/* __RADEON_CONTEXT_H__ */
diff --git a/src/mesa/drivers/dri/r300/radeon_mesa_to_rc.c b/src/mesa/drivers/dri/r300/radeon_mesa_to_rc.c
new file mode 100644
index 0000000000..9f9dec840b
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_mesa_to_rc.c
@@ -0,0 +1,223 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_mesa_to_rc.h"
+
+#include "main/mtypes.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+
+#include "compiler/radeon_compiler.h"
+#include "compiler/radeon_program.h"
+
+
+static rc_opcode translate_opcode(gl_inst_opcode opcode)
+{
+	switch(opcode) {
+	case OPCODE_NOP: return RC_OPCODE_NOP;
+	case OPCODE_ABS: return RC_OPCODE_ABS;
+	case OPCODE_ADD: return RC_OPCODE_ADD;
+	case OPCODE_ARL: return RC_OPCODE_ARL;
+	case OPCODE_CMP: return RC_OPCODE_CMP;
+	case OPCODE_COS: return RC_OPCODE_COS;
+	case OPCODE_DDX: return RC_OPCODE_DDX;
+	case OPCODE_DDY: return RC_OPCODE_DDY;
+	case OPCODE_DP3: return RC_OPCODE_DP3;
+	case OPCODE_DP4: return RC_OPCODE_DP4;
+	case OPCODE_DPH: return RC_OPCODE_DPH;
+	case OPCODE_DST: return RC_OPCODE_DST;
+	case OPCODE_EX2: return RC_OPCODE_EX2;
+	case OPCODE_EXP: return RC_OPCODE_EXP;
+	case OPCODE_FLR: return RC_OPCODE_FLR;
+	case OPCODE_FRC: return RC_OPCODE_FRC;
+	case OPCODE_KIL: return RC_OPCODE_KIL;
+	case OPCODE_LG2: return RC_OPCODE_LG2;
+	case OPCODE_LIT: return RC_OPCODE_LIT;
+	case OPCODE_LOG: return RC_OPCODE_LOG;
+	case OPCODE_LRP: return RC_OPCODE_LRP;
+	case OPCODE_MAD: return RC_OPCODE_MAD;
+	case OPCODE_MAX: return RC_OPCODE_MAX;
+	case OPCODE_MIN: return RC_OPCODE_MIN;
+	case OPCODE_MOV: return RC_OPCODE_MOV;
+	case OPCODE_MUL: return RC_OPCODE_MUL;
+	case OPCODE_POW: return RC_OPCODE_POW;
+	case OPCODE_RCP: return RC_OPCODE_RCP;
+	case OPCODE_RSQ: return RC_OPCODE_RSQ;
+	case OPCODE_SCS: return RC_OPCODE_SCS;
+	case OPCODE_SEQ: return RC_OPCODE_SEQ;
+	case OPCODE_SFL: return RC_OPCODE_SFL;
+	case OPCODE_SGE: return RC_OPCODE_SGE;
+	case OPCODE_SGT: return RC_OPCODE_SGT;
+	case OPCODE_SIN: return RC_OPCODE_SIN;
+	case OPCODE_SLE: return RC_OPCODE_SLE;
+	case OPCODE_SLT: return RC_OPCODE_SLT;
+	case OPCODE_SNE: return RC_OPCODE_SNE;
+	case OPCODE_SUB: return RC_OPCODE_SUB;
+	case OPCODE_SWZ: return RC_OPCODE_SWZ;
+	case OPCODE_TEX: return RC_OPCODE_TEX;
+	case OPCODE_TXB: return RC_OPCODE_TXB;
+	case OPCODE_TXD: return RC_OPCODE_TXD;
+	case OPCODE_TXL: return RC_OPCODE_TXL;
+	case OPCODE_TXP: return RC_OPCODE_TXP;
+	case OPCODE_XPD: return RC_OPCODE_XPD;
+	default: return RC_OPCODE_ILLEGAL_OPCODE;
+	}
+}
+
+static rc_saturate_mode translate_saturate(unsigned int saturate)
+{
+	switch(saturate) {
+	default:
+	case SATURATE_OFF: return RC_SATURATE_NONE;
+	case SATURATE_ZERO_ONE: return RC_SATURATE_ZERO_ONE;
+	}
+}
+
+static rc_register_file translate_register_file(unsigned int file)
+{
+	switch(file) {
+	case PROGRAM_TEMPORARY: return RC_FILE_TEMPORARY;
+	case PROGRAM_INPUT: return RC_FILE_INPUT;
+	case PROGRAM_OUTPUT: return RC_FILE_OUTPUT;
+	case PROGRAM_LOCAL_PARAM:
+	case PROGRAM_ENV_PARAM:
+	case PROGRAM_STATE_VAR:
+	case PROGRAM_NAMED_PARAM:
+	case PROGRAM_CONSTANT:
+	case PROGRAM_UNIFORM: return RC_FILE_CONSTANT;
+	case PROGRAM_ADDRESS: return RC_FILE_ADDRESS;
+	default: return RC_FILE_NONE;
+	}
+}
+
+static void translate_srcreg(struct rc_src_register * dest, struct prog_src_register * src)
+{
+	dest->File = translate_register_file(src->File);
+	dest->Index = src->Index;
+	dest->RelAddr = src->RelAddr;
+	dest->Swizzle = src->Swizzle;
+	dest->Abs = src->Abs;
+	dest->Negate = src->Negate;
+}
+
+static void translate_dstreg(struct rc_dst_register * dest, struct prog_dst_register * src)
+{
+	dest->File = translate_register_file(src->File);
+	dest->Index = src->Index;
+	dest->RelAddr = src->RelAddr;
+	dest->WriteMask = src->WriteMask;
+}
+
+static rc_texture_target translate_tex_target(gl_texture_index target)
+{
+	switch(target) {
+	case TEXTURE_2D_ARRAY_INDEX: return RC_TEXTURE_2D_ARRAY;
+	case TEXTURE_1D_ARRAY_INDEX: return RC_TEXTURE_1D_ARRAY;
+	case TEXTURE_CUBE_INDEX: return RC_TEXTURE_CUBE;
+	case TEXTURE_3D_INDEX: return RC_TEXTURE_3D;
+	case TEXTURE_RECT_INDEX: return RC_TEXTURE_RECT;
+	default:
+	case TEXTURE_2D_INDEX: return RC_TEXTURE_2D;
+	case TEXTURE_1D_INDEX: return RC_TEXTURE_1D;
+	}
+}
+
+static void translate_instruction(struct radeon_compiler * c,
+		struct rc_instruction * dest, struct prog_instruction * src)
+{
+	const struct rc_opcode_info * opcode;
+	unsigned int i;
+
+	dest->U.I.Opcode = translate_opcode(src->Opcode);
+	if (dest->U.I.Opcode == RC_OPCODE_ILLEGAL_OPCODE) {
+		rc_error(c, "Unsupported opcode %i\n", src->Opcode);
+		return;
+	}
+	dest->U.I.SaturateMode = translate_saturate(src->SaturateMode);
+
+	opcode = rc_get_opcode_info(dest->U.I.Opcode);
+
+	for(i = 0; i < opcode->NumSrcRegs; ++i)
+		translate_srcreg(&dest->U.I.SrcReg[i], &src->SrcReg[i]);
+
+	if (opcode->HasDstReg)
+		translate_dstreg(&dest->U.I.DstReg, &src->DstReg);
+
+	if (opcode->HasTexture) {
+		dest->U.I.TexSrcUnit = src->TexSrcUnit;
+		dest->U.I.TexSrcTarget = translate_tex_target(src->TexSrcTarget);
+		dest->U.I.TexShadow = src->TexShadow;
+	}
+}
+
+void radeon_mesa_to_rc_program(struct radeon_compiler * c, struct gl_program * program)
+{
+	struct prog_instruction *source;
+	unsigned int i;
+
+	for(source = program->Instructions; source->Opcode != OPCODE_END; ++source) {
+		struct rc_instruction * dest = rc_insert_new_instruction(c, c->Program.Instructions.Prev);
+		translate_instruction(c, dest, source);
+	}
+
+	c->Program.ShadowSamplers = program->ShadowSamplers;
+	c->Program.InputsRead = program->InputsRead;
+	c->Program.OutputsWritten = program->OutputsWritten;
+
+	int isNVProgram = 0;
+
+	if (program->Target == GL_VERTEX_PROGRAM_ARB) {
+		struct gl_vertex_program * vp = (struct gl_vertex_program *) program;
+		isNVProgram = vp->IsNVProgram;
+	}
+
+	if (isNVProgram) {
+		/* NV_vertex_program has a fixed-sized constant environment.
+		 * This could be handled more efficiently for programs that
+		 * do not use relative addressing.
+		 */
+		for(i = 0; i < 96; ++i) {
+			struct rc_constant constant;
+
+			constant.Type = RC_CONSTANT_EXTERNAL;
+			constant.Size = 4;
+			constant.u.External = i;
+
+			rc_constants_add(&c->Program.Constants, &constant);
+		}
+	} else {
+		for(i = 0; i < program->Parameters->NumParameters; ++i) {
+			struct rc_constant constant;
+
+			constant.Type = RC_CONSTANT_EXTERNAL;
+			constant.Size = 4;
+			constant.u.External = i;
+
+			rc_constants_add(&c->Program.Constants, &constant);
+		}
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/radeon_mesa_to_rc.h b/src/mesa/drivers/dri/r300/radeon_mesa_to_rc.h
new file mode 100644
index 0000000000..9511a04f36
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_mesa_to_rc.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2009 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef RADEON_MESA_TO_RC_H
+#define RADEON_MESA_TO_RC_H
+
+struct gl_program;
+struct radeon_compiler;
+
+void radeon_mesa_to_rc_program(struct radeon_compiler * c, struct gl_program * program);
+
+#endif /* RADEON_MESA_TO_RC_H */
diff --git a/src/mesa/drivers/dri/r600/Makefile b/src/mesa/drivers/dri/r600/Makefile
index d925a2dfe3..9b7c42042e 100644
--- a/src/mesa/drivers/dri/r600/Makefile
+++ b/src/mesa/drivers/dri/r600/Makefile
@@ -29,6 +29,7 @@ COMMON_SOURCES = \
 RADEON_COMMON_SOURCES = \
 	radeon_bo_legacy.c \
 	radeon_common_context.c \
+	radeon_buffer_objects.c \
 	radeon_common.c \
 	radeon_cs_legacy.c \
 	radeon_dma.c \
@@ -65,8 +66,7 @@ DRIVER_SOURCES = \
 
 C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES)
 
-DRIVER_DEFINES = -DCOMPILE_R600 -DR200_MERGED=0 \
-	-DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R600 \
+DRIVER_DEFINES = -DRADEON_R600 \
 #	-DRADEON_BO_TRACK \
 	-Wall
 
@@ -76,4 +76,3 @@ DRI_LIB_DEPS += $(RADEON_LDFLAGS)
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c
index e0b77d4385..ba0d450cbb 100644
--- a/src/mesa/drivers/dri/r600/r600_context.c
+++ b/src/mesa/drivers/dri/r600/r600_context.c
@@ -59,6 +59,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_debug.h"
 #include "r600_context.h"
 #include "radeon_common_context.h"
+#include "radeon_buffer_objects.h"
 #include "radeon_span.h"
 #include "r600_cmdbuf.h"
 #include "r600_emit.h"
@@ -85,6 +86,7 @@ int hw_tcl_on = 1;
 #define need_GL_EXT_framebuffer_object
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_gpu_program_parameters
+#define need_GL_EXT_provoking_vertex
 #define need_GL_EXT_secondary_color
 #define need_GL_EXT_stencil_two_side
 #define need_GL_ATI_separate_stencil
@@ -117,6 +119,7 @@ const struct dri_extension card_extensions[] = {
   {"GL_EXT_packed_depth_stencil",	NULL},
   {"GL_EXT_fog_coord",			GL_EXT_fog_coord_functions },
   {"GL_EXT_gpu_program_parameters",     GL_EXT_gpu_program_parameters_functions},
+  {"GL_EXT_provoking_vertex",           GL_EXT_provoking_vertex_functions },
   {"GL_EXT_secondary_color", 		GL_EXT_secondary_color_functions},
   {"GL_EXT_shadow_funcs",		NULL},
   {"GL_EXT_stencil_two_side",		GL_EXT_stencil_two_side_functions},
@@ -128,6 +131,8 @@ const struct dri_extension card_extensions[] = {
   {"GL_EXT_texture_lod_bias",		NULL},
   {"GL_EXT_texture_mirror_clamp",	NULL},
   {"GL_EXT_texture_rectangle",		NULL},
+  {"GL_EXT_vertex_array_bgra",          NULL},
+  {"GL_EXT_texture_sRGB",               NULL},
   {"GL_ATI_separate_stencil",		GL_ATI_separate_stencil_functions},
   {"GL_ATI_texture_env_combine3",	NULL},
   {"GL_ATI_texture_mirror_once",	NULL},
@@ -254,6 +259,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 	r600InitTextureFuncs(&functions);
 	r700InitShaderFuncs(&functions);
 	r700InitIoctlFuncs(&functions);
+    radeonInitBufferObjectFuncs(&functions);
 
 	if (!radeonInitContext(&r600->radeon, &functions,
 			       glVisual, driContextPriv,
@@ -372,6 +378,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 		_mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 	}
 
+    r700InitDraw(ctx);
+
 	radeon_fbo_init(&r600->radeon);
    	radeonInitSpanFuncs( ctx );
 
@@ -383,9 +391,6 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 
 	if (driQueryOptionb(&r600->radeon.optionCache, "no_rast")) {
 		radeon_warning("disabling 3D acceleration\n");
-#if R200_MERGED
-		FALLBACK(&r600->radeon, RADEON_FALLBACK_DISABLE, 1);
-#endif
 	}
 
 	return GL_TRUE;
diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h
index 9397ecde81..7f68820fda 100644
--- a/src/mesa/drivers/dri/r600/r600_context.h
+++ b/src/mesa/drivers/dri/r600/r600_context.h
@@ -126,6 +126,32 @@ struct r600_hw_state {
 	struct radeon_state_atom tx_brdr_clr;
 };
 
+typedef struct StreamDesc
+{
+	GLint   size;   //number of data element
+	GLenum  type;  //data element type
+	GLsizei stride;
+
+	struct radeon_bo *bo;
+	GLint  bo_offset;
+
+	GLuint    dwords;
+	GLuint    dst_loc;
+	GLuint    _signed;
+	GLboolean normalize;
+	GLboolean is_named_bo;
+	GLubyte   element;
+} StreamDesc;
+
+typedef struct r700_index_buffer
+{
+	struct radeon_bo *bo;
+	int    bo_offset;
+
+	GLboolean is_32bit;
+	GLuint    count;
+} r700_index_buffer;
+
 /**
  * \brief R600 context structure.
  */
@@ -144,6 +170,9 @@ struct r600_context {
 	GLvector4f dummy_attrib[_TNL_ATTRIB_MAX];
 	GLvector4f *temp_attrib[_TNL_ATTRIB_MAX];
 
+	GLint      nNumActiveAos;
+	StreamDesc stream_desc[VERT_ATTRIB_MAX];
+    struct r700_index_buffer ind_buf;
 };
 
 #define R700_CONTEXT(ctx)		((context_t *)(ctx->DriverCtx))
@@ -177,6 +206,7 @@ extern GLboolean r700SyncSurf(context_t *context,
 extern void r700SetupStreams(GLcontext * ctx);
 extern void r700Start3D(context_t *context);
 extern void r600InitAtoms(context_t *context);
+extern void r700InitDraw(GLcontext *ctx);
 
 #define RADEON_D_CAPTURE 0
 #define RADEON_D_PLAYBACK 1
diff --git a/src/mesa/drivers/dri/r600/r600_texstate.c b/src/mesa/drivers/dri/r600/r600_texstate.c
index a30703e41b..ba1371bf95 100644
--- a/src/mesa/drivers/dri/r600/r600_texstate.c
+++ b/src/mesa/drivers/dri/r600/r600_texstate.c
@@ -531,6 +531,49 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 			return GL_FALSE;
 		}
 		break;
+	/* EXT_texture_sRGB */
+	case MESA_FORMAT_SRGBA8:
+		SETfield(t->SQ_TEX_RESOURCE1, FMT_8_8_8_8,
+			 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
+
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_W,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_X_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_X_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_Z,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Y_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Y_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_Y,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_mask);
+		SETbit(t->SQ_TEX_RESOURCE4, SQ_TEX_RESOURCE_WORD4_0__FORCE_DEGAMMA_bit);
+		break;
+	case MESA_FORMAT_SLA8:
+		SETfield(t->SQ_TEX_RESOURCE1, FMT_8_8,
+			 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
+
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_X_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_X_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Y_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Y_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_Y,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_mask);
+		SETbit(t->SQ_TEX_RESOURCE4, SQ_TEX_RESOURCE_WORD4_0__FORCE_DEGAMMA_bit);
+		break;
+	case MESA_FORMAT_SL8: /* X, X, X, ONE */
+		SETfield(t->SQ_TEX_RESOURCE1, FMT_8,
+			 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
+
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_X_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_X_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Y_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Y_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_X,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_mask);
+		SETfield(t->SQ_TEX_RESOURCE4, SQ_SEL_1,
+			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_mask);
+		SETbit(t->SQ_TEX_RESOURCE4, SQ_TEX_RESOURCE_WORD4_0__FORCE_DEGAMMA_bit);
+		break;
 	default:
 		/* Not supported format */
 		return GL_FALSE;
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c
index 6ff08e1cfb..917318c02a 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.c
+++ b/src/mesa/drivers/dri/r600/r700_assembler.c
@@ -791,6 +791,133 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm,
 	return GL_TRUE;
 }
 
+GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm,
+                                       GLuint              destination_register,								       
+                                       GLenum              type,
+                                       GLint               size,
+                                       GLubyte             element,
+                                       GLuint              _signed,
+                                       GLboolean           normalize,
+                                       VTX_FETCH_METHOD  * pFetchMethod)
+{
+    GLuint client_size_inbyte;
+	GLuint data_format;
+    GLuint mega_fetch_count;
+	GLuint is_mega_fetch_flag;
+
+	R700VertexGenericFetch*   vfetch_instruction_ptr;
+	R700VertexGenericFetch*   assembled_vfetch_instruction_ptr 
+                                     = pAsm->vfetch_instruction_ptr_array[element];
+
+	if (assembled_vfetch_instruction_ptr == NULL) 
+	{
+		vfetch_instruction_ptr = (R700VertexGenericFetch*) CALLOC_STRUCT(R700VertexGenericFetch);
+		if (vfetch_instruction_ptr == NULL) 
+		{
+			return GL_FALSE;
+		}
+        Init_R700VertexGenericFetch(vfetch_instruction_ptr);
+    }
+	else 
+	{
+		vfetch_instruction_ptr = assembled_vfetch_instruction_ptr;
+	}
+
+    data_format = GetSurfaceFormat(type, size, &client_size_inbyte);	
+
+	if(GL_TRUE == pFetchMethod->bEnableMini) //More conditions here
+	{
+		//TODO : mini fetch
+	}
+	else
+	{
+		mega_fetch_count = MEGA_FETCH_BYTES - 1;
+		is_mega_fetch_flag       = 0x1;
+		pFetchMethod->mega_fetch_remainder = MEGA_FETCH_BYTES - client_size_inbyte;
+	}
+
+	vfetch_instruction_ptr->m_Word0.f.vtx_inst         = SQ_VTX_INST_FETCH;
+	vfetch_instruction_ptr->m_Word0.f.fetch_type       = SQ_VTX_FETCH_VERTEX_DATA;
+	vfetch_instruction_ptr->m_Word0.f.fetch_whole_quad = 0x0;
+
+	vfetch_instruction_ptr->m_Word0.f.buffer_id        = element;
+	vfetch_instruction_ptr->m_Word0.f.src_gpr          = 0x0; 
+	vfetch_instruction_ptr->m_Word0.f.src_rel          = SQ_ABSOLUTE;
+	vfetch_instruction_ptr->m_Word0.f.src_sel_x        = SQ_SEL_X;
+	vfetch_instruction_ptr->m_Word0.f.mega_fetch_count = mega_fetch_count;
+
+	vfetch_instruction_ptr->m_Word1.f.dst_sel_x        = (size < 1) ? SQ_SEL_0 : SQ_SEL_X;
+	vfetch_instruction_ptr->m_Word1.f.dst_sel_y        = (size < 2) ? SQ_SEL_0 : SQ_SEL_Y;
+	vfetch_instruction_ptr->m_Word1.f.dst_sel_z        = (size < 3) ? SQ_SEL_0 : SQ_SEL_Z;
+	vfetch_instruction_ptr->m_Word1.f.dst_sel_w        = (size < 4) ? SQ_SEL_1 : SQ_SEL_W;
+
+	vfetch_instruction_ptr->m_Word1.f.use_const_fields = 1;
+    vfetch_instruction_ptr->m_Word1.f.data_format      = data_format;
+    vfetch_instruction_ptr->m_Word2.f.endian_swap      = SQ_ENDIAN_NONE;
+
+    if(1 == _signed)
+    {
+        vfetch_instruction_ptr->m_Word1.f.format_comp_all  = SQ_FORMAT_COMP_SIGNED;
+    }
+    else
+    {
+        vfetch_instruction_ptr->m_Word1.f.format_comp_all  = SQ_FORMAT_COMP_UNSIGNED;
+    }
+
+    if(GL_TRUE == normalize)
+    {
+        vfetch_instruction_ptr->m_Word1.f.num_format_all   = SQ_NUM_FORMAT_NORM;
+    }
+    else
+    {
+        vfetch_instruction_ptr->m_Word1.f.num_format_all   = SQ_NUM_FORMAT_INT;
+    }
+
+	// Destination register
+	vfetch_instruction_ptr->m_Word1_GPR.f.dst_gpr = destination_register; 
+	vfetch_instruction_ptr->m_Word1_GPR.f.dst_rel = SQ_ABSOLUTE;
+
+	vfetch_instruction_ptr->m_Word2.f.offset              = 0;
+	vfetch_instruction_ptr->m_Word2.f.const_buf_no_stride = 0x0;
+
+	vfetch_instruction_ptr->m_Word2.f.mega_fetch          = is_mega_fetch_flag;
+
+	if (assembled_vfetch_instruction_ptr == NULL) 
+	{
+		if ( GL_FALSE == add_vfetch_instruction(pAsm, (R700VertexInstruction *)vfetch_instruction_ptr) ) 
+        {   
+			return GL_FALSE;
+		}
+
+		if (pAsm->vfetch_instruction_ptr_array[element] != NULL) 
+		{
+			return GL_FALSE;
+		}
+		else 
+		{
+			pAsm->vfetch_instruction_ptr_array[element] = vfetch_instruction_ptr;
+		}
+	}
+
+	return GL_TRUE;
+}
+
+GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm)
+{
+    GLint i;
+    pAsm->cf_current_clause_type    = CF_EMPTY_CLAUSE;
+    pAsm->cf_current_vtx_clause_ptr = NULL;
+
+    for (i=0; i<VERT_ATTRIB_MAX; i++) 
+	{
+		pAsm->vfetch_instruction_ptr_array[ i ] = NULL;
+	}
+
+    cleanup_vfetch_shaderinst(pAsm->pR700Shader);
+    
+    return GL_TRUE;
+}
+
 GLuint gethelpr(r700_AssemblerBase* pAsm) 
 {
     GLuint r = pAsm->uHelpReg;
@@ -1963,9 +2090,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
     GLuint contiguous_slots_needed;
 
     GLuint    uNumSrc = r700GetNumOperands(pAsm);
-    GLuint    channel_swizzle, j;
-    GLuint    chan_counter[4] = {0, 0, 0, 0};
-    PVSSRC *  pSource[3];
+    //GLuint    channel_swizzle, j;
+    //GLuint    chan_counter[4] = {0, 0, 0, 0};
+    //PVSSRC *  pSource[3];
     GLboolean bSplitInst = GL_FALSE;
 
     if (1 == pAsm->D.dst.math) 
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h
index d639592702..8cbca066e9 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.h
+++ b/src/mesa/drivers/dri/r600/r700_assembler.h
@@ -415,6 +415,15 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm,
 								GLuint number_of_elements,
                                 GLenum dataElementType,
 								VTX_FETCH_METHOD* pFetchMethod);
+GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm,
+                                       GLuint              destination_register,								       
+                                       GLenum              type,
+                                       GLint               size,
+                                       GLubyte             element,
+                                       GLuint              _signed,
+                                       GLboolean           normalize,
+                                       VTX_FETCH_METHOD  * pFetchMethod);
+GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm);
 GLuint gethelpr(r700_AssemblerBase* pAsm);
 void resethelpr(r700_AssemblerBase* pAsm);
 void checkop_init(r700_AssemblerBase* pAsm);
diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c
index 06d7e9c9ab..3b7f6fffe0 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.c
+++ b/src/mesa/drivers/dri/r600/r700_chip.c
@@ -208,6 +208,93 @@ static void r700SetupVTXConstants(GLcontext  * ctx,
 
 }
 
+extern int getTypeSize(GLenum type);
+static void r700SetupVTXConstants2(GLcontext  * ctx,
+				                   void *       pAos,
+                                   StreamDesc * pStreamDesc)     
+{
+    context_t *context = R700_CONTEXT(ctx);
+    struct radeon_aos * paos = (struct radeon_aos *)pAos;
+    unsigned int nVBsize;
+    BATCH_LOCALS(&context->radeon);
+
+    unsigned int uSQ_VTX_CONSTANT_WORD0_0;
+    unsigned int uSQ_VTX_CONSTANT_WORD1_0;
+    unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0;
+    unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0;
+    unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0;
+
+    if (!paos->bo)
+	    return;
+
+    if ((context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV610) ||
+	(context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV620) ||
+	(context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS780) ||
+	(context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS880) ||
+	(context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV710))
+	    r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, TC_ACTION_ENA_bit);
+    else
+	    r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, VC_ACTION_ENA_bit);
+
+    if(0 == pStreamDesc->stride)
+    {
+        nVBsize = paos->count * pStreamDesc->size * getTypeSize(pStreamDesc->type);
+    }
+    else
+    {
+        nVBsize = paos->count * pStreamDesc->stride;
+    }
+
+    uSQ_VTX_CONSTANT_WORD0_0 = paos->offset;
+    uSQ_VTX_CONSTANT_WORD1_0 = nVBsize - 1;
+
+    SETfield(uSQ_VTX_CONSTANT_WORD2_0, 0, BASE_ADDRESS_HI_shift, BASE_ADDRESS_HI_mask); /* TODO */
+    SETfield(uSQ_VTX_CONSTANT_WORD2_0, pStreamDesc->stride, SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift,
+	     SQ_VTX_CONSTANT_WORD2_0__STRIDE_mask);
+    SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(pStreamDesc->type, pStreamDesc->size, NULL),
+	     SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift,
+	     SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_mask); /* TODO : trace back api for initial data type, not only GL_FLOAT */
+    
+    if(GL_TRUE == pStreamDesc->normalize)
+    {
+        SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_NORM,
+	             SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
+    }
+    //else
+    //{
+    //    SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_INT,
+	//             SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
+    //}
+
+    if(1 == pStreamDesc->_signed)
+    {
+        SETbit(uSQ_VTX_CONSTANT_WORD2_0, SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit);
+    }
+
+    SETfield(uSQ_VTX_CONSTANT_WORD3_0, 1, MEM_REQUEST_SIZE_shift, MEM_REQUEST_SIZE_mask);
+    SETfield(uSQ_VTX_CONSTANT_WORD6_0, SQ_TEX_VTX_VALID_BUFFER,
+	     SQ_TEX_RESOURCE_WORD6_0__TYPE_shift, SQ_TEX_RESOURCE_WORD6_0__TYPE_mask);
+
+    BEGIN_BATCH_NO_AUTOSTATE(9 + 2);
+
+    R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7));
+    R600_OUT_BATCH((pStreamDesc->element + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE);
+    R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD0_0);
+    R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0);
+    R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0);
+    R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0);
+    R600_OUT_BATCH(0);
+    R600_OUT_BATCH(0);
+    R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD6_0);
+    R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD0_0,
+                         paos->bo,
+                         uSQ_VTX_CONSTANT_WORD0_0,
+                         RADEON_GEM_DOMAIN_GTT, 0, 0);
+    END_BATCH();
+    COMMIT_BATCH();
+
+}
+
 void r700SetupStreams(GLcontext *ctx)
 {
     context_t         *context = R700_CONTEXT(ctx);
@@ -256,14 +343,24 @@ static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom)
     COMMIT_BATCH();
 
     for(i=0; i<VERT_ATTRIB_MAX; i++) {
-	    if(vp->mesa_program->Base.InputsRead & (1 << i)) {
-		    /* currently aos are packed */
-		    r700SetupVTXConstants(ctx,
-					  i,
-					  (void*)(&context->radeon.tcl.aos[j]),
-					  (unsigned int)context->radeon.tcl.aos[j].components,
-					  (unsigned int)context->radeon.tcl.aos[j].stride * 4,
-					  (unsigned int)context->radeon.tcl.aos[j].count);
+	    if(vp->mesa_program->Base.InputsRead & (1 << i)) 
+        {
+		    if(1 == context->selected_vp->uiVersion)
+            {
+		        /* currently aos are packed */
+		        r700SetupVTXConstants(ctx,
+					      i,
+					      (void*)(&context->radeon.tcl.aos[j]),
+					      (unsigned int)context->radeon.tcl.aos[j].components,
+					      (unsigned int)context->radeon.tcl.aos[j].stride * 4,
+					      (unsigned int)context->radeon.tcl.aos[j].count);
+            }
+            else
+            {   /* context->selected_vp->uiVersion == 2 : aos not always packed */                
+                r700SetupVTXConstants2(ctx,					    
+					      (void*)(&context->radeon.tcl.aos[j]),
+					      &(context->stream_desc[j]));
+            }
 		    j++;
 	    }
     }
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index b1c3648ca5..4f39d9f1bd 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -43,6 +43,7 @@
 #include "tnl/t_context.h"
 #include "tnl/t_vertex.h"
 #include "tnl/t_pipeline.h"
+#include "vbo/vbo_context.h"
 
 #include "r600_context.h"
 #include "r600_cmdbuf.h"
@@ -53,6 +54,7 @@
 #include "r700_fragprog.h"
 #include "r700_state.h"
 
+#include "radeon_buffer_objects.h"
 #include "radeon_common_context.h"
 
 void r700WaitForIdle(context_t *context);
@@ -249,78 +251,134 @@ static int r700NumVerts(int num_verts, int prim)
 
 static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim)
 {
-	context_t *context = R700_CONTEXT(ctx);
-	BATCH_LOCALS(&context->radeon);
-	int type, i, total_emit;
-	int num_indices;
-	uint32_t vgt_draw_initiator = 0;
-	uint32_t vgt_index_type     = 0;
-	uint32_t vgt_primitive_type = 0;
-	uint32_t vgt_num_indices    = 0;
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *vb = &tnl->vb;
-
-	type = r700PrimitiveType(prim);
-	num_indices = r700NumVerts(end - start, prim);
-
-	radeon_print(RADEON_RENDER, RADEON_TRACE,
-		"%s type %x num_indices %d\n",
-		__func__, type, num_indices);
-
-	if (type < 0 || num_indices <= 0)
-		return;
+    context_t *context = R700_CONTEXT(ctx);
+    BATCH_LOCALS(&context->radeon);
+    int type, i, total_emit;
+    int num_indices;
+    uint32_t vgt_draw_initiator = 0;
+    uint32_t vgt_index_type     = 0;
+    uint32_t vgt_primitive_type = 0;
+    uint32_t vgt_num_indices    = 0;
+    TNLcontext *tnl = TNL_CONTEXT(ctx);
+    struct vertex_buffer *vb = &tnl->vb;
+    GLboolean bUseDrawIndex;
+
+    if(NULL != context->ind_buf.bo)
+    {
+        bUseDrawIndex = GL_TRUE;
+    }
+    else
+    {
+        bUseDrawIndex = GL_FALSE;
+    }
+
+    type = r700PrimitiveType(prim);
+    num_indices = r700NumVerts(end - start, prim);
+
+    radeon_print(RADEON_RENDER, RADEON_TRACE,
+		 "%s type %x num_indices %d\n",
+		 __func__, type, num_indices);
+
+    if (type < 0 || num_indices <= 0)
+	    return;
 
+    if(GL_TRUE == bUseDrawIndex)
+    {
+        total_emit =   3  /* VGT_PRIMITIVE_TYPE */
+		     + 2  /* VGT_INDEX_TYPE */
+		     + 2  /* NUM_INSTANCES */
+                     + 5 + 2; /* DRAW_INDEX */
+    }
+    else
+    {
         total_emit =   3 /* VGT_PRIMITIVE_TYPE */
 		     + 2 /* VGT_INDEX_TYPE */
-		     + 2 /* NUM_INSTANCES */
+	             + 2 /* NUM_INSTANCES */
                      + num_indices + 3; /* DRAW_INDEX_IMMD */
+    }
 
-        BEGIN_BATCH_NO_AUTOSTATE(total_emit);
-	// prim
-        SETfield(vgt_primitive_type, type,
-		 VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask);
-        R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1));
-        R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX);
-        R600_OUT_BATCH(vgt_primitive_type);
+    BEGIN_BATCH_NO_AUTOSTATE(total_emit);
+    // prim
+    SETfield(vgt_primitive_type, type,
+	     VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask);
+    R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1));
+    R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX);
+    R600_OUT_BATCH(vgt_primitive_type);
 
 	// index type
-        SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
-        R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0));
-        R600_OUT_BATCH(vgt_index_type);
+    SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
 
-	// num instances
-	R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0));
-        R600_OUT_BATCH(1);
+    if(GL_TRUE == bUseDrawIndex)
+    {
+        if(GL_TRUE != context->ind_buf.is_32bit)
+        {
+            SETfield(vgt_index_type, DI_INDEX_SIZE_16_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
+        }
+    }
+
+    R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0));
+    R600_OUT_BATCH(vgt_index_type);
+
+    // num instances
+    R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0));
+    R600_OUT_BATCH(1);
+
+    // draw packet
+    vgt_num_indices = num_indices;
 
-	// draw packet
-        vgt_num_indices = num_indices;
+    if(GL_TRUE == bUseDrawIndex)
+    {
+        SETfield(vgt_draw_initiator, DI_SRC_SEL_DMA, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
+    }
+    else
+    {
         SETfield(vgt_draw_initiator, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
+    }
+
 	SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask);
 
+    if(GL_TRUE == bUseDrawIndex)
+    {
+        R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX, 3));
+        R600_OUT_BATCH(context->ind_buf.bo_offset);
+        R600_OUT_BATCH(0);
+        R600_OUT_BATCH(vgt_num_indices);
+        R600_OUT_BATCH(vgt_draw_initiator);
+        R600_OUT_BATCH_RELOC(context->ind_buf.bo_offset,
+                             context->ind_buf.bo,
+                             context->ind_buf.bo_offset,
+                             RADEON_GEM_DOMAIN_GTT, 0, 0);
+    }
+    else
+    {
         R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (num_indices + 1)));
         R600_OUT_BATCH(vgt_num_indices);
         R600_OUT_BATCH(vgt_draw_initiator);
 
-        for (i = start; i < (start + num_indices); i++) {
-		if(vb->Elts)
-			R600_OUT_BATCH(vb->Elts[i]);
-		else
-			R600_OUT_BATCH(i);
+        for (i = start; i < (start + num_indices); i++)
+	{
+            if(vb->Elts)
+            {
+                R600_OUT_BATCH(vb->Elts[i]);
+            }
+            else
+	    {
+                R600_OUT_BATCH(i);
+	    }
         }
-        END_BATCH();
-        COMMIT_BATCH();
+    }
 
+    END_BATCH();
+    COMMIT_BATCH();
 }
 
 /* start 3d, idle, cb/db flush */
 #define PRE_EMIT_STATE_BUFSZ 10 + 5 + 14
 
-static GLuint r700PredictRenderSize(GLcontext* ctx)
+static GLuint r700PredictRenderSize(GLcontext* ctx, GLuint nr_prims)
 {
     context_t *context = R700_CONTEXT(ctx);
-    TNLcontext *tnl = TNL_CONTEXT(ctx);
     struct r700_vertex_program *vp = context->selected_vp;
-    struct vertex_buffer *vb = &tnl->vb;
     GLboolean flushed;
     GLuint dwords, i;
     GLuint state_size;
@@ -328,8 +386,15 @@ static GLuint r700PredictRenderSize(GLcontext* ctx)
     context->radeon.tcl.aos_count = _mesa_bitcount(vp->mesa_program->Base.InputsRead);
 
     dwords = PRE_EMIT_STATE_BUFSZ;
-    for (i = 0; i < vb->PrimitiveCount; i++)
-        dwords += vb->Primitive[i].count + 10;
+    if (nr_prims)
+	    dwords += nr_prims * 14;
+    else {
+	    TNLcontext *tnl = TNL_CONTEXT(ctx);
+	    struct vertex_buffer *vb = &tnl->vb;
+
+	    for (i = 0; i < vb->PrimitiveCount; i++)
+		    dwords += vb->Primitive[i].count + 10;
+    }
     state_size = radeonCountStateEmitSize(&context->radeon);
     flushed = rcommonEnsureCmdBufSpace(&context->radeon,
             dwords + state_size, __FUNCTION__);
@@ -369,7 +434,7 @@ static GLboolean r700RunRender(GLcontext * ctx,
     r700SetupFragmentProgram(ctx);
     r600UpdateTextureState(ctx);
 
-    GLuint emit_end = r700PredictRenderSize(ctx) 
+    GLuint emit_end = r700PredictRenderSize(ctx, 0)
         + context->radeon.cmdbuf.cs->cdw;
     r700SetupStreams(ctx);
 
@@ -477,4 +542,544 @@ const struct tnl_pipeline_stage *r700_pipeline[] =
     0,
 };
 
+#define CONVERT( TYPE, MACRO ) do {		\
+	GLuint i, j, sz;				\
+	sz = input->Size;				\
+	if (input->Normalized) {			\
+		for (i = 0; i < count; i++) {		\
+			const TYPE *in = (TYPE *)src_ptr;		\
+			for (j = 0; j < sz; j++) {		\
+				*dst_ptr++ = MACRO(*in);		\
+				in++;				\
+			}					\
+			src_ptr += stride;			\
+		}						\
+	} else {					\
+		for (i = 0; i < count; i++) {		\
+			const TYPE *in = (TYPE *)src_ptr;		\
+			for (j = 0; j < sz; j++) {		\
+				*dst_ptr++ = (GLfloat)(*in);		\
+				in++;				\
+			}					\
+			src_ptr += stride;			\
+		}						\
+	}						\
+} while (0)
+
+/**
+ * Convert attribute data type to float
+ * If the attribute uses named buffer object replace the bo with newly allocated bo
+ */
+static void r700ConvertAttrib(GLcontext *ctx, int count, 
+                              const struct gl_client_array *input, 
+                              struct StreamDesc *attr)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    const GLvoid *src_ptr;
+    GLboolean mapped_named_bo = GL_FALSE;
+    GLfloat *dst_ptr;
+    GLuint stride;
+
+    stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size : input->StrideB;
+
+    /* Convert value for first element only */
+    if (input->StrideB == 0)
+    {
+        count = 1;
+    }
+
+    if (input->BufferObj->Name) 
+    {
+        if (!input->BufferObj->Pointer) 
+        {
+            ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj);
+            mapped_named_bo = GL_TRUE;
+        }
+
+        src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr);
+    } 
+    else 
+    {
+        src_ptr = input->Ptr;
+    }
+
+    radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, 
+                         sizeof(GLfloat) * input->Size * count, 32);
+    dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset);
+
+    assert(src_ptr != NULL);
+
+    switch (input->Type) 
+    {
+        case GL_DOUBLE:
+            CONVERT(GLdouble, (GLfloat));
+            break;
+        case GL_UNSIGNED_INT:
+            CONVERT(GLuint, UINT_TO_FLOAT);
+            break;
+        case GL_INT:
+            CONVERT(GLint, INT_TO_FLOAT);
+            break;
+        case GL_UNSIGNED_SHORT:
+            CONVERT(GLushort, USHORT_TO_FLOAT);
+            break;
+        case GL_SHORT:
+            CONVERT(GLshort, SHORT_TO_FLOAT);
+            break;
+        case GL_UNSIGNED_BYTE:
+            assert(input->Format != GL_BGRA);
+            CONVERT(GLubyte, UBYTE_TO_FLOAT);
+            break;
+        case GL_BYTE:
+            CONVERT(GLbyte, BYTE_TO_FLOAT);
+            break;
+        default:
+            assert(0);
+            break;
+    }
+
+    if (mapped_named_bo) 
+    {
+        ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
+    }
+}
+
+static void r700AlignDataToDword(GLcontext *ctx, 
+                                 const struct gl_client_array *input, 
+                                 int count, 
+                                 struct StreamDesc *attr)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    const int dst_stride = (input->StrideB + 3) & ~3;
+    const int size = getTypeSize(input->Type) * input->Size * count;
+    GLboolean mapped_named_bo = GL_FALSE;
+
+    radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, size, 32);
+
+    if (!input->BufferObj->Pointer) 
+    {
+        ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj);
+        mapped_named_bo = GL_TRUE;
+    }
+
+    {
+        GLvoid *src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr);
+        GLvoid *dst_ptr = ADD_POINTERS(attr->bo->ptr, attr->bo_offset);
+        int i;
+
+        for (i = 0; i < count; ++i) 
+        {
+            _mesa_memcpy(dst_ptr, src_ptr, input->StrideB);
+            src_ptr += input->StrideB;
+            dst_ptr += dst_stride;
+        }
+    }
+
+    if (mapped_named_bo) 
+    {
+        ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
+    }
+
+    attr->stride = dst_stride;
+}
+
+static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *input[], int count)
+{
+	context_t *context = R700_CONTEXT(ctx);
+    GLuint stride;
+    int ret;
+    int i, index;
+
+    R600_STATECHANGE(context, vtx);
+
+    for(index = 0; index < context->nNumActiveAos; index++) 
+    {
+        struct radeon_aos *aos = &context->radeon.tcl.aos[index];
+        i = context->stream_desc[index].element;
+
+        stride = (input[i]->StrideB == 0) ? getTypeSize(input[i]->Type) * input[i]->Size : input[i]->StrideB;
+
+        if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT ||
+#if MESA_BIG_ENDIAN
+            getTypeSize(input[i]->Type) != 4 || 
+#endif
+            stride < 4) 
+        {
+            r700ConvertAttrib(ctx, count, input[i], &context->stream_desc[index]);
+        } 
+        else 
+        {
+            if (input[i]->BufferObj->Name) 
+            {
+                if (stride % 4 != 0) 
+                {
+                    assert(((intptr_t) input[i]->Ptr) % input[i]->StrideB == 0);
+                    r700AlignDataToDword(ctx, input[i], count, &context->stream_desc[index]);
+                    context->stream_desc[index].is_named_bo = GL_FALSE;
+                } 
+                else 
+                {
+                    context->stream_desc[index].stride = input[i]->StrideB;
+                    context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr;
+                    context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo;
+                    context->stream_desc[index].is_named_bo = GL_TRUE;
+                }
+            } 
+            else 
+            {
+                int size;
+                int local_count = count;
+                uint32_t *dst;
+
+                if (input[i]->StrideB == 0) 
+                {
+                    size = getTypeSize(input[i]->Type) * input[i]->Size;
+                    local_count = 1;
+                } 
+                else 
+                {
+                    size = getTypeSize(input[i]->Type) * input[i]->Size * local_count;
+                }
+
+                radeonAllocDmaRegion(&context->radeon, &context->stream_desc[index].bo, 
+                                     &context->stream_desc[index].bo_offset, size, 32);
+                assert(context->stream_desc[index].bo->ptr != NULL);
+                dst = (uint32_t *)ADD_POINTERS(context->stream_desc[index].bo->ptr, 
+                                               context->stream_desc[index].bo_offset);
+
+                switch (context->stream_desc[index].dwords) 
+                {
+                case 1:                     
+                    radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count);                         
+					context->stream_desc[index].stride = 4; 
+                    break;
+                case 2: 
+                    radeonEmitVec8(dst, input[i]->Ptr, input[i]->StrideB, local_count); 
+					context->stream_desc[index].stride = 8; 
+                    break;
+                case 3: 
+                    radeonEmitVec12(dst, input[i]->Ptr, input[i]->StrideB, local_count); 
+					context->stream_desc[index].stride = 12; 
+                    break;
+                case 4: 
+                    radeonEmitVec16(dst, input[i]->Ptr, input[i]->StrideB, local_count); 
+					context->stream_desc[index].stride = 16; 
+                    break;
+                default: 
+                    assert(0); 
+                    break;
+                }
+            }
+        }
+
+        aos->count = context->stream_desc[index].stride == 0 ? 1 : count;
+        aos->stride = context->stream_desc[index].stride / sizeof(float);
+        aos->components = context->stream_desc[index].dwords;
+        aos->bo = context->stream_desc[index].bo;
+        aos->offset = context->stream_desc[index].bo_offset;
+
+        if(context->stream_desc[index].is_named_bo) 
+        {
+            radeon_cs_space_add_persistent_bo(context->radeon.cmdbuf.cs, 
+                                              context->stream_desc[index].bo, 
+                                              RADEON_GEM_DOMAIN_GTT, 0);
+        }
+    }
+
+    context->radeon.tcl.aos_count = context->nNumActiveAos;
+    ret = radeon_cs_space_check_with_bo(context->radeon.cmdbuf.cs, 
+                                        first_elem(&context->radeon.dma.reserved)->bo, 
+                                        RADEON_GEM_DOMAIN_GTT, 0);    
+}
+
+static void r700FreeData(GLcontext *ctx)
+{
+    /* Need to zero tcl.aos[n].bo and tcl.elt_dma_bo
+     * to prevent double unref in radeonReleaseArrays
+     * called during context destroy
+     */
+    context_t *context = R700_CONTEXT(ctx);
+
+    int i;
+
+    for (i = 0; i < context->nNumActiveAos; i++)
+    {
+        if (!context->stream_desc[i].is_named_bo)
+        {
+	        radeon_bo_unref(context->stream_desc[i].bo);
+        }
+        context->radeon.tcl.aos[i].bo = NULL;
+    }
+
+    if (context->ind_buf.bo != NULL)
+    {
+            radeon_bo_unref(context->ind_buf.bo);
+    }
+}
+
+static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    GLvoid *src_ptr;
+    GLuint *out;
+    int i;
+    GLboolean mapped_named_bo = GL_FALSE;
+
+    if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer)
+    {
+        ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj);
+        mapped_named_bo = GL_TRUE;
+        assert(mesa_ind_buf->obj->Pointer != NULL);
+    }
+    src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr);
+
+    if (mesa_ind_buf->type == GL_UNSIGNED_BYTE)
+    {
+        GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1);
+        GLubyte *in = (GLubyte *)src_ptr;
+
+	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
+			     &context->ind_buf.bo_offset, size, 4);
+
+	assert(context->ind_buf.bo->ptr != NULL);
+	out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
+
+        for (i = 0; i + 1 < mesa_ind_buf->count; i += 2)
+        {
+            *out++ = in[i] | in[i + 1] << 16;
+        }
+
+        if (i < mesa_ind_buf->count)
+        {
+            *out++ = in[i];
+        }
+
+#if MESA_BIG_ENDIAN
+    }
+    else
+    { /* if (mesa_ind_buf->type == GL_UNSIGNED_SHORT) */
+        GLushort *in = (GLushort *)src_ptr;
+        GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1);
+
+	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
+			     &context->ind_buf.bo_offset, size, 4);
+
+	assert(context->ind_buf.bo->ptr != NULL);
+	out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
+
+        for (i = 0; i + 1 < mesa_ind_buf->count; i += 2)
+        {
+            *out++ = in[i] | in[i + 1] << 16;
+        }
+
+        if (i < mesa_ind_buf->count)
+        {
+            *out++ = in[i];
+        }
+#endif
+    }
+
+    context->ind_buf.is_32bit = GL_FALSE;
+    context->ind_buf.count = mesa_ind_buf->count;
+
+    if (mapped_named_bo)
+    {
+        ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj);
+    }
+}
+
+static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf)
+{
+    context_t *context = R700_CONTEXT(ctx);
+
+    if (!mesa_ind_buf) {
+        context->ind_buf.bo = NULL;
+        return;
+    }
+
+#if MESA_BIG_ENDIAN
+    if (mesa_ind_buf->type == GL_UNSIGNED_INT)
+    {
+#else
+    if (mesa_ind_buf->type != GL_UNSIGNED_BYTE)
+    {
+#endif
+        const GLvoid *src_ptr;
+        GLvoid *dst_ptr;
+        GLboolean mapped_named_bo = GL_FALSE;
+
+        if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer)
+        {
+	        ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj);
+	        assert(mesa_ind_buf->obj->Pointer != NULL);
+	        mapped_named_bo = GL_TRUE;
+        }
+
+        src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr);
+
+        const GLuint size = mesa_ind_buf->count * getTypeSize(mesa_ind_buf->type);
+
+	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
+			     &context->ind_buf.bo_offset, size, 4);
+	assert(context->ind_buf.bo->ptr != NULL);
+	dst_ptr = ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
+
+        _mesa_memcpy(dst_ptr, src_ptr, size);
+
+        context->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT);
+        context->ind_buf.count = mesa_ind_buf->count;
+
+        if (mapped_named_bo)
+        {
+	        ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj);
+        }
+    }
+    else
+    {
+	    r700FixupIndexBuffer(ctx, mesa_ind_buf);
+    }
+}
+
+static GLboolean r700TryDrawPrims(GLcontext *ctx,
+					 const struct gl_client_array *arrays[],
+					 const struct _mesa_prim *prim,
+					 GLuint nr_prims,
+					 const struct _mesa_index_buffer *ib,
+					 GLuint min_index,
+					 GLuint max_index )
+{
+    context_t *context = R700_CONTEXT(ctx);
+    radeonContextPtr radeon = &context->radeon;
+    GLuint i, id = 0;
+    struct radeon_renderbuffer *rrb;
+
+    if (ctx->NewState)
+    {
+        _mesa_update_state( ctx );
+    }
+
+    _tnl_UpdateFixedFunctionProgram(ctx);
+    r700SetVertexFormat(ctx, arrays, max_index + 1);
+    /* shaders need to be updated before buffers are validated */
+    r700UpdateShaders2(ctx);
+    if (!r600ValidateBuffers(ctx))
+	    return GL_FALSE;
+
+    /* always emit CB base to prevent
+     * lock ups on some chips.
+     */
+    R600_STATECHANGE(context, cb_target);
+    /* mark vtx as dirty since it changes per-draw */
+    R600_STATECHANGE(context, vtx);
+
+    r700SetScissor(context);
+    r700SetupVertexProgram(ctx);
+    r700SetupFragmentProgram(ctx);
+    r600UpdateTextureState(ctx);
+
+    GLuint emit_end = r700PredictRenderSize(ctx, nr_prims)
+                    + context->radeon.cmdbuf.cs->cdw;
+
+    r700SetupIndexBuffer(ctx, ib);
+    r700SetupStreams2(ctx, arrays, max_index + 1);
+
+    radeonEmitState(radeon);
+
+    radeon_debug_add_indent();
+    for (i = 0; i < nr_prims; ++i)
+    {
+	    r700RunRenderPrimitive(ctx,
+                               prim[i].start,
+                               prim[i].start + prim[i].count,
+                               prim[i].mode);
+    }
+    radeon_debug_remove_indent();
+
+    /* Flush render op cached for last several quads. */
+    r700WaitForIdleClean(context);
+
+    rrb = radeon_get_colorbuffer(&context->radeon);
+    if (rrb && rrb->bo)
+	    r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM,
+			 CB_ACTION_ENA_bit | (1 << (id + 6)));
+
+    rrb = radeon_get_depthbuffer(&context->radeon);
+    if (rrb && rrb->bo)
+	    r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM,
+			 DB_ACTION_ENA_bit | DB_DEST_BASE_ENA_bit);
+
+    r700FreeData(ctx);
+
+    if (emit_end < context->radeon.cmdbuf.cs->cdw)
+    {
+        WARN_ONCE("Rendering was %d commands larger than predicted size."
+            " We might overflow  command buffer.\n", context->radeon.cmdbuf.cs->cdw - emit_end);
+    }
+
+    return GL_TRUE;
+}
+
+static void r700DrawPrimsRe(GLcontext *ctx,
+			 const struct gl_client_array *arrays[],
+			 const struct _mesa_prim *prim,
+			 GLuint nr_prims,
+			 const struct _mesa_index_buffer *ib,
+			 GLboolean index_bounds_valid,
+			 GLuint min_index,
+			 GLuint max_index)
+{
+    GLboolean retval = GL_FALSE;
+
+    /* This check should get folded into just the places that
+	 * min/max index are really needed.
+	 */
+	if (!index_bounds_valid) {
+		vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index);
+	}
+
+	if (min_index) {
+		vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r700DrawPrimsRe );
+		return;
+	}
+
+	/* Make an attempt at drawing */
+	retval = r700TryDrawPrims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
+
+	/* If failed run tnl pipeline - it should take care of fallbacks */
+	if (!retval)
+		_tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
+}
+
+static void r700DrawPrims(GLcontext *ctx,
+			 const struct gl_client_array *arrays[],
+			 const struct _mesa_prim *prim,
+			 GLuint nr_prims,
+			 const struct _mesa_index_buffer *ib,
+			 GLboolean index_bounds_valid,
+			 GLuint min_index,
+			 GLuint max_index)
+{
+    context_t *context = R700_CONTEXT(ctx);
+
+    /* For non indexed drawing, using tnl pipe. */
+    if(!ib)
+    {
+        context->ind_buf.bo = NULL;
+
+        _tnl_vbo_draw_prims(ctx, arrays, prim, nr_prims, ib,
+                            index_bounds_valid, min_index, max_index);
+        return;
+    }
+
+	r700DrawPrimsRe(ctx, arrays, prim, nr_prims, ib, index_bounds_valid, min_index, max_index);
+}
+
+void r700InitDraw(GLcontext *ctx)
+{
+	struct vbo_context *vbo = vbo_context(ctx);
+
+	/* to be enabled */
+	vbo->draw_prims = r700DrawPrims;
+}
+
 
diff --git a/src/mesa/drivers/dri/r600/r700_shader.c b/src/mesa/drivers/dri/r600/r700_shader.c
index b4fd51c137..955ea4e4e1 100644
--- a/src/mesa/drivers/dri/r600/r700_shader.c
+++ b/src/mesa/drivers/dri/r600/r700_shader.c
@@ -60,6 +60,55 @@ void AddInstToList(TypedShaderList * plstCFInstructions, R700ShaderInstruction *
 	plstCFInstructions->uNumOfNode++;
 }
 
+void TakeInstOutFromList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * pInst)
+{
+    GLuint    ulIndex = 0;
+    GLboolean bFound  = GL_FALSE;
+    R700ShaderInstruction * pPrevInst = NULL;
+    R700ShaderInstruction * pCurInst = plstCFInstructions->pHead;
+
+    /* Need go thro list to make sure pInst is there. */
+    while(NULL != pCurInst)
+    {
+        if(pCurInst == pInst)
+        {                        
+            bFound  = GL_TRUE;
+            break;
+        }
+
+        pPrevInst = pCurInst;
+        pCurInst  = pCurInst->pNextInst;
+    }
+    if(GL_TRUE == bFound)
+    {
+        plstCFInstructions->uNumOfNode--;
+
+        pCurInst = pInst->pNextInst;
+        ulIndex  = pInst->m_uIndex;
+        while(NULL != pCurInst)
+        {
+            pCurInst->m_uIndex = ulIndex;
+            ulIndex++;
+            pCurInst = pCurInst->pNextInst;
+        }
+
+        if(plstCFInstructions->pHead == pInst)
+        {
+            plstCFInstructions->pHead = pInst->pNextInst;
+        }
+        if(plstCFInstructions->pTail == pInst)
+        {
+            plstCFInstructions->pTail = pPrevInst;
+        }
+        if(NULL != pPrevInst)
+        {
+            pPrevInst->pNextInst = pInst->pNextInst;
+        }
+
+        FREE(pInst);
+    }
+}
+
 void Init_R700_Shader(R700_Shader * pShader)
 {
 	pShader->Type = R700_SHADER_INVALID;
@@ -488,6 +537,47 @@ void DebugPrint(void)
 {
 }
 
+void cleanup_vfetch_shaderinst(R700_Shader *pShader)
+{
+    R700ShaderInstruction      *pInst;
+    R700ShaderInstruction      *pInstToFree;
+    R700VertexInstruction      *pVTXInst;
+    R700ControlFlowInstruction *pCFInst;
+
+    pInst = pShader->lstVTXInstructions.pHead;
+    while(NULL != pInst)
+    {
+        pVTXInst = (R700VertexInstruction  *)pInst;        
+        pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pVTXInst->m_ShaderInstType);
+
+        if(NULL != pVTXInst->m_pLinkedGenericClause)
+        {
+            pCFInst = (R700ControlFlowInstruction*)(pVTXInst->m_pLinkedGenericClause);
+
+            TakeInstOutFromList(&(pShader->lstCFInstructions), 
+                                 (R700ShaderInstruction*)pCFInst);
+
+            pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pCFInst->m_ShaderInstType);
+        }
+
+        pInst = pInst->pNextInst;
+    };
+
+    //destroy each item in pShader->lstVTXInstructions;
+    pInst = pShader->lstVTXInstructions.pHead;
+    while(NULL != pInst)
+    {
+        pInstToFree = pInst;
+        pInst = pInst->pNextInst;
+        FREE(pInstToFree);
+    };
+
+    //set NULL pShader->lstVTXInstructions
+    pShader->lstVTXInstructions.pHead=NULL; 
+	pShader->lstVTXInstructions.pTail=NULL; 
+	pShader->lstVTXInstructions.uNumOfNode=0;
+}
+
 void Clean_Up_Shader(R700_Shader *pShader)
 {
     FREE(pShader->pProgram);
diff --git a/src/mesa/drivers/dri/r600/r700_shader.h b/src/mesa/drivers/dri/r600/r700_shader.h
index bfd01e1a93..997cb05aaf 100644
--- a/src/mesa/drivers/dri/r600/r700_shader.h
+++ b/src/mesa/drivers/dri/r600/r700_shader.h
@@ -143,6 +143,7 @@ void LoadProgram(R700_Shader *pShader);
 void UpdateShaderRegisters(R700_Shader *pShader);
 void DeleteInstructions(R700_Shader *pShader);
 void DebugPrint(void);
+void cleanup_vfetch_shaderinst(R700_Shader *pShader);
 
 void Clean_Up_Shader(R700_Shader *pShader);
 
diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c
index 244a016e07..3d3c8b958f 100644
--- a/src/mesa/drivers/dri/r600/r700_state.c
+++ b/src/mesa/drivers/dri/r600/r700_state.c
@@ -92,7 +92,25 @@ void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
 	    }
     }
 
-    r700SelectVertexShader(ctx);
+    r700SelectVertexShader(ctx, 1);
+    r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS);
+    context->radeon.NewGLState = 0;
+}
+
+void r700UpdateShaders2(GLcontext * ctx)  
+{
+    context_t *context = R700_CONTEXT(ctx);
+
+    /* should only happenen once, just after context is created */
+    /* TODO: shouldn't we fallback to sw here? */
+    if (!ctx->FragmentProgram._Current) {
+	    _mesa_fprintf(stderr, "No ctx->FragmentProgram._Current!!\n");
+	    return;
+    }
+
+    r700SelectFragmentShader(ctx);
+
+    r700SelectVertexShader(ctx, 2);
     r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS);
     context->radeon.NewGLState = 0;
 }
@@ -171,6 +189,14 @@ static void r700InvalidateState(GLcontext * ctx, GLuint new_state) //-----------
 	    R600_STATECHANGE(context, db_target);
     }
 
+    if (new_state & (_NEW_LIGHT)) {
+	    R600_STATECHANGE(context, su);
+	    if (ctx->Light.ProvokingVertex == GL_LAST_VERTEX_CONVENTION)
+		    SETbit(r700->PA_SU_SC_MODE_CNTL.u32All, PROVOKING_VTX_LAST_bit);
+	    else
+		    CLEARbit(r700->PA_SU_SC_MODE_CNTL.u32All, PROVOKING_VTX_LAST_bit);
+    }
+
     r700UpdateStateParameters(ctx, new_state);
 
     R600_STATECHANGE(context, cl);
diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h
index 0f53d5b4c5..209189d8d7 100644
--- a/src/mesa/drivers/dri/r600/r700_state.h
+++ b/src/mesa/drivers/dri/r600/r700_state.h
@@ -35,6 +35,7 @@
 
 extern void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state);
 extern void r700UpdateShaders (GLcontext * ctx);
+extern void r700UpdateShaders2(GLcontext * ctx);
 
 extern void r700UpdateViewportOffset(GLcontext * ctx);
 
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c
index 9ee26286d9..d12c39c9f7 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -159,7 +159,35 @@ GLboolean Process_Vertex_Program_Vfetch_Instructions(
 	return GL_TRUE;
 }
 
-void Map_Vertex_Program(struct r700_vertex_program *vp,
+GLboolean Process_Vertex_Program_Vfetch_Instructions2(
+    GLcontext *ctx,
+	struct r700_vertex_program *vp,
+	struct gl_vertex_program   *mesa_vp)
+{
+    int i;
+    context_t *context = R700_CONTEXT(ctx);
+
+    VTX_FETCH_METHOD vtxFetchMethod;
+	vtxFetchMethod.bEnableMini          = GL_FALSE;
+	vtxFetchMethod.mega_fetch_remainder = 0;
+
+    for(i=0; i<context->nNumActiveAos; i++)
+    {
+        assemble_vfetch_instruction2(&vp->r700AsmCode,
+                                      vp->r700AsmCode.ucVP_AttributeMap[context->stream_desc[i].element],
+                                      context->stream_desc[i].type,
+                                      context->stream_desc[i].size,
+                                      context->stream_desc[i].element,
+                                      context->stream_desc[i]._signed,
+                                      context->stream_desc[i].normalize,						            
+                                     &vtxFetchMethod);
+    }
+
+    return GL_TRUE;
+}
+
+void Map_Vertex_Program(GLcontext *ctx,
+                        struct r700_vertex_program *vp,
 						struct gl_vertex_program   *mesa_vp)
 {
     GLuint ui;
@@ -175,11 +203,22 @@ void Map_Vertex_Program(struct r700_vertex_program *vp,
 	pAsm->number_used_registers += num_inputs;
 
 	// Create VFETCH instructions for inputs
-	if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) 
-	{
-		radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n");
-		return; //error
-	}
+	if(1 == vp->uiVersion) 
+    {
+	    if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) 
+	    {
+		    radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n");
+		    return; 
+	    }
+    }
+    else
+    {
+        if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions2(ctx, vp, mesa_vp) ) 
+	    {
+		    radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions2 return error. \n");
+		    return; 
+	    }
+    }
 
 	// Map Outputs
 	pAsm->number_of_exports = Map_Vertex_Output(pAsm, mesa_vp, pAsm->number_used_registers);
@@ -261,7 +300,8 @@ GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp,
 }
 
 struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
-						struct gl_vertex_program *mesa_vp)
+						struct gl_vertex_program *mesa_vp,
+                        GLint nVer)
 {
 	context_t *context = R700_CONTEXT(ctx);
 	struct r700_vertex_program *vp;
@@ -271,6 +311,7 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 	unsigned int i;
 
 	vp = _mesa_calloc(sizeof(*vp));
+    vp->uiVersion = nVer;
 	vp->mesa_program = (struct gl_vertex_program *)_mesa_clone_program(ctx, &mesa_vp->Base);
 
 	if (mesa_vp->IsPositionInvariant)
@@ -278,8 +319,10 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
                 _mesa_insert_mvp_code(ctx, vp->mesa_program);
         }
 
-	for(i=0; i<VERT_ATTRIB_MAX; i++)
+	if( 1 == nVer )
 	{
+	    for(i=0; i<VERT_ATTRIB_MAX; i++)
+	    {
 		unBit = 1 << i;
 		if(vp->mesa_program->Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */
 		{
@@ -287,7 +330,17 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 			vp->aos_desc[i].stride = vb->AttribPtr[i]->size * sizeof(GL_FLOAT);/* when emit array, data is packed. vb->AttribPtr[i]->stride;*/
 			vp->aos_desc[i].type   = GL_FLOAT;
 		}
+	    }
 	}
+	else
+	{
+		for(i=0; i<context->nNumActiveAos; i++)
+		{
+			vp->aos_desc[i].size   = context->stream_desc[i].size;
+			vp->aos_desc[i].stride = context->stream_desc[i].stride;
+			vp->aos_desc[i].type   = context->stream_desc[i].type;
+		}
+	}	
 	
 	if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
 	{
@@ -296,7 +349,7 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 
 	//Init_Program
 	Init_r700_AssemblerBase(SPT_VP, &(vp->r700AsmCode), &(vp->r700Shader) );
-	Map_Vertex_Program( vp, vp->mesa_program );
+	Map_Vertex_Program(ctx, vp, vp->mesa_program );
 
 	if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, vp->mesa_program))
 	{
@@ -325,7 +378,7 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 	return vp;
 }
 
-void r700SelectVertexShader(GLcontext *ctx)
+void r700SelectVertexShader(GLcontext *ctx, GLint nVersion)
 {
     context_t *context = R700_CONTEXT(ctx);
     struct r700_vertex_program_cont *vpc;
@@ -347,17 +400,35 @@ void r700SelectVertexShader(GLcontext *ctx)
     
     for (vp = vpc->progs; vp; vp = vp->next)
     {
+	if (vp->uiVersion != nVersion ) 
+	    continue;
 	match = GL_TRUE;	
-	for(i=0; i<VERT_ATTRIB_MAX; i++)
+	if ( 1 == nVersion ) 
 	{
+	    for(i=0; i<VERT_ATTRIB_MAX; i++)
+	    {
 		unBit = 1 << i;
 		if(InputsRead & unBit)
 		{
-			if (vp->aos_desc[i].size != vb->AttribPtr[i]->size)
-				match = GL_FALSE;
-				break;
+		    if (vp->aos_desc[i].size != vb->AttribPtr[i]->size)
+		    {
+			match = GL_FALSE;
+			break;
+		    }
 		}
+	    }
 	}
+	else
+	{
+	    for(i=0; i<context->nNumActiveAos; i++)
+	    {
+		if (vp->aos_desc[i].size != context->stream_desc[i].size)
+		{
+		    match = GL_FALSE;
+		    break;
+		}
+	    }
+	}	
 	if (match) 
 	{
 		context->selected_vp = vp;
@@ -365,7 +436,7 @@ void r700SelectVertexShader(GLcontext *ctx)
 	}
     }
 
-    vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program) );
+    vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program), nVersion);
     if(!vp)
     {
 	radeon_error("Failed to translate vertex shader. \n");
@@ -377,6 +448,140 @@ void r700SelectVertexShader(GLcontext *ctx)
     return;
 }
 
+int getTypeSize(GLenum type)
+{
+    switch (type) 
+    {
+    case GL_DOUBLE:
+        return sizeof(GLdouble);
+    case GL_FLOAT:
+        return sizeof(GLfloat);
+    case GL_INT:
+        return sizeof(GLint);
+    case GL_UNSIGNED_INT:
+        return sizeof(GLuint);
+    case GL_SHORT:
+        return sizeof(GLshort);
+    case GL_UNSIGNED_SHORT:
+        return sizeof(GLushort);
+    case GL_BYTE:
+        return sizeof(GLbyte);
+    case GL_UNSIGNED_BYTE:
+        return sizeof(GLubyte);
+    default:
+        assert(0);
+        return 0;
+    }
+}
+
+static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const struct gl_client_array *input)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    
+    StreamDesc * pStreamDesc = &(context->stream_desc[context->nNumActiveAos]);
+
+	GLuint stride;
+
+	stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size 
+                                   : input->StrideB;
+
+    if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT ||
+#if MESA_BIG_ENDIAN
+        getTypeSize(input->Type) != 4 ||
+#endif
+        stride < 4) 
+    {
+        pStreamDesc->type = GL_FLOAT;
+
+        if (input->StrideB == 0) 
+        {
+	        pStreamDesc->stride = 0;
+        } 
+        else 
+        {
+	        pStreamDesc->stride = sizeof(GLfloat) * input->Size;
+        }
+        pStreamDesc->dwords = input->Size;
+        pStreamDesc->is_named_bo = GL_FALSE;
+    } 
+    else 
+    {
+        pStreamDesc->type = input->Type;
+        pStreamDesc->dwords = (getTypeSize(input->Type) * input->Size + 3)/ 4;
+        if (!input->BufferObj->Name) 
+        {
+            if (input->StrideB == 0) 
+            {
+                pStreamDesc->stride = 0;
+            } 
+            else 
+            {
+                pStreamDesc->stride = (getTypeSize(pStreamDesc->type) * input->Size + 3) & ~3;
+            }
+
+            pStreamDesc->is_named_bo = GL_FALSE;
+        }
+    }
+
+	pStreamDesc->size = input->Size;
+	pStreamDesc->dst_loc = context->nNumActiveAos;
+	pStreamDesc->element = unLoc;
+
+	switch (pStreamDesc->type) 
+	{ //GetSurfaceFormat
+	case GL_FLOAT:
+		pStreamDesc->_signed = 0;
+		pStreamDesc->normalize = GL_FALSE;
+		break;
+	case GL_SHORT:
+		pStreamDesc->_signed = 1;
+		pStreamDesc->normalize = input->Normalized;
+		break;
+	case GL_BYTE:
+		pStreamDesc->_signed = 1;
+		pStreamDesc->normalize = input->Normalized;
+		break;
+	case GL_UNSIGNED_SHORT:
+		pStreamDesc->_signed = 0;
+		pStreamDesc->normalize = input->Normalized;
+		break;
+	case GL_UNSIGNED_BYTE:
+		pStreamDesc->_signed = 0;
+		pStreamDesc->normalize = input->Normalized;
+		break;
+	default:
+	case GL_INT:
+	case GL_UNSIGNED_INT:
+	case GL_DOUBLE: 
+		assert(0);
+		break;
+	}
+	context->nNumActiveAos++;
+}
+
+void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    struct r700_vertex_program *vpc
+           = (struct r700_vertex_program *)ctx->VertexProgram._Current;
+
+    struct gl_vertex_program * mesa_vp = (struct gl_vertex_program *)&(vpc->mesa_program);
+    unsigned int unLoc = 0;
+    unsigned int unBit = mesa_vp->Base.InputsRead;
+    context->nNumActiveAos = 0;
+
+    while(unBit) 
+    {
+        if(unBit & 1)
+        {
+            r700TranslateAttrib(ctx, unLoc, count, arrays[unLoc]);
+        }
+
+        unBit >>= 1;
+        ++unLoc;
+    }
+}
+
 void * r700GetActiveVpShaderBo(GLcontext * ctx)
 {
     context_t *context = R700_CONTEXT(ctx);
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h
index c48764c43b..f9a3e395ee 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.h
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.h
@@ -52,7 +52,7 @@ struct r700_vertex_program
 
     GLboolean translated;
     GLboolean loaded;
-    GLboolean needUpdateVF;
+    GLint     uiVersion;
 	
     void * shaderbo;
 
@@ -76,19 +76,28 @@ unsigned int Map_Vertex_Input(r700_AssemblerBase       *pAsm,
 GLboolean Process_Vertex_Program_Vfetch_Instructions(
 	struct r700_vertex_program *vp,
 	struct gl_vertex_program   *mesa_vp);
-void Map_Vertex_Program(struct r700_vertex_program *vp,
+GLboolean Process_Vertex_Program_Vfetch_Instructions2(
+    GLcontext *ctx,
+	struct r700_vertex_program *vp,
+	struct gl_vertex_program   *mesa_vp);
+void Map_Vertex_Program(GLcontext *ctx,
+            struct r700_vertex_program *vp,
 			struct gl_vertex_program   *mesa_vp);
 GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp,
 					   struct gl_vertex_program   *mesa_vp);
 
 struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
-				    struct gl_vertex_program   *mesa_vp);
+				    struct gl_vertex_program   *mesa_vp,
+                    GLint nVer);
 
 /* Interface */
-extern void r700SelectVertexShader(GLcontext *ctx);
+extern void r700SelectVertexShader(GLcontext *ctx, GLint nVersion);
+extern void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count);
 
 extern GLboolean r700SetupVertexProgram(GLcontext * ctx);
 
 extern void *    r700GetActiveVpShaderBo(GLcontext * ctx);
 
+extern int getTypeSize(GLenum type);
+
 #endif /* _R700_VERTPROG_H_ */
diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.c b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c
new file mode 120000
index 0000000000..f6a5f66470
--- /dev/null
+++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c
@@ -0,0 +1 @@
+../radeon/radeon_buffer_objects.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.h b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h
new file mode 120000
index 0000000000..2f134fd17b
--- /dev/null
+++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h
@@ -0,0 +1 @@
+../radeon/radeon_buffer_objects.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/radeon/Makefile b/src/mesa/drivers/dri/radeon/Makefile
index 1f286776b5..ae2e695bfc 100644
--- a/src/mesa/drivers/dri/radeon/Makefile
+++ b/src/mesa/drivers/dri/radeon/Makefile
@@ -47,8 +47,7 @@ C_SOURCES = \
 	$(DRIVER_SOURCES) \
 	$(CS_SOURCES)
 
-DRIVER_DEFINES = -DRADEON_COMMON=0 \
-				 -Wall
+DRIVER_DEFINES = -DRADEON_R100 -Wall
 
 DRI_LIB_DEPS += $(RADEON_LDFLAGS)
 
@@ -56,4 +55,3 @@ X86_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c
index 8032cbcd69..097ab7cf61 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common.c
@@ -1344,5 +1344,5 @@ void rcommonBeginBatch(radeonContextPtr rmesa, int n,
 
 void radeonUserClear(GLcontext *ctx, GLuint mask)
 {
-   _mesa_meta_clear(ctx, mask);
+   _mesa_meta_Clear(ctx, mask);
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c
index 330721acee..6b9b1e3c5e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c
@@ -47,7 +47,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) /* +r6/r7 */
+#if defined(RADEON_R600)
 #include "r600_context.h"
 #endif
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c
index 2012cbcf83..0ae906a6ce 100644
--- a/src/mesa/drivers/dri/radeon/radeon_fbo.c
+++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c
@@ -584,7 +584,7 @@ void radeon_fbo_init(struct radeon_context *radeon)
   radeon->glCtx->Driver.FinishRenderTexture = radeon_finish_render_texture;
   radeon->glCtx->Driver.ResizeBuffers = radeon_resize_buffers;
   radeon->glCtx->Driver.ValidateFramebuffer = radeon_validate_framebuffer;
-  radeon->glCtx->Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer;
+  radeon->glCtx->Driver.BlitFramebuffer = _mesa_meta_BlitFramebuffer;
 }
 
   
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 5ffb55db5e..573eb6c9c1 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -48,17 +48,17 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_screen.h"
 #include "radeon_common.h"
 #include "radeon_span.h"
-#if !RADEON_COMMON
+#if defined(RADEON_R100)
 #include "radeon_context.h"
 #include "radeon_tex.h"
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_tex.h"
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+#elif defined(RADEON_R300)
 #include "r300_context.h"
 #include "r300_tex.h"
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #include "r600_context.h"
 #include "r700_driconf.h" /* +r6/r7 */
 #include "r600_tex.h"     /* +r6/r7 */
@@ -82,7 +82,7 @@ DRI_CONF_OPT_BEGIN_V(command_buffer_size,int,def, # min ":" # max ) \
         DRI_CONF_DESC(de,"Grösse des Befehlspuffers (in KB)") \
 DRI_CONF_OPT_END
 
-#if !RADEON_COMMON	/* R100 */
+#if defined(RADEON_R100)	/* R100 */
 PUBLIC const char __driConfigOptions[] =
 DRI_CONF_BEGIN
     DRI_CONF_SECTION_PERFORMANCE
@@ -109,7 +109,7 @@ DRI_CONF_BEGIN
 DRI_CONF_END;
 static const GLuint __driNConfigOptions = 15;
 
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 
 PUBLIC const char __driConfigOptions[] =
 DRI_CONF_BEGIN
@@ -147,7 +147,7 @@ extern const struct dri_extension NV_vp_extension[];
 extern const struct dri_extension ATI_fs_extension[];
 extern const struct dri_extension point_extensions[];
 
-#elif RADEON_COMMON && (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600))
+#elif defined(RADEON_R300) || defined(RADEON_R600)
 
 #define DRI_CONF_FP_OPTIMIZATION_SPEED   0
 #define DRI_CONF_FP_OPTIMIZATION_QUALITY 1
@@ -220,7 +220,7 @@ static const GLuint __driNConfigOptions = 17;
 
 extern const struct dri_extension gl_20_extension[];
 
-#endif /* RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) */
+#endif
 
 extern const struct dri_extension card_extensions[];
 extern const struct dri_extension mm_extensions[];
@@ -337,7 +337,7 @@ radeonFillInModes( __DRIscreenPrivate *psp,
     return (const __DRIconfig **) configs;
 }
 
-#if !RADEON_COMMON
+#if defined(RADEON_R100)
 static const __DRItexOffsetExtension radeonTexOffsetExtension = {
     { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
     radeonSetTexOffset,
@@ -350,7 +350,7 @@ static const __DRItexBufferExtension radeonTexBufferExtension = {
 };
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
 static const __DRIallocateExtension r200AllocateExtension = {
     { __DRI_ALLOCATE, __DRI_ALLOCATE_VERSION },
     r200AllocateMemoryMESA,
@@ -370,7 +370,7 @@ static const __DRItexBufferExtension r200TexBufferExtension = {
 };
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+#if defined(RADEON_R300)
 static const __DRItexOffsetExtension r300texOffsetExtension = {
     { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
    r300SetTexOffset,
@@ -383,7 +383,7 @@ static const __DRItexBufferExtension r300TexBufferExtension = {
 };
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 static const __DRItexOffsetExtension r600texOffsetExtension = {
     { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
    r600SetTexOffset, /* +r6/r7 */
@@ -1222,22 +1222,22 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
        screen->extensions[i++] = &driMediaStreamCounterExtension.base;
    }
 
-#if !RADEON_COMMON
+#if defined(RADEON_R100)
    screen->extensions[i++] = &radeonTexOffsetExtension.base;
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
    if (IS_R200_CLASS(screen))
       screen->extensions[i++] = &r200AllocateExtension.base;
 
    screen->extensions[i++] = &r200texOffsetExtension.base;
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+#if defined(RADEON_R300)
    screen->extensions[i++] = &r300texOffsetExtension.base;
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
    screen->extensions[i++] = &r600texOffsetExtension.base;
 #endif
 
@@ -1376,22 +1376,22 @@ radeonCreateScreen2(__DRIscreenPrivate *sPriv)
        screen->extensions[i++] = &driMediaStreamCounterExtension.base;
    }
 
-#if !RADEON_COMMON
+#if defined(RADEON_R100)
    screen->extensions[i++] = &radeonTexBufferExtension.base;
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
    if (IS_R200_CLASS(screen))
        screen->extensions[i++] = &r200AllocateExtension.base;
 
    screen->extensions[i++] = &r200TexBufferExtension.base;
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+#if defined(RADEON_R300)
    screen->extensions[i++] = &r300TexBufferExtension.base;
 #endif
 
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
    screen->extensions[i++] = &r600TexBufferExtension.base;
 #endif
 
@@ -1589,22 +1589,22 @@ radeonDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
 static const __DRIconfig **
 radeonInitScreen(__DRIscreenPrivate *psp)
 {
-#if !RADEON_COMMON
+#if defined(RADEON_R100)
    static const char *driver_name = "Radeon";
    static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 1, 6, 0 };
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
    static const char *driver_name = "R200";
    static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 1, 6, 0 };
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+#elif defined(RADEON_R300)
    static const char *driver_name = "R300";
    static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 1, 24, 0 };
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
    static const char *driver_name = "R600";
    static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
@@ -1630,13 +1630,13 @@ radeonInitScreen(__DRIscreenPrivate *psp)
     * Hello chicken.  Hello egg.  How are you two today?
     */
    driInitExtensions( NULL, card_extensions, GL_FALSE );
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
    driInitExtensions( NULL, blend_extensions, GL_FALSE );
    driInitSingleExtension( NULL, ARB_vp_extension );
    driInitSingleExtension( NULL, NV_vp_extension );
    driInitSingleExtension( NULL, ATI_fs_extension );
    driInitExtensions( NULL, point_extensions, GL_FALSE );
-#elif (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600))
+#elif (defined(RADEON_R300) || defined(RADEON_R600))
    driInitSingleExtension( NULL, gl_20_extension );
 #endif
 
@@ -1684,13 +1684,13 @@ __DRIconfig **radeonInitScreen2(__DRIscreenPrivate *psp)
     */
    driInitExtensions( NULL, card_extensions, GL_FALSE );
    driInitExtensions( NULL, mm_extensions, GL_FALSE );
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
    driInitExtensions( NULL, blend_extensions, GL_FALSE );
    driInitSingleExtension( NULL, ARB_vp_extension );
    driInitSingleExtension( NULL, NV_vp_extension );
    driInitSingleExtension( NULL, ATI_fs_extension );
    driInitExtensions( NULL, point_extensions, GL_FALSE );
-#elif (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600))
+#elif (defined(RADEON_R300) || defined(RADEON_R600))
    driInitSingleExtension( NULL, gl_20_extension );
 #endif
 
@@ -1772,13 +1772,13 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
 const struct __DriverAPIRec driDriverAPI = {
    .InitScreen      = radeonInitScreen,
    .DestroyScreen   = radeonDestroyScreen,
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
    .CreateContext   = r200CreateContext,
    .DestroyContext  = r200DestroyContext,
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
    .CreateContext   = r600CreateContext,
    .DestroyContext  = radeonDestroyContext,
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+#elif defined(RADEON_R300)
    .CreateContext   = r300CreateContext,
    .DestroyContext  = radeonDestroyContext,
 #else
diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c
index d603f52df7..0c49c3713a 100644
--- a/src/mesa/drivers/dri/radeon/radeon_span.c
+++ b/src/mesa/drivers/dri/radeon/radeon_span.c
@@ -55,7 +55,7 @@ static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb);
 /* r200 depth buffer is always tiled - this is the formula
    according to the docs unless I typo'ed in it
 */
-#if defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
 static GLubyte *r200_depth_2byte(const struct radeon_renderbuffer * rrb,
 				 GLint x, GLint y)
 {
@@ -112,7 +112,7 @@ static GLubyte *r200_depth_4byte(const struct radeon_renderbuffer * rrb,
  * - 2D (akin to macro-tiled/micro-tiled on older asics)
  * only 1D tiling is implemented below
  */
-#if defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 static inline GLint r600_1d_tile_helper(const struct radeon_renderbuffer * rrb,
 					GLint x, GLint y, GLint is_depth, GLint is_stencil)
 {
@@ -334,8 +334,8 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 
 #endif
 
-#ifndef COMPILE_R300
-#ifndef COMPILE_R600
+#ifndef RADEON_R300
+#ifndef RADEON_R600
 static uint32_t
 z24s8_to_s8z24(uint32_t val)
 {
@@ -409,7 +409,7 @@ s8z24_to_z24s8(uint32_t val)
 
 #define TAG(x)    radeon##x##_RGB565
 #define TAG2(x,y) radeon##x##_RGB565##y
-#if defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 #define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
 #else
 #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off)
@@ -423,7 +423,7 @@ s8z24_to_z24s8(uint32_t val)
 
 #define TAG(x)    radeon##x##_ARGB1555
 #define TAG2(x,y) radeon##x##_ARGB1555##y
-#if defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 #define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
 #else
 #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off)
@@ -437,7 +437,7 @@ s8z24_to_z24s8(uint32_t val)
 
 #define TAG(x)    radeon##x##_ARGB4444
 #define TAG2(x,y) radeon##x##_ARGB4444##y
-#if defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 #define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
 #else
 #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off)
@@ -451,7 +451,7 @@ s8z24_to_z24s8(uint32_t val)
 
 #define TAG(x)    radeon##x##_xRGB8888
 #define TAG2(x,y) radeon##x##_xRGB8888##y
-#if defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 #define GET_VALUE(_x, _y) ((*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off)) | 0xff000000))
 #define PUT_VALUE(_x, _y, d) { \
    GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off );		\
@@ -473,7 +473,7 @@ s8z24_to_z24s8(uint32_t val)
 
 #define TAG(x)    radeon##x##_ARGB8888
 #define TAG2(x,y) radeon##x##_ARGB8888##y
-#if defined(RADEON_COMMON_FOR_R600)
+#if defined(RADEON_R600)
 #define GET_VALUE(_x, _y) (*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off)))
 #define PUT_VALUE(_x, _y, d) { \
    GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off );		\
@@ -506,10 +506,10 @@ s8z24_to_z24s8(uint32_t val)
  */
 #define VALUE_TYPE GLushort
 
-#if defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
 #define WRITE_DEPTH( _x, _y, d )					\
    *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) = d
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
    *(GLushort *)r600_ptr_depth(rrb, _x + x_off, _y + y_off) = d
 #else
@@ -517,10 +517,10 @@ s8z24_to_z24s8(uint32_t val)
    *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) = d
 #endif
 
-#if defined(RADEON_COMMON_FOR_R200)
+#if defined(RADEON_R200)
 #define READ_DEPTH( d, _x, _y )						\
    d = *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define READ_DEPTH( d, _x, _y )						\
    d = *(GLushort *)r600_ptr_depth(rrb, _x + x_off, _y + y_off)
 #else
@@ -538,7 +538,7 @@ s8z24_to_z24s8(uint32_t val)
  */
 #define VALUE_TYPE GLuint
 
-#if defined(COMPILE_R300)
+#if defined(RADEON_R300)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
@@ -547,7 +547,7 @@ do {									\
    tmp |= ((d << 8) & 0xffffff00);					\
    *_ptr = tmp;					\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off );		\
@@ -556,7 +556,7 @@ do {									\
    tmp |= ((d) & 0x00ffffff);					\
    *_ptr = tmp;					\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
@@ -576,17 +576,17 @@ do {									\
 } while (0)
 #endif
 
-#if defined(COMPILE_R300)
+#if defined(RADEON_R300)
 #define READ_DEPTH( d, _x, _y )						\
   do {									\
     d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) & 0xffffff00) >> 8; \
   }while(0)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define READ_DEPTH( d, _x, _y )						\
   do {									\
     d = (*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off)) & 0x00ffffff); \
   }while(0)
-#elif defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #define READ_DEPTH( d, _x, _y )						\
   do {									\
     d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; \
@@ -607,13 +607,13 @@ do {									\
  */
 #define VALUE_TYPE GLuint
 
-#if defined(COMPILE_R300)
+#if defined(RADEON_R300)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
    *_ptr = d;								\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off );		\
@@ -627,7 +627,7 @@ do {									\
    tmp |= (d) & 0xff;							\
    *_ptr = tmp;					\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
@@ -643,18 +643,18 @@ do {									\
 } while (0)
 #endif
 
-#if defined(COMPILE_R300)
+#if defined(RADEON_R300)
 #define READ_DEPTH( d, _x, _y )						\
   do { \
     d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)));	\
   }while(0)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define READ_DEPTH( d, _x, _y )						\
   do { \
     d = ((*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off))) << 8) & 0xffffff00; \
     d |= (*(GLuint*)(r600_ptr_stencil(rrb, _x + x_off, _y + y_off))) & 0x000000ff;	\
   }while(0)
-#elif defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #define READ_DEPTH( d, _x, _y )						\
   do { \
     d = s8z24_to_z24s8(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)));	\
@@ -674,7 +674,7 @@ do {									\
 
 /* 24 bit depth, 8 bit stencil depthbuffer functions
  */
-#ifdef COMPILE_R300
+#ifdef RADEON_R300
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off);		\
@@ -683,7 +683,7 @@ do {									\
    tmp |= (d) & 0xff;							\
    *_ptr = tmp;					\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r600_ptr_stencil(rrb, _x + x_off, _y + y_off);		\
@@ -692,7 +692,7 @@ do {									\
    tmp |= (d) & 0xff;							\
    *_ptr = tmp;					\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte(rrb, _x + x_off, _y + y_off);		\
@@ -712,21 +712,21 @@ do {									\
 } while (0)
 #endif
 
-#ifdef COMPILE_R300
+#ifdef RADEON_R300
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
    GLuint tmp = *_ptr;				\
    d = tmp & 0x000000ff;						\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R600)
+#elif defined(RADEON_R600)
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)r600_ptr_stencil( rrb, _x + x_off, _y + y_off );		\
    GLuint tmp = *_ptr;				\
    d = tmp & 0x000000ff;						\
 } while (0)
-#elif defined(RADEON_COMMON_FOR_R200)
+#elif defined(RADEON_R200)
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c
index 9d252aa74c..ae41b90efe 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texstate.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c
@@ -833,11 +833,14 @@ static void import_tex_obj_state( r100ContextPtr rmesa,
    cmd[TEX_PP_TXFORMAT] |= texobj->pp_txformat & TEXOBJ_TXFORMAT_MASK;
    cmd[TEX_PP_BORDER_COLOR] = texobj->pp_border_color;
 
-   if (texobj->base.Target == GL_TEXTURE_RECTANGLE_NV) {
-      GLuint *txr_cmd = RADEON_DB_STATE( txr[unit] );
+   if (texobj->pp_txformat & RADEON_TXFORMAT_NON_POWER2) {
+      uint32_t *txr_cmd = &rmesa->hw.txr[unit].cmd[TXR_CMD_0];
       txr_cmd[TXR_PP_TEX_SIZE] = texobj->pp_txsize; /* NPOT only! */
       txr_cmd[TXR_PP_TEX_PITCH] = texobj->pp_txpitch; /* NPOT only! */
-      RADEON_DB_STATECHANGE( rmesa, &rmesa->hw.txr[unit] );
+      RADEON_STATECHANGE( rmesa, txr[unit] );
+   }
+
+   if (texobj->base.Target == GL_TEXTURE_RECTANGLE_NV) {
       se_coord_fmt |= RADEON_VTX_ST0_NONPARAMETRIC << unit;
    }
    else {
@@ -1114,7 +1117,6 @@ static GLboolean radeon_validate_texture(GLcontext *ctx, struct gl_texture_objec
    RADEON_STATECHANGE( rmesa, ctx );
    rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= 
      (RADEON_TEX_0_ENABLE | RADEON_TEX_BLEND_0_ENABLE) << unit;
-
    RADEON_STATECHANGE( rmesa, tcl );
    rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_ST_BIT(unit);
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c
index fad3d1ceda..7b7392b217 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texture.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texture.c
@@ -101,7 +101,12 @@ void radeonFreeTexImageData(GLcontext *ctx, struct gl_texture_image *timage)
 /* Set Data pointer and additional data for mapped texture image */
 static void teximage_set_map_data(radeon_texture_image *image)
 {
-	radeon_mipmap_level *lvl = &image->mt->levels[image->mtlevel];
+	radeon_mipmap_level *lvl;
+
+	if (!image->mt)
+		return;
+
+	lvl = &image->mt->levels[image->mtlevel];
 
 	image->base.Data = image->mt->bo->ptr + lvl->faces[image->mtface].offset;
 	image->base.RowStride = lvl->rowstride / image->mt->bpp;
@@ -659,11 +664,6 @@ static void radeon_teximage(
 			if (dims == 3)
 				_mesa_free(dstImageOffsets);
 		}
-
-		/* SGIS_generate_mipmap */
-		if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-			radeon_generate_mipmap(ctx, target, texObj);
-		}
 	}
 
 	_mesa_unmap_teximage_pbo(ctx, packing);
@@ -792,11 +792,6 @@ static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int leve
 							     format, type, pixels, packing))
 				_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
 		}
-
-		/* GL_SGIS_generate_mipmap */
-		if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-			radeon_generate_mipmap(ctx, target, texObj);
-		}
 	}
 
 	radeon_teximage_unmap(image);
@@ -979,7 +974,7 @@ int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *t
 			radeon_texture_image *image = get_radeon_texture_image(texObj->Image[face][level]);
 			if (RADEON_DEBUG & RADEON_TEXTURE)
 				fprintf(stderr, " face %i, level %i... %p vs %p ", face, level, t->mt, image->mt);
-			if (t->mt == image->mt) {
+			if (t->mt == image->mt || (!image->mt && !image->base.Data)) {
 				if (RADEON_DEBUG & RADEON_TEXTURE)
 					fprintf(stderr, "OK\n");
 
diff --git a/src/mesa/drivers/dri/s3v/Makefile b/src/mesa/drivers/dri/s3v/Makefile
deleted file mode 100644
index 9bd7973154..0000000000
--- a/src/mesa/drivers/dri/s3v/Makefile
+++ /dev/null
@@ -1,36 +0,0 @@
-# src/mesa/drivers/dri/s3v/Makefile
-
-TOP = ../../../../..
-include $(TOP)/configs/current
-
-LIBNAME = s3v_dri.so
-
-# Doesn't exist yet.
-#MINIGLX_SOURCES = server/savage_dri.c 
-
-DRIVER_SOURCES = \
-	s3v_context.c \
-        s3v_dd.c \
-        s3v_inithw.c \
-        s3v_lock.c \
-        s3v_render.c \
-        s3v_screen.c \
-        s3v_span.c \
-        s3v_state.c \
-        s3v_tex.c \
-        s3v_texmem.c \
-        s3v_texstate.c \
-        s3v_tris.c \
-        s3v_vb.c \
-        s3v_xmesa.c
-
-C_SOURCES = \
-	$(COMMON_SOURCES) \
-	$(DRIVER_SOURCES) 
-
-ASM_SOURCES = 
-
-
-include ../Makefile.template
-
-symlinks:
diff --git a/src/mesa/drivers/dri/s3v/s3v_common.h b/src/mesa/drivers/dri/s3v/s3v_common.h
deleted file mode 100644
index b66cdf1df0..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_common.h
+++ /dev/null
@@ -1,83 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-/* WARNING: If you change any of these defines, make sure to change
- * the kernel include file as well (s3v_drm.h)
- */
-
-#ifndef _XF86DRI_S3V_H_
-#define _XF86DRI_S3V_H_
-
-#ifndef _S3V_DEFINES_
-#define _S3V_DEFINES_
-#define S3V_USE_BATCH 1
-
-/* #define S3V_BUF_4K 1 */
-
-#ifdef S3V_BUF_4K
-#define S3V_DMA_BUF_ORDER 12
-#define S3V_DMA_BUF_NR    256
-#else
-#define S3V_DMA_BUF_ORDER 16 /* -much- better */
-#define S3V_DMA_BUF_NR    16
-#endif
-/* on s3virge you can only choose between *
- * 4k (2^12) and 64k (2^16) dma bufs      */
-#define S3V_DMA_BUF_SZ        (1<<S3V_DMA_BUF_ORDER)
-
-#define S3V_NR_SAREA_CLIPRECTS 8
-
-/* Each region is a minimum of 16k (64*64@4bpp)
- * and there are at most 40 of them.
- */
-#define S3V_NR_TEX_REGIONS 64 /* was 40 */
-#define S3V_LOG_TEX_GRANULARITY 16 /* was 4 */
-/* 40 * (2 ^ 4) = 640k, that's all we have for tex on 4mb gfx card */
-/* FIXME: will it work with card with less than 4mb? */
-/* FIXME: we should set this at run time */
-
-#endif  /* _S3V_DEFINES */
-
-/*
- * WARNING: If you change any of these defines, make sure to change
- * the kernel include file as well (gamma_drm.h)
- */
-
-/* Driver specific DRM command indices
- * NOTE: these are not OS specific, but they are driver specific
- */
-#define DRM_S3V_INIT_DMA              0x00
-#define DRM_S3V_CLEANUP_DMA           0x01
-
-typedef struct _drmS3VInit {
-    enum {
-        S3V_INIT_DMA = 0x01,
-        S3V_CLEANUP_DMA = 0x02
-    } func;
-
-    unsigned int pcimode;   /* bool: 1=pci 0=agp */
-
-    unsigned int mmio_offset;
-    unsigned int buffers_offset;
-    unsigned int sarea_priv_offset;
-
-    unsigned int front_offset;
-    unsigned int front_width;
-    unsigned int front_height;
-    unsigned int front_pitch;
-
-    unsigned int back_offset;
-    unsigned int back_width;
-    unsigned int back_height;
-    unsigned int back_pitch;
-
-    unsigned int depth_offset;
-    unsigned int depth_width;
-    unsigned int depth_height;
-    unsigned int depth_pitch;
-
-    unsigned int texture_offset;
-} drmS3VInit;
-
-#endif
diff --git a/src/mesa/drivers/dri/s3v/s3v_context.c b/src/mesa/drivers/dri/s3v/s3v_context.c
deleted file mode 100644
index 0a3bf7258d..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_context.c
+++ /dev/null
@@ -1,260 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "vbo/vbo.h"
-
-#include "tnl/tnl.h"
-#include "tnl/t_pipeline.h"
-
-#include "main/context.h"
-#include "main/simple_list.h"
-#include "main/matrix.h"
-#include "main/extensions.h"
-#if defined(USE_X86_ASM)
-#include "x86/common_x86_asm.h"
-#endif
-#include "main/simple_list.h"
-#include "main/mm.h"
-
-#include "drivers/common/driverfuncs.h"
-#include "s3v_vb.h"
-#include "s3v_tris.h"
-
-#if 0
-extern const struct tnl_pipeline_stage _s3v_render_stage;
-
-static const struct tnl_pipeline_stage *s3v_pipeline[] = {
-   &_tnl_vertex_transform_stage,
-   &_tnl_normal_transform_stage,
-   &_tnl_lighting_stage,
-   &_tnl_fog_coordinate_stage,
-   &_tnl_texgen_stage,
-   &_tnl_texture_transform_stage,
-				/* REMOVE: point attenuation stage */
-#if 1
-   &_s3v_render_stage,	/* ADD: unclipped rastersetup-to-dma */
-#endif
-   &_tnl_render_stage,
-   0,
-};
-#endif
-
-GLboolean s3vCreateContext(const __GLcontextModes *glVisual,
-			   __DRIcontextPrivate *driContextPriv,
-                     	   void *sharedContextPrivate)
-{
-	GLcontext *ctx, *shareCtx;
-	__DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-	s3vContextPtr vmesa;
-	s3vScreenPtr s3vScrn;
-	S3VSAREAPtr saPriv=(S3VSAREAPtr)(((char*)sPriv->pSAREA) + 
-                            sizeof(drm_sarea_t));
-        struct dd_function_table functions;
-
-	DEBUG_WHERE(("*** s3vCreateContext ***\n"));
-
-	vmesa = (s3vContextPtr) CALLOC( sizeof(*vmesa) );
-	if ( !vmesa ) return GL_FALSE;
-
-	/* Allocate the Mesa context */
-	if (sharedContextPrivate)
-		shareCtx = ((s3vContextPtr) sharedContextPrivate)->glCtx;
-	else
-		shareCtx = NULL;
-
-        _mesa_init_driver_functions(&functions);
-
-	vmesa->glCtx = _mesa_create_context(glVisual, shareCtx, &functions,
-                                            (void *)vmesa);
-	if (!vmesa->glCtx) {
-		FREE(vmesa);
-		return GL_FALSE;
-	}
-
-	vmesa->driContext = driContextPriv;
-	vmesa->driScreen = sPriv;
-	vmesa->driDrawable = NULL; /* Set by XMesaMakeCurrent */
-
-	vmesa->hHWContext = driContextPriv->hHWContext;
-	vmesa->driHwLock = (drmLock *)&sPriv->pSAREA->lock;
-	vmesa->driFd = sPriv->fd;
-	vmesa->sarea = saPriv;
-
-	s3vScrn = vmesa->s3vScreen = (s3vScreenPtr)(sPriv->private);
-
-	ctx = vmesa->glCtx;
-
-	ctx->Const.MaxTextureLevels = 11;  /* it is (11-1) -> 1024 * 1024 FIXME */
-
-	ctx->Const.MaxTextureUnits = 1; /* FIXME: or 2 ? */
-
-	/* No wide points.
-	 */
-	ctx->Const.MinPointSize = 1.0;
-	ctx->Const.MinPointSizeAA = 1.0;
-	ctx->Const.MaxPointSize = 1.0;
-	ctx->Const.MaxPointSizeAA = 1.0;
-
-	/* No wide lines.
-	 */
-	ctx->Const.MinLineWidth = 1.0;
-	ctx->Const.MinLineWidthAA = 1.0;
-	ctx->Const.MaxLineWidth = 1.0;
-	ctx->Const.MaxLineWidthAA = 1.0;
-	ctx->Const.LineWidthGranularity = 1.0;
-
-	ctx->Const.MaxDrawBuffers = 1;
-
-	vmesa->texHeap = mmInit( 0, vmesa->s3vScreen->textureSize );
-	DEBUG(("vmesa->s3vScreen->textureSize = 0x%x\n",
-		vmesa->s3vScreen->textureSize));
-	
-	/* NOTE */
-	/* mmInit(offset, size); */
-
-	/* allocates a structure like this:
-
-	struct mem_block_t {
-		struct mem_block_t *next;
-		struct mem_block_t *heap;
-		int ofs,size;
-		int align;
-		int free:1;
-		int reserved:1;
-	};
-
-	*/
-
-	make_empty_list(&vmesa->TexObjList);
-	make_empty_list(&vmesa->SwappedOut);
-
-	vmesa->CurrentTexObj[0] = 0;
-	vmesa->CurrentTexObj[1] = 0; /* FIXME */
-
-	vmesa->RenderIndex = ~0;
-
-	/* Initialize the software rasterizer and helper modules.
-	 */
-	_swrast_CreateContext( ctx );
-	_vbo_CreateContext( ctx );
-	_tnl_CreateContext( ctx );
-	_swsetup_CreateContext( ctx );
-
-	/* Install the customized pipeline:
-	 */
-#if 0
-	_tnl_destroy_pipeline( ctx );
-	_tnl_install_pipeline( ctx, s3v_pipeline );
-#endif
-	/* Configure swrast to match hardware characteristics:
-	 */
-#if 0
-	_swrast_allow_pixel_fog( ctx, GL_FALSE );
-	_swrast_allow_vertex_fog( ctx, GL_TRUE );
-#endif
-	vmesa->_3d_mode = 0;
-
-	/* 3D lines / gouraud tris */
-	vmesa->CMD = ( AUTO_EXEC_ON | HW_CLIP_ON | DEST_COL_1555
-			| FOG_OFF | ALPHA_OFF | Z_OFF | Z_UPDATE_OFF
-			| Z_LESS | TEX_WRAP_ON | TEX_MODULATE | LINEAR
-			| TEX_COL_ARGB1555 | CMD_3D );
-
-	vmesa->_alpha[0] = vmesa->_alpha[1] = ALPHA_OFF;
-	vmesa->alpha_cmd = vmesa->_alpha[0];
-	vmesa->_tri[0] = DO_GOURAUD_TRI;
-	vmesa->_tri[1] = DO_TEX_LIT_TRI;
-	vmesa->prim_cmd = vmesa->_tri[0];
-
-	/* printf("first vmesa->CMD = 0x%x\n", vmesa->CMD); */
-
-	vmesa->TexOffset = vmesa->s3vScreen->texOffset;
-
-	s3vInitVB( ctx );
-	s3vInitExtensions( ctx );
-	s3vInitDriverFuncs( ctx );
-	s3vInitStateFuncs( ctx );
-	s3vInitSpanFuncs( ctx );
-	s3vInitTextureFuncs( ctx );
-	s3vInitTriFuncs( ctx );
-	s3vInitState( vmesa );
-
-	driContextPriv->driverPrivate = (void *)vmesa;
-
-	/* HACK */
-	vmesa->bufSize = S3V_DMA_BUF_SZ;
-
-	DEBUG(("vmesa->bufSize = %i\n", vmesa->bufSize));
-	DEBUG(("vmesa->bufCount = %i\n", vmesa->bufCount));
-
-
-	/* dma init */
-	DEBUG_BUFS(("GET_FIRST_DMA\n"));
-	
-	vmesa->_bufNum = 0;
-
-	GET_FIRST_DMA(vmesa->driFd, vmesa->hHWContext,
-	1, &(vmesa->bufIndex[0]), &(vmesa->bufSize),
-	&vmesa->_buf[0], &vmesa->bufCount, s3vScrn);
-
-	GET_FIRST_DMA(vmesa->driFd, vmesa->hHWContext,
-    1, &(vmesa->bufIndex[1]), &(vmesa->bufSize),
-    &vmesa->_buf[1], &vmesa->bufCount, s3vScrn);
-
-	vmesa->buf = vmesa->_buf[vmesa->_bufNum];
-	
-/*
-	vmesa->CMD = (AUTO_EXEC_ON | HW_CLIP_ON | DEST_COL_1555
-	| FOG_OFF | ALPHA_OFF | Z_OFF | Z_UPDATE_OFF
-	| DO_GOURAUD_TRI | CMD_3D);
-
-	vmesa->TexOffset = vmesa->s3vScreen->texOffset;
-*/
-
-/* ... but we should support only 15 bit in virge (out of 8/15/24)... */
-
-	DEBUG(("glVisual->depthBits = %i\n", glVisual->depthBits));
-
-	switch (glVisual->depthBits) {
-	case 8:
-		break;
-	
-	case 15:
-	case 16:
-		vmesa->depth_scale = 1.0f / 0xffff; 
-		break;
-	case 24:
-		vmesa->depth_scale = 1.0f / 0xffffff;
-		break;
-	default:
-		break;
-	}
-
-	vmesa->cull_zero = 0.0f;
-
-	vmesa->DepthSize = glVisual->depthBits;
-	vmesa->Flags  = S3V_FRONT_BUFFER;
-	vmesa->Flags |= (glVisual->doubleBufferMode ? S3V_BACK_BUFFER : 0);
-	vmesa->Flags |= (vmesa->DepthSize > 0 ? S3V_DEPTH_BUFFER : 0);
-
-	vmesa->EnabledFlags = S3V_FRONT_BUFFER;
-	vmesa->EnabledFlags |= (glVisual->doubleBufferMode ? S3V_BACK_BUFFER : 0);
-
-
-	if (vmesa->Flags & S3V_BACK_BUFFER) {
-       	vmesa->readOffset = vmesa->drawOffset = vmesa->s3vScreen->backOffset;
-	} else {
-	   	vmesa->readOffset = vmesa->drawOffset = 0;
-	}
-
-	s3vInitHW( vmesa );
-
-	driContextPriv->driverPrivate = (void *)vmesa;
-
-	return GL_TRUE;
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_context.h b/src/mesa/drivers/dri/s3v/s3v_context.h
deleted file mode 100644
index 671ba90d78..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_context.h
+++ /dev/null
@@ -1,443 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef _S3V_CONTEXT_H_
-#define _S3V_CONTEXT_H_
-
-#include "dri_util.h"
-
-#include "s3v_dri.h"
-#include "s3v_regs.h"
-#include "s3v_macros.h"
-#include "s3v_screen.h"
-#include "main/colormac.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-#include "drm.h"
-#include "main/mm.h"
-#include "drirenderbuffer.h"
-
-/* Flags for context */
-#define S3V_FRONT_BUFFER    0x00000001
-#define S3V_BACK_BUFFER     0x00000002
-#define S3V_DEPTH_BUFFER    0x00000004
-
-	/* FIXME: check */
-#define S3V_MAX_TEXTURE_SIZE    2048
-
-/* These are the minimum requirements and should probably be increased */
-#define MAX_MODELVIEW_STACK    16
-#define MAX_PROJECTION_STACK    2
-#define MAX_TEXTURE_STACK       2
-
-extern void	  	s3vDDUpdateHWState(GLcontext *ctx);
-extern s3vScreenPtr	s3vCreateScreen(__DRIscreenPrivate *sPriv);
-extern void	  	s3vDestroyScreen(__DRIscreenPrivate *sPriv);
-extern GLboolean 	s3vCreateContext(const __GLcontextModes *glVisual,
-                                     __DRIcontextPrivate *driContextPriv,
-                                     void *sharedContextPrivate);
-
-#define S3V_UPLOAD_ALL			0xffffffff
-/* #define S3V_UPLOAD_CLIPRECTS		0x00000002 */
-#define S3V_UPLOAD_ALPHA		0x00000004
-#define S3V_UPLOAD_BLEND		0x00000008
-#define S3V_UPLOAD_DEPTH		0x00000010
-#define S3V_UPLOAD_VIEWPORT		0x00000020
-#define S3V_UPLOAD_SHADE		0x00000040
-#define S3V_UPLOAD_CLIP			0x00000080
-#define S3V_UPLOAD_MASKS		0x00000100
-#define S3V_UPLOAD_WINDOW		0x00000200 /* defunct */
-#define S3V_UPLOAD_GEOMETRY		0x00000400
-#define S3V_UPLOAD_POLYGON		0x00000800
-#define S3V_UPLOAD_DITHER		0x00001000
-#define S3V_UPLOAD_LOGICOP		0x00002000
-#define S3V_UPLOAD_FOG			0x00004000
-#define S3V_UPLOAD_LIGHT		0x00008000
-#define S3V_UPLOAD_CONTEXT		0x00010000
-#define S3V_UPLOAD_TEX0			0x00020000
-#define S3V_UPLOAD_STIPPLE		0x00040000
-#define S3V_UPLOAD_TRANSFORM		0x00080000
-#define S3V_UPLOAD_LINEMODE		0x00100000
-#define S3V_UPLOAD_POINTMODE		0x00200000
-#define S3V_UPLOAD_TRIMODE		0x00400000
-
-#define S3V_NEW_CLIP			0x00000001
-#define S3V_NEW_WINDOW			0x00000002
-#define S3V_NEW_CONTEXT			0x00000004
-#define S3V_NEW_TEXTURE			0x00000008 /* defunct */
-#define S3V_NEW_ALPHA			0x00000010
-#define S3V_NEW_DEPTH			0x00000020
-#define S3V_NEW_MASKS			0x00000040
-#define S3V_NEW_POLYGON			0x00000080
-#define S3V_NEW_CULL			0x00000100
-#define S3V_NEW_LOGICOP			0x00000200
-#define S3V_NEW_FOG			0x00000400
-#define S3V_NEW_LIGHT			0x00000800
-#define S3V_NEW_STIPPLE			0x00001000
-#define S3V_NEW_ALL			0xffffffff
-
-#define S3V_FALLBACK_TRI		0x00000001
-#define S3V_FALLBACK_TEXTURE		0x00000002
-
-struct s3v_context;
-typedef struct s3v_context s3vContextRec;
-typedef struct s3v_context *s3vContextPtr;
-typedef struct s3v_texture_object_t *s3vTextureObjectPtr;
-
-#define VALID_S3V_TEXTURE_OBJECT(tobj)  (tobj) 
-
-#define S3V_TEX_MAXLEVELS 12
-
-/* For shared texture space managment, these texture objects may also
- * be used as proxies for regions of texture memory containing other
- * client's textures.  Such proxy textures (not to be confused with GL
- * proxy textures) are subject to the same LRU aging we use for our
- * own private textures, and thus we have a mechanism where we can
- * fairly decide between kicking out our own textures and those of
- * other clients.
- *
- * Non-local texture objects have a valid MemBlock to describe the
- * region managed by the other client, and can be identified by
- * 't->globj == 0' 
- */
-struct s3v_texture_object_t {
-	struct s3v_texture_object_t *next, *prev;
-
-	GLuint age;   
-	struct gl_texture_object *globj;
-     
-	int Pitch;
-	int Height;
-	int WidthLog2;
-	int texelBytes;
-	int totalSize;
-	int bound;
-
-	struct mem_block *MemBlock;   
-	GLuint BufAddr;
-   
-	GLuint min_level;
-	GLuint max_level;
-	GLuint dirty_images;
-
-	GLint firstLevel, lastLevel;  /* upload tObj->Image[first .. lastLevel] */
-
-	struct { 
-		const struct gl_texture_image *image;
-      		int offset;		/* into BufAddr */
-      		int height;
-      		int internalFormat;
-   	} image[S3V_TEX_MAXLEVELS];
-
-	GLuint TextureCMD;
-
-	GLuint TextureColorMode;
-	GLuint TextureFilterMode;
-	GLuint TextureBorderColor;
-	GLuint TextureWrap;
-	GLuint TextureMipSize;
-
-	GLuint TextureBaseAddr[S3V_TEX_MAXLEVELS];
-	GLuint TextureFormat;
-	GLuint TextureReadMode;
-};		
-
-#define S3V_NO_PALETTE        0x0
-#define S3V_USE_PALETTE       0x1
-#define S3V_UPDATE_PALETTE    0x2
-#define S3V_FALLBACK_PALETTE  0x4
-
-void s3vUpdateTextureState( GLcontext *ctx );
-
-void s3vDestroyTexObj( s3vContextPtr vmesa, s3vTextureObjectPtr t);
-void s3vUploadTexImages( s3vContextPtr vmesa, s3vTextureObjectPtr t );
-
-void s3vResetGlobalLRU( s3vContextPtr vmesa );
-void s3vTexturesGone( s3vContextPtr vmesa, 
-		       GLuint start, GLuint end, 
-		       GLuint in_use ); 
-
-void s3vEmitHwState( s3vContextPtr vmesa );
-void s3vGetLock( s3vContextPtr vmesa, GLuint flags );
-void s3vInitExtensions( GLcontext *ctx );
-void s3vInitDriverFuncs( GLcontext *ctx );
-void s3vSetSpanFunctions(driRenderbuffer *rb, const GLvisual *vis);
-void s3vInitState( s3vContextPtr vmesa );
-void s3vInitHW( s3vContextPtr vmesa );
-void s3vInitStateFuncs( GLcontext *ctx );
-void s3vInitTextureFuncs( GLcontext *ctx );
-void s3vInitTriFuncs( GLcontext *ctx );
-
-void s3vUpdateWindow( GLcontext *ctx );
-void s3vUpdateViewportOffset( GLcontext *ctx );
-
-void s3vPrintLocalLRU( s3vContextPtr vmesa );
-void s3vPrintGlobalLRU( s3vContextPtr vmesa );
-
-extern void s3vFallback( s3vContextPtr vmesa, GLuint bit, GLboolean mode );
-#define FALLBACK( imesa, bit, mode ) s3vFallback( imesa, bit, mode )
-
-/* Use the templated vertex formats.  Only one of these is used in s3v.
- */
-#define TAG(x) s3v##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-typedef void (*s3v_quad_func)( s3vContextPtr, 
-				const s3vVertex *, 
-				const s3vVertex *,
-				const s3vVertex *,
-				const s3vVertex * );
-typedef void (*s3v_tri_func)( s3vContextPtr, 
-				const s3vVertex *, 
-				const s3vVertex *,
-				const s3vVertex * );
-typedef void (*s3v_line_func)( s3vContextPtr, 
-				const s3vVertex *, 
-				const s3vVertex * );
-typedef void (*s3v_point_func)( s3vContextPtr, 
-				 const s3vVertex * );
-
-
-/* static void s3v_lines_emit(GLcontext *ctx, GLuint start, GLuint end); */
-typedef void (*emit_func)( GLcontext *, GLuint, GLuint);
-
-struct s3v_context {
-	GLcontext 		*glCtx;		/* Mesa context */
-
-	__DRIcontextPrivate	*driContext;
-	__DRIscreenPrivate	*driScreen;
-	__DRIdrawablePrivate	*driDrawable;
-
-	GLuint new_gl_state;
-	GLuint new_state;
-	GLuint dirty;
-
-	S3VSAREAPtr	sarea; 
-
-	/* Temporaries for translating away float colors
-	 */
-	struct gl_client_array UbyteColor;
-	struct gl_client_array UbyteSecondaryColor;
-
-   	/* Mirrors of some DRI state
-    	 */
-
-	drm_context_t hHWContext;
-	drmLock *driHwLock;
-	int driFd;
-
-	GLuint numClipRects;		/* Cliprects for the draw buffer */
-	drm_clip_rect_t *pClipRects;
-
-	GLuint*	buf;			/* FIXME */
-	GLuint*	_buf[2];
-	int		_bufNum;
-	int		bufIndex[2];
-	int		bufSize;
-	int		bufCount;
-
-	s3vScreenPtr 	s3vScreen;		/* Screen private DRI data */
-
-	int		drawOffset;
-	int		readOffset;
-
-	s3v_point_func	draw_point;
-	s3v_line_func	draw_line;
-	s3v_tri_func	draw_tri;
-	s3v_quad_func	draw_quad;
-
-	GLuint Fallback;
-	GLuint RenderIndex;
-	GLuint SetupNewInputs;
-	GLuint SetupIndex;
-
-	GLuint vertex_format;
-	GLuint vertex_size;
-	GLuint vertex_stride_shift;
-	char *verts;
-
-	GLfloat hw_viewport[16];
-	GLuint hw_primitive;
-	GLenum render_primitive;
-
-	GLfloat	depth_scale;
-
-	s3vTextureObjectPtr CurrentTexObj[2];
-	struct s3v_texture_object_t TexObjList;
-	struct s3v_texture_object_t SwappedOut; 
-	GLenum TexEnvImageFmt[2];
-
-	struct mem_block *texHeap;
-
-   	int lastSwap;
-   	int texAge;
-   	int ctxAge;
-	int dirtyAge;
-	int lastStamp;
-
-	/* max was here: don't touch */
-   
-	unsigned int S3V_REG[S3V_REGS_NUM];
-
-	GLuint texMode;
-	GLuint alphaMode;
-	GLuint lightMode;
-
-	GLuint SrcBase;
-	GLuint DestBase;
-	GLuint DestBlit;
-	GLuint ScissorLR;
-	GLuint ScissorTB;
-	GLuint ScissorWH; /* SubScissorWH */ /* RectWH */
-	GLuint FrontStride;
-	GLuint BackStride;
-	GLuint SrcStride;
-	GLuint DestStride;
-	GLuint SrcXY;
-	GLuint DestXY;
-
-   	GLuint ClearColor;
-	GLuint Color;
-	GLuint DitherMode;
-   	GLuint ClearDepth;
-
-	GLuint TextureBorderColor;
-	GLuint TexOffset;
-	GLuint TexStride;
-
-	GLuint CMD;
-	GLuint prim_cmd;
-	GLuint _tri[2]; /* 0 = gouraud; 1 = tex (lit or unlit) */
-	GLuint alpha_cmd; /* actual alpha cmd */
-	GLuint _alpha[2];
-	GLuint _alpha_tex; /* tex alpha type */
-	/* (3d_mode) 0 = 3d line/gourad tri; 1 = 3d tex tri */
-	GLuint _3d_mode;
-	
-	GLfloat backface_sign;
-	GLfloat cull_zero;
-
-	int restore_primitive;
-
-/* *** 2check *** */
-
-	GLuint		FogMode;
-	GLuint		AreaStippleMode;
-	GLuint		LBReadFormat;
-	GLuint		LBWriteFormat;
-	GLuint		LineMode;
-	GLuint		PointMode;
-	GLuint		TriangleMode;
-	GLuint		AntialiasMode;
-	GLfloat		ViewportScaleX;
-	GLfloat		ViewportScaleY;
-	GLfloat		ViewportScaleZ;
-	GLfloat		ViewportOffsetX;
-	GLfloat		ViewportOffsetY;
-	GLfloat		ViewportOffsetZ;
-	int		MatrixMode;
-	int		DepthMode;
-	int		TransformMode;
-	int		LBReadMode;
-	int		FBReadMode;
-	int		FBWindowBase;
-	int		LBWindowBase;
-	int		ColorDDAMode;
-	int		GeometryMode;
-	int		AlphaTestMode;
-	int		AlphaBlendMode;
-	int		AB_FBReadMode;
-	int		AB_FBReadMode_Save;
-	int		DeltaMode;
-	int		ColorMaterialMode;
-	int		FBHardwareWriteMask;
-	int		MaterialMode;
-	int		NormalizeMode;
-	int		LightingMode;
-	int		Light0Mode;
-	int		Light1Mode;
-	int		Light2Mode;
-	int		Light3Mode;
-	int		Light4Mode;
-	int		Light5Mode;
-	int		Light6Mode;
-	int		Light7Mode;
-	int		Light8Mode;
-	int		Light9Mode;
-	int		Light10Mode;
-	int		Light11Mode;
-	int		Light12Mode;
-	int		Light13Mode;
-	int		Light14Mode;
-	int		Light15Mode;
-	int		LogicalOpMode;
-	int		ScissorMode;
-	int		ScissorMaxXY;
-	int		ScissorMinXY;
-	int		Window; /* GID part probably should be in draw priv */
-	int		WindowOrigin;
-	int		x, y, w, h; /* Probably should be in drawable priv */
-	int		FrameCount; /* Probably should be in drawable priv */
-	int		NotClipped; /* Probably should be in drawable priv */
-	int		WindowChanged; /* Probably should be in drawabl... */
-	int		Flags;
-	int		EnabledFlags;
-	int		DepthSize;
-	int		Begin;
-	GLenum		ErrorValue;
-	int		Texture1DEnabled;
-	int		Texture2DEnabled;
-
-	float		ModelView[16];
-	float		Proj[16];
-	float		ModelViewProj[16];
-	float		Texture[16];
-
-	float		ModelViewStack[(MAX_MODELVIEW_STACK-1)*16];
-	int		ModelViewCount;
-	float		ProjStack[(MAX_PROJECTION_STACK-1)*16];
-	int		ProjCount;
-	float		TextureStack[(MAX_TEXTURE_STACK-1)*16];
-	int		TextureCount;
-};
-
-#define S3VIRGEPACKCOLOR555( r, g, b, a ) \
-    ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-     ((a) ? 0x8000 : 0))
-
-#define S3VIRGEPACKCOLOR565( r, g, b ) \
-    ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define S3VIRGEPACKCOLOR888( r, g, b ) \
-    (((r) << 16) | ((g) << 8) | (b))
-
-#define S3VIRGEPACKCOLOR8888( r, g, b, a ) \
-    (((a) << 24) | ((r) << 16) | ((g) << 8) | (b))
-
-#define S3VIRGEPACKCOLOR4444( r, g, b, a ) \
-    ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-static INLINE GLuint s3vPackColor( GLuint cpp,
-                                   GLubyte r, GLubyte g,
-                                   GLubyte b, GLubyte a )
-{
-	unsigned int ret;
-	DEBUG(("cpp = %i, r=0x%x, g=0x%x, b=0x%x, a=0x%x\n", cpp, r, g, b, a));
-
-	switch ( cpp ) {
-	case 2:
-		ret = S3VIRGEPACKCOLOR555( r, g, b, a );
-		DEBUG(("ret = 0x%x\n", ret));
-		return ret;
-	case 4:
-		return PACK_COLOR_8888( a, r, g, b );
-	default:
-    	return 0;
-	}
-}
-
-#define S3V_CONTEXT(ctx)	((s3vContextPtr)(ctx->DriverCtx))
-
-#endif /* _S3V_CONTEXT_H_ */
diff --git a/src/mesa/drivers/dri/s3v/s3v_dd.c b/src/mesa/drivers/dri/s3v/s3v_dd.c
deleted file mode 100644
index e340116f5e..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_dd.c
+++ /dev/null
@@ -1,90 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-#include "s3v_vb.h"
-#include "s3v_lock.h"
-#if defined(USE_X86_ASM)
-#include "x86/common_x86_asm.h"
-#endif
-
-#include "main/context.h"
-#include "main/framebuffer.h"
-#include "swrast/swrast.h"
-
-#define S3V_DATE "20020207"
-
-
-/* Return the width and height of the current color buffer.
- */
-static void s3vDDGetBufferSize( GLframebuffer *buffer,
-				 GLuint *width, GLuint *height )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-/* S3VHW_LOCK( vmesa ); */
-   *width  = vmesa->driDrawable->w;
-   *height = vmesa->driDrawable->h;
-/* S3VHW_UNLOCK( vmesa ); */
-}
-
-
-/* Return various strings for glGetString().
- */
-static const GLubyte *s3vDDGetString( GLcontext *ctx, GLenum name )
-{
-   static char buffer[128];
-
-   switch ( name ) {
-   case GL_VENDOR:
-      return (GLubyte *)"Max Lingua (ladybug)";
-
-   case GL_RENDERER:
-      sprintf( buffer, "Mesa DRI S3 Virge " S3V_DATE );
-
-      /* Append any CPU-specific information.
-       */
-#ifdef USE_X86_ASM
-      if ( _mesa_x86_cpu_features ) {
-	 strncat( buffer, " x86", 4 );
-      
-}
-#ifdef USE_MMX_ASM
-      if ( cpu_has_mmx ) {
-	 strncat( buffer, "/MMX", 4 );
-      }
-#endif
-#ifdef USE_3DNOW_ASM
-      if ( cpu_has_3dnow ) {
-	 strncat( buffer, "/3DNow!", 7 );
-      }
-#endif
-#ifdef USE_SSE_ASM
-      if ( cpu_has_xmm ) {
-	 strncat( buffer, "/SSE", 4 );
-      }
-#endif
-#endif
-      return (GLubyte *)buffer;
-
-   default:
-      return NULL;
-   }
-}
-
-/* Enable the extensions supported by this driver.
- */
-void s3vInitExtensions( GLcontext *ctx )
-{
-   /* None... */
-}
-
-/* Initialize the driver's misc functions.
- */
-void s3vInitDriverFuncs( GLcontext *ctx )
-{
-   ctx->Driver.GetBufferSize = s3vDDGetBufferSize;
-   ctx->Driver.GetString = s3vDDGetString;
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_dri.h b/src/mesa/drivers/dri/s3v/s3v_dri.h
deleted file mode 100644
index 339c579f7f..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_dri.h
+++ /dev/null
@@ -1,143 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef _S3V_DRI
-#define _S3V_DRI
-
-#include "s3v_common.h"
-
-#define S3V_MAX_DRAWABLES (S3V_DMA_BUF_NR/2) /* 32 */ /* 256 */ /* FIXME */
-
-typedef struct
-{
-   int deviceID;
-   int width;
-   int height;
-   int mem;
-   int cpp;
-   int bitsPerPixel;
-
-   int fbOffset;
-   int fbStride;
-
-   int logTextureGranularity;
-   int textureOffset; 
-
-   drm_handle_t regs;
-   drmSize regsSize;
-
-   unsigned int sarea_priv_offset;
-/*
-   drmAddress regsMap;
-
-   drmSize textureSize;
-   drm_handle_t textures;
-*/
-
-#if 0
-   drm_handle_t agp_buffers;
-   drmSize agp_buf_size;
-#endif
-
-/*
-   drmBufMapPtr drmBufs;
-   int irq;
-   unsigned int sarea_priv_offset;
-*/
-
-/* FIXME: cleanup ! */
-
-   drmSize            registerSize; /* == S3V_MMIO_REGSIZE */
-   drm_handle_t       registerHandle;
-
-   drmSize            pciSize;
-   drm_handle_t       pciMemHandle;
-
-   drmSize            frontSize;    /* == videoRambytes */
-/* drm_handle_t       frontHandle; */
-   unsigned long      frontOffset;  /* == fbOffset */
-   int                frontPitch;
-/* unsigned char      *front; */
-
-   unsigned int       bufferSize; /* size of depth/back buffer */
-
-   drmSize            backSize;
-/* drm_handle_t       backHandle; */
-   unsigned long      backOffset;
-   int                backPitch;
-/* unsigned char      *back; */
-
-   drmSize            depthSize;
-/* drm_handle_t       depthHandle; */
-   unsigned long      depthOffset;
-   int                depthPitch;
-/* unsigned char      *depth; */
-
-   drmSize            texSize;
-/* drm_handle_t       texHandle; */
-   unsigned long      texOffset;
-   int                texPitch;
-/* unsigned char      *tex; */
-
-   drmSize            dmaBufSize;       /* Size of buffers (in bytes) */
-   drm_handle_t       dmaBufHandle;     /* Handle from drmAddMap */
-   unsigned long      dmaBufOffset;     /* Offset/Start */
-   int                dmaBufPitch;      /* Pitch */
-   unsigned char      *dmaBuf;          /* Map */
-   int                bufNumBufs;       /* Number of buffers */
-   drmBufMapPtr       buffers;          /* Buffer map */
-
-} S3VDRIRec, *S3VDRIPtr;
-
-/* WARNING: Do not change the SAREA structure without changing the kernel
- * as well */
-
-typedef struct {
-   unsigned char next, prev; /* indices to form a circular LRU  */
-   unsigned char in_use;   /* owned by a client, or free? */
-   int age;                /* tracked by clients to update local LRU's */
-} S3VTexRegionRec, *S3VTexRegionPtr;
-
-typedef struct {
-   unsigned int nbox;
-   drm_clip_rect_t boxes[S3V_NR_SAREA_CLIPRECTS];
-   
-   /* Maintain an LRU of contiguous regions of texture space.  If
-    * you think you own a region of texture memory, and it has an
-    * age different to the one you set, then you are mistaken and
-    * it has been stolen by another client.  If global texAge
-    * hasn't changed, there is no need to walk the list.
-    *
-    * These regions can be used as a proxy for the fine-grained
-    * texture information of other clients - by maintaining them
-    * in the same lru which is used to age their own textures,
-    * clients have an approximate lru for the whole of global
-    * texture space, and can make informed decisions as to which
-    * areas to kick out.  There is no need to choose whether to
-    * kick out your own texture or someone else's - simply eject
-    * them all in LRU order.  
-    */
-   S3VTexRegionRec texList[S3V_NR_TEX_REGIONS+1]; /* Last elt is sentinal */
-   
-   int texAge;             /* last time texture was uploaded */
-   
-   int last_enqueue;       /* last time a buffer was enqueued */
-   int last_dispatch;      /* age of the most recently dispatched buffer */
-   int last_quiescent;     /*  */
-   
-   int ctxOwner;           /* last context to upload state */
-} S3VSAREARec, *S3VSAREAPtr;
-
-typedef struct {
-  /* Nothing here yet */
-  int dummy;
-} S3VConfigPrivRec, *S3VConfigPrivPtr;
-
-typedef struct {
-  /* Nothing here yet */
-  int dummy;
-} S3VDRIContextRec, *S3VDRIContextPtr;
-
-
-#endif
diff --git a/src/mesa/drivers/dri/s3v/s3v_inithw.c b/src/mesa/drivers/dri/s3v/s3v_inithw.c
deleted file mode 100644
index bdc9effb79..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_inithw.c
+++ /dev/null
@@ -1,79 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include <sys/ioctl.h>
-
-#include "s3v_context.h"
-
-void s3vInitHW( s3vContextPtr vmesa )
-{
-	int i;
-	static short _reset = 1;
-
-	DEBUG(("vmesa->driDrawable = %p\n", vmesa->driDrawable));
-	DEBUG(("stride = %i\n",
-		vmesa->driScreen->fbWidth*vmesa->s3vScreen->cpp));
-	DEBUG(("frontOffset = 0x%x\n", vmesa->s3vScreen->frontOffset));
-	DEBUG(("backOffset = 0x%x\n", vmesa->s3vScreen->backOffset));
-	DEBUG(("depthOffset = 0x%x\n", vmesa->s3vScreen->depthOffset));
-	DEBUG(("textureOffset = 0x%x\n", vmesa->s3vScreen->texOffset));
-
-/*	if (_reset) { */
-/*	ioctl(vmesa->driFd, 0x4a); */
-		ioctl(vmesa->driFd, 0x41); /* reset */
-		_reset = 0;
-/*	ioctl(vmesa->driFd, 0x4c); */
-/*	} */
-
-	/* FIXME */
-	switch (vmesa->s3vScreen->cpp) {
-		case 2:
-			break;
-		case 4:
-			break;
-	}
-
-	/* FIXME for stencil, gid, etc */
-	switch (vmesa->DepthSize) {
-		case 15:
-		case 16:
-			break;
-		case 24:
-			break;
-		case 32:
-			break;
-	}
-
-	vmesa->FogMode = 1;
-	vmesa->ClearDepth = 0xffff;
-	vmesa->x = 0;
-	vmesa->y = 0;
-	vmesa->w = 0;
-	vmesa->h = 0;
-	vmesa->FrameCount = 0;
-	vmesa->MatrixMode = GL_MODELVIEW;
-	vmesa->ModelViewCount = 0;
-	vmesa->ProjCount = 0;
-	vmesa->TextureCount = 0;
-
-
-	/* FIXME: do we need the following? */
-
-	for (i = 0; i < 16; i++)
-		if (i % 5 == 0)
-			vmesa->ModelView[i] =
-			vmesa->Proj[i] =
-			vmesa->ModelViewProj[i] =
-			vmesa->Texture[i] = 1.0;
-		else
-			vmesa->ModelView[i] =
-			vmesa->Proj[i] =
-			vmesa->ModelViewProj[i] =
-			vmesa->Texture[i] = 0.0;
-
-	vmesa->LBWindowBase = vmesa->driScreen->fbWidth *
-				(vmesa->driScreen->fbHeight - 1);
-	vmesa->FBWindowBase = vmesa->driScreen->fbWidth * 
-				(vmesa->driScreen->fbHeight - 1);
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_lock.c b/src/mesa/drivers/dri/s3v/s3v_lock.c
deleted file mode 100644
index 52bb87ecec..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_lock.c
+++ /dev/null
@@ -1,62 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-
-#if DEBUG_LOCKING
-char *prevLockFile = NULL;
-int prevLockLine = 0;
-#endif
-
-
-/* Update the hardware state.  This is called if another context has
- * grabbed the hardware lock, which includes the X server.  This
- * function also updates the driver's window state after the X server
- * moves, resizes or restacks a window -- the change will be reflected
- * in the drawable position and clip rects.  Since the X server grabs
- * the hardware lock when it changes the window state, this routine will
- * automatically be called after such a change.
- */
-void s3vGetLock( s3vContextPtr vmesa, GLuint flags )
-{
-   __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-/*   __DRIscreenPrivate *sPriv = vmesa->driScreen; */
-
-   printf("s3vGetLock <- ***\n");
-
-   drmGetLock( vmesa->driFd, vmesa->hHWContext, flags );
-
-   /* The window might have moved, so we might need to get new clip
-    * rects.
-    *
-    * NOTE: This releases and regrabs the hw lock to allow the X server
-    * to respond to the DRI protocol request for new drawable info.
-    * Since the hardware state depends on having the latest drawable
-    * clip rects, all state checking must be done _after_ this call.
-    */
-   /* DRI_VALIDATE_DRAWABLE_INFO( vmesa->display, sPriv, dPriv ); */
-
-   if ( vmesa->lastStamp != dPriv->lastStamp ) {
-      vmesa->lastStamp = dPriv->lastStamp;
-      vmesa->new_state |= S3V_NEW_WINDOW | S3V_NEW_CLIP;
-   }
-
-   vmesa->numClipRects = dPriv->numClipRects;
-   vmesa->pClipRects = dPriv->pClipRects;
-
-#if 0
-   vmesa->dirty = ~0;
-
-   if ( sarea->ctxOwner != vmesa->hHWContext ) {
-      sarea->ctxOwner = vmesa->hHWContext;
-      vmesa->dirty = S3V_UPLOAD_ALL;
-   }
-
-   for ( i = 0 ; i < vmesa->lastTexHeap ; i++ ) {
-      if ( sarea->texAge[i] != vmesa->lastTexAge[i] ) {
-	 s3vAgeTextures( vmesa, i );
-      }
-   }
-#endif
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_lock.h b/src/mesa/drivers/dri/s3v/s3v_lock.h
deleted file mode 100644
index c39d24a38a..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_lock.h
+++ /dev/null
@@ -1,99 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef __S3V_LOCK_H__
-#define __S3V_LOCK_H__
-
-#include <sys/ioctl.h>
-
-extern void s3vGetLock( s3vContextPtr vmesa, GLuint flags );
-
-/* Turn DEBUG_LOCKING on to find locking conflicts.
- */
-#define DEBUG_LOCKING	0
-
-#if DEBUG_LOCKING
-extern char *prevLockFile;
-extern int prevLockLine;
-
-#define DEBUG_LOCK() \
-   do {	\
-      prevLockFile = (__FILE__); \
-      prevLockLine = (__LINE__); \
-   } while (0)
-
-#define DEBUG_RESET() \
-   do {	\
-      prevLockFile = 0;	\
-      prevLockLine = 0;	\
-   } while (0)
-
-#define DEBUG_CHECK_LOCK() \
-   do {	\
-      if ( prevLockFile ) { \
-	 fprintf( stderr, \
-		  "LOCK SET!\n\tPrevious %s:%d\n\tCurrent: %s:%d\n", \
-		  prevLockFile, prevLockLine, __FILE__, __LINE__ );	\
-	 exit(1); \
-      }	\
-   } while (0)
-
-#else
-
-#define DEBUG_LOCK()
-#define DEBUG_RESET()
-#define DEBUG_CHECK_LOCK()
-
-#endif
-
-/*
- * !!! We may want to separate locks from locks with validation.  This
- * could be used to improve performance for those things commands that
- * do not do any drawing !!!
- */
-
-/* Lock the hardware and validate our state.
- */
-#define LOCK_HARDWARE( vmesa ) \
-   do {	\
-      char __ret = 0; \
-      DEBUG_CHECK_LOCK(); \
-      DRM_CAS( vmesa->driHwLock, vmesa->hHWContext, \
-	       (DRM_LOCK_HELD | vmesa->hHWContext), __ret ); \
-      if ( __ret ) \
-         s3vGetLock( vmesa, 0 ); \
-      DEBUG_LOCK(); \
-   } while (0)
-
-/* Unlock the hardware.
- */
-#define UNLOCK_HARDWARE( vmesa ) \
-   do { \
-      DRM_UNLOCK( vmesa->driFd, \
-		  vmesa->driHwLock, \
-		  vmesa->hHWContext ); \
-      DEBUG_RESET(); \
-   } while (0)
-
-#define S3VHW_LOCK( vmesa )	\
-   DRM_UNLOCK(vmesa->driFd, vmesa->driHwLock, vmesa->hHWContext); \
-   DRM_SPINLOCK(&vmesa->driScreen->pSAREA->drawable_lock, \
-		 vmesa->driScreen->drawLockID); \
-   /* VALIDATE_DRAWABLE_INFO_NO_LOCK(vmesa); */
-
-#define S3VHW_UNLOCK( vmesa ) \
-    DRM_SPINUNLOCK(&vmesa->driScreen->pSAREA->drawable_lock, \
-		   vmesa->driScreen->drawLockID); \
-    /* VALIDATE_DRAWABLE_INFO_NO_LOCK_POST(vmesa); */
-
-#define S3V_SIMPLE_LOCK( vmesa ) \
-	ioctl(vmesa->driFd, 0x4a) 
-
-#define S3V_SIMPLE_FLUSH_LOCK( vmesa ) \
-	ioctl(vmesa->driFd, 0x4b) 
-
-#define S3V_SIMPLE_UNLOCK( vmesa ) \
-	ioctl(vmesa->driFd, 0x4c) 
-
-#endif /* __S3V_LOCK_H__ */
diff --git a/src/mesa/drivers/dri/s3v/s3v_macros.h b/src/mesa/drivers/dri/s3v/s3v_macros.h
deleted file mode 100644
index 7e9b4529df..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_macros.h
+++ /dev/null
@@ -1,230 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef _S3V_MACROS_H_
-#define _S3V_MACROS_H_
-
-/**************/
-/* DRI macros */
-/**************/
-
-#define GENERIC_DEBUG 0
-#define FLOW_DEBUG    0
-#define DMABUFS_DEBUG 0
-
-/* Note: The argument to DEBUG*() _must_ be enclosed in parenthesis */
-
-#if (GENERIC_DEBUG || FLOW_DEBUG || DMABUFS_DEBUG)
-#include <stdio.h>
-#endif
-
-#undef DEBUG
-#if GENERIC_DEBUG
-#define DEBUG(str) printf str
-#else
-#define DEBUG(str)
-#endif
-
-#if FLOW_DEBUG
-#define DEBUG_WHERE(str) printf str
-#else
-#define DEBUG_WHERE(str)
-#endif
-
-#if DMABUFS_DEBUG
-#define DEBUG_BUFS(str) printf str
-#else
-#define DEBUG_BUFS(str)
-#endif
-
-
-#if 0
-#define S3V_DMA_SEND_FLAGS    DRM_DMA_PRIORITY
-#define S3V_DMA_SEND_FLAGS    DRM_DMA_BLOCK
-#else
-#define S3V_DMA_SEND_FLAGS    0
-#endif
-
-#if 0
-#define S3V_DMA_GET_FLAGS     \
-    (DRM_DMA_SMALLER_OK | DRM_DMA_LARGER_OK | DRM_DMA_WAIT)
-#else
-#define S3V_DMA_GET_FLAGS     DRM_DMA_WAIT
-#endif
-
-
-#define DMAOUT_CHECK(reg,len) \
-do { \
-	DEBUG(("DMAOUT_CHECK: reg = 0x%x\n", S3V_##reg##_REG)); \
-	DEBUG_BUFS(("DMAOUT_CHECK (was): ")); \
-	DEBUG_BUFS(("vmesa->bufCount=%i of vmesa->bufSize=%i\n", \
-		vmesa->bufCount, vmesa->bufSize)); \
-	/* FIXME: > or >= */ \
-	if (vmesa->bufCount+(len+1) >= vmesa->bufSize) \
-		DMAFLUSH(); \
-\
-	vmesa->bufCount += (len+1); \
-	DEBUG_BUFS(("DMAOUT_CHECK (is): vmesa->bufCount=%i len=%i, reg=%x\n", \
-		vmesa->bufCount, len, S3V_##reg##_REG)); \
-	DMAOUT(	((len & 0xffff) | ((S3V_##reg##_REG & 0xfffc) << 14)) );  \
-} while (0)
-
-#define DMAOUT(val) \
-do { \
-	*(vmesa->buf++)=val; \
-	DEBUG_BUFS(("DMAOUT: val=0x%x\n", (unsigned int)val)); \
-} while(0)
-
-#define DMAFINISH()	\
-do { \
-	/* NOTE: it does nothing - it just prints some summary infos */ \
-	DEBUG(("DMAFINISH: vmesa->bufCount=%i\n", vmesa->bufCount)); \
-	DEBUG(("buf: index=%i; addr=%p\n", vmesa->bufIndex[vmesa->_bufNum], \
-	vmesa->s3vScreen->bufs->list[vmesa->bufIndex[vmesa->_bufNum]].address)); \
-} while(0)
-
-#define DMAFLUSH() \
-do { \
-	if (vmesa->bufCount) { \
-		SEND_DMA(vmesa->driFd, vmesa->hHWContext, 1, \
-			&vmesa->bufIndex[vmesa->_bufNum], &vmesa->bufCount); \
-/*
-		GET_DMA(vmesa->driFd, vmesa->hHWContext, 1, \
-			&vmesa->bufIndex, &vmesa->bufSize); \
-*/ \
-		vmesa->_bufNum = !(vmesa->_bufNum); \
-		vmesa->buf = vmesa->_buf[vmesa->_bufNum]; \
-/* 
-		vmesa->buf = \
-			vmesa->s3vScreen->bufs->list[vmesa->bufIndex].address; \
-*/ \
-		vmesa->bufCount = 0; \
-	} \
-} while (0)
-
-#define CMDCHANGE() \
-do { \
-	DMAOUT_CHECK(3DTRI_CMDSET, 1); /* FIXME: TRI/LINE */ \
-		DMAOUT(vmesa->CMD); \
-	DMAFINISH(); \
-} while (0)
-
-#ifdef DONT_SEND_DMA
-#define GET_DMA(fd, hHWCtx, n, idx, size)
-#define SEND_DMA(fd, hHWCtx,n, idx, cnt)
-#else
-#define GET_DMA(fd, hHWCtx, n, idx, size) \
-do { \
-	drmDMAReq dma; \
-	int retcode, i; \
-\
-	DEBUG(("GET_DMA: ")); \
-	DEBUG(("req_count=%i; req_list[#0]=%i; req_size[#0]=%i\n", \
-		n, (idx)[n-1], (size)[n-1])); \
-\
-	dma.context       = (hHWCtx); \
-	dma.send_count    = 0; \
-	dma.send_list     = NULL; \
-	dma.send_sizes    = NULL; \
-	dma.flags         = S3V_DMA_GET_FLAGS; \
-	dma.request_count = (n); \
-	dma.request_size  = S3V_DMA_BUF_SZ; \
-	dma.request_list  = (idx); \
-	dma.request_sizes = (size); \
-\
-	do { \
-		if ((retcode = drmDMA((fd), &dma))) { \
-		DEBUG_BUFS(("drmDMA (get) returned %d\n", retcode)); \
-	} \
-} while (!(dma).granted_count); \
-\
-	for (i = 0; i < (n); i++) { \
-		DEBUG(("Got buffer %i (index #%i)\n", (idx)[i], i)); \
-		DEBUG(("of %i bytes (%i words) size\n", \
-			(size)[i], (size)[i] >>2)); \
-		/* Convert from bytes to words */ \
-		(size)[i] >>= 2; \
-	} \
-} while (0)
-
-#define SEND_DMA(fd, hHWCtx, n, idx, cnt) \
-do { \
-	drmDMAReq dma; \
-	int retcode, i; \
-\
-	DEBUG(("SEND_DMA: ")); \
-	DEBUG(("send_count=%i; send_list[#0]=%i; send_sizes[#0]=%i\n", \
-		n, (idx)[n-1], (cnt)[n-1])); \
-\
-	for (i = 0; i < (n); i++) { \
-		/* Convert from words to bytes */ \
-		(cnt)[i] <<= 2; \
-	} \
-\
-	dma.context       = (hHWCtx); \
-	dma.send_count    = (n); \
-	dma.send_list     = (idx); \
-	dma.send_sizes    = (cnt); \
-	dma.flags         = S3V_DMA_SEND_FLAGS; \
-	dma.request_count = 0; \
-	dma.request_size  = 0; \
-	dma.request_list  = NULL; \
-	dma.request_sizes = NULL; \
-\
-	if ((retcode = drmDMA((fd), &dma))) { \
-		DEBUG_BUFS(("drmDMA (send) returned %d\n", retcode)); \
-	} \
-\
-	for (i = 0; i < (n); i++) { \
-		DEBUG(("Sent buffer %i (index #%i)\n", (idx)[i], i)); \
-		DEBUG(("of %i bytes (%i words) size\n", \
-			(cnt)[i], (cnt)[i] >>2)); \
-		(cnt)[i] = 0; \
-	} \
-} while (0)
-#endif /* DONT_SEND_DMA */
-
-#define GET_FIRST_DMA(fd, hHWCtx, n, idx, size, buf, cnt, vPriv) \
-do { \
-	int i; \
-	DEBUG_BUFS(("GET_FIRST_DMA\n")); \
-	DEBUG_BUFS(("n=%i idx=%i size=%i\n", n, *idx, *size)); \
-	DEBUG_BUFS(("going to GET_DMA\n")); \
-	GET_DMA(fd, hHWCtx, n, idx, size); \
-	DEBUG_BUFS(("coming from GET_DMA\n")); \
-	DEBUG_BUFS(("n=%i idx=%i size=%i\n", n, (idx)[0], (size)[0])); \
-	for (i = 0; i < (n); i++) { \
-		DEBUG_BUFS(("buf #%i @%p\n", \
-			i, (vPriv)->bufs->list[(idx)[i]].address)); \
-		(buf)[i] = (vPriv)->bufs->list[(idx)[i]].address; \
-		(cnt)[i] = 0; \
-	} \
-	DEBUG(("GOING HOME\n")); \
-} while (0)
-
-/**************************/
-/* generic, global macros */
-/**************************/
-
-#define CALC_LOG2(l2,s) \
-do { \
-	int __s = s; \
-	l2 = 0; \
-	while (__s > 1) { ++l2; __s >>= 1; } \
-} while (0)
-
-#define PrimType_Null               0x00000000
-#define PrimType_Points             0x10000000
-#define PrimType_Lines              0x20000000
-#define PrimType_LineLoop           0x30000000
-#define PrimType_LineStrip          0x40000000
-#define PrimType_Triangles          0x50000000
-#define PrimType_TriangleStrip      0x60000000
-#define PrimType_TriangleFan        0x70000000
-#define PrimType_Quads              0x80000000
-#define PrimType_QuadStrip          0x90000000
-#define PrimType_Polygon            0xa0000000
-#define PrimType_Mask               0xf0000000
-
-#endif /* _S3V_MACROS_H_ */
diff --git a/src/mesa/drivers/dri/s3v/s3v_regs.h b/src/mesa/drivers/dri/s3v/s3v_regs.h
deleted file mode 100644
index 26a7c54af5..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_regs.h
+++ /dev/null
@@ -1,367 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef _S3V_REG_H
-#define _S3V_REG_H
-
-#define S3V_REGS_NUM 256
-
-/************
- * DMA REGS *
- ************/
-
-#define S3V_DMA_ID				0
-#define S3V_DMA_REG				0x8590
-#define S3V_DMA_WRITEP_ID			1
-#define S3V_DMA_WRITEP_REG			0x8594
-#define S3V_DMA_READP_ID			2
-#define S3V_DMA_READP_REG			0x8598
-#define S3V_DMA_ENABLE_ID			3
-#define S3V_DMA_ENABLE_REG			0x859C
-#define S3V_DMA_UPDATE_ID			4
-#define S3V_DMA_UPDATE_REG			0x10000
-
-/***************
- * STATUS REGS *
- ***************/
-
-#define S3V_STAT_ID				10
-#define S3V_STAT_REG				0x8504
-#define S3V_STAT_VSYNC_ID			11
-#define S3V_STAT_VSYNC_REG			0x8505
-#define S3V_STAT_3D_DONE_ID			12
-#define S3V_STAT_3D_DONE_REG			0x8506
-#define S3V_STAT_FIFO_OVER_ID			13
-#define S3V_STAT_FIFO_OVER_REG			0x8508
-#define S3V_STAT_FIFO_EMPTY_ID			14
-#define S3V_STAT_FIFO_EMPTY_REG			0x850C
-#define S3V_STAT_HDMA_DONE_ID			15
-#define S3V_STAT_HDMA_DONE_REG			0x8514
-#define S3V_STAT_CDMA_DONE_ID			16
-#define S3V_STAT_CDMA_DONE_REG			0x8524
-#define S3V_STAT_3D_FIFO_EMPTY_ID		17
-#define S3V_STAT_3D_FIFO_EMPTY_REG		0x8544
-#define S3V_STAT_LPB_ID				18
-#define S3V_STAT_LPB_REG			0x8584
-#define S3V_STAT_3D_BUSY_ID			19
-#define S3V_STAT_3D_BUSY_REG			0x8704
-
-/***********
- * 2D REGS *
- ***********/
-
-#define S3V_BITBLT_ID				30
-#define S3V_BITBLT_REG				0xA400
-#define S3V_BITBLT_SRC_BASE_ID			31
-#define S3V_BITBLT_SRC_BASE_REG			0xA4D4
-#define S3V_BITBLT_DEST_BASE_ID			32
-#define S3V_BITBLT_DEST_BASE_REG		0xA4D8
-#define S3V_BITBLT_CLIP_L_R_ID			33
-#define S3V_BITBLT_CLIP_L_R_REG			0xA4DC
-#define S3V_BITBLT_CLIP_T_B_ID			34
-#define S3V_BITBLT_CLIP_T_B_REG			0xA4E0
-#define S3V_BITBLT_DEST_SRC_STRIDE_ID		35
-#define S3V_BITBLT_DEST_SRC_STRIDE_REG		0xA4E4
-#define S3V_BITBLT_MONO_PAT0_ID			36
-#define S3V_BITBLT_MONO_PAT0_REG		0xA4E8
-#define S3V_BITBLT_MONO_PAT1_ID			37
-#define S3V_BITBLT_MONO_PAT1_REG		0xA4EC
-#define S3V_BITBLT_PAT_BG_COLOR_ID		38
-#define S3V_BITBLT_PAT_BG_COLOR_REG		0xA4F0
-#define S3V_BITBLT_PAT_FG_COLOR_ID		39
-#define S3V_BITBLT_PAT_FG_COLOR_REG		0xA4F4
-#define S3V_BITBLT_CMDSET_ID			40
-#define S3V_BITBLT_CMDSET_REG			0xA500
-#define S3V_BITBLT_WIDTH_HEIGHT_ID		41
-#define S3V_BITBLT_WIDTH_HEIGHT_REG		0xA504
-#define S3V_BITBLT_SRC_X_Y_ID			42
-#define S3V_BITBLT_SRC_X_Y_REG			0xA508
-#define S3V_BITBLT_DEST_X_Y_ID			43
-#define S3V_BITBLT_DEST_X_Y_REG			0xA50C
-#define S3V_2DLINE_ID				44
-#define S3V_2DLINE_REG				0xA800
-#define S3V_2DPOLY_ID				45
-#define S3V_2DPOLY_REG				0xAC00
-
-/***************
- * 3DLINE REGS *
- ***************/
-/* base regs */
-#define S3V_3DLINE_ID                  		50
-#define S3V_3DLINE_REG				0xB000
-#define S3V_3DLINE_Z_BASE_ID           		51
-#define S3V_3DLINE_Z_BASE_REG			0xB0D4
-#define S3V_3DLINE_SRC_BASE_ID         		52   /* it is the same reg */
-#define S3V_3DLINE_SRC_BASE_REG			0xB0D4
-#define S3V_3DLINE_DEST_BASE_ID        		53
-#define S3V_3DLINE_DEST_BASE_REG		0xB0D8
-#define S3V_3DLINE_CLIP_L_R_ID         		54
-#define S3V_3DLINE_CLIP_L_R_REG			0xB0DC
-#define S3V_3DLINE_CLIP_T_B_ID			55
-#define S3V_3DLINE_CLIP_T_B_REG         	0xB0E0
-#define S3V_3DLINE_DEST_SRC_STRIDE_ID		56
-#define S3V_3DLINE_DEST_SRC_STRIDE_REG  	0xB0E4
-#define S3V_3DLINE_Z_STRIDE_ID       		57
-#define S3V_3DLINE_Z_STRIDE_REG  		0xB0E8
-#define S3V_3DLINE_TEX_BASE_ID			58
-#define S3V_3DLINE_TEX_BASE_REG         	0xB0EC
-#define S3V_3DLINE_TEX_B_COLOR_ID		59
-#define S3V_3DLINE_TEX_B_COLOR_REG      	0xB0F0
-#define S3V_3DLINE_FOG_COLOR_ID  		60
-#define S3V_3DLINE_FOG_COLOR_REG      		0xB0F4
-#define S3V_3DLINE_COLOR0_ID   			61
-#define S3V_3DLINE_COLOR0_REG        		0xB0F8
-#define S3V_3DLINE_COLOR1_ID			62
-#define S3V_3DLINE_COLOR1_REG           	0xB0FC
-#define S3V_3DLINE_CMDSET_ID			63
-#define S3V_3DLINE_CMDSET_REG            	0xB100 /* special */
-/* tex regs */
-/* FIXME: shouldn't it be a 1D tex for lines? */
-#define S3V_3DLINE_BASEV_ID 			64
-#define S3V_3DLINE_BASEV_REG          		0xB104 
-#define S3V_3DLINE_BASEU_ID			65
-#define S3V_3DLINE_BASEU_REG            	0xB108
-#define S3V_3DLINE_WXD_ID  			66
-#define S3V_3DLINE_WXD_REG             		0xB10C
-#define S3V_3DLINE_WYD_ID 			67
-#define S3V_3DLINE_WYD_REG              	0xB110
-#define S3V_3DLINE_WSTART_ID			68
-#define S3V_3DLINE_WSTART_REG          		0xB114
-#define S3V_3DLINE_DXD_ID   			69
-#define S3V_3DLINE_DXD_REG          		0xB118
-#define S3V_3DLINE_VXD_ID			70
-#define S3V_3DLINE_VXD_REG              	0xB11C
-#define S3V_3DLINE_UXD_ID			71
-#define S3V_3DLINE_UXD_REG              	0xB120
-#define S3V_3DLINE_DYD_ID			72
-#define S3V_3DLINE_DYD_REG              	0xB124
-#define S3V_3DLINE_VYD_ID			73
-#define S3V_3DLINE_VYD_REG              	0xB128
-#define S3V_3DLINE_UYD_ID			74
-#define S3V_3DLINE_UYD_REG              	0xB12C
-#define S3V_3DLINE_DSTART_ID			75
-#define S3V_3DLINE_DSTART_REG           	0xB130
-#define S3V_3DLINE_VSTART_ID			76
-#define S3V_3DLINE_VSTART_REG           	0xB134
-#define S3V_3DLINE_USTART_ID			77
-#define S3V_3DLINE_USTART_REG           	0xB138
-/* gourad regs */
-#define S3V_3DLINE_GBD_ID   			78
-#define S3V_3DLINE_GBD_REG            		0xB144
-#define S3V_3DLINE_ARD_ID 			79
-#define S3V_3DLINE_ARD_REG             		0xB148
-#define S3V_3DLINE_GS_BS_ID			80
-#define S3V_3DLINE_GS_BS_REG            	0xB14C
-#define S3V_3DLINE_AS_RS_ID			81
-#define S3V_3DLINE_AS_RS_REG             	0xB150
-/* vertex regs */
-#define S3V_3DLINE_DZ_ID    			82
-#define S3V_3DLINE_DZ_REG            		0xB158
-#define S3V_3DLINE_ZSTART_ID			83
-#define S3V_3DLINE_ZSTART_REG            	0xB15C
-#define S3V_3DLINE_XEND0_END1_ID		84
-#define S3V_3DLINE_XEND0_END1_REG       	0xB16C
-#define S3V_3DLINE_DX_ID        		85
-#define S3V_3DLINE_DX_REG        		0xB170
-#define S3V_3DLINE_XSTART_ID			86
-#define S3V_3DLINE_XSTART_REG           	0xB174
-#define S3V_3DLINE_YSTART_ID 			87
-#define S3V_3DLINE_YSTART_REG          		0xB178
-#define S3V_3DLINE_YCNT_ID  			88
-#define S3V_3DLINE_YCNT_REG           		0xB17C
-
-/**************
- * 3DTRI REGS *
- **************/
-/* base regs */
-#define S3V_3DTRI_ID                   		100
-#define S3V_3DTRI_REG				0xB400
-#define S3V_3DTRI_Z_BASE_ID			101
-#define S3V_3DTRI_Z_BASE_REG            	0xB4D4
-#define S3V_3DTRI_SRC_BASE_ID          		102 /* it is the same reg */
-#define S3V_3DTRI_SRC_BASE_REG			0xB4D4
-#define S3V_3DTRI_DEST_BASE_ID         		103
-#define S3V_3DTRI_DEST_BASE_REG			0xB4D8
-#define S3V_3DTRI_CLIP_L_R_ID          		104
-#define S3V_3DTRI_CLIP_L_R_REG 			0xB4DC
-#define S3V_3DTRI_CLIP_T_B_ID          		105
-#define S3V_3DTRI_CLIP_T_B_REG			0xB4E0
-#define S3V_3DTRI_DEST_SRC_STRIDE_ID   		106
-#define S3V_3DTRI_DEST_SRC_STRIDE_REG		0xB4E4
-#define S3V_3DTRI_Z_STRIDE_ID          		107
-#define S3V_3DTRI_Z_STRIDE_REG			0xB4E8
-#define S3V_3DTRI_TEX_BASE_ID          		108
-#define S3V_3DTRI_TEX_BASE_REG			0xB4EC
-#define S3V_3DTRI_TEX_B_COLOR_ID       		109
-#define S3V_3DTRI_TEX_B_COLOR_REG		0xB4F0
-#define S3V_3DTRI_FOG_COLOR_ID         		110
-#define S3V_3DTRI_FOG_COLOR_REG			0xB4F4
-#define S3V_3DTRI_COLOR0_ID            		111
-#define S3V_3DTRI_COLOR0_REG			0xB4F8
-#define S3V_3DTRI_COLOR1_ID            		112
-#define S3V_3DTRI_COLOR1_REG			0xB4FC
-#define S3V_3DTRI_CMDSET_ID            		113  /* special */
-#define S3V_3DTRI_CMDSET_REG			0xB500
-/* tex regs */
-#define S3V_3DTRI_BASEV_ID             		114
-#define S3V_3DTRI_BASEV_REG			0xB504
-#define S3V_3DTRI_BASEU_ID             		115
-#define S3V_3DTRI_BASEU_REG			0xB508
-#define S3V_3DTRI_WXD_ID               		116
-#define S3V_3DTRI_WXD_REG			0xB50C
-#define S3V_3DTRI_WYD_ID               		117
-#define S3V_3DTRI_WYD_REG			0xB510
-#define S3V_3DTRI_WSTART_ID            		118
-#define S3V_3DTRI_WSTART_REG 			0xB514
-#define S3V_3DTRI_DXD_ID               		119
-#define S3V_3DTRI_DXD_REG			0xB518
-#define S3V_3DTRI_VXD_ID               		120
-#define S3V_3DTRI_VXD_REG			0xB51C
-#define S3V_3DTRI_UXD_ID            		121
-#define S3V_3DTRI_UXD_REG			0xB520
-#define S3V_3DTRI_DYD_ID            		122
-#define S3V_3DTRI_DYD_REG			0xB524
-#define S3V_3DTRI_VYD_ID            		123
-#define S3V_3DTRI_VYD_REG			0xB528
-#define S3V_3DTRI_UYD_ID            		124
-#define S3V_3DTRI_UYD_REG			0xB52C
-#define S3V_3DTRI_DSTART_ID         		125
-#define S3V_3DTRI_DSTART_REG			0xB530
-#define S3V_3DTRI_VSTART_ID         		126
-#define S3V_3DTRI_VSTART_REG			0xB534
-#define S3V_3DTRI_USTART_ID         		127
-#define S3V_3DTRI_USTART_REG			0xB538
-/* gourad regs */
-#define S3V_3DTRI_GBX_ID               		128
-#define S3V_3DTRI_GBX_REG			0xB53C
-#define S3V_3DTRI_ARX_ID               		129
-#define S3V_3DTRI_ARX_REG			0xB540
-#define S3V_3DTRI_GBY_ID			130
-#define S3V_3DTRI_GBY_REG               	0xB544
-#define S3V_3DTRI_ARY_ID			131
-#define S3V_3DTRI_ARY_REG               	0xB548
-#define S3V_3DTRI_GS_BS_ID			132
-#define S3V_3DTRI_GS_BS_REG             	0xB54C
-#define S3V_3DTRI_AS_RS_ID			133
-#define S3V_3DTRI_AS_RS_REG             	0xB550
-/* vertex regs */
-#define S3V_3DTRI_ZXD_ID               		134
-#define S3V_3DTRI_ZXD_REG			0xB554
-#define S3V_3DTRI_ZYD_ID               		135
-#define S3V_3DTRI_ZYD_REG			0xB558
-#define S3V_3DTRI_ZSTART_ID            		136
-#define S3V_3DTRI_ZSTART_REG			0xB55C
-#define S3V_3DTRI_TXDELTA12_ID         		137
-#define S3V_3DTRI_TXDELTA12_REG			0xB560
-#define S3V_3DTRI_TXEND12_ID           		138
-#define S3V_3DTRI_TXEND12_REG			0xB564
-#define S3V_3DTRI_TXDELTA01_ID         		139
-#define S3V_3DTRI_TXDELTA01_REG			0xB568
-#define S3V_3DTRI_TXEND01_ID           		140
-#define S3V_3DTRI_TXEND01_REG			0xB56C
-#define S3V_3DTRI_TXDELTA02_ID         		141
-#define S3V_3DTRI_TXDELTA02_REG			0xB570
-#define S3V_3DTRI_TXSTART02_ID         		142
-#define S3V_3DTRI_TXSTART02_REG			0xB574
-#define S3V_3DTRI_TYS_ID               		143
-#define S3V_3DTRI_TYS_REG			0xB578
-#define S3V_3DTRI_TY01_Y12_ID          		144
-#define S3V_3DTRI_TY01_Y12_REG			0xB57C
-
-/* COMMANDS (to 0xB100 [lines] or 0xB500 [tris]) */
-
-/* Auto execute */
-#define AUTO_EXEC_MASK		0x00000001
-#define AUTO_EXEC_OFF		(0x0)
-#define AUTO_EXEC_ON		(0x1)
-/* HW clipping */
-#define HW_CLIP_MASK		0x00000002
-#define HW_CLIP_OFF		(0x0 << 1)
-#define HW_CLIP_ON		(0x1 << 1)
-/* Destination color */
-#define DEST_COL_MASK		0x0000001c
-#define DEST_COL_PAL		(0x0 << 2)	/* 8 bpp - palettized */
-#define DEST_COL_1555		(0x1 << 2)	/* 16 bpp - ZRGB */
-#define DEST_COL_888  		(0x2 << 2)	/* 24 bpp - RGB */
-/* Texture color */
-#define TEX_COL_MASK		0x000000e0
-#define TEX_COL_ARGB8888	(0x0 << 5)	/* 32 bpp - ARGB */
-#define TEX_COL_ARGB4444	(0x1 << 5)  	/* 16 bpp - ARGB */
-#define TEX_COL_ARGB1555	(0x2 << 5)  	/* 16 bpp - ARGB */
-#define TEX_COL_ALPHA4		(0x3 << 5)  	/* 8 bpp - ALPHA4 */
-#define TEX_COL_BLEND4_LOW 	(0x4 << 5)  	/* 4 bpp - BLEND4 low nibble */
-#define TEX_COL_BLEND4_HIGH	(0x5 << 5)  	/* 4 bpp - BLEND4 high nibble */
-#define TEX_COL_PAL		(0x6 << 5)	/* 8 bpp - palettized */
-#define TEX_COL_YUV		(0x7 << 5)	/* 16 bpp - YUV */
-/* Mipmap level */
-#define MIP_MASK		0x00000f00
-#define MIPMAP_LEVEL(s)		(s << 8)	/* 8 -> 11 bits */
-/* Texture filtering */
-#define TEX_FILTER_MASK		0x00007000
-#define MIP_NEAREST		(0x0 << 12)
-#define LINEAR_MIP_NEAREST  	(0x1 << 12)
-#define MIP_LINEAR	        (0x2 << 12)
-#define LINEAR_MIP_LINEAR   	(0x3 << 12)
-#define NEAREST		        (0x4 << 12)
-#define FAST_BILINEAR       	(0x5 << 12)
-#define LINEAR		        (0x6 << 12)
-/* Texture blending */
-#define TEX_BLEND_MAKS		0x00018000
-#define TEX_REFLECT		(0x0 << 15)
-#define TEX_MODULATE    	(0x1 << 15)
-#define TEX_DECAL	        (0x2 << 15)
-/* Fog */
-#define FOG_MASK		0x00020000
-#define FOG_OFF			(0x0 << 17)
-#define FOG_ON              	(0x1 << 17)
-/* Alpha blending */
-#define ALPHA_BLEND_MASK	0x000c0000
-#define ALPHA_OFF		(0x0 << 18) | (0x0 << 19)
-#define ALPHA_TEX		(0x2 << 18)
-#define ALPHA_SRC           	(0x3 << 18)
-/* Depth compare mode */
-#define Z_MODE_MASK		0x00700000
-#define Z_NEVER     		(0x0 << 20)
-#define Z_GREATER   		(0x1 << 20)
-#define Z_EQUAL     		(0x2 << 20)
-#define Z_GEQUAL    		(0x3 << 20)
-#define Z_LESS      		(0x4 << 20)
-#define Z_NOTEQUAL  		(0x5 << 20)
-#define Z_LEQUAL    		(0x6 << 20) 
-#define Z_ALWAYS   		(0x7 << 20)
-/* Depth update */
-#define Z_UPDATE_MASK		0x00800000
-#define Z_UPDATE_OFF		(0x0 << 23)	/* disable z update */
-#define Z_UPDATE_ON         	(0x1 << 23)
-/* Depth buffering mode */
-#define Z_BUFFER_MASK		0x03000000
-#define Z_BUFFER		(0x0 << 24) | (0x0 << 25)
-#define Z_MUX_BUF		(0x1 << 24) | (0x0 << 25)
-#define Z_MUX_DRAW		(0x2 << 24)
-#define Z_OFF			(0x3 << 24) /* no z buffering */
-/* Texture wrapping */
-#define TEX_WRAP_MASK		0x04000000
-#define TEX_WRAP_OFF		(0x0 << 26)
-#define TEX_WRAP_ON		(0x1 << 26)
-/* 3d command */
-#define DO_MASK			0x78000000
-#define DO_GOURAUD_TRI		(0x0 << 27)
-#define DO_TEX_LIT_TRI_OLD	(0x1 << 27) 
-#define DO_TEX_UNLIT_TRI_OLD 	(0x2 << 27)
-#define DO_TEX_LIT_TRI		(0x5 << 27)
-#define DO_TEX_UNLIT_TRI	(0x6 << 27)
-#define DO_3D_LINE		(0x8 << 27)
-#define	DO_NOP			(0xf << 27) /* turn on autoexec */
-/* status */
-#define CMD_MASK		0x80000000
-#define CMD_2D			(0x0 << 31) /* execute a 2d cmd */
-#define CMD_3D			(0x1 << 31) /* execute a 3d cmd */
-
-/* global masks */
-#define TEX_MASK		( TEX_COL_MASK | TEX_WRAP_MASK | MIP_MASK \
-				| TEX_FILTER_MASK | TEX_BLEND_MAKS \
-				| TEX_WRAP_MASK )
-#define Z_MASK			( Z_MODE_MASK | Z_UPDATE_MASK | Z_BUFFER_MASK )
-
-#endif /* _S3V_REG_H */
diff --git a/src/mesa/drivers/dri/s3v/s3v_render.c b/src/mesa/drivers/dri/s3v/s3v_render.c
deleted file mode 100644
index 5023f3c464..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_render.c
+++ /dev/null
@@ -1,203 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-
-#include "tnl/t_context.h"
-
-#include "s3v_context.h"
-#include "s3v_tris.h"
-#include "s3v_vb.h"
-
-
-#define HAVE_POINTS      0
-#define HAVE_LINES       0
-#define HAVE_LINE_STRIPS 0
-#define HAVE_TRIANGLES   0
-#define HAVE_TRI_STRIPS  0
-#define HAVE_TRI_STRIP_1 0
-#define HAVE_TRI_FANS    0
-#define HAVE_QUADS       0
-#define HAVE_QUAD_STRIPS 0
-#define HAVE_POLYGONS    0
-
-#define HAVE_ELTS        0
-
-#if 0
-static void VERT_FALLBACK( GLcontext *ctx,
-			   GLuint start,
-			   GLuint count,
-			   GLuint flags )
-{
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-/*	s3vContextPtr vmesa = S3V_CONTEXT(ctx); */
-	int _flags;
-   
-	DEBUG(("VERT_FALLBACK: flags & PRIM_MODE_MASK = %i\n",
-		flags & PRIM_MODE_MASK));
-	DEBUG(("VERT_FALLBACK: flags=%i PRIM_MODE_MASK=%i\n",
-		flags, PRIM_MODE_MASK));
-#if 0
-	tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK );
-#endif
-	tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 );
-
-	_flags = flags & PRIM_MODE_MASK;
-
-	tnl->Driver.Render.PrimTabVerts[_flags]( ctx, start, count, flags );
-	S3V_CONTEXT(ctx)->SetupNewInputs = VERT_BIT_POS;
-}
-#endif
-
-static const GLuint hw_prim[GL_POLYGON+1] = {
-	PrimType_Points,
-	PrimType_Lines,
-	PrimType_LineLoop,
-	PrimType_LineStrip,
-	PrimType_Triangles,
-	PrimType_TriangleStrip,
-	PrimType_TriangleFan,
-	PrimType_Quads,
-	PrimType_QuadStrip,
-	PrimType_Polygon
-};
-
-static INLINE void s3vStartPrimitive( s3vContextPtr vmesa, GLenum prim )
-{
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-
-	int _hw_prim = hw_prim[prim];
-
-	DEBUG(("s3vStartPrimitive (new #%i) ", prim));
-
-	if (_hw_prim != vmesa->restore_primitive) {
-
-		if (prim == 4) { /* TRI */
-			DEBUG(("switching to tri\n"));
-			vmesa->prim_cmd = vmesa->_tri[vmesa->_3d_mode];
-			vmesa->alpha_cmd = vmesa->_alpha[vmesa->_3d_mode];
-			DMAOUT_CHECK(3DTRI_Z_BASE, 12);
-		} else if (prim == 1) { /* LINE */
-			DEBUG(("switching to line\n"));
-			vmesa->prim_cmd = DO_3D_LINE;
-			vmesa->alpha_cmd = vmesa->_alpha[0];
-			DMAOUT_CHECK(3DLINE_Z_BASE, 12);
-		} else {
-			DEBUG(("Never mind the bollocks!\n"));
-		}
-
-        	DMAOUT(vmesa->s3vScreen->depthOffset & 0x003FFFF8);
-	        DMAOUT(vmesa->DestBase);
-		/* DMAOUT(vmesa->ScissorLR); */
-		/* DMAOUT(vmesa->ScissorTB); */
-		DMAOUT( (0 << 16) | (dPriv->w-1) );
-		DMAOUT( (0 << 16) | (dPriv->h-1) );
-		DMAOUT( (vmesa->SrcStride << 16) | vmesa->TexStride );
-		DMAOUT(vmesa->SrcStride);
-		DMAOUT(vmesa->TexOffset);
-		DMAOUT(vmesa->TextureBorderColor);
-		DMAOUT(0); /* FOG */
-		DMAOUT(0);
-		DMAOUT(0);
-	        DMAOUT(vmesa->CMD | vmesa->prim_cmd | vmesa->alpha_cmd);
-		DMAFINISH();
-   	}
-
-	vmesa->restore_primitive = _hw_prim;
-}
-
-static INLINE void s3vEndPrimitive( s3vContextPtr vmesa )
-{
-/*	GLcontext *ctx = vmesa->glCtx; */
-	DEBUG(("s3vEndPrimitive\n"));
-}
-
-#define LOCAL_VARS s3vContextPtr vmesa = S3V_CONTEXT(ctx)
-#define INIT( prim ) s3vStartPrimitive( vmesa, prim )
-#define FINISH s3vEndPrimitive( vmesa )
-#define NEW_PRIMITIVE() (void) vmesa
-#define NEW_BUFFER() (void) vmesa
-#define FIRE_VERTICES() (void) vmesa
-#define GET_CURRENT_VB_MAX_VERTS() \
-	(vmesa->bufSize - vmesa->bufCount) / 2
-#define GET_SUBSEQUENT_VB_MAX_VERTS() \
-	S3V_DMA_BUF_SZ / 2
-/* XXX */
-#define ALLOC_VERTS(nr) NULL
-#define EMIT_VERTS(ctx, start, count, buf) NULL
-#define FLUSH() s3vEndPrimitive( vmesa )
-
-#define TAG(x) s3v_##x
-
-#include "tnl_dd/t_dd_dmatmp.h"
-
-/**********************************************************************/
-/*                          Render pipeline stage                     */
-/**********************************************************************/
-
-
-static GLboolean s3v_run_render( GLcontext *ctx,
-				  struct tnl_pipeline_stage *stage )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *VB = &tnl->vb;
-	GLuint i;
-	tnl_render_func *tab;
-
-	DEBUG(("s3v_run_render\n"));
-	
-	/* FIXME: hw clip */
-	if (VB->ClipOrMask || vmesa->RenderIndex != 0) {
-		DEBUG(("*** CLIPPED in render ***\n"));
-#if 1
-		return GL_TRUE;	/* don't handle clipping here */
-#endif
-	}
-
-
-	/* We don't do elts */
-	if (VB->Elts)
-		return GL_TRUE;
-
-	tab = TAG(render_tab_verts);
-
-	tnl->Driver.Render.Start( ctx );
-
-	for (i = 0 ; i < VB->PrimitiveCount ; i++ )
-	{
-		GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
-		GLuint start = VB->Primitive[i].start;
-		GLuint length = VB->Primitive[i].count;
-
-		DEBUG(("s3v_run_render (loop=%i) (lenght=%i)\n", i, length));
-
-		if (length) {
-			tnl->Driver.Render.BuildVertices( ctx, start,
-                                start+length, ~0 /*stage->inputs*/); /* XXX */
-			tnl->Driver.Render.PrimTabVerts[prim & PRIM_MODE_MASK]
-				( ctx, start, start + length, prim );
-			vmesa->SetupNewInputs = VERT_BIT_POS;
-		}
-	}
-	
-	tnl->Driver.Render.Finish( ctx );
-
-	return GL_FALSE; /* finished the pipe */
-}
-
-
-
-const struct tnl_pipeline_stage _s3v_render_stage =
-{
-	"s3v render",
-	NULL,
-	NULL,
-	NULL,
-	NULL,
-	 s3v_run_render		/* run */
-};
diff --git a/src/mesa/drivers/dri/s3v/s3v_screen.c b/src/mesa/drivers/dri/s3v/s3v_screen.c
deleted file mode 100644
index f1810597e6..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_screen.c
+++ /dev/null
@@ -1,99 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-#include "s3v_vb.h"
-#include "s3v_dri.h" 
-
-s3vScreenPtr s3vCreateScreen( __DRIscreenPrivate *sPriv )
-{
-   s3vScreenPtr s3vScreen;
-   S3VDRIPtr vDRIPriv = (S3VDRIPtr)sPriv->pDevPriv;
-
-/*   int i; */
-
-   DEBUG(("s3vCreateScreen\n"));
-   DEBUG(("sPriv->pDevPriv at %p\n", sPriv->pDevPriv));
-   DEBUG(("size = %i\n", sizeof(*vDRIPriv)));
-
-   if (sPriv->devPrivSize != sizeof(S3VDRIRec)) {
-      fprintf(stderr,"\nERROR!  sizeof(S3VDRIRec) does not match passed size from device driver\n");
-      return GL_FALSE;
-   }
-
-   /* Allocate the private area */
-   s3vScreen = (s3vScreenPtr) CALLOC( sizeof(*s3vScreen) );
-   if ( !s3vScreen ) return NULL;
-
-   s3vScreen->regionCount  = 4;	/* Magic number.  Can we fix this? */
-    
-   s3vScreen->regions = _mesa_malloc(s3vScreen->regionCount * 
-							sizeof(s3vRegion));
-   DEBUG(("sPriv->fd = %i\nvDRIPriv->dmaBufHandle = %x\n",
-      sPriv->fd, vDRIPriv->dmaBufHandle));
-
-   DEBUG(("vDRIPriv->dmaBufSize=%i\nvDRIPriv->dmaBuf=%p\n",
-      vDRIPriv->dmaBufSize, vDRIPriv->dmaBuf));
-
-
-   /* Get the list of dma buffers */
-   s3vScreen->bufs = drmMapBufs(sPriv->fd);
-
-   if (!s3vScreen->bufs) {
-      DEBUG(("Helter/skelter with drmMapBufs\n"));
-      return GL_FALSE; 
-   }
-
-   s3vScreen->textureSize		    = vDRIPriv->texSize;
-   s3vScreen->logTextureGranularity = vDRIPriv->logTextureGranularity;
-   s3vScreen->cpp 					= vDRIPriv->cpp;
-   s3vScreen->frontOffset			= vDRIPriv->frontOffset;
-   s3vScreen->frontPitch			= vDRIPriv->frontPitch;
-   s3vScreen->backOffset			= vDRIPriv->backOffset;
-   s3vScreen->backPitch				= vDRIPriv->frontPitch; /* FIXME: check */
-   s3vScreen->depthOffset			= vDRIPriv->depthOffset;
-   s3vScreen->depthPitch			= vDRIPriv->frontPitch;
-   s3vScreen->texOffset				= vDRIPriv->texOffset;
-
-   s3vScreen->driScreen = sPriv;
-
-   DEBUG(("vDRIPriv->width =%i; vDRIPriv->deviceID =%x\n", vDRIPriv->width,
-		  vDRIPriv->deviceID));
-   DEBUG(("vDRIPriv->mem =%i\n", vDRIPriv->mem));
-   DEBUG(("vDRIPriv->fbOffset =%i\n", vDRIPriv->fbOffset));
-   DEBUG((" ps3vDRI->fbStride =%i\n", vDRIPriv->fbStride));
-   DEBUG(("s3vScreen->cpp = %i\n", s3vScreen->cpp));
-   DEBUG(("s3vScreen->backOffset = %x\n", s3vScreen->backOffset));
-   DEBUG(("s3vScreen->depthOffset = %x\n", s3vScreen->depthOffset));
-   DEBUG(("s3vScreen->texOffset = %x\n", s3vScreen->texOffset));
-   DEBUG(("I will return from s3vCreateScreen now\n"));
-   
-   DEBUG(("s3vScreen->bufs = 0x%x\n", s3vScreen->bufs));
-   return s3vScreen;
-}
-
-/* Destroy the device specific screen private data struct.
- */
-void s3vDestroyScreen( __DRIscreenPrivate *sPriv )
-{
-    s3vScreenPtr s3vScreen = (s3vScreenPtr)sPriv->private;
-
-    DEBUG(("s3vDestroyScreen\n"));
-
-    /* First, unmap the dma buffers */
-/*
-    drmUnmapBufs( s3vScreen->bufs );
-*/
-    /* Next, unmap all the regions */
-/*    while (s3vScreen->regionCount > 0) { 
-
-	(void)drmUnmap(s3vScreen->regions[s3vScreen->regionCount].map,
-		       s3vScreen->regions[s3vScreen->regionCount].size);
-	s3vScreen->regionCount--;
-
-    }
-    FREE(s3vScreen->regions); */
-	if (s3vScreen)
-	    FREE(s3vScreen);
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_screen.h b/src/mesa/drivers/dri/s3v/s3v_screen.h
deleted file mode 100644
index c49bc8587d..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_screen.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "main/mtypes.h"
-
-typedef struct _s3vRegion {
-    drm_handle_t  handle;
-    drmSize    size;
-    drmAddress map;
-} s3vRegion, *s3vRegionPtr;
-
-typedef struct {
-
-    int          regionCount;  	   /* Count of register regions */
-    s3vRegion  	 *regions;         /* Vector of mapped region info */
-
-    drmBufMapPtr bufs;             /* Map of DMA buffers */
-
-    __DRIscreenPrivate *driScreen; /* Back pointer to DRI screen */
-
-    int		cpp;
-    int		frontPitch;
-    int		frontOffset;
-
-    int		backPitch;
-    int		backOffset;
-    int		backX;
-    int		backY;
-
-    int		depthOffset;
-    int		depthPitch;
-
-    int		texOffset;
-    int		textureOffset;
-    int		textureSize;
-    int		logTextureGranularity;
-} s3vScreenRec, *s3vScreenPtr;
-
diff --git a/src/mesa/drivers/dri/s3v/s3v_span.c b/src/mesa/drivers/dri/s3v/s3v_span.c
deleted file mode 100644
index f9f7c0d1ee..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_span.c
+++ /dev/null
@@ -1,219 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-#include "s3v_lock.h"
-
-#include "swrast/swrast.h"
-
-#define _SPANLOCK 1
-#define DBG 0
-
-#define LOCAL_VARS \
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);				\
-	__DRIscreenPrivate *sPriv = vmesa->driScreen;			\
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;		\
-	driRenderbuffer *drb = (driRenderbuffer *) rb;			\
-	GLuint cpp = drb->cpp;						\
-	GLuint pitch = ( (drb->backBuffer) ?				\
-			((dPriv->w+31)&~31) * cpp			\
-			: sPriv->fbWidth * cpp);			\
-	GLuint height = dPriv->h;					\
-	char *buf = (char *)(sPriv->pFB + drb->offset			\
-	   + (drb->backBuffer ?	0 : dPriv->x * cpp + dPriv->y * pitch));\
-	GLuint p; \
-	(void) p
-
-/* FIXME! Depth/Stencil read/writes don't work ! */
-#define LOCAL_DEPTH_VARS					\
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;	\
-	__DRIscreenPrivate *sPriv = vmesa->driScreen;		\
-	driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-	GLuint pitch = drb->pitch;				\
-	GLuint height = dPriv->h;				\
-	char *buf = (char *)(sPriv->pFB + drb->offset);		\
-	(void) pitch
-
-#define LOCAL_STENCIL_VARS	LOCAL_DEPTH_VARS
-
-#define Y_FLIP( _y )	(height - _y - 1)
-
-#if _SPANLOCK	/* OK, we lock */
-
-#define HW_LOCK() \
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);	\
-	(void) vmesa; \
-	DMAFLUSH(); \
-	S3V_SIMPLE_FLUSH_LOCK(vmesa);
-#define HW_UNLOCK() S3V_SIMPLE_UNLOCK(vmesa);
-
-#else			/* plz, don't lock */
-
-#define HW_LOCK() \
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx); \
-    (void) vmesa; \
-	DMAFLUSH(); 
-#define HW_UNLOCK()
-
-#endif
-
-
-/* ================================================================
- * Color buffer
- */
-
-/* 16 bit, RGB565 color spanline and pixel functions
- */
-#define INIT_MONO_PIXEL(p, color) \
-  p = S3VIRGEPACKCOLOR555( color[0], color[1], color[2], color[3] )
-
-#define WRITE_RGBA( _x, _y, r, g, b, a ) \
-do { \
-   *(GLushort *)(buf + _x*2 + _y*pitch) = ((((int)r & 0xf8) << 7) | \
-					   (((int)g & 0xf8) << 2) | \
-					   (((int)b & 0xf8) >> 3)); \
-   DEBUG(("buf=0x%x drawOffset=0x%x dPriv->x=%i drb->cpp=%i dPriv->y=%i pitch=%i\n", \
-   	sPriv->pFB, vmesa->drawOffset, dPriv->x, drb->cpp, dPriv->y, pitch)); \
-   DEBUG(("dPriv->w = %i\n", dPriv->w)); \
-} while(0)
-
-#define WRITE_PIXEL( _x, _y, p ) \
-   *(GLushort *)(buf + _x*2 + _y*pitch) = p
-
-#define READ_RGBA( rgba, _x, _y ) \
-   do { \
-      GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch); \
-      rgba[0] = (p >> 7) & 0xf8; \
-      rgba[1] = (p >> 2) & 0xf8; \
-      rgba[2] = (p << 3) & 0xf8; \
-      rgba[3] = 0xff; /*
-      if ( rgba[0] & 0x08 ) rgba[0] |= 0x07; \ 
-      if ( rgba[1] & 0x04 ) rgba[1] |= 0x03; \
-      if ( rgba[2] & 0x08 ) rgba[2] |= 0x07; */ \
-   } while (0)
-
-#define TAG(x) s3v##x##_RGB555
-#include "spantmp.h"
-
-
-/* 32 bit, ARGB8888 color spanline and pixel functions
- */
-
-#undef INIT_MONO_PIXEL
-#define INIT_MONO_PIXEL(p, color) \
-  p = PACK_COLOR_8888( color[3], color[0], color[1], color[2] )
-
-#define WRITE_RGBA( _x, _y, r, g, b, a ) \
-   *(GLuint *)(buf + _x*4 + _y*pitch) = ((b <<  0) | \
-					 (g <<  8) | \
-					 (r << 16) | \
-					 (a << 24) )
-
-#define WRITE_PIXEL( _x, _y, p ) \
-   *(GLuint *)(buf + _x*4 + _y*pitch) = p
-
-#define READ_RGBA( rgba, _x, _y ) \
-do { \
-   GLuint p = *(GLuint *)(buf + _x*4 + _y*pitch); \
-   rgba[0] = (p >> 16) & 0xff; \
-   rgba[1] = (p >>  8) & 0xff; \
-   rgba[2] = (p >>  0) & 0xff; \
-   rgba[3] = (p >> 24) & 0xff; \
-} while (0)
-
-#define TAG(x) s3v##x##_ARGB8888
-#include "spantmp.h"
-
-
-/* 16 bit depthbuffer functions.
- */
-#define VALUE_TYPE GLushort
-
-#define WRITE_DEPTH( _x, _y, d ) \
-   *(GLushort *)(buf + _x*2 + _y*dPriv->w*2) = d
-
-#define READ_DEPTH( d, _x, _y ) \
-   d = *(GLushort *)(buf + _x*2 + _y*dPriv->w*2);
-
-#define TAG(x) s3v##x##_z16
-#include "depthtmp.h"
-
-
-
-
-/* 32 bit depthbuffer functions.
- */
-#if 0
-#define VALUE_TYPE GLuint
-
-#define WRITE_DEPTH( _x, _y, d )	\
-   *(GLuint *)(buf + _x*4 + _y*pitch) = d;
-
-#define READ_DEPTH( d, _x, _y )		\
-   d = *(GLuint *)(buf + _x*4 + _y*pitch);	
-
-#define TAG(x) s3v##x##_32
-#include "depthtmp.h"
-#endif
-
-
-/* 24/8 bit interleaved depth/stencil functions
- */
-#if 0
-#define VALUE_TYPE GLuint
-
-#define WRITE_DEPTH( _x, _y, d ) { \
-   GLuint tmp = *(GLuint *)(buf + _x*4 + _y*pitch);	\
-   tmp &= 0xff; \
-   tmp |= (d) & 0xffffff00; \
-   *(GLuint *)(buf + _x*4 + _y*pitch) = tmp; \
-}
-
-#define READ_DEPTH( d, _x, _y ) \
-   d = *(GLuint *)(buf + _x*4 + _y*pitch) & ~0xff	
-
-
-#define TAG(x) s3v##x##_24_8
-#include "depthtmp.h"
-
-#define WRITE_STENCIL( _x, _y, d ) { \
-   GLuint tmp = *(GLuint *)(buf + _x*4 + _y*pitch);	\
-   tmp &= 0xffffff00; \
-   tmp |= d & 0xff; \
-   *(GLuint *)(buf + _x*4 + _y*pitch) = tmp; \
-}
-
-#define READ_STENCIL( d, _x, _y ) \
-   d = *(GLuint *)(buf + _x*4 + _y*pitch) & 0xff	
-
-#define TAG(x) s3v##x##_24_8
-#include "stenciltmp.h"
-
-#endif
-
-
-/**
- * Plug in the Get/Put routines for the given driRenderbuffer.
- */
-void
-s3vSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
-{
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         s3vInitPointers_RGB555(&drb->Base);
-      }
-      else {
-         s3vInitPointers_ARGB8888(&drb->Base);
-      }
-   }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
-      s3vInitDepthPointers_z16(&drb->Base);
-   }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
-      /* not done yet */
-   }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
-      /* not done yet */
-   }
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_state.c b/src/mesa/drivers/dri/s3v/s3v_state.c
deleted file mode 100644
index 561f42c705..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_state.c
+++ /dev/null
@@ -1,888 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-#include "s3v_macros.h"
-#include "s3v_dri.h"
-#include "main/macros.h"
-#include "main/colormac.h"
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "vbo/vbo.h"
-#include "tnl/tnl.h"
-
-/* #define DEBUG(str) printf str */
-#define ENABLELIGHTING 0
-
-
-/* =============================================================
- * Alpha blending
- */
-
-static void s3vUpdateAlphaMode( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	uint32_t cmd = vmesa->CMD;
-	cmd &= ~ALPHA_BLEND_MASK;
-
-	if ( ctx->Color.BlendEnabled ) {
-		DEBUG(("ctx->Color.AlphaEnabled = 1"));
-		vmesa->_alpha[0] = ALPHA_SRC;
-		vmesa->_alpha[1] = vmesa->_alpha_tex; /* FIXME: not all tex modes
-							 support alpha */
-	} else {
-		DEBUG(("ctx->Color.AlphaEnabled = 0"));
-		vmesa->_alpha[0] = vmesa->_alpha[1] = ALPHA_OFF;
-	}
-#if 1
-	if ((cmd & DO_MASK) & DO_3D_LINE) { 	/* we are drawing 3d lines */
-						/* which don't support tex */
-		cmd |= vmesa->_alpha[0];
-	} else {
-		cmd |= vmesa->_alpha[vmesa->_3d_mode];
-	}
-
-	vmesa->CMD = cmd; /* FIXME: enough? */
-#else
-	vmesa->restore_primitive = -1;
-#endif
-	
-}
-
-static void s3vDDAlphaFunc( GLcontext *ctx, GLenum func, GLfloat ref )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-   DEBUG(("s3vDDAlphaFunc\n"));
-
-   vmesa->new_state |= S3V_NEW_ALPHA;
-}
-
-static void s3vDDBlendFunc( GLcontext *ctx, GLenum sfactor, GLenum dfactor )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-   DEBUG(("s3vDDBlendFunc\n"));
-
-   vmesa->new_state |= S3V_NEW_ALPHA;
-}
-
-/* ================================================================
- * Buffer clear
- */
-
-static void s3vDDClear( GLcontext *ctx, GLbitfield mask )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	unsigned int _stride;
-        GLint cx = ctx->DrawBuffer->_Xmin;
-        GLint cy = ctx->DrawBuffer->_Ymin;
-        GLint cw = ctx->DrawBuffer->_Xmax - cx;
-        GLint ch = ctx->DrawBuffer->_Ymax - cy;
-
-        /* XXX FIX ME: the cx,cy,cw,ch vars are currently ignored! */
-
-	vmesa->restore_primitive = -1;
-
-	/* Update and emit any new state.  We need to do this here to catch
-	 * changes to the masks.
-	 * FIXME: Just update the masks?
-	 */
-
-	if ( vmesa->new_state )
-		s3vDDUpdateHWState( ctx );
-
-/*	s3vUpdateMasks( ctx ); */
-/* 	s3vUpdateClipping( ctx ); */
-/*	s3vEmitHwState( vmesa ); */
-	
-
-#if 1 /* soft (0)/hw (1)*/
-
-	DEBUG(("*** s3vDDClear ***\n"));
-
-	DMAOUT_CHECK(BITBLT_SRC_BASE, 15);
-		DMAOUT(vmesa->SrcBase);
-		DMAOUT(vmesa->DestBlit);
-		DMAOUT( vmesa->ScissorLR );
-		DMAOUT( vmesa->ScissorTB );
-		DMAOUT( (vmesa->SrcStride << 16) | vmesa->SrcStride );	/* FIXME: unify */
-		DMAOUT( (~(0)) ); /* masks */
-		DMAOUT( (~(0)) );
-		DMAOUT(0);
-		DMAOUT(vmesa->ClearColor);
-		DMAOUT(0);
-		DMAOUT(0);
-		/* FIXME */
-		DMAOUT(0x16000122 | 0x5 | (0xF0 << 17));    /* black magic to me */
-		DMAOUT(vmesa->ScissorWH);
-		DMAOUT(vmesa->SrcXY);
-		DMAOUT(vmesa->DestXY);
-	DMAFINISH();
-
-	if (mask & BUFFER_BIT_DEPTH) { /* depth */
-		DEBUG(("BUFFER_BIT_DEPTH\n"));
-		
-		_stride = ((cw+31)&~31) * 2; /* XXX cw or Buffer->Width??? */
-
-		DMAOUT_CHECK(BITBLT_SRC_BASE, 15);
-	        	DMAOUT(0);
-	        	DMAOUT(vmesa->s3vScreen->depthOffset);
-			DMAOUT( (0 << 16) | cw );
-                	DMAOUT( (0 << 16) | ch );
-			DMAOUT( (vmesa->SrcStride << 16) | vmesa->DestStride );
-	        	DMAOUT( (~(0)) ); /* masks */
-		        DMAOUT( (~(0)) );
-		        DMAOUT(0);
-		        DMAOUT(vmesa->ClearDepth); /* 0x7FFF */
-		        /* FIXME */
-		        DMAOUT(0);
-		        DMAOUT(0);
-		        DMAOUT(0x16000122 | 0x5 | (0xF0 << 17));
-		        DMAOUT( ((cw-1) << 16) | (ch-1) );
-	        	DMAOUT(0);
-		        DMAOUT( (0 << 16) | 0 );
-		DMAFINISH();		
-
-		DEBUG(("vmesa->ClearDepth = 0x%x\n", vmesa->ClearDepth));
-		mask &= ~BUFFER_BIT_DEPTH;
-	}
-
-	if (!vmesa->NotClipped) {
-		DEBUG(("vmesa->NotClipped\n")); /* yes */
-	}
-
-	if (!(vmesa->EnabledFlags & S3V_BACK_BUFFER)) {
-		DEBUG(("!S3V_BACK_BUFFER -> flush\n"));
-		DMAFLUSH();
-	}
-/*
-	if ( mask )
-		DEBUG(("still masked ;3(\n")); */ /* yes */
-#else
-      _swrast_Clear( ctx, mask );
-#endif
-}
-
-/* =============================================================
- * Depth testing
- */
-
-static void s3vUpdateZMode( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	uint32_t cmd = vmesa->CMD;
-
-	DEBUG(("Depth.Test = %i\n", ctx->Depth.Test));
-	DEBUG(("CMD was = 0x%x ", cmd));
-
-/*	printf("depth --- CMD was = 0x%x \n", cmd); */
-
-	cmd &= ~Z_MASK; /*  0xfc0fffff; */
-	/* Z_BUFFER */ /* 000 mode */ /* Z_UPDATE_OFF */
-
-	if (!ctx->Depth.Test)
-		cmd |= Z_OFF;
-
-	if ( ctx->Depth.Mask )
-		cmd |= Z_UPDATE_ON;
-			
-	switch ( ctx->Depth.Func ) {
-		case GL_NEVER:
-			cmd |= Z_NEVER;
-			break;
-		case GL_ALWAYS:
-			cmd |= Z_ALWAYS;
-			break;
-		case GL_LESS:
-			cmd |= Z_LESS;
-			break;
-		case GL_LEQUAL:
-			cmd |= Z_LEQUAL;
-			break;
-		case GL_EQUAL:
-			cmd |= Z_EQUAL;
-			break;
-		case GL_GEQUAL:
-			cmd |= Z_GEQUAL;
-			break;
-		case GL_GREATER:
-			cmd |= Z_GREATER;
-			break;
-		case GL_NOTEQUAL:
-			cmd |= Z_NOTEQUAL;
-			break;
-	}
-
-	DEBUG(("CMD is 0x%x\n", cmd));
-
-	vmesa->dirty |= S3V_UPLOAD_DEPTH;
-	vmesa->CMD = cmd;
-}
-
-static void s3vDDDepthFunc( GLcontext *ctx, GLenum func )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-/*	FLUSH_BATCH( vmesa ); */
-	DEBUG(("s3vDDDepthFunc\n"));
-	vmesa->new_state |= S3V_NEW_DEPTH;
-}
-
-static void s3vDDDepthMask( GLcontext *ctx, GLboolean flag )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	/* FLUSH_BATCH( vmesa ); */
-	DEBUG(("s3vDDDepthMask\n"));
-	vmesa->new_state |= S3V_NEW_DEPTH;
-}
-
-static void s3vDDClearDepth( GLcontext *ctx, GLclampd d )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	switch ( vmesa->DepthSize ) {
-	case 15:
-	case 16:
-		vmesa->ClearDepth = d * 0x0000ffff;	/* 65536 */
-		DEBUG(("GLclampd d = %f\n", d));
-		DEBUG(("ctx->Depth.Clear = %f\n", ctx->Depth.Clear));
-		DEBUG(("(They should be the same)\n"));
-		break;
-	case 24:
-		vmesa->ClearDepth = d * 0x00ffffff;
-		break;
-	case 32:
-		vmesa->ClearDepth = d * 0xffffffff;
-		break;
-   }
-}
-
-static void s3vDDFinish( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	DMAFLUSH(); 
-}
-
-static void s3vDDFlush( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	DMAFLUSH();
-}
-
-/* =============================================================
- * Fog
- */
-
-static void s3vUpdateFogAttrib( GLcontext *ctx )
-{
-/*	s3vContextPtr vmesa = S3V_CONTEXT(ctx); */
-
-	if (ctx->Fog.Enabled) {
-	} else {
-	}
-
-	switch (ctx->Fog.Mode) {
-		case GL_LINEAR:
-			break;
-		case GL_EXP:
-			break;
-		case GL_EXP2:
-			break;
-	}
-}
-
-static void s3vDDFogfv( GLcontext *ctx, GLenum pname, const GLfloat *param )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   vmesa->new_state |= S3V_NEW_FOG;
-}
-
-/* =============================================================
- * Lines
- */
-static void s3vDDLineWidth( GLcontext *ctx, GLfloat width )
-{
-	/* FIXME: on virge you only have one size of 3d lines	 *
-	 * if we wanted more, we should start using tris instead *
-	 * but virge has problem with some tris when all of the  *
-	 * vertices stay on a line */
-}
-
-/* =============================================================
- * Points
- */
-static void s3vDDPointSize( GLcontext *ctx, GLfloat size )
-{
-	/* FIXME: we use 3d line to fake points. So same limitations
-	 * as above apply */
-}
-
-/* =============================================================
- * Polygon 
- */
-
-static void s3vUpdatePolygon( GLcontext *ctx )
-{
-	/* FIXME: I don't think we could do much here */
-
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	vmesa->dirty |= S3V_UPLOAD_POLYGON;
-}
-
-/* =============================================================
- * Clipping
- */
-
-static void s3vUpdateClipping( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-
-	int x0,y0,x1,y1;
-
-	DEBUG((">>> s3vUpdateClipping <<<\n"));
-/*
-	if ( vmesa->driDrawable ) {
-		DEBUG(("s3vUpdateClipping\n"));
-*/
-	if (vmesa->EnabledFlags & S3V_BACK_BUFFER) {
-		DEBUG(("S3V_BACK_BUFFER\n"));
-
-		x0 = 0;
-		y0 = 0;
-		x1 = dPriv->w - 1;
-		y1 = dPriv->h - 1;
-
-		vmesa->SrcBase = 0;
-		vmesa->DestBase =  vmesa->s3vScreen->backOffset;
-		vmesa->DestBlit = vmesa->DestBase;
-	        vmesa->ScissorLR = ( (0 << 16) | (dPriv->w-1) );
-	        vmesa->ScissorTB = ( (0 << 16) | (dPriv->h-1) );
-/*
-	        vmesa->ScissorLR = ( (x0 << 16) | x1 );
-	        vmesa->ScissorTB = ( (y0 << 16) | y1 );
-*/
-	        vmesa->SrcStride = ( ((dPriv->w+31)&~31) * vmesa->s3vScreen->cpp );
-		vmesa->DestStride = vmesa->driScreen->fbWidth*vmesa->s3vScreen->cpp;
-		vmesa->ScissorWH = ( (dPriv->w << 16) | dPriv->h );
-		vmesa->SrcXY = 0;
-/*		vmesa->DestXY = ( (dPriv->x << 16) | dPriv->y );  */
-		vmesa->DestXY = ( (0 << 16) | 0 );
-	} else {
-		DEBUG(("S3V_FRONT_BUFFER\n"));
-
-		x0 = dPriv->x;
-	        y0 = dPriv->y;
-	        x1 = x0 + dPriv->w - 1;
-	        y1 = y0 + dPriv->h - 1;
-
-		vmesa->SrcBase = 0;
-		vmesa->DestBase = 0;
-		vmesa->ScissorLR = ( (x0 << 16) | x1 );
-		vmesa->ScissorTB = ( (y0 << 16) | y1 );
-		vmesa->DestStride = vmesa->driScreen->fbWidth*vmesa->s3vScreen->cpp;
-		vmesa->SrcStride = vmesa->DestStride;
-		vmesa->DestBase = (y0 * vmesa->DestStride)
-				+ x0*vmesa->s3vScreen->cpp;
-		vmesa->DestBlit = 0;
-		vmesa->ScissorWH = ( (x1 << 16) | y1 );
-		vmesa->SrcXY = 0;
-		vmesa->DestXY = ( (0 << 16) | 0 );
-/*		vmesa->DestXY = ( (dPriv->x << 16) | dPriv->y ); */
-    	}
-
-	DEBUG(("x0=%i y0=%i x1=%i y1=%i\n", x0, y0, x1, y1));
-	DEBUG(("stride=%i rectWH=0x%x\n\n", vmesa->DestStride, vmesa->ScissorWH));
-
-	/* FIXME: how could we use the following info? */
-	/* if (ctx->Scissor.Enabled) {} */
-
-	vmesa->dirty |= S3V_UPLOAD_CLIP; 
-/*	}  */
-}
-
-static void s3vDDScissor( GLcontext *ctx,
-			   GLint x, GLint y, GLsizei w, GLsizei h )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	DEBUG((">>> s3vDDScissor <<<"));
-	/* FLUSH_BATCH( vmesa ); */
-	vmesa->new_state |= S3V_NEW_CLIP;
-}
-
-/* =============================================================
- * Culling
- */
-
-static void s3vUpdateCull( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	GLfloat backface_sign = 1;
-
-	DEBUG(("s3vUpdateCull\n"));
-	/* FIXME: GL_FRONT_AND_BACK */
-
-	switch ( ctx->Polygon.CullFaceMode ) {
-	case GL_BACK:
-		if (ctx->Polygon.FrontFace == GL_CCW)
-			backface_sign = -1;
-		break;
-
-	case GL_FRONT:
-		if (ctx->Polygon.FrontFace != GL_CCW)
-			backface_sign = -1;
-		break;
-
-	default:
-		break;
-	}
-
-	vmesa->backface_sign = backface_sign;
-	vmesa->dirty |= S3V_UPLOAD_GEOMETRY;
-}
-
-
-static void s3vDDCullFace( GLcontext *ctx, GLenum mode )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	/* FLUSH_BATCH( vmesa ); */
-	vmesa->new_state |= S3V_NEW_CULL;
-}
-
-static void s3vDDFrontFace( GLcontext *ctx, GLenum mode )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	/* FLUSH_BATCH( vmesa ); */
-	vmesa->new_state |= S3V_NEW_CULL;
-}
-
-/* =============================================================
- * Masks
- */
-
-static void s3vUpdateMasks( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	GLuint mask = s3vPackColor( vmesa->s3vScreen->cpp,
-				ctx->Color.ColorMask[RCOMP],
-				ctx->Color.ColorMask[GCOMP],
-				ctx->Color.ColorMask[BCOMP],
-				ctx->Color.ColorMask[ACOMP] );
-
-	if (vmesa->s3vScreen->cpp == 2) mask |= mask << 16;
-
-	/* FIXME: can we do something in virge? */
-}
-/*
-static void s3vDDColorMask( GLcontext *ctx, GLboolean r, GLboolean g,
-			      GLboolean b, GLboolean a)
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-   FLUSH_BATCH( vmesa );
-   vmesa->new_state |= S3V_NEW_MASKS;
-}
-*/
-/* =============================================================
- * Rendering attributes
- */
-
-/* =============================================================
- * Miscellaneous
- */
-
-static void s3vDDClearColor( GLcontext *ctx, const GLfloat color[4])
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-   DEBUG(("*** s3vDDClearColor\n"));
-
-   vmesa->ClearColor = s3vPackColor( 2, /* vmesa->s3vScreen->cpp, */
-				      color[0], color[1], color[2], color[3] );
-
-#if 0
-   if (vmesa->s3vScreen->cpp == 2) vmesa->ClearColor |= vmesa->ClearColor<<16;
-#endif
-}
-
-static void s3vDDSetDrawBuffer( GLcontext *ctx, GLenum mode )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	int found = GL_TRUE;
-
-	DEBUG(("*** s3vDDSetDrawBuffer ***\n"));
-
-	/* FLUSH_BATCH( vmesa ); */
-
-	switch ( mode ) {
-		case GL_FRONT_LEFT:
-			vmesa->drawOffset = vmesa->s3vScreen->frontOffset;
-			break;
-		case GL_BACK_LEFT:
-			vmesa->drawOffset = vmesa->s3vScreen->backOffset;
-			/* vmesa->driScreen->fbHeight *
-			 * vmesa->driScreen->fbWidth *
-			 * vmesa->s3vScreen->cpp; */
-			break;
-		default:
-			found = GL_FALSE;
-			break;
-	}
-
-	DEBUG(("vmesa->drawOffset = 0x%x\n", vmesa->drawOffset));
-/*	return GL_TRUE; */
-}
-
-/* =============================================================
- * Window position and viewport transformation
- */
-
-void s3vUpdateWindow( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-	GLfloat xoffset = (GLfloat)dPriv->x;
-	GLfloat yoffset =
-		vmesa->driScreen->fbHeight - (GLfloat)dPriv->y - dPriv->h;
-	const GLfloat *v = ctx->Viewport._WindowMap.m;
-
-	GLfloat sx = v[MAT_SX];
-	GLfloat tx = v[MAT_TX] + xoffset;
-	GLfloat sy = v[MAT_SY];
-	GLfloat ty = v[MAT_TY] + yoffset;
-	GLfloat sz = v[MAT_SZ] * vmesa->depth_scale;
-	GLfloat tz = v[MAT_TZ] * vmesa->depth_scale;
-
-	vmesa->dirty |= S3V_UPLOAD_VIEWPORT;
-
-	vmesa->ViewportScaleX = sx;
-	vmesa->ViewportScaleY = sy;
-	vmesa->ViewportScaleZ = sz;
-	vmesa->ViewportOffsetX = tx;
-	vmesa->ViewportOffsetY = ty;
-	vmesa->ViewportOffsetZ = tz;
-}
-
-
-/*
-static void s3vDDViewport( GLcontext *ctx, GLint x, GLint y,
-			    GLsizei width, GLsizei height )
-{
-	s3vUpdateWindow( ctx );
-}
-
-static void s3vDDDepthRange( GLcontext *ctx, GLclampd nearval,
-			      GLclampd farval )
-{
-	s3vUpdateWindow( ctx );
-}
-*/
-void s3vUpdateViewportOffset( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-	GLfloat xoffset = (GLfloat)dPriv->x;
-	GLfloat yoffset =
-		vmesa->driScreen->fbHeight - (GLfloat)dPriv->y - dPriv->h;
-	const GLfloat *v = ctx->Viewport._WindowMap.m;
-
-	GLfloat tx = v[MAT_TX] + xoffset;
-	GLfloat ty = v[MAT_TY] + yoffset;
-
-	DEBUG(("*** s3vUpdateViewportOffset ***\n"));
-
-	if ( vmesa->ViewportOffsetX != tx ||
-		vmesa->ViewportOffsetY != ty )
-	{
-		vmesa->ViewportOffsetX = tx;
-		vmesa->ViewportOffsetY = ty;
-
-		vmesa->new_state |= S3V_NEW_WINDOW;
-	}
-
-/*	vmesa->new_state |= S3V_NEW_CLIP; */
-}
-
-/* =============================================================
- * State enable/disable
- */
-
-static void s3vDDEnable( GLcontext *ctx, GLenum cap, GLboolean state )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	switch ( cap ) {
-	case GL_ALPHA_TEST:
-	case GL_BLEND:
-		vmesa->new_state |= S3V_NEW_ALPHA;
-		DEBUG(("s3vDDEnable: GL_BLEND\n"));
-	break;
-
-	case GL_CULL_FACE:
-		vmesa->new_state |= S3V_NEW_CULL;
-		DEBUG(("s3vDDEnable: GL_CULL_FACE\n"));
-	break;
-
-	case GL_DEPTH_TEST:
-		vmesa->new_state |= S3V_NEW_DEPTH;
-		DEBUG(("s3vDDEnable: GL_DEPTH\n"));
-	break;
-#if 0
-	case GL_FOG:
-		vmesa->new_state |= S3V_NEW_FOG;
-	break;
-#endif
-
-	case GL_SCISSOR_TEST:
-		vmesa->new_state |= S3V_NEW_CLIP;
-	break;
-
-	case GL_TEXTURE_2D:
-	   	DEBUG(("*** GL_TEXTURE_2D: %i\n", state));
-		vmesa->_3d_mode = state;
-		vmesa->restore_primitive = -1;
-	break;
-	
-	default:
-		return;
-	}
-}
-
-/* =============================================================
- * State initialization, management
- */
-
-
-/*
- * Load the current context's state into the hardware.
- *
- * NOTE: Be VERY careful about ensuring the context state is marked for
- * upload, the only place it shouldn't be uploaded is when the setup
- * state has changed in ReducedPrimitiveChange as this comes right after
- * a state update.
- *
- * Blits of any type should always upload the context and masks after
- * they are done.
- */
-void s3vEmitHwState( s3vContextPtr vmesa )
-{
-    if (!vmesa->driDrawable) return;
-    if (!vmesa->dirty) return; 
-
-	DEBUG(("**********************\n"));
-	DEBUG(("*** s3vEmitHwState ***\n"));
-	DEBUG(("**********************\n"));
-
-    if (vmesa->dirty & S3V_UPLOAD_VIEWPORT) {
-	vmesa->dirty &= ~S3V_UPLOAD_VIEWPORT;
-	DEBUG(("S3V_UPLOAD_VIEWPORT\n"));
-    }
-   
-    if ( (vmesa->dirty & S3V_UPLOAD_POINTMODE) ||
-	 (vmesa->dirty & S3V_UPLOAD_LINEMODE) ||
-	 (vmesa->dirty & S3V_UPLOAD_TRIMODE) ) {
-
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_POINTMODE) {
-	vmesa->dirty &= ~S3V_UPLOAD_POINTMODE;
-    }
-
-    if (vmesa->dirty & S3V_UPLOAD_LINEMODE) {
-	vmesa->dirty &= ~S3V_UPLOAD_LINEMODE;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_TRIMODE) {
-	vmesa->dirty &= ~S3V_UPLOAD_TRIMODE;
-    }
-
-    if (vmesa->dirty & S3V_UPLOAD_FOG) {
-	GLchan c[3], col;
-   	UNCLAMPED_FLOAT_TO_RGB_CHAN( c, vmesa->glCtx->Fog.Color );
-	DEBUG(("uploading ** FOG **\n"));
-	col = s3vPackColor(2, c[0], c[1], c[2], 0);
-	vmesa->dirty &= ~S3V_UPLOAD_FOG;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_DITHER) {
-	vmesa->dirty &= ~S3V_UPLOAD_DITHER;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_LOGICOP) {
-	vmesa->dirty &= ~S3V_UPLOAD_LOGICOP;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_CLIP) {
-	vmesa->dirty &= ~S3V_UPLOAD_CLIP;
-	DEBUG(("S3V_UPLOAD_CLIP\n"));
-	DEBUG(("vmesa->ScissorLR: %i\n",  vmesa->ScissorLR));
-	DEBUG(("vmesa->ScissorTB: %i\n",  vmesa->ScissorTB));
-    }
-
-    if (vmesa->dirty & S3V_UPLOAD_MASKS) {
-	vmesa->dirty &= ~S3V_UPLOAD_MASKS;
-	DEBUG(("S3V_UPLOAD_BLEND\n"));
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_ALPHA) {
-	vmesa->dirty &= ~S3V_UPLOAD_ALPHA;
-	DEBUG(("S3V_UPLOAD_ALPHA\n"));
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_SHADE) {
-	vmesa->dirty &= ~S3V_UPLOAD_SHADE;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_POLYGON) {
-	vmesa->dirty &= ~S3V_UPLOAD_POLYGON;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_DEPTH) {
-	vmesa->dirty &= ~S3V_UPLOAD_DEPTH;
-	DEBUG(("S3V_UPLOAD_DEPTH: DepthMode = 0x%x08\n", vmesa->DepthMode));
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_GEOMETRY) {
-	vmesa->dirty &= ~S3V_UPLOAD_GEOMETRY;
-    }
-
-    if (vmesa->dirty & S3V_UPLOAD_TRANSFORM) {
-	vmesa->dirty &= ~S3V_UPLOAD_TRANSFORM;
-    }
-    
-    if (vmesa->dirty & S3V_UPLOAD_TEX0) {
-	s3vTextureObjectPtr curTex = vmesa->CurrentTexObj[0];
-	vmesa->dirty &= ~S3V_UPLOAD_TEX0;
-	DEBUG(("S3V_UPLOAD_TEX0\n"));
-	if (curTex) {
-		DEBUG(("S3V_UPLOAD_TEX0: curTex\n"));
-	} else {
-		DEBUG(("S3V_UPLOAD_TEX0: !curTex\n"));
-	}
-    }
-}
-
-void s3vDDUpdateHWState( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	int new_state = vmesa->new_state;
-
-	/* s3vUpdateClipping( ctx ); */
-
-	if ( new_state )
-	{
-
-		vmesa->new_state = 0;
-
-		/* Update the various parts of the context's state.
-		 */
-		if ( new_state & S3V_NEW_ALPHA )
-			s3vUpdateAlphaMode( ctx );
-
-		if ( new_state & S3V_NEW_DEPTH )
-			s3vUpdateZMode( ctx );
-
-		if ( new_state & S3V_NEW_FOG )
-			s3vUpdateFogAttrib( ctx );
-
-		if ( new_state & S3V_NEW_CLIP )
-		{
-			DEBUG(("---> going to s3vUpdateClipping\n"));
-			s3vUpdateClipping( ctx );
-		}
-
-		if ( new_state & S3V_NEW_POLYGON )
-			s3vUpdatePolygon( ctx );
-
-		if ( new_state & S3V_NEW_CULL )
-			s3vUpdateCull( ctx );
-
-		if ( new_state & S3V_NEW_MASKS )
-			s3vUpdateMasks( ctx );
-
-		if ( new_state & S3V_NEW_WINDOW )
-			s3vUpdateWindow( ctx );
-/*
-		if ( new_state & S3_NEW_TEXTURE )
-			s3vUpdateTextureState( ctx );		
-*/
-		CMDCHANGE();
-	}
-
-	/* HACK ! */
-	s3vEmitHwState( vmesa );
-}
-
-
-static void s3vDDUpdateState( GLcontext *ctx, GLuint new_state )
-{
-	_swrast_InvalidateState( ctx, new_state );
-	_swsetup_InvalidateState( ctx, new_state );
-	_vbo_InvalidateState( ctx, new_state );
-	_tnl_InvalidateState( ctx, new_state );
-	S3V_CONTEXT(ctx)->new_gl_state |= new_state;
-}
-
-
-/* Initialize the context's hardware state.
- */
-void s3vInitState( s3vContextPtr vmesa )
-{
-	vmesa->new_state = 0;
-}
-
-/* Initialize the driver's state functions.
- */
-void s3vInitStateFuncs( GLcontext *ctx )
-{
-   ctx->Driver.UpdateState		= s3vDDUpdateState;
-
-   ctx->Driver.Clear			= s3vDDClear;
-   ctx->Driver.ClearIndex		= NULL;
-   ctx->Driver.ClearColor		= s3vDDClearColor;
-   ctx->Driver.DrawBuffer		= s3vDDSetDrawBuffer; 
-   ctx->Driver.ReadBuffer               = NULL; /* XXX */
-
-   ctx->Driver.IndexMask		= NULL;
-   ctx->Driver.ColorMask		= NULL; /* s3vDDColorMask; */ /* FIXME */
-
-   ctx->Driver.AlphaFunc		= s3vDDAlphaFunc; /* FIXME */
-#if 0
-   ctx->Driver.BlendEquation		= NULL; /* s3vDDBlendEquation; */
-   ctx->Driver.BlendFunc		= s3vDDBlendFunc; /* FIXME */
-#endif
-   ctx->Driver.BlendFuncSeparate	= NULL; /* s3vDDBlendFuncSeparate; */
-   ctx->Driver.ClearDepth		= s3vDDClearDepth;
-   ctx->Driver.CullFace			= s3vDDCullFace; 
-   ctx->Driver.FrontFace		= s3vDDFrontFace;
-   ctx->Driver.DepthFunc		= s3vDDDepthFunc;	/* FIXME */
-   ctx->Driver.DepthMask		= s3vDDDepthMask;	/* FIXME */
-   ctx->Driver.DepthRange		= NULL; /* s3vDDDepthRange; */
-   ctx->Driver.Enable			= s3vDDEnable;		/* FIXME */
-   ctx->Driver.Finish			= s3vDDFinish;
-   ctx->Driver.Flush			= s3vDDFlush;
-#if 1
-   ctx->Driver.Fogfv			= NULL; /* s3vDDFogfv; */
-#endif
-   ctx->Driver.Hint			= NULL;
-   ctx->Driver.LineWidth		= NULL; /* s3vDDLineWidth; */
-   ctx->Driver.LineStipple		= NULL; /* s3vDDLineStipple; */
-#if ENABLELIGHTING
-   ctx->Driver.Lightfv			= NULL; /* s3vDDLightfv; */
-
-   ctx->Driver.LightModelfv		= NULL; /* s3vDDLightModelfv; */
-#endif
-   ctx->Driver.LogicOpcode		= NULL; /* s3vDDLogicalOpcode; */
-   ctx->Driver.PointSize		= NULL; /* s3vDDPointSize; */
-   ctx->Driver.PolygonMode		= NULL; /* s3vDDPolygonMode; */
-   ctx->Driver.PolygonStipple		= NULL; /* s3vDDPolygonStipple; */
-   ctx->Driver.Scissor			= s3vDDScissor; /* ScissorLR / ScissorTB */
-   ctx->Driver.ShadeModel		= NULL; /* s3vDDShadeModel; */
-   ctx->Driver.Viewport			= NULL; /* s3vDDViewport; */
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_tex.c b/src/mesa/drivers/dri/s3v/s3v_tex.c
deleted file mode 100644
index 9b92519862..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_tex.c
+++ /dev/null
@@ -1,559 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include <stdlib.h>
-#include <stdio.h>
-
-#include "main/glheader.h"
-#include "main/mtypes.h"
-#include "main/simple_list.h"
-#include "main/enums.h"
-#include "main/mm.h"
-#include "main/texstore.h"
-#include "main/texformat.h"
-#include "main/teximage.h"
-#include "swrast/swrast.h"
-
-#include "s3v_context.h"
-#include "s3v_tex.h"
-
-
-extern void s3vSwapOutTexObj(s3vContextPtr vmesa, s3vTextureObjectPtr t);
-extern void s3vDestroyTexObj(s3vContextPtr vmesa, s3vTextureObjectPtr t);
-
-/*
-static GLuint s3vComputeLodBias(GLfloat bias)
-{
-#if TEX_DEBUG_ON
-	DEBUG_TEX(("*** s3vComputeLodBias ***\n"));
-#endif
-	return bias;
-}
-*/
-
-static void s3vSetTexWrapping(s3vContextPtr vmesa,
-                               s3vTextureObjectPtr t, 
-			       GLenum wraps, GLenum wrapt)
-{
-	GLuint t0 = t->TextureCMD;
-	GLuint cmd = vmesa->CMD;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vSetTexWrapping: #%i ***\n", ++times));
-#endif
-
-
-	t0 &= ~TEX_WRAP_MASK;
-	cmd &= ~TEX_WRAP_MASK;
-
-	if ((wraps != GL_CLAMP) || (wrapt != GL_CLAMP)) {
-		DEBUG(("TEX_WRAP_ON\n"));
-		t0 |= TEX_WRAP_ON;
-		cmd |= TEX_WRAP_ON; 
-	}
-
-	cmd |= TEX_WRAP_ON; /* FIXME: broken if off */
-	t->TextureCMD = t0;
-	vmesa->CMD = cmd;
-}
-
-
-static void s3vSetTexFilter(s3vContextPtr vmesa, 
-			     s3vTextureObjectPtr t, 
-			     GLenum minf, GLenum magf)
-{
-	GLuint t0 = t->TextureCMD;
-	GLuint cmd = vmesa->CMD;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vSetTexFilter: #%i ***\n", ++times));
-#endif
-
-	t0 &= ~TEX_FILTER_MASK;
-	cmd &= ~TEX_FILTER_MASK;
-
-	switch (minf) {
-		case GL_NEAREST:
-			DEBUG(("GL_NEAREST\n"));
-			t0 |= NEAREST;
-			cmd |= NEAREST;
-			break;
-		case GL_LINEAR:
-			DEBUG(("GL_LINEAR\n"));
-			t0 |= LINEAR;
-			cmd |= LINEAR;
-			break;
-		case GL_NEAREST_MIPMAP_NEAREST:
-			DEBUG(("GL_MIPMAP_NEAREST\n"));
-			t0 |= MIP_NEAREST;
-			cmd |= MIP_NEAREST;
-			break;
-		case GL_LINEAR_MIPMAP_NEAREST:
-			DEBUG(("GL_LINEAR_MIPMAP_NEAREST\n"));
-			t0 |= LINEAR_MIP_NEAREST;
-			cmd |= LINEAR_MIP_NEAREST;
-			break;
-		case GL_NEAREST_MIPMAP_LINEAR:
-			DEBUG(("GL_NEAREST_MIPMAP_LINEAR\n"));
-			t0 |= MIP_LINEAR;
-			cmd |= MIP_LINEAR;
-			break;
-		case GL_LINEAR_MIPMAP_LINEAR:
-			DEBUG(("GL_LINEAR_MIPMAP_LINEAR\n"));
-			t0 |= LINEAR_MIP_LINEAR;
-			cmd |= LINEAR_MIP_LINEAR;
-			break;
-		default:
-			break;
-	}
-	/* FIXME: bilinear? */
-
-#if 0
-	switch (magf) {
-		case GL_NEAREST:
-			break;
-		case GL_LINEAR:
-			break;
-		default:
-			break;
-	}  
-#endif
-
-	t->TextureCMD = t0;
-
-	DEBUG(("CMD was = 0x%x\n", vmesa->CMD));
-	DEBUG(("CMD is = 0x%x\n", cmd));
-
-	vmesa->CMD = cmd; 
-	/* CMDCHANGE(); */
-}
-
-
-static void s3vSetTexBorderColor(s3vContextPtr vmesa,
-				  s3vTextureObjectPtr t, 
-				  const GLfloat color[4])
-{
-	GLubyte c[4];
-	CLAMPED_FLOAT_TO_UBYTE(c[0], color[0]);
-	CLAMPED_FLOAT_TO_UBYTE(c[1], color[1]);
-	CLAMPED_FLOAT_TO_UBYTE(c[2], color[2]);
-	CLAMPED_FLOAT_TO_UBYTE(c[3], color[3]);
-
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vSetTexBorderColor: #%i ***\n", ++times));
-#endif
-
-	/*FIXME: it should depend on tex col format */
-	/* switch(t0 ... t->TextureColorMode) */
-
-	/* case TEX_COL_ARGB1555: */
-	t->TextureBorderColor =	S3VIRGEPACKCOLOR555(c[0], c[1], c[2], c[3]);
-
-	DEBUG(("TextureBorderColor = 0x%x\n", t->TextureBorderColor));
-
-	vmesa->TextureBorderColor = t->TextureBorderColor;
-}
-
-static void s3vTexParameter( GLcontext *ctx, GLenum target,
-			      struct gl_texture_object *tObj,
-			      GLenum pname, const GLfloat *params )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr) tObj->DriverData;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vTexParameter: #%i ***\n", ++times));
-#endif
-   
-	if (!t) return;
-
-	/* Can't do the update now as we don't know whether to flush
-	 * vertices or not.  Setting vmesa->new_state means that
-	 * s3vUpdateTextureState() will be called before any triangles are
-	 * rendered.  If a statechange has occurred, it will be detected at
-	 * that point, and buffered vertices flushed.  
-	*/
-	switch (pname) {
-	case GL_TEXTURE_MIN_FILTER:
-	case GL_TEXTURE_MAG_FILTER:
-		s3vSetTexFilter( vmesa, t, tObj->MinFilter, tObj->MagFilter );
-		break;
-
-	case GL_TEXTURE_WRAP_S:
-	case GL_TEXTURE_WRAP_T:
-		s3vSetTexWrapping( vmesa, t, tObj->WrapS, tObj->WrapT );
-		break;
-  
-	case GL_TEXTURE_BORDER_COLOR:
-		s3vSetTexBorderColor( vmesa, t, tObj->BorderColor );
-		break;
-
-	case GL_TEXTURE_BASE_LEVEL:
-	case GL_TEXTURE_MAX_LEVEL:
-	case GL_TEXTURE_MIN_LOD:
-	case GL_TEXTURE_MAX_LOD:
-	/* This isn't the most efficient solution but there doesn't appear to
-	 * be a nice alternative for Virge.  Since there's no LOD clamping,
-	 * we just have to rely on loading the right subset of mipmap levels
-	 * to simulate a clamped LOD.
-	 */
-		s3vSwapOutTexObj( vmesa, t ); 
-		break;
-
-	default:
-		return;
-	}
-
-	if (t == vmesa->CurrentTexObj[0])
-		vmesa->dirty |= S3V_UPLOAD_TEX0;
-
-#if 0
-	if (t == vmesa->CurrentTexObj[1]) {
-		vmesa->dirty |= S3V_UPLOAD_TEX1;
-	}
-#endif
-}
-
-
-static void s3vTexEnv( GLcontext *ctx, GLenum target, 
-			GLenum pname, const GLfloat *param )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	GLuint unit = ctx->Texture.CurrentUnit;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vTexEnv: #%i ***\n", ++times));
-#endif
-
-	/* Only one env color.  Need a fallback if env colors are different
-	 * and texture setup references env color in both units.  
-	 */
-	switch (pname) {
-	case GL_TEXTURE_ENV_COLOR: {
-		struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-		GLfloat *fc = texUnit->EnvColor;
-		GLuint r, g, b, a, col;
-		CLAMPED_FLOAT_TO_UBYTE(r, fc[0]);
-		CLAMPED_FLOAT_TO_UBYTE(g, fc[1]);
-		CLAMPED_FLOAT_TO_UBYTE(b, fc[2]);
-		CLAMPED_FLOAT_TO_UBYTE(a, fc[3]);
-
-		col =  ((a << 24) | 
-			(r << 16) | 
-			(g <<  8) | 
-			(b <<  0));
-
-		break;
-	}
-	case GL_TEXTURE_ENV_MODE:
-		vmesa->TexEnvImageFmt[unit] = 0; /* force recalc of env state */
-		break;
-	case GL_TEXTURE_LOD_BIAS_EXT: {
-/*
-		struct gl_texture_object *tObj =
-			ctx->Texture.Unit[unit]._Current;
-
-		s3vTextureObjectPtr t = (s3vTextureObjectPtr) tObj->DriverData;
-*/
-		break;
-	}
-	default:
-		break;
-	}
-} 
-
-static void s3vTexImage1D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint border,
-			    GLenum format, GLenum type, 
-			    const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *pack,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr) texObj->DriverData;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vTexImage1D: #%i ***\n", ++times));
-#endif
-
-#if 1 
-	if (t) {
-#if _TEXFLUSH
-		DMAFLUSH();
-#endif
-		s3vSwapOutTexObj( vmesa, t );
-/*
-		s3vDestroyTexObj( vmesa, t );
-		texObj->DriverData = 0;
-*/
-	}
-#endif
-	_mesa_store_teximage1d( ctx, target, level, internalFormat,
-				width, border, format, type,
-				pixels, pack, texObj, texImage );
-}
-
-static void s3vTexSubImage1D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset,
-			       GLsizei width,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *pack,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr) texObj->DriverData;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vTexSubImage1D: #%i ***\n", ++times));
-#endif
-
-#if 1
-	if (t) {
-#if _TEXFLUSH
-		DMAFLUSH();
-#endif
-		s3vSwapOutTexObj( vmesa, t );
-/*
-		s3vDestroyTexObj( vmesa, t );
-		texObj->DriverData = 0;
-*/
-	}
-#endif
-	_mesa_store_texsubimage1d(ctx, target, level, xoffset, width, 
-				format, type, pixels, pack, texObj,
-				texImage);
-}
-
-static void s3vTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint height, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr) texObj->DriverData;
-
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vTexImage2D: #%i ***\n", ++times));
-#endif
-
-#if 1 
-	if (t) {
-#if _TEXFLUSH
-		DMAFLUSH();
-#endif
-		s3vSwapOutTexObj( vmesa, t ); 
-/*
-		s3vDestroyTexObj( vmesa, t );
-		texObj->DriverData = 0;
-*/
-	}
-#endif
-	_mesa_store_teximage2d( ctx, target, level, internalFormat,
-				width, height, border, format, type,
-				pixels, packing, texObj, texImage );
-}
-
-static void s3vTexSubImage2D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset, GLint yoffset,
-			       GLsizei width, GLsizei height,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr) texObj->DriverData;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vTexSubImage2D: #%i ***\n", ++times));
-#endif
-
-#if 1
-	if (t) {
-#if _TEXFLUSH
-		DMAFLUSH();
-#endif
-		s3vSwapOutTexObj( vmesa, t );
-/* 
-		s3vDestroyTexObj( vmesa, t );
-		texObj->DriverData = 0;
-*/
-	}
-#endif
-	_mesa_store_texsubimage2d(ctx, target, level, xoffset, yoffset, width, 
-				height, format, type, pixels, packing, texObj,
-				texImage);
-}
-
-
-static void s3vBindTexture( GLcontext *ctx, GLenum target,
-			     struct gl_texture_object *tObj )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr) tObj->DriverData;
-	GLuint cmd = vmesa->CMD;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vBindTexture: #%i ***\n", ++times));
-#endif
-
-	if (!t) {
-/*
-		GLfloat bias = ctx->Texture.Unit[ctx->Texture.CurrentUnit].LodBias;
-*/
-		t = CALLOC_STRUCT(s3v_texture_object_t);
-
-		/* Initialize non-image-dependent parts of the state:
-		 */
-		t->globj = tObj;
-#if 0
-		if (target == GL_TEXTURE_2D) {
-		} else
-		if (target == GL_TEXTURE_1D) {
-		}
-
-#if X_BYTE_ORDER == X_LITTLE_ENDIAN
-		t->TextureFormat = (TF_LittleEndian |
-#else
-		t->TextureFormat = (TF_BigEndian |
-#endif
-#endif
-		t->dirty_images = ~0;
-
-		tObj->DriverData = t;
-		make_empty_list( t );
-#if 0
-		s3vSetTexWrapping( vmesa, t, tObj->WrapS, tObj->WrapT );
-		s3vSetTexFilter( vmesa, t, tObj->MinFilter, tObj->MagFilter );
-		s3vSetTexBorderColor( vmesa, t, tObj->BorderColor );
-#endif
-	}
-
-	cmd = vmesa->CMD & ~MIP_MASK;
-    vmesa->dirty |= S3V_UPLOAD_TEX0;
-    vmesa->TexOffset = t->TextureBaseAddr[tObj->BaseLevel];
-    vmesa->TexStride = t->Pitch;
-    cmd |= MIPMAP_LEVEL(t->WidthLog2);
-	vmesa->CMD = cmd;
-	vmesa->restore_primitive = -1;
-#if 0
-	printf("t->TextureBaseAddr[0] = 0x%x\n", t->TextureBaseAddr[0]);
-	printf("t->TextureBaseAddr[1] = 0x%x\n", t->TextureBaseAddr[1]);
-	printf("t->TextureBaseAddr[2] = 0x%x\n", t->TextureBaseAddr[2]);
-#endif
-}
-
-
-static void s3vDeleteTexture( GLcontext *ctx, struct gl_texture_object *tObj )
-{
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr)tObj->DriverData;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vDeleteTexture: #%i ***\n", ++times));
-#endif
-
-	if (t) {
-		s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-
-#if _TEXFLUSH
-		if (vmesa) {
-			DMAFLUSH();
-		}
-#endif
-
-		s3vDestroyTexObj( vmesa, t );
-		tObj->DriverData = 0;
-
-	}
-}
-
-static GLboolean s3vIsTextureResident( GLcontext *ctx, 
-					struct gl_texture_object *tObj )
-{
-	s3vTextureObjectPtr t = (s3vTextureObjectPtr)tObj->DriverData;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vIsTextureResident: #%i ***\n", ++times));
-#endif
-   
-	return (t && t->MemBlock);
-}
-
-static void s3vInitTextureObjects( GLcontext *ctx )
-{
-	/* s3vContextPtr vmesa = S3V_CONTEXT(ctx); */
-	struct gl_texture_object *texObj;
-	GLuint tmp = ctx->Texture.CurrentUnit;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vInitTextureObjects: #%i ***\n", ++times));
-#endif
-
-#if 1
-	ctx->Texture.CurrentUnit = 0;
-
-	texObj = ctx->Texture.Unit[0].CurrentTex[TEXTURE_1D_INDEX];
-	s3vBindTexture( ctx, GL_TEXTURE_1D, texObj );
-
-	texObj = ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];
-	s3vBindTexture( ctx, GL_TEXTURE_2D, texObj );
-#endif
-
-#if 0
-	ctx->Texture.CurrentUnit = 1;
-
-	texObj = ctx->Texture.Unit[1].CurrentTex[TEXTURE_1D_INDEX];
-	s3vBindTexture( ctx, GL_TEXTURE_1D, texObj );
-
-	texObj = ctx->Texture.Unit[1].CurrentTex[TEXTURE_2D_INDEX];
-	s3vBindTexture( ctx, GL_TEXTURE_2D, texObj );
-#endif
-
-	ctx->Texture.CurrentUnit = tmp;
-}
-
-
-void s3vInitTextureFuncs( GLcontext *ctx )
-{
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	DEBUG_TEX(("*** s3vInitTextureFuncs: #%i ***\n", ++times));
-#endif
-
-	ctx->Driver.TexEnv = s3vTexEnv;
-	ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format;
-	ctx->Driver.TexImage1D = _mesa_store_teximage1d;
-	ctx->Driver.TexImage2D = s3vTexImage2D;
-	ctx->Driver.TexImage3D = _mesa_store_teximage3d;
-	ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d;
-	ctx->Driver.TexSubImage2D = s3vTexSubImage2D;
-	ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d;
-	ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d;
-	ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d;
-	ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d;
-	ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d;
-	ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d;
-	ctx->Driver.BindTexture = s3vBindTexture;
-	ctx->Driver.DeleteTexture = s3vDeleteTexture;
-	ctx->Driver.TexParameter = s3vTexParameter;
-	ctx->Driver.UpdateTexturePalette = 0;
-	ctx->Driver.IsTextureResident = s3vIsTextureResident;
-	ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage;
-
-	s3vInitTextureObjects( ctx );
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_tex.h b/src/mesa/drivers/dri/s3v/s3v_tex.h
deleted file mode 100644
index a823fe2453..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_tex.h
+++ /dev/null
@@ -1,28 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef _S3V_TEX_H
-#define _S3V_TEX_H
-
-#define TEX_DEBUG_ON 0
-
-extern void s3vUpdateTexLRU( s3vContextPtr vmesa, s3vTextureObjectPtr t );
-
-#if TEX_DEBUG_ON
-#define DEBUG_TEX(str) printf str
-#else
-#define DEBUG_TEX(str) /* str */
-#endif
-
-#define _TEXFLUSH 1 /* flush before uploading */
-#define _TEXLOCK  1 /* lock before writing new texures to card mem */
-					/* 	if you turn it on you will gain stability and image
-						quality, but you will loose performance (~10%) */
-#define _TEXFALLBACK 0 /* fallback to software for -big- textures (slow) */
-					/* turning this off, you will lose some tex (e.g. mountains
-					   on tuxracer) but you will increase average playability */
-
-#define _TEXALIGN 0x00000007
-
-#endif
diff --git a/src/mesa/drivers/dri/s3v/s3v_texmem.c b/src/mesa/drivers/dri/s3v/s3v_texmem.c
deleted file mode 100644
index 705d105f55..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_texmem.c
+++ /dev/null
@@ -1,582 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include <stdlib.h>
-#include <stdio.h>
-
-#include "main/glheader.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-#include "main/simple_list.h"
-#include "main/enums.h"
-
-#include "main/mm.h"
-#include "s3v_context.h"
-#include "s3v_lock.h"
-#include "s3v_tex.h"
-
-void s3vSwapOutTexObj(s3vContextPtr vmesa, s3vTextureObjectPtr t);
-void s3vUpdateTexLRU( s3vContextPtr vmesa, s3vTextureObjectPtr t );
-
-
-void s3vDestroyTexObj(s3vContextPtr vmesa, s3vTextureObjectPtr t)
-{
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vDestroyTexObj: #%i ***\n", ++times));
-#endif
-
-   if (!t) return;
-
-/* FIXME: useful? */
-#if _TEXFLUSH
-	if (vmesa)
-		DMAFLUSH();
-#endif
-
-   /* This is sad - need to sync *in case* we upload a texture
-    * to this newly free memory...
-    */
-   if (t->MemBlock) {
-      mmFreeMem(t->MemBlock);
-      t->MemBlock = 0;
-
-      if (vmesa && t->age > vmesa->dirtyAge)
-	     vmesa->dirtyAge = t->age;
-   }
-
-   if (t->globj)
-      t->globj->DriverData = NULL;
-
-   if (vmesa) {
-      if (vmesa->CurrentTexObj[0] == t) {
-         	vmesa->CurrentTexObj[0] = 0;
-        	vmesa->dirty &= ~S3V_UPLOAD_TEX0;
-      }
-
-#if 0
-      if (vmesa->CurrentTexObj[1] == t) {
-         vmesa->CurrentTexObj[1] = 0;
-         vmesa->dirty &= ~S3V_UPLOAD_TEX1;
-      }
-#endif
-   }
-
-   remove_from_list(t);
-   FREE(t);
-}
-
-
-void s3vSwapOutTexObj(s3vContextPtr vmesa, s3vTextureObjectPtr t)
-{
-/*   int i; */
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vSwapOutTexObj: #%i ***\n", ++times));
-#endif
-
-   if (t->MemBlock) {
-
-      mmFreeMem(t->MemBlock);
-      t->MemBlock = 0;
-
-      if (t->age > vmesa->dirtyAge)
-         vmesa->dirtyAge = t->age;
-   
-      t->dirty_images = ~0; 
-      move_to_tail(&(vmesa->SwappedOut), t);
-   }
-}
-
-
-/* Upload an image from mesa's internal copy.
- */
-
-static void s3vUploadTexLevel( s3vContextPtr vmesa, s3vTextureObjectPtr t,
-				int level )
-{
-	__DRIscreenPrivate *sPriv = vmesa->driScreen;
-	const struct gl_texture_image *image = t->image[level].image;
-	int i,j;
-	int l2d;
-	/* int offset = 0; */
-	int words;
-	GLuint* dest;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-#endif
-	if ( !image ) return;
-	if (image->Data == 0) return;
-
-	DEBUG_TEX(("*** s3vUploadTexLevel: #%i ***\n", ++times));
-	DEBUG_TEX(("level = %i\n", level));
-
-	l2d = 5; /* 32bits per texel == 1<<5 */
-/*
-	if (level == 0) 
-		;
-*/
-	DEBUG_TEX(("t->image[%i].offset = 0x%x\n",
-		level, t->image[level].offset));
-		
-	t->TextureBaseAddr[level] = (GLuint)(t->BufAddr + t->image[level].offset
-		+ _TEXALIGN) & (GLuint)(~_TEXALIGN);
-	dest = (GLuint*)(sPriv->pFB + t->TextureBaseAddr[level]); 
-
-	DEBUG_TEX(("sPriv->pFB = 0x%x\n", sPriv->pFB));
-	DEBUG_TEX(("dest = 0x%x\n", dest));
-	DEBUG_TEX(("dest - sPriv->pFB = 0x%x\n", ((int)dest - (int)sPriv->pFB)));
-
-	/* NOTE: we implicitly suppose t->texelBytes == 2 */
-
-	words = (image->Width * image->Height) >> 1;
-
-	DEBUG_TEX(("\n\n"));
-
-	switch (t->image[level].internalFormat) {
-	case GL_RGB:
-	case 3:
-	{
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_RGB:\n"));
-/*
-		if (level == 0)
-     			;
-*/
-		/* The UGLY way, and SLOW : use DMA FIXME ! */
-
-		for (i = 0; i < words; i++) {
-		unsigned int data;
-		/* data = PACK_COLOR_565(src[0],src[1],src[2]); */
-		data = S3VIRGEPACKCOLOR555(src[0],src[1],src[2],255)
-			|(S3VIRGEPACKCOLOR555(src[3],src[4],src[5],255)<<16);
-
-		*dest++ = data;
-	 	/* src += 3; */
-		src +=6;
-      	}
-	}
-	break;
-
-	case GL_RGBA:
-	case 4:
-	{
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_RGBA:\n"));
-/*
-		if (level == 0)
-			;
-*/
-		for (i = 0; i < words; i++) {		
-		unsigned int data;
-		
-		/* data = PACK_COLOR_8888(src[0],src[1],src[2],src[3]); */
-		data = S3VIRGEPACKCOLOR4444(src[0], src[1],src[2], src[3])
-		| (S3VIRGEPACKCOLOR4444(src[4], src[5], src[6], src[7]) << 16);
-		
-		*dest++ = data;
-		/* src += 4; */
-		src += 8;
-		}
-	}
-	break;
-
-	case GL_LUMINANCE:
-	{
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_LUMINANCE:\n"));
-/*
-		if (level == 0)
-			;
-*/
-		for (i = 0; i < words; i++) {
-		unsigned int data;
-		
-		/* data = PACK_COLOR_888(src[0],src[0],src[0]); */
-		data = S3VIRGEPACKCOLOR4444(src[0],src[0],src[0],src[0])
-		| (S3VIRGEPACKCOLOR4444(src[1],src[1],src[1],src[1]) << 16);
-		 
-		*dest++ = data;
-		/* src ++; */
-		src +=2;
-		}
-	}
-	break;
-
-	case GL_INTENSITY:
-	{
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_INTENSITY:\n"));
-/*	
-		if (level == 0)
-			;
-*/
-		for (i = 0; i < words; i++) {
-		unsigned int data;
-		
-		/* data = PACK_COLOR_8888(src[0],src[0],src[0],src[0]); */
-		data = S3VIRGEPACKCOLOR4444(src[0],src[0],src[0],src[0])
-	        | (S3VIRGEPACKCOLOR4444(src[1],src[1],src[1],src[1]) << 16);
-
-		*dest++ = data; 
-		/* src ++; */
-		src += 2;
-		}
-	}
-	break;
-
-	case GL_LUMINANCE_ALPHA:
-	{
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_LUMINANCE_ALPHA:\n"));
-/*
-		if (level == 0)
-			;
-*/
-		for (i = 0; i < words; i++) {
-		unsigned int data;
-		
-		/* data = PACK_COLOR_8888(src[0],src[0],src[0],src[1]); */
-		data = S3VIRGEPACKCOLOR4444(src[0],src[0],src[0],src[1])
-	        | (S3VIRGEPACKCOLOR4444(src[2],src[2],src[2],src[3]) << 16);
-		
-		*dest++ = data;
-		/* src += 2; */
-		src += 4;
-		}
-	}
-	break;
-
-	case GL_ALPHA:
-	{
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_ALPHA:\n"));
-/*
-		if (level == 0)
-			;
-*/
-		for (i = 0; i < words; i++) {
-		unsigned int data;
-		
-		/* data = PACK_COLOR_8888(255,255,255,src[0]); */
-		data = S3VIRGEPACKCOLOR4444(255,255,255,src[0])
-		| (S3VIRGEPACKCOLOR4444(255,255,255,src[1]) << 16);
-		
-		*dest++ = data;
-		/* src += 1; */
-		src += 2;
-		}
-	}
-	break;
-
-	/* TODO: Translate color indices *now*:
-	 */
-	case GL_COLOR_INDEX:
-	{
-	
-		GLubyte *dst = (GLubyte *)(t->BufAddr + t->image[level].offset);
-		GLubyte *src = (GLubyte *)image->Data;
-
-		DEBUG_TEX(("GL_COLOR_INDEX:\n"));
-
-		for (j = 0 ; j < image->Height ; j++, dst += t->Pitch) {
-			for (i = 0 ; i < image->Width ; i++) {
-				dst[i] = src[0];
-				src += 1;
-			}
-		}
-	}
-	break;
-
-	default:
-		fprintf(stderr, "Not supported texture format %s\n",
-			_mesa_lookup_enum_by_nr(image->_BaseFormat));
-	}
-
-	DEBUG_TEX(("words = %i\n\n", words));
-}
-
-void s3vPrintLocalLRU( s3vContextPtr vmesa )
-{
-   s3vTextureObjectPtr t;
-   int sz = 1 << (vmesa->s3vScreen->logTextureGranularity);
-
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vPrintLocalLRU: #%i ***\n", ++times));
-#endif
-
-   foreach( t, &vmesa->TexObjList ) {
-      if (!t->globj)
-         fprintf(stderr, "Placeholder %d at %x sz %x\n",
-		    t->MemBlock->ofs / sz,
-		    t->MemBlock->ofs,
-		    t->MemBlock->size);
-      else
-         fprintf(stderr, "Texture at %x sz %x\n",
-		    t->MemBlock->ofs,
-		    t->MemBlock->size);
-
-   }
-}
-
-void s3vPrintGlobalLRU( s3vContextPtr vmesa )
-{
-   int i, j;
-   S3VTexRegionPtr list = vmesa->sarea->texList;
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vPrintGlobalLRU: #%i ***\n", ++times));
-#endif
-
-   for (i = 0, j = S3V_NR_TEX_REGIONS ; i < S3V_NR_TEX_REGIONS ; i++) {
-      fprintf(stderr, "list[%d] age %d next %d prev %d\n",
-	      j, list[j].age, list[j].next, list[j].prev);
-      j = list[j].next;
-      if (j == S3V_NR_TEX_REGIONS) break;
-   }
-
-   if (j != S3V_NR_TEX_REGIONS)
-      fprintf(stderr, "Loop detected in global LRU\n");
-}
-
-
-void s3vResetGlobalLRU( s3vContextPtr vmesa )
-{
-   S3VTexRegionPtr list = vmesa->sarea->texList;
-   int sz = 1 << vmesa->s3vScreen->logTextureGranularity;
-   int i;
-
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vResetGlobalLRU: #%i ***\n", ++times));
-#endif
-
-   /* (Re)initialize the global circular LRU list.  The last element
-    * in the array (S3V_NR_TEX_REGIONS) is the sentinal.  Keeping it
-    * at the end of the array allows it to be addressed rationally
-    * when looking up objects at a particular location in texture
-    * memory.
-    */
-   for (i = 0 ; (i+1) * sz <= vmesa->s3vScreen->textureSize ; i++) {
-      list[i].prev = i-1;
-      list[i].next = i+1;
-      list[i].age = 0;
-   }
-
-   i--;
-   list[0].prev = S3V_NR_TEX_REGIONS;
-   list[i].prev = i-1;
-   list[i].next = S3V_NR_TEX_REGIONS;
-   list[S3V_NR_TEX_REGIONS].prev = i;
-   list[S3V_NR_TEX_REGIONS].next = 0;
-   vmesa->sarea->texAge = 0;
-}
-
-
-void s3vUpdateTexLRU( s3vContextPtr vmesa, s3vTextureObjectPtr t )
-{
-/*
-   int i;
-   int logsz = vmesa->s3vScreen->logTextureGranularity;
-   int start = t->MemBlock->ofs >> logsz;
-   int end = (t->MemBlock->ofs + t->MemBlock->size - 1) >> logsz;
-   S3VTexRegionPtr list = vmesa->sarea->texList;
-*/
-
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vUpdateTexLRU: #%i ***\n", ++times));
-#endif
-
-   vmesa->texAge = ++vmesa->sarea->texAge;
-
-   /* Update our local LRU
-    */
-   move_to_head( &(vmesa->TexObjList), t );
-
-   /* Update the global LRU
-    */
-#if 0
-   for (i = start ; i <= end ; i++) {
-
-      list[i].in_use = 1;
-      list[i].age = vmesa->texAge;
-
-      /* remove_from_list(i)
-       */
-      list[(unsigned)list[i].next].prev = list[i].prev;
-      list[(unsigned)list[i].prev].next = list[i].next;
-
-      /* insert_at_head(list, i)
-       */
-      list[i].prev = S3V_NR_TEX_REGIONS;
-      list[i].next = list[S3V_NR_TEX_REGIONS].next;
-      list[(unsigned)list[S3V_NR_TEX_REGIONS].next].prev = i;
-      list[S3V_NR_TEX_REGIONS].next = i;
-   }
-#endif
-}
-
-
-/* Called for every shared texture region which has increased in age
- * since we last held the lock.
- *
- * Figures out which of our textures have been ejected by other clients,
- * and pushes a placeholder texture onto the LRU list to represent
- * the other client's textures.
- */
-void s3vTexturesGone( s3vContextPtr vmesa,
-		       GLuint offset,
-		       GLuint size,
-		       GLuint in_use )
-{
-   s3vTextureObjectPtr t, tmp;
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vTexturesGone: #%i ***\n", ++times));
-#endif
-
-   foreach_s ( t, tmp, &vmesa->TexObjList ) {
-
-      if (t->MemBlock->ofs >= offset + size ||
-	  t->MemBlock->ofs + t->MemBlock->size <= offset)
-         continue;
-
-      /* It overlaps - kick it off.  Need to hold onto the currently bound
-       * objects, however.
-       */
-	  s3vSwapOutTexObj( vmesa, t );
-   }
-
-   if (in_use) {
-		   t = (s3vTextureObjectPtr) calloc(1,sizeof(*t));
-		   if (!t) return;
-
-		   t->MemBlock = mmAllocMem( vmesa->texHeap, size, 0, offset);
-		   insert_at_head( &vmesa->TexObjList, t );
-   }
-
-   /* Reload any lost textures referenced by current vertex buffer.
-	*/
-#if 0
-   if (vmesa->vertex_buffer) {
-		   int i, j;
-
-		   fprintf(stderr, "\n\nreload tex\n");
-
-		   for (i = 0 ; i < vmesa->statenr ; i++) {
-				   for (j = 0 ; j < 2 ; j++) {
-						   s3vTextureObjectPtr t = vmesa->state_tex[j][i];
-						   if (t) {
-								   if (t->MemBlock == 0)
-										   s3vUploadTexImages( vmesa, t );
-						   }
-				   }
-		   }
-
-		   /* Hard to do this with the lock held:
-			*/
-		   /*        S3V_FIREVERTICES( vmesa ); */
-   }
-#endif
-}
-
-
-/* This is called with the lock held.  May have to eject our own and/or
- * other client's texture objects to make room for the upload.
- */
-void s3vUploadTexImages( s3vContextPtr vmesa, s3vTextureObjectPtr t )
-{
-	int i;
-	int ofs;
-	int numLevels;
-#if TEX_DEBUG_ON
-	static unsigned int times=0;
-	static unsigned int try=0;
-
-	DEBUG_TEX(("*** s3vUploadTexImages: #%i ***\n", ++times));
-	DEBUG_TEX(("vmesa->texHeap = 0x%x; t->totalSize = %i\n",
-		(unsigned int)vmesa->texHeap, t->totalSize));
-#endif
-
-	/* Do we need to eject LRU texture objects?
-	 */
-	if (!t->MemBlock) {
-
-		while (1)
-		{
-			/* int try = 0; */
-			DEBUG_TEX(("trying to alloc mem for tex (try %i)\n", ++try));
-
-			t->MemBlock = mmAllocMem( vmesa->texHeap, t->totalSize, 12, 0 );
-
-			if (t->MemBlock)
-				break;
-
-			if (vmesa->TexObjList.prev == vmesa->CurrentTexObj[0]) {
-/*			    || vmesa->TexObjList.prev == vmesa->CurrentTexObj[1]) {
-				fprintf(stderr, "Hit bound texture in upload\n");
-				s3vPrintLocalLRU( vmesa ); */
-				return;
-			}
-
-			if (vmesa->TexObjList.prev == &(vmesa->TexObjList)) {
-/*				fprintf(stderr, "Failed to upload texture, sz %d\n",
-					t->totalSize);
-				mmDumpMemInfo( vmesa->texHeap ); */
-				return;
-			}
-
-			DEBUG_TEX(("swapping out: %p\n", vmesa->TexObjList.prev));
-			s3vSwapOutTexObj( vmesa, vmesa->TexObjList.prev );
-		}
-
-	ofs = t->MemBlock->ofs;
-
-	t->BufAddr = vmesa->s3vScreen->texOffset + ofs;
-
-	DEBUG_TEX(("ofs = 0x%x\n", ofs));
-	DEBUG_TEX(("t->BufAddr = 0x%x\n", t->BufAddr));
-
-/* FIXME: check if we need it */
-#if 0
-	if (t == vmesa->CurrentTexObj[0]) {
-		vmesa->dirty |= S3V_UPLOAD_TEX0; 
-		vmesa->restore_primitive = -1; 
-	}
-#endif
-
-#if 0
-	if (t == vmesa->CurrentTexObj[1])
-		vmesa->dirty |= S3V_UPLOAD_TEX1;
-#endif
-
-	s3vUpdateTexLRU( vmesa, t );
-	}
-
-#if 0
-	if (vmesa->dirtyAge >= GET_DISPATCH_AGE(vmesa))
-		s3vWaitAgeLocked( vmesa, vmesa->dirtyAge );
-#endif
-
-#if _TEXLOCK
-	S3V_SIMPLE_FLUSH_LOCK(vmesa);
-#endif
-	numLevels = t->lastLevel - t->firstLevel + 1;
-	for (i = 0 ; i < numLevels ; i++)
-		if (t->dirty_images & (1<<i))
-			s3vUploadTexLevel( vmesa, t, i );
-
-	t->dirty_images = 0;
-#if _TEXLOCK
-	S3V_SIMPLE_UNLOCK(vmesa);
-#endif
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_texstate.c b/src/mesa/drivers/dri/s3v/s3v_texstate.c
deleted file mode 100644
index 455bae6301..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_texstate.c
+++ /dev/null
@@ -1,300 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include <stdlib.h>
-#include <stdio.h>
-
-#include "main/glheader.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-#include "main/simple_list.h"
-#include "main/enums.h"
-
-#include "main/mm.h"
-#include "s3v_context.h"
-#include "s3v_tex.h"
-
-
-static void s3vSetTexImages( s3vContextPtr vmesa, 
-			      struct gl_texture_object *tObj )
-{
-   GLuint height, width, pitch, i, /*textureFormat,*/ log_pitch;
-   s3vTextureObjectPtr t = (s3vTextureObjectPtr) tObj->DriverData;
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   GLint firstLevel, lastLevel, numLevels;
-   GLint log2Width, log2Height;
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vSetTexImages: #%i ***\n", ++times));
-#endif
-
-   t->texelBytes = 2; /* FIXME: always 2 ? */
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    * This depends on the base image size, GL_TEXTURE_MIN_LOD,
-    * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
-    * Yes, this looks overly complicated, but it's all needed.
-    */
-   if (tObj->MinFilter == GL_LINEAR || tObj->MinFilter == GL_NEAREST) {
-      firstLevel = lastLevel = tObj->BaseLevel;
-   }
-   else {
-      firstLevel = tObj->BaseLevel + (GLint) (tObj->MinLod + 0.5);
-      firstLevel = MAX2(firstLevel, tObj->BaseLevel);
-      lastLevel = tObj->BaseLevel + (GLint) (tObj->MaxLod + 0.5);
-	  lastLevel = MAX2(lastLevel, tObj->BaseLevel);
-      lastLevel = MIN2(lastLevel, tObj->BaseLevel + baseImage->MaxLog2);
-      lastLevel = MIN2(lastLevel, tObj->MaxLevel);
-      lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
-   }
-
-   /* save these values */
-   t->firstLevel = firstLevel;
-   t->lastLevel = lastLevel;
-
-   numLevels = lastLevel - firstLevel + 1;
-
-   log2Width = tObj->Image[0][firstLevel]->WidthLog2;
-   log2Height = tObj->Image[0][firstLevel]->HeightLog2;
-
-
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   width = tObj->Image[0][firstLevel]->Width * t->texelBytes;
-   for (pitch = 32, log_pitch=2 ; pitch < width ; pitch *= 2 )
-      log_pitch++;
-   
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   for ( height = i = 0 ; i < numLevels ; i++ ) {
-      t->image[i].image = tObj->Image[0][firstLevel + i];
-      t->image[i].offset = height * pitch;
-      t->image[i].internalFormat = baseImage->_BaseFormat;
-      height += t->image[i].image->Height;
-      t->TextureBaseAddr[i] = (t->BufAddr + t->image[i].offset +
-         _TEXALIGN) & (GLuint)(~_TEXALIGN);
-   }
-
-   t->Pitch = pitch;
-   t->WidthLog2 = log2Width;
-   t->totalSize = height*pitch;
-   t->max_level = i-1;
-   vmesa->dirty |= S3V_UPLOAD_TEX0 /* | S3V_UPLOAD_TEX1*/;   
-   vmesa->restore_primitive = -1;
-   DEBUG(("<><>pitch = TexStride = %i\n", pitch));
-   DEBUG(("log2Width = %i\n", log2Width));
-
-   s3vUploadTexImages( vmesa, t );
-}
-
-static void s3vUpdateTexEnv( GLcontext *ctx, GLuint unit )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   const struct gl_texture_object *tObj = texUnit->_Current;
-   const GLuint format = tObj->Image[0][tObj->BaseLevel]->_BaseFormat;
-/*
-   s3vTextureObjectPtr t = (s3vTextureObjectPtr)tObj->DriverData;
-   GLuint tc;
-*/
-   GLuint alpha = 0;
-   GLuint cmd = vmesa->CMD;
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vUpdateTexEnv: %i ***\n", ++times));
-#endif
-
-   cmd &= ~TEX_COL_MASK;
-   cmd &= ~TEX_BLEND_MAKS;
-/*   cmd &= ~ALPHA_BLEND_MASK; */
-
-   DEBUG(("format = "));
-
-   switch (format) {
-   case GL_RGB:
-      DEBUG_TEX(("GL_RGB\n"));
-      cmd |= TEX_COL_ARGB1555;
-      break;
-   case GL_LUMINANCE:
-      DEBUG_TEX(("GL_LUMINANCE\n"));
-      cmd |= TEX_COL_ARGB4444;
-      alpha = 1; /* FIXME: check */
-      break;
-   case GL_ALPHA:
-      DEBUG_TEX(("GL_ALPHA\n"));
-      cmd |= TEX_COL_ARGB4444;
-      alpha = 1;
-      break;
-   case GL_LUMINANCE_ALPHA:
-      DEBUG_TEX(("GL_LUMINANCE_ALPHA\n"));
-      cmd |= TEX_COL_ARGB4444;
-      alpha = 1;
-      break;
-   case GL_INTENSITY:
-      DEBUG_TEX(("GL_INTENSITY\n"));
-      cmd |= TEX_COL_ARGB4444;
-      alpha = 1;
-      break;
-   case GL_RGBA:
-      DEBUG_TEX(("GL_RGBA\n"));
-      cmd |= TEX_COL_ARGB4444;
-      alpha = 1;
-      break;
-   case GL_COLOR_INDEX:
-      DEBUG_TEX(("GL_COLOR_INDEX\n"));
-      cmd |= TEX_COL_PAL;
-      break;
-   }
-
-   DEBUG_TEX(("EnvMode = "));
-
-   switch (texUnit->EnvMode) {
-   case GL_REPLACE:
-      DEBUG_TEX(("GL_REPLACE\n"));
-      cmd |= TEX_REFLECT; /* FIXME */
-      vmesa->_tri[1] = DO_TEX_UNLIT_TRI; /* FIXME: white tri hack */
-      vmesa->_alpha_tex = ALPHA_TEX /* * alpha */;
-      break;
-   case GL_MODULATE:
-      DEBUG_TEX(("GL_MODULATE\n"));
-      cmd |= TEX_MODULATE;
-      vmesa->_tri[1] = DO_TEX_LIT_TRI;
-#if 0
-	if (alpha)
-		vmesa->_alpha_tex = ALPHA_TEX /* * alpha */;
-	else
-		vmesa->_alpha_tex = ALPHA_SRC /* * alpha */;
-#else
-	vmesa->_alpha_tex = ALPHA_TEX ;
-#endif
-      break;
-   case GL_ADD:
-      DEBUG_TEX(("DEBUG_TEX\n"));
-      /* do nothing ???*/
-      break;
-   case GL_DECAL:
-      DEBUG_TEX(("GL_DECAL\n"));
-      cmd |= TEX_DECAL;
-      vmesa->_tri[1] = DO_TEX_LIT_TRI;
-      vmesa->_alpha_tex = ALPHA_OFF;
-      break;
-   case GL_BLEND:
-      DEBUG_TEX(("GL_BLEND\n"));
-      cmd |= TEX_DECAL;
-      vmesa->_tri[1] = DO_TEX_LIT_TRI;
-      vmesa->_alpha_tex = ALPHA_OFF; /* FIXME: sure? */
-      break;
-   default:
-      fprintf(stderr, "unknown tex env mode");
-      return;
-   }
-  
-   DEBUG_TEX(("\n\n    vmesa->CMD was 0x%x\n", vmesa->CMD));   
-   DEBUG_TEX((	  "    vmesa->CMD is 0x%x\n\n", cmd ));
-
-   vmesa->_alpha[1] = vmesa->_alpha_tex;
-   vmesa->CMD = cmd; /* | MIPMAP_LEVEL(8); */
-   vmesa->restore_primitive = -1; 
-}
-
-static void s3vUpdateTexUnit( GLcontext *ctx, GLuint unit )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   GLuint cmd = vmesa->CMD;
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vUpdateTexUnit: %i ***\n", ++times));
-   DEBUG_TEX(("and vmesa->CMD was 0x%x\n", vmesa->CMD));
-#endif
-
-   if (texUnit->_ReallyEnabled == TEXTURE_2D_BIT) 
-   {
-      struct gl_texture_object *tObj = texUnit->_Current;
-      s3vTextureObjectPtr t = (s3vTextureObjectPtr)tObj->DriverData;
-
-      /* Upload teximages (not pipelined)
-       */
-      if (t->dirty_images) {
-#if _TEXFLUSH
-         DMAFLUSH();
-#endif
-         s3vSetTexImages( vmesa, tObj ); 
-         if (!t->MemBlock) {
-#if _TEXFALLBACK
-            FALLBACK( vmesa, S3V_FALLBACK_TEXTURE, GL_TRUE );
-#endif
-            return;
-         }
-      }
-
-      /* Update state if this is a different texture object to last
-       * time.
-       */
-#if 1
-      if (vmesa->CurrentTexObj[unit] != t) {
-         vmesa->dirty |= S3V_UPLOAD_TEX0 /* << unit */;
-         vmesa->CurrentTexObj[unit] = t;
-         s3vUpdateTexLRU( vmesa, t ); /* done too often */
-      }
-#endif
-      
-      /* Update texture environment if texture object image format or 
-       * texture environment state has changed.
-       */
-      if (tObj->Image[0][tObj->BaseLevel]->_BaseFormat !=
-          vmesa->TexEnvImageFmt[unit]) {
-         vmesa->TexEnvImageFmt[unit] = tObj->Image[0][tObj->BaseLevel]->_BaseFormat;
-         s3vUpdateTexEnv( ctx, unit );
-      }
-#if 1
-      cmd = vmesa->CMD & ~MIP_MASK;
-      vmesa->dirty |= S3V_UPLOAD_TEX0 /* << unit */;
-      vmesa->CurrentTexObj[unit] = t;
-      vmesa->TexOffset = t->TextureBaseAddr[tObj->BaseLevel];
-      vmesa->TexStride = t->Pitch;
-      cmd |= MIPMAP_LEVEL(t->WidthLog2);
-	
-      DEBUG_TEX(("\n\n>>  vmesa->CMD was 0x%x\n", vmesa->CMD));
-      DEBUG_TEX((    ">>  vmesa->CMD is 0x%x\n\n", cmd ));
-      DEBUG_TEX(("t->WidthLog2 = %i\n", t->WidthLog2));
-      DEBUG_TEX(("MIPMAP_LEVEL(t->WidthLog2) = 0x%x\n", MIPMAP_LEVEL(t->WidthLog2)));
-
-      vmesa->CMD = cmd;
-      vmesa->restore_primitive = -1;
-#endif
-   }
-   else if (texUnit->_ReallyEnabled) { /* _ReallyEnabled but != TEXTURE0_2D */
-#if _TEXFALLBACK
-      FALLBACK( vmesa, S3V_FALLBACK_TEXTURE, GL_TRUE );
-#endif
-   }
-   else /*if (vmesa->CurrentTexObj[unit])*/ { /* !_ReallyEnabled */
-      vmesa->CurrentTexObj[unit] = 0;
-      vmesa->TexEnvImageFmt[unit] = 0;	
-      vmesa->dirty &= ~(S3V_UPLOAD_TEX0<<unit); 
-   }
-}
-
-
-void s3vUpdateTextureState( GLcontext *ctx )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   (void) vmesa;
-#if TEX_DEBUG_ON
-   static unsigned int times=0;
-   DEBUG_TEX(("*** s3vUpdateTextureState: #%i ***\n", ++times));
-#endif
-
-#if _TEXFALLBACK   
-   FALLBACK( vmesa, S3V_FALLBACK_TEXTURE, GL_FALSE );
-#endif
-   s3vUpdateTexUnit( ctx, 0 );
-#if 0
-   s3vUpdateTexUnit( ctx, 1 );
-#endif
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_tris.c b/src/mesa/drivers/dri/s3v/s3v_tris.c
deleted file mode 100644
index fafd38480c..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_tris.c
+++ /dev/null
@@ -1,850 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include <stdio.h>
-#include <stdlib.h>
-
-#include <sys/ioctl.h>
-
-#include "s3v_context.h"
-#include "s3v_vb.h"
-#include "s3v_tris.h"
-
-#include "main/glheader.h"
-#include "main/mtypes.h"
-#include "main/macros.h"
-#include "main/colormac.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-#include "tnl/t_context.h"
-#include "tnl/t_pipeline.h"
-
-
-/***********************************************************************
- *                   Build hardware rasterization functions            *
- ***********************************************************************/
-
-#define DO_TRI			1
-#define HAVE_RGBA		1
-#define HAVE_SPEC		0
-#define HAVE_BACK_COLORS	0
-#define HAVE_HW_FLATSHADE	1
-#define VERTEX			s3vVertex
-#define TAB			rast_tab
-
-#define VERT_SET_RGBA( v, c ) \
-do { \
-	UNCLAMPED_FLOAT_TO_RGBA_CHAN( v->ub4[4], c); \
-/*	*(v->ub4[4]) = c; \ */ \
-} while (0)
-#define VERT_COPY_RGBA( v0, v1 ) v0->ui[4] = v1->ui[4]
-/*
-#define VERT_COPY_RGBA1( v0, v1 ) v0->ui[4] = v1->ui[4]
-*/
-#define VERT_SAVE_RGBA( idx )    color[idx] = v[idx]->ui[4]
-#define VERT_RESTORE_RGBA( idx ) v[idx]->ui[4] = color[idx]
-
-#define S3V_OFFSET_BIT  	0x01
-#define S3V_TWOSIDE_BIT 	0x02
-#define S3V_UNFILLED_BIT        0x04
-#define S3V_FALLBACK_BIT        0x08
-#define S3V_MAX_TRIFUNC 	0x10
-
-
-static struct {
-	tnl_points_func		points;
-	tnl_line_func		line;
-	tnl_triangle_func	triangle;
-	tnl_quad_func		quad;
-} rast_tab[S3V_MAX_TRIFUNC];
-
-#define S3V_RAST_CULL_BIT	0x01
-#define S3V_RAST_FLAT_BIT	0x02
-#define S3V_RAST_TEX_BIT	0x04
-
-static s3v_point_func s3v_point_tab[0x8];
-static s3v_line_func s3v_line_tab[0x8];
-static s3v_tri_func s3v_tri_tab[0x8];
-static s3v_quad_func s3v_quad_tab[0x8];
-
-#define IND (0)
-#define TAG(x) x
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_CULL_BIT)
-#define TAG(x) x##_cull
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_FLAT_BIT)
-#define TAG(x) x##_flat
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_CULL_BIT|S3V_RAST_FLAT_BIT)
-#define TAG(x) x##_cull_flat
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_TEX_BIT)
-#define TAG(x) x##_tex
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_CULL_BIT|S3V_RAST_TEX_BIT)
-#define TAG(x) x##_cull_tex
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_FLAT_BIT|S3V_RAST_TEX_BIT)
-#define TAG(x) x##_flat_tex
-#include "s3v_tritmp.h"
-
-#define IND (S3V_RAST_CULL_BIT|S3V_RAST_FLAT_BIT|S3V_RAST_TEX_BIT)
-#define TAG(x) x##_cull_flat_tex
-#include "s3v_tritmp.h"
-
-static void init_rast_tab( void )
-{
-	DEBUG(("*** init_rast_tab ***\n"));
-
-	s3v_init();
-	s3v_init_cull();
-	s3v_init_flat();
-	s3v_init_cull_flat();
-	s3v_init_tex();
-	s3v_init_cull_tex();
-	s3v_init_flat_tex();
-	s3v_init_cull_flat_tex();
-}
-
-/***********************************************************************
- *                    Rasterization fallback helpers                   *
- ***********************************************************************/
-
-
-/* This code is hit only when a mix of accelerated and unaccelerated
- * primitives are being drawn, and only for the unaccelerated
- * primitives.  
- */
-
-#if 0
-static void 
-s3v_fallback_quad( s3vContextPtr vmesa, 
-		    const s3vVertex *v0, 
-		    const s3vVertex *v1, 
-		    const s3vVertex *v2, 
-		    const s3vVertex *v3 )
-{
-   GLcontext *ctx = vmesa->glCtx;
-   SWvertex v[4];
-   s3v_translate_vertex( ctx, v0, &v[0] );
-   s3v_translate_vertex( ctx, v1, &v[1] );
-   s3v_translate_vertex( ctx, v2, &v[2] );
-   s3v_translate_vertex( ctx, v3, &v[3] );
-   DEBUG(("s3v_fallback_quad\n"));
-/*   _swrast_Quad( ctx, &v[0], &v[1], &v[2], &v[3] ); */
-}
-
-static void 
-s3v_fallback_tri( s3vContextPtr vmesa, 
-		    const s3vVertex *v0, 
-		    const s3vVertex *v1, 
-		    const s3vVertex *v2 )
-{
-   GLcontext *ctx = vmesa->glCtx;
-   SWvertex v[3];
-   s3v_translate_vertex( ctx, v0, &v[0] );
-   s3v_translate_vertex( ctx, v1, &v[1] );
-   s3v_translate_vertex( ctx, v2, &v[2] );
-   DEBUG(("s3v_fallback_tri\n"));
-/*   _swrast_Triangle( ctx, &v[0], &v[1], &v[2] ); */
-}
-
-static void
-s3v_fallback_line( s3vContextPtr vmesa,
-		     const s3vVertex *v0,
-		     const s3vVertex *v1 )
-{
-   GLcontext *ctx = vmesa->glCtx;
-   SWvertex v[2];
-   s3v_translate_vertex( ctx, v0, &v[0] );
-   s3v_translate_vertex( ctx, v1, &v[1] );
-   DEBUG(("s3v_fallback_line\n"));
-   _swrast_Line( ctx, &v[0], &v[1] );
-}
-
-/*
-static void 
-s3v_fallback_point( s3vContextPtr vmesa, 
-		      const s3vVertex *v0 )
-{
-   GLcontext *ctx = vmesa->glCtx;
-   SWvertex v[1];
-   s3v_translate_vertex( ctx, v0, &v[0] );
-   _swrast_Point( ctx, &v[0] );
-}
-*/
-#endif
-
-/***********************************************************************
- *                    Choose rasterization functions                   *
- ***********************************************************************/
-
-#define _S3V_NEW_RASTER_STATE 	(_NEW_FOG | \
-				 _NEW_TEXTURE | \
-				 _DD_NEW_TRI_SMOOTH | \
-				 _DD_NEW_LINE_SMOOTH | \
-				 _DD_NEW_POINT_SMOOTH | \
-				 _DD_NEW_TRI_STIPPLE | \
-				 _DD_NEW_LINE_STIPPLE)
-
-#define LINE_FALLBACK (0)
-#define TRI_FALLBACK (0)
-
-static void s3v_nodraw_triangle(GLcontext *ctx, s3vVertex *v0,
-                                s3vVertex *v1, s3vVertex *v2)
-{
-	(void) (ctx && v0 && v1 && v2);
-}
-
-static void s3v_nodraw_quad(GLcontext *ctx,
-                            s3vVertex *v0, s3vVertex *v1,
-                            s3vVertex *v2, s3vVertex *v3)
-{
-	(void) (ctx && v0 && v1 && v2 && v3);
-}
-
-void s3vChooseRasterState(GLcontext *ctx);
-
-void s3vChooseRasterState(GLcontext *ctx)
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	GLuint flags = ctx->_TriangleCaps;
-	GLuint ind = 0;
-
-	DEBUG(("*** s3vChooseRasterState ***\n"));
-
-	if (ctx->Polygon.CullFlag) {
-		if (ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) {
-			vmesa->draw_tri = (s3v_tri_func)s3v_nodraw_triangle;
-			vmesa->draw_quad = (s3v_quad_func)s3v_nodraw_quad;
-			return;
-		}
-		ind |= S3V_RAST_CULL_BIT;
-		/* s3v_update_cullsign(ctx); */
-	} /* else vmesa->backface_sign = 0; */
-
-	if ( flags & DD_FLATSHADE )
-		ind |= S3V_RAST_FLAT_BIT;
-
-	if ( ctx->Texture.Unit[0]._ReallyEnabled ) {
-		ind |= S3V_RAST_TEX_BIT;
-	}
-
-	DEBUG(("ind = %i\n", ind));
-
-	vmesa->draw_line = s3v_line_tab[ind];
-	vmesa->draw_tri = s3v_tri_tab[ind];
-	vmesa->draw_quad = s3v_quad_tab[ind];
-	vmesa->draw_point = s3v_point_tab[ind];
-
-#if 0
-	/* Hook in fallbacks for specific primitives.  CURRENTLY DISABLED
-	 */
-	
-	if (flags & LINE_FALLBACK) 
-		vmesa->draw_line = s3v_fallback_line;
-	 
-	if (flags & TRI_FALLBACK) {
-		DEBUG(("TRI_FALLBACK\n"));
-		vmesa->draw_tri = s3v_fallback_tri;
-		vmesa->draw_quad = s3v_fallback_quad;
-	}
-#endif
-}
-
-
-
-
-/***********************************************************************
- *          Macros for t_dd_tritmp.h to draw basic primitives          *
- ***********************************************************************/
-
-#define TRI( v0, v1, v2 ) \
-do { \
-	/*
-	if (DO_FALLBACK) \
-		vmesa->draw_tri( vmesa, v0, v1, v2 ); \
-	else */ \
-   	DEBUG(("TRI: max was here\n")); /* \
-	s3v_draw_tex_triangle( vmesa, v0, v1, v2 ); */ \
-	vmesa->draw_tri( vmesa, v0, v1, v2 ); \
-} while (0)
-
-#define QUAD( v0, v1, v2, v3 ) \
-do { \
-	DEBUG(("QUAD: max was here\n")); \
-	vmesa->draw_quad( vmesa, v0, v1, v2, v3 ); \
-} while (0)
-
-#define LINE( v0, v1 ) \
-do { \
-	DEBUG(("LINE: max was here\n")); \
-	vmesa->draw_line( vmesa, v0, v1 ); \
-} while (0)
-
-#define POINT( v0 ) \
-do { \
-	vmesa->draw_point( vmesa, v0 ); \
-} while (0)
-
-
-/***********************************************************************
- *              Build render functions from dd templates               *
- ***********************************************************************/
-
-/*
-#define S3V_OFFSET_BIT 	0x01
-#define S3V_TWOSIDE_BIT	0x02
-#define S3V_UNFILLED_BIT	0x04
-#define S3V_FALLBACK_BIT	0x08
-#define S3V_MAX_TRIFUNC	0x10
-
-
-static struct {
-   points_func		points;
-   line_func		line;
-   triangle_func	triangle;
-   quad_func		quad;
-} rast_tab[S3V_MAX_TRIFUNC];
-*/
-
-#define DO_FALLBACK  (IND & S3V_FALLBACK_BIT)
-#define DO_OFFSET    (IND & S3V_OFFSET_BIT)
-#define DO_UNFILLED  (IND & S3V_UNFILLED_BIT)
-#define DO_TWOSIDE   (IND & S3V_TWOSIDE_BIT)
-#define DO_FLAT      0
-#define DO_TRI       1
-#define DO_QUAD      1
-#define DO_LINE      1
-#define DO_POINTS    1
-#define DO_FULL_QUAD 1
-
-#define HAVE_RGBA         1
-#define HAVE_SPEC         0
-#define HAVE_BACK_COLORS  0
-#define HAVE_HW_FLATSHADE 1
-#define VERTEX            s3vVertex
-#define TAB               rast_tab
-
-#define DEPTH_SCALE 1.0
-#define UNFILLED_TRI unfilled_tri
-#define UNFILLED_QUAD unfilled_quad
-#define VERT_X(_v) _v->v.x
-#define VERT_Y(_v) _v->v.y
-#define VERT_Z(_v) _v->v.z
-#define AREA_IS_CCW( a ) (a > 0)
-#define GET_VERTEX(e) (vmesa->verts + (e<<vmesa->vertex_stride_shift))
-
-#if 0
-#define VERT_SET_RGBA( v, c ) \
-do { \
-/*    UNCLAMPED_FLOAT_TO_RGBA_CHAN( v->ub4[4], c) */ \
-} while (0)
-
-#define VERT_COPY_RGBA( v0, v1 ) v0->ui[4] = v1->ui[4]
-/*
-#define VERT_COPY_RGBA1( v0, v1 ) v0->ui[4] = v1->ui[4]
-*/
-#define VERT_SAVE_RGBA( idx )    color[idx] = v[idx]->ui[4]
-#define VERT_RESTORE_RGBA( idx ) v[idx]->ui[4] = color[idx]   
-#endif
-
-#define LOCAL_VARS(n) \
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx); \
-	GLuint color[n]; \
-	(void) color;
-
-
-/***********************************************************************
- *                Helpers for rendering unfilled primitives            *
- ***********************************************************************/
-
-static const GLuint hw_prim[GL_POLYGON+1] = {
-	PrimType_Points,
-	PrimType_Lines,
-	PrimType_Lines,
-	PrimType_Lines,
-	PrimType_Triangles,
-	PrimType_Triangles,
-	PrimType_Triangles,
-	PrimType_Triangles,
-	PrimType_Triangles,
-	PrimType_Triangles
-};
-
-static void s3vResetLineStipple( GLcontext *ctx );
-static void s3vRasterPrimitive( GLcontext *ctx, GLuint hwprim );
-static void s3vRenderPrimitive( GLcontext *ctx, GLenum prim );
-/*
-extern static void s3v_lines_emit(GLcontext *ctx, GLuint start, GLuint end);
-extern static void s3v_tris_emit(GLcontext *ctx, GLuint start, GLuint end);
-*/
-#define RASTERIZE(x) if (vmesa->hw_primitive != hw_prim[x]) \
-                        s3vRasterPrimitive( ctx, hw_prim[x] )
-#define RENDER_PRIMITIVE vmesa->render_primitive
-#define TAG(x) x
-#define IND S3V_FALLBACK_BIT
-#include "tnl_dd/t_dd_unfilled.h"
-#undef IND
-
-/***********************************************************************
- *                      Generate GL render functions                   *
- ***********************************************************************/
-
-#define IND (0)
-#define TAG(x) x
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_OFFSET_BIT)
-#define TAG(x) x##_offset
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_TWOSIDE_BIT)
-#define TAG(x) x##_twoside
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_TWOSIDE_BIT|S3V_OFFSET_BIT)
-#define TAG(x) x##_twoside_offset
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_UNFILLED_BIT)
-#define TAG(x) x##_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_OFFSET_BIT|S3V_UNFILLED_BIT)
-#define TAG(x) x##_offset_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_TWOSIDE_BIT|S3V_UNFILLED_BIT)
-#define TAG(x) x##_twoside_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (S3V_TWOSIDE_BIT|S3V_OFFSET_BIT|S3V_UNFILLED_BIT)
-#define TAG(x) x##_twoside_offset_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-
-static void init_render_tab( void )
-{
-	DEBUG(("*** init_render_tab ***\n"));
-
-	init();
-	init_offset();
-	init_twoside();
-	init_twoside_offset();
-	init_unfilled();
-	init_offset_unfilled();
-	init_twoside_unfilled();
-	init_twoside_offset_unfilled();
-}
-
-
-/**********************************************************************/
-/*               Render unclipped begin/end objects                   */
-/**********************************************************************/
-
-#define VERT(x) (s3vVertex *)(s3vverts + (x << shift))
-
-#define RENDER_POINTS( start, count ) \
-	DEBUG(("RENDER_POINTS...(ok)\n")); \
-	for ( ; start < count ; start++) \
-		vmesa->draw_line( vmesa, VERT(start), VERT(start) )
-	/*      vmesa->draw_point( vmesa, VERT(start) ) */
-
-#define RENDER_LINE( v0, v1 ) \
-	/* DEBUG(("RENDER_LINE...(ok)\n")); \ */ \
-	vmesa->draw_line( vmesa, VERT(v0), VERT(v1) ); \
-	DEBUG(("RENDER_LINE...(ok)\n"))
-
-#define RENDER_TRI( v0, v1, v2 )  \
-	DEBUG(("RENDER_TRI...(ok)\n")); \
-	vmesa->draw_tri( vmesa, VERT(v0), VERT(v1), VERT(v2) )
-
-#define RENDER_QUAD( v0, v1, v2, v3 ) \
-	DEBUG(("RENDER_QUAD...(ok)\n")); \
-	/* s3v_draw_quad( vmesa, VERT(v0), VERT(v1), VERT(v2),VERT(v3) ) */\
-	/* s3v_draw_triangle( vmesa, VERT(v0), VERT(v1), VERT(v2) ); \
-	s3v_draw_triangle( vmesa, VERT(v0), VERT(v2), VERT(v3) ) */ \
-	vmesa->draw_quad( vmesa, VERT(v0), VERT(v1), VERT(v2), VERT(v3) ) 
-	
-#define INIT(x) s3vRenderPrimitive( ctx, x );
-#undef LOCAL_VARS
-#define LOCAL_VARS						\
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);		\
-   const GLuint shift = vmesa->vertex_stride_shift;		\
-   const char *s3vverts = (char *)vmesa->verts;		\
-   const GLboolean stipple = ctx->Line.StippleFlag;		\
-   (void) stipple;
-#define RESET_STIPPLE	if ( stipple ) s3vResetLineStipple( ctx );
-#define RESET_OCCLUSION
-#define PRESERVE_VB_DEFS
-#define ELT(x) (x)
-#define TAG(x) s3v_##x##_verts
-#include "tnl_dd/t_dd_rendertmp.h"
-
-
-/**********************************************************************/
-/*                   Render clipped primitives                        */
-/**********************************************************************/
-
-static void s3vRenderClippedPoly( GLcontext *ctx, const GLuint *elts, 
-				   GLuint n )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	GLuint prim = vmesa->render_primitive;
-
-	DEBUG(("I AM in: s3vRenderClippedPoly\n"));
-
-	/* Render the new vertices as an unclipped polygon. 
-	 */
-	if (1)
-	{
-	GLuint *tmp = VB->Elts;
-	VB->Elts = (GLuint *)elts;
-	tnl->Driver.Render.PrimTabElts[GL_POLYGON]
-		( ctx, 0, n, PRIM_BEGIN|PRIM_END );
-
-	VB->Elts = tmp;
-	}
-
-	/* Restore the render primitive
-	 */
-#if 1
-	if (prim != GL_POLYGON) {
-		DEBUG(("and prim != GL_POLYGON\n"));
-		tnl->Driver.Render.PrimitiveNotify( ctx, prim );
-	}
-	
-#endif
-}
-
-static void s3vRenderClippedLine( GLcontext *ctx, GLuint ii, GLuint jj )
-{
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   /*tnl->Driver.LineFunc = s3v_line_tab[2];*/ /* _swsetup_Line; */
-
-   DEBUG(("I AM in: s3vRenderClippedLine\n"));
-   tnl->Driver.Render.Line( ctx, ii, jj );
-}
-
-
-/**********************************************************************/
-/*                    Choose render functions                         */
-/**********************************************************************/
-
-
-
-#define _S3V_NEW_RENDERSTATE (_DD_NEW_TRI_UNFILLED |		\
-			       _DD_NEW_TRI_LIGHT_TWOSIDE |	\
-			       _DD_NEW_TRI_OFFSET)
-
-#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE|DD_TRI_OFFSET|DD_TRI_UNFILLED)
-
-static void s3vChooseRenderState(GLcontext *ctx)
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint flags = ctx->_TriangleCaps;
-   GLuint index = 0;
-
-   DEBUG(("s3vChooseRenderState\n"));
-
-   if (flags & ANY_RASTER_FLAGS) {
-      if (flags & DD_TRI_LIGHT_TWOSIDE)       index |= S3V_TWOSIDE_BIT;
-      if (flags & DD_TRI_OFFSET)	      index |= S3V_OFFSET_BIT;
-      if (flags & DD_TRI_UNFILLED)	      index |= S3V_UNFILLED_BIT;
-   }
-
-   DEBUG(("vmesa->RenderIndex = %i\n", vmesa->RenderIndex));
-   DEBUG(("index = %i\n", index));
-
-   if (vmesa->RenderIndex != index) {
-      vmesa->RenderIndex = index;
-
-      tnl->Driver.Render.Points = rast_tab[index].points;
-      tnl->Driver.Render.Line = rast_tab[index].line;
-      tnl->Driver.Render.Triangle = rast_tab[index].triangle;
-      tnl->Driver.Render.Quad = rast_tab[index].quad;
-
-      if (vmesa->RenderIndex == 0)
-         tnl->Driver.Render.PrimTabVerts = s3v_render_tab_verts;
-      else
-         tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
-      tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
-      tnl->Driver.Render.ClippedLine = s3vRenderClippedLine;
-      tnl->Driver.Render.ClippedPolygon = s3vRenderClippedPoly;
-   }
-}
-
-
-/**********************************************************************/
-/*                 High level hooks for t_vb_render.c                 */
-/**********************************************************************/
-
-
-
-/* Determine the rasterized primitive when not drawing unfilled 
- * polygons.
- *
- * Used only for the default render stage which always decomposes
- * primitives to trianges/lines/points.  For the accelerated stage,
- * which renders strips as strips, the equivalent calculations are
- * performed in s3v_render.c.
- */
-
-static void s3vRasterPrimitive( GLcontext *ctx, GLuint hwprim )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-/*	__DRIdrawablePrivate *dPriv = vmesa->driDrawable; */
-	GLuint cmd = vmesa->CMD;
-	
-	unsigned int _hw_prim = hwprim;
-
-	DEBUG(("s3vRasterPrimitive: hwprim = 0x%x ", _hw_prim));
-
-/*	printf("* vmesa->CMD = 0x%x\n", vmesa->CMD); */
-
-	if (vmesa->hw_primitive != _hw_prim)
-	{
-		DEBUG(("(new one) ***\n"));
-		cmd &= ~DO_MASK;
-		cmd &= ~ALPHA_BLEND_MASK;
-		vmesa->hw_primitive = _hw_prim;
-
-		if (_hw_prim == PrimType_Triangles) {
-		/* TRI */
-	                DEBUG(("->switching to tri\n"));
-			cmd |= (vmesa->_tri[vmesa->_3d_mode] | vmesa->_alpha[vmesa->_3d_mode]);
-	        } else if (_hw_prim == PrimType_Lines
-			|| _hw_prim == PrimType_Points) {
-		/* LINE */
-        	        DEBUG(("->switching to line\n"));
-			cmd |= (DO_3D_LINE | vmesa->_alpha[0]);
-		} else  {
-		/* ugh? */
-			DEBUG(("->switching to your sis'ass\n"));
-		}
-        
-		DEBUG(("\n"));
-
-		vmesa->restore_primitive = _hw_prim;
-		/* 0xacc16827: good value -> lightened newave!!! */
-		vmesa->CMD = cmd;
-		CMDCHANGE();
-	}
-}
-
-static void s3vRenderPrimitive( GLcontext *ctx, GLenum prim )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-	GLuint cmd = vmesa->CMD;
-
-	unsigned int _hw_prim = hw_prim[prim];
-
-	vmesa->render_primitive = prim;
-	vmesa->hw_primitive = _hw_prim;
-   
-	DEBUG(("s3vRenderPrimitive #%i ", prim));
-	DEBUG(("_hw_prim = 0x%x\n", _hw_prim));
-
-/*	printf(" vmesa->CMD = 0x%x\n", vmesa->CMD); */
-
-	if (_hw_prim != vmesa->restore_primitive) {
-		DEBUG(("_hw_prim != vmesa->restore_primitive (was 0x%x)\n",
-			vmesa->restore_primitive));
-#if 1
-		cmd &= ~DO_MASK;
-		cmd &= ~ALPHA_BLEND_MASK;
-/*		
-		printf(" cmd = 0x%x\n", cmd);
-		printf(" vmesa->_3d_mode=%i; vmesa->_tri[vmesa->_3d_mode]=0x%x\n",
-			vmesa->_3d_mode, vmesa->_tri[vmesa->_3d_mode]);
-		printf("vmesa->alpha[0] = 0x%x; vmesa->alpha[1] = 0x%x\n",
-			vmesa->_alpha[0], vmesa->_alpha[1]);
-*/		
-	   	if (_hw_prim == PrimType_Triangles) { /* TRI */
-			DEBUG(("->switching to tri\n"));
-   			cmd |= (vmesa->_tri[vmesa->_3d_mode] | vmesa->_alpha[vmesa->_3d_mode]);
-			DEBUG(("vmesa->TexStride = %i\n", vmesa->TexStride));
-			DEBUG(("vmesa->TexOffset = %i\n", vmesa->TexOffset));
-			DMAOUT_CHECK(3DTRI_Z_BASE, 12);
-		} else { /* LINE */
-			DEBUG(("->switching to line\n"));
-			cmd |= (DO_3D_LINE | vmesa->_alpha[0]);
-			DMAOUT_CHECK(3DLINE_Z_BASE, 12);
-		}
-
-		DMAOUT(vmesa->s3vScreen->depthOffset & 0x003FFFF8);
-		DMAOUT(vmesa->DestBase);
-		/* DMAOUT(vmesa->ScissorLR); */
-		/* DMAOUT(vmesa->ScissorTB); */
-
-		/* NOTE: we need to restore all these values since we
-		 * are coming back from a vmesa->restore_primitive */
-		DMAOUT( (0 << 16) | (dPriv->w-1) );
-		DMAOUT( (0 << 16) | (dPriv->h-1) );
-		DMAOUT( (vmesa->SrcStride << 16) | vmesa->TexStride );
-		DMAOUT(vmesa->SrcStride);
-		DMAOUT(vmesa->TexOffset);
-		DMAOUT(vmesa->TextureBorderColor);
-		DMAOUT(0); /* FOG */
-		DMAOUT(0);
-		DMAOUT(0);
-	        DMAOUT(cmd);
-		/* 0xacc16827: good value -> lightened newave!!! */
-        DMAFINISH();
-
-	vmesa->CMD = cmd;
-#endif
-	}
-
-	DEBUG(("\n"));
-
-	vmesa->restore_primitive = _hw_prim;
-}
-
-static void s3vRunPipeline( GLcontext *ctx )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-
-	DEBUG(("*** s3vRunPipeline ***\n"));
-
-	if ( vmesa->new_state )
-		s3vDDUpdateHWState( ctx );
-
-	if (vmesa->new_gl_state) {
-
-		if (vmesa->new_gl_state & _NEW_TEXTURE) {
-			s3vUpdateTextureState( ctx );
-		}
-
-		if (!vmesa->Fallback) {
-			if (vmesa->new_gl_state & _S3V_NEW_VERTEX)
-				s3vChooseVertexState( ctx );
-      
-			if (vmesa->new_gl_state & _S3V_NEW_RASTER_STATE)
-				s3vChooseRasterState( ctx );
-      
-			if (vmesa->new_gl_state & _S3V_NEW_RENDERSTATE)
-				s3vChooseRenderState( ctx );
-		}
-
-	vmesa->new_gl_state = 0;
-	
-	}
-
-	_tnl_run_pipeline( ctx );
-}
-
-static void s3vRenderStart( GLcontext *ctx )
-{
-	/* Check for projective texturing.  Make sure all texcoord
-	 * pointers point to something.  (fix in mesa?)  
-	 */
-
-	DEBUG(("s3vRenderStart\n"));
-	/* s3vCheckTexSizes( ctx ); */
-}
-
-static void s3vRenderFinish( GLcontext *ctx )
-{
-   if (0)
-      _swrast_flush( ctx );	/* never needed */
-}
-
-static void s3vResetLineStipple( GLcontext *ctx )
-{
-/*   s3vContextPtr vmesa = S3V_CONTEXT(ctx); */
-
-   /* Reset the hardware stipple counter.
-    */
-/*
-   CHECK_DMA_BUFFER(vmesa, 1);
-   WRITE(vmesa->buf, UpdateLineStippleCounters, 0);
-*/
-}
-
-
-/**********************************************************************/
-/*           Transition to/from hardware rasterization.               */
-/**********************************************************************/
-
-
-void s3vFallback( s3vContextPtr vmesa, GLuint bit, GLboolean mode )
-{
-   GLcontext *ctx = vmesa->glCtx;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint oldfallback = vmesa->Fallback;
-
-   DEBUG(("*** s3vFallback: "));
-
-   if (mode) {
-      vmesa->Fallback |= bit;
-      if (oldfallback == 0) {
-		 DEBUG(("oldfallback == 0 ***\n"));
-		 _swsetup_Wakeup( ctx );
-		 _tnl_need_projected_coords( ctx, GL_TRUE );
-		 vmesa->RenderIndex = ~0;
-      }
-   }
-   else {
-      DEBUG(("***\n"));
-      vmesa->Fallback &= ~bit;
-      if (oldfallback == bit) {
-	 	_swrast_flush( ctx );
-	 	tnl->Driver.Render.Start = s3vRenderStart;
-	 	tnl->Driver.Render.PrimitiveNotify = s3vRenderPrimitive;
-	 	tnl->Driver.Render.Finish = s3vRenderFinish;
-	 	tnl->Driver.Render.BuildVertices = s3vBuildVertices;
-   		tnl->Driver.Render.ResetLineStipple = s3vResetLineStipple;
-	 	vmesa->new_gl_state |= (_S3V_NEW_RENDERSTATE|
-					_S3V_NEW_RASTER_STATE|
-					_S3V_NEW_VERTEX);
-      }
-   }
-}
-
-
-/**********************************************************************/
-/*                            Initialization.                         */
-/**********************************************************************/
-
-
-void s3vInitTriFuncs( GLcontext *ctx )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   static int firsttime = 1;
-
-   if (firsttime) {
-      init_rast_tab();
-      init_render_tab();
-      firsttime = 0;
-   }
-   
-   vmesa->RenderIndex = ~0;
-
-   tnl->Driver.RunPipeline = s3vRunPipeline;
-   tnl->Driver.Render.Start = s3vRenderStart;
-   tnl->Driver.Render.Finish = s3vRenderFinish; 
-   tnl->Driver.Render.PrimitiveNotify = s3vRenderPrimitive;
-   tnl->Driver.Render.ResetLineStipple = s3vResetLineStipple;
-/*
-   tnl->Driver.RenderInterp = _swsetup_RenderInterp;
-   tnl->Driver.RenderCopyPV = _swsetup_RenderCopyPV;
-*/
-   tnl->Driver.Render.BuildVertices = s3vBuildVertices;
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_tris.h b/src/mesa/drivers/dri/s3v/s3v_tris.h
deleted file mode 100644
index 0010a7fe0a..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_tris.h
+++ /dev/null
@@ -1,11 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef _S3V_TRIS_H
-#define _S3V_TRIS_H
-
-extern void s3vDDTrifuncInit(void);
-extern void s3vDDChooseTriRenderState(GLcontext *);
-
-#endif /* !(_S3V_TRIS_H) */
diff --git a/src/mesa/drivers/dri/s3v/s3v_tritmp.h b/src/mesa/drivers/dri/s3v/s3v_tritmp.h
deleted file mode 100644
index 2321bd414f..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_tritmp.h
+++ /dev/null
@@ -1,958 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-/**** MACROS start ****/
-
-/* point/line macros */
-
-#define LINE_VERT_VARS \
-        SWvertex v[3]; \
-	s3vVertex* vvv[2]; \
-        int x[3], y[3], z[3]; \
-        int idx[3]; \
-        int dx01, dy01; \
-        int delt02; \
-        int deltzy, zstart; \
-        int start02, end01; \
-        int ystart, y01y12; \
-        int i, tmp, tmp2, tmp3; \
-        GLfloat ydiff, fy[3]
-#define LINE_VERT_VARS_VOIDS \
-        (void) v; (void) vvv; (void) x; (void) y; (void) z; (void) idx; \
-        (void) dx01; (void) dy01; (void) delt02; (void) deltzy; \
-        (void) zstart; (void) start02; (void) ystart; (void) y01y12; \
-        (void) i; (void) tmp; (void) tmp2; (void) tmp3; (void) ydiff; (void) fy
-
-#define LINE_FLAT_VARS \
-        int arstart, gbstart; \
-        int deltarx, deltgbx, deltary, deltgby; \
-        GLubyte *(col)[3]
-#define LINE_FLAT_VARS_VOIDS \
-        (void) arstart; (void) gbstart; (void) deltarx; (void) deltgbx; \
-        (void) deltary; (void) deltgby; (void) col
-
-#define LINE_GOURAUD_VARS \
-        int arstart, gbstart; \
-        int deltary, deltgby; \
-        int ctmp, ctmp2, ctmp3, ctmp4; \
-        GLubyte *(col)[3]
-#define LINE_GOURAUD_VARS_VOIDS \
-        (void) arstart; (void) gbstart; (void) deltary; (void) deltgby; \
-        (void) ctmp; (void) ctmp2; (void) ctmp3; (void) ctmp4; (void) col
-
-#define SORT_LINE_VERT() \
-do { \
-	if(v[0].attrib[FRAG_ATTRIB_WPOS][1] <= v[1].attrib[FRAG_ATTRIB_WPOS][1]) { \
-\
-                idx[0] = 0; \
-                idx[1] = 1; \
-\
-        } else if (v[0].attrib[FRAG_ATTRIB_WPOS][1] > v[1].attrib[FRAG_ATTRIB_WPOS][1]) { \
-\
-                idx[0] = 1; \
-                idx[1] = 0; \
-\
-        } \
-} while(0)
-
-#define SET_LINE_VERT() \
-do { \
-        x[0] = (v[idx[0]].attrib[FRAG_ATTRIB_WPOS][0] * 1024.0f * 1024.0f); /* 0x100000 */ \
-        y[0] = fy[0] = dPriv->h - v[idx[0]].attrib[FRAG_ATTRIB_WPOS][1]; \
-        z[0] = (v[idx[0]].attrib[FRAG_ATTRIB_WPOS][2]) * 1024.0f * 32.0f; /* 0x8000; */ \
-\
-        x[1] = (v[idx[1]].attrib[FRAG_ATTRIB_WPOS][0] * 1024.0f * 1024.0f); /* 0x100000 */ \
-        y[1] = dPriv->h - v[idx[1]].attrib[FRAG_ATTRIB_WPOS][1]; \
-        z[1] = (v[idx[1]].attrib[FRAG_ATTRIB_WPOS][2]) * 1024.0f * 32.0f; /* 0x8000 */ \
-} while(0)
-
-#define SET_LINE_XY() \
-do { \
-	tmp = v[idx[0]].attrib[FRAG_ATTRIB_WPOS][0]; \
-        tmp2 = v[idx[1]].attrib[FRAG_ATTRIB_WPOS][0]; \
-\
-	dx01 = x[0] - x[1]; \
-        dy01 = y[0] - y[1]; \
-\
-        ydiff = fy[0] - (float)y[0]; \
-        ystart = y[0]; \
-        y01y12 = dy01 + 1; \
-} while (0)
-
-#define SET_LINE_DIR() \
-do { \
-        if (tmp2 > tmp) { \
-                y01y12 |= 0x80000000; \
-                tmp3 = tmp2-tmp; \
-        } else { \
-                tmp3 = tmp-tmp2; \
-        } \
-\
-        end01 = ((tmp << 16) | tmp2); \
-\
-	if (dy01) \
-                delt02 = -(dx01/dy01); \
-        else \
-		delt02 = 0; \
-\
-        if (dy01 > tmp3) { /* Y MAJ */ \
-	/* NOTE: tmp3 always >=0 */ \
-                start02 = x[0]; \
-        } else if (delt02 >= 0){ /* X MAJ - positive delta */ \
-                start02 = x[0] + delt02/2; \
-                dy01 = tmp3; /* could be 0 */ \
-        } else { /* X MAJ - negative delta */ \
-                start02 = x[0] + delt02/2 + ((1 << 20) - 1); \
-                dy01 = tmp3; /* could be 0 */ \
-        } \
-} while(0)
-
-#define SET_LINE_Z() \
-do { \
-	zstart = z[0]; \
-\
-	if (dy01) { \
-                deltzy = (z[1] - z[0])/dy01; \
-        } else { \
-                deltzy = 0; /* dy01 = tmp3 = 0 (it's a point)*/ \
-        } \
-} while (0)
-
-#define SET_LINE_FLAT_COL() \
-do { \
-        col[0] = &(v[idx[0]].color[0]); \
-        deltarx = deltary = deltgbx = deltgby = 0; \
-        gbstart = (((col[0][1]) << 23) | ((col[0][2]) << 7)); \
-        arstart = (((col[0][3]) << 23) | ((col[0][0]) << 7)); \
-} while(0)
-
-#define SET_LINE_GOURAUD_COL() \
-do { \
-        col[0] = &(v[idx[0]].color[0]); \
-        col[1] = &(v[idx[1]].color[0]); \
-\
-        vvv[0] = _v0; \
-        vvv[1] = _v1; \
-\
-        for (i=0; i<2; i++) { \
-	/* FIXME: swapped ! */ \
-                col[i][0] = vvv[!idx[i]]->v.color.red; \
-                col[i][1] = vvv[!idx[i]]->v.color.green; \
-                col[i][2] = vvv[!idx[i]]->v.color.blue; \
-                col[i][3] = vvv[!idx[i]]->v.color.alpha; \
-        } \
-\
-	if (dy01) { \
-\
-        ctmp = ((col[0][1] - col[1][1]) << 7) / dy01; \
-        ctmp2 = ((col[0][2] - col[1][2]) << 7) / dy01; \
-        deltgby = ((ctmp << 16) & 0xFFFF0000) | (ctmp2 & 0xFFFF); \
-\
-        ctmp3 = ((col[0][3] - col[1][3]) << 7) / dy01; \
-        ctmp4 = ((col[0][0] - col[1][0]) << 7) / dy01; \
-        deltary = ((ctmp3 << 16) & 0xFFFF0000) | (ctmp4 & 0xFFFF); \
-        } else { \
-        ctmp = ((col[1][1] - col[0][1]) << 7); \
-        ctmp2 = ((col[1][2] - col[0][2]) << 7); \
-        deltgby = ((ctmp << 16) & 0xFFFF0000) | (ctmp2 & 0xFFFF); \
-\
-        ctmp3 = ((col[1][3] - col[0][3]) << 7); \
-        ctmp4 = ((col[1][0] - col[0][0]) << 7); \
-        deltary = ((ctmp3 << 16) & 0xFFFF0000) | (ctmp4 & 0xFFFF); \
-        deltgby = deltary = 0; \
-        } \
-\
-	idx[0] = 1; /* FIXME: swapped */ \
-\
-        gbstart = \
-	(((int)((ydiff * ctmp) + (col[idx[0]][1] << 7)) << 16) & 0x7FFF0000) \
-	| ((int)((ydiff * ctmp2) + (col[idx[0]][2] << 7)) & 0x7FFF); \
-        arstart = \
-	(((int)((ydiff * ctmp3) + (col[idx[0]][3] << 7)) << 16) & 0x7FFF0000) \
-	| ((int)((ydiff * ctmp4) + (col[idx[0]][0] << 7)) & 0x7FFF); \
-} while(0)
-
-#define SEND_LINE_COL() \
-do { \
-	DMAOUT(deltgby); \
-	DMAOUT(deltary); \
-	DMAOUT(gbstart); \
-	DMAOUT(arstart); \
-} while (0)
-
-#define SEND_LINE_VERT() \
-do { \
-	DMAOUT(deltzy); \
-	DMAOUT(zstart); \
-	DMAOUT(0); \
-	DMAOUT(0); \
-	DMAOUT(0); \
-	DMAOUT(end01); \
-	DMAOUT(delt02); \
-	DMAOUT(start02); \
-	DMAOUT(ystart); \
-	DMAOUT(y01y12); \
-} while (0)
-
-
-/* tri macros (mostly stolen from utah-glx...) */
-
-#define VERT_VARS \
-        SWvertex v[3]; \
-        int x[3], y[3], z[3]; \
-        int idx[3]; \
-        int dx01, dy01; \
-        int dx02, dy02; \
-        int dx12, dy12; \
-        int delt01, delt02, delt12; \
-        int deltzx, deltzy, zstart; \
-        int start02, end01, end12; \
-        int ystart, y01y12; \
-        int i, tmp, lr; \
-        GLfloat ydiff, fy[3]
-#define VERT_VARS_VOIDS \
-        (void) v; (void) x; (void) y; (void) z; (void) idx; (void) dx01; \
-        (void) dy01; (void) dx02; (void) dy02; (void) dx12; (void) dy12; \
-        (void) delt01; (void) delt02; (void) delt12; (void) deltzx; \
-        (void) deltzy; (void) zstart; (void) start02; (void) end01; \
-        (void) end12; (void) ystart; (void) y01y12; (void) i; (void) tmp; \
-        (void) lr; (void) ydiff; (void) fy
-
-#define GOURAUD_VARS \
-        int arstart, gbstart; \
-        int deltarx, deltgbx, deltary, deltgby; \
-        int ctmp, ctmp2, ctmp3, ctmp4; \
-        GLubyte *(col)[3]
-#define GOURAUD_VARS_VOIDS \
-        (void) arstart; (void) gbstart; (void) deltarx; (void) deltgbx; \
-        (void) deltary; (void) deltgby; (void) ctmp; (void) ctmp2; \
-        (void) ctmp3; (void) ctmp4; (void) col
-
-#define FLAT_VARS \
-        int arstart, gbstart; \
-        int deltarx, deltgbx, deltary, deltgby; \
-        GLubyte *(col)[3]
-#define FLAT_VARS_VOIDS \
-        (void) arstart; (void) gbstart; (void) deltarx; (void) deltgbx; \
-        (void) deltary; (void) deltgby; (void) col
-
-#define TEX_VARS \
-        int u0, u1, u2; \
-        GLfloat ru0, ru1, ru2; \
-        int v0, v1, v2; \
-        GLfloat rv0, rv1, rv2; \
-        GLfloat w0, w1, w2; \
-        GLfloat rw0, rw1, rw2; \
-        int baseu, basev; \
-        int d0, d1, d2; \
-        int deltdx, deltvx, deltux, deltdy, deltvy, deltuy; \
-        int deltwx, deltwy; \
-        int rbaseu, rbasev; \
-        int dstart, ustart, wstart, vstart; \
-        static int stmp = 0; \
-        s3vTextureObjectPtr t
-#define TEX_VARS_VOIDS \
-        (void) u0; (void) u1; (void) u2; (void) ru0; (void) ru1; (void) ru2; \
-        (void) v0; (void) v1; (void) v2; (void) rv0; (void) rv1; (void) rv2; \
-        (void) w0; (void) w1; (void) w2; (void) rw0; (void) rw1; (void) rw2; \
-        (void) baseu; (void) basev; (void) d0; (void) d1; (void) d2; \
-        (void) deltdx; (void) deltvx; (void) deltux; (void) deltdy; \
-        (void) deltuy; (void) deltwx; (void) deltwy; (void) rbaseu; \
-        (void) rbasev; (void) dstart; (void) ustart; (void) wstart; \
-        (void) vstart; (void) stmp; (void) t
-
-#define SORT_VERT() \
-do { \
-	for (i=0; i<3; i++) \
-		fy[i] = v[i].attrib[FRAG_ATTRIB_WPOS][1]; \
-\
-		if (fy[1] > fy[0]) {  /* (fy[1] > fy[0]) */ \
-\
-			if (fy[2] > fy[0]) { \
-				idx[0] = 0; \
-				if (fy[1] > fy[2]) { \
-					idx[1] = 2; \
-					idx[2] = 1; \
-				} else { \
-					idx[1] = 1; \
-					idx[2] = 2; \
-				} \
-			} else { \
-				idx[0] = 2; \
-				idx[1] = 0; \
-				idx[2] = 1; \
-			} \
-		} else { /* (fy[1] < y[0]) */ \
-			if (fy[2] > fy[0]) { \
-				idx[0] = 1; \
-				idx[1] = 0; \
-				idx[2] = 2; \
-		} else { \
-			idx[2] = 0; \
-			if (fy[2] > fy[1]) { \
-				idx[0] = 1; \
-				idx[1] = 2; \
-			} else { \
-				idx[0] = 2; \
-				idx[1] = 1; \
-			} \
-		} \
-	} \
-} while(0)
-
-#define SET_VERT() \
-do { \
-	for (i=0; i<3; i++) \
-	{ \
-		x[i] = ((v[idx[i]].attrib[FRAG_ATTRIB_WPOS][0]) * /* 0x100000*/  1024.0 * 1024.0); \
-		y[i] = fy[i] = (dPriv->h - v[idx[i]].attrib[FRAG_ATTRIB_WPOS][1]); \
-		z[i] = ((v[idx[i]].attrib[FRAG_ATTRIB_WPOS][2]) * /* 0x8000 */ 1024.0 * 32.0); \
-	} \
-\
-	ydiff = fy[0] - (float)y[0]; \
-\
-	ystart = y[0]; \
-\
-	dx12 = x[2] - x[1]; \
-	dy12 = y[1] - y[2]; \
-	dx01 = x[1] - x[0]; \
-	dy01 = y[0] - y[1]; \
-	dx02 = x[2] - x[0]; \
-	dy02 = y[0] - y[2]; \
-\
-	delt01 = delt02 = delt12 = 0; \
-} while (0)
-
-
-#define SET_XY() \
-do { \
-	if (dy01) delt01 = dx01 / dy01; \
-	if (dy12) delt12 = dx12 / dy12; \
-	delt02 = dx02 / dy02; \
-\
-	start02 = x[0] + (ydiff * delt02); \
-	end01 = x[0] + (ydiff * delt01); \
-	end12 = x[1] + ((fy[1] - (GLfloat)y[1]) * delt12); \
-} while (0)
-
-#define SET_DIR() \
-do { \
-	tmp = x[1] - (dy01 * delt02 + x[0]); \
-	if (tmp > 0) { \
-		lr = 0x80000000; \
-	} else { \
-		tmp *= -1; \
-		lr = 0; \
-	} \
-	tmp >>= 20; \
-\
-        y01y12 = ((((y[0] - y[1]) & 0x7FF) << 16) \
-                 | ((y[1] - y[2]) & 0x7FF) | lr); \
-} while (0)
-
-#define SET_Z() \
-do { \
-	deltzy = (z[2] - z[0]) / dy02; \
-	if (tmp) { \
-		deltzx = (z[1] - (dy01 * deltzy + z[0])) / tmp; \
-	} else { \
-		deltzx = 0; \
-	} \
-	zstart = (deltzy * ydiff) + z[0]; \
-} while (0)
-
-#define SET_FLAT_COL() \
-do { \
-	col[0] = &(v[0].color[0]); \
-	deltarx = deltary = deltgbx = deltgby = 0; \
-	gbstart = (((col[0][1]) << 23) | ((col[0][2]) << 7)); \
-	arstart = (((col[0][3]) << 23) | ((col[0][0]) << 7)); \
-} while(0)
-
-#define SET_GOURAUD_COL() \
-do { \
-	col[0] = &(v[idx[0]].color[0]); \
-	col[1] = &(v[idx[1]].color[0]); \
-	col[2] = &(v[idx[2]].color[0]); \
-\
-	ctmp = ((col[2][3] - col[0][3]) << 7) / dy02; \
-	ctmp2 = ((col[2][0] - col[0][0]) << 7) / dy02; \
-	deltary = ((ctmp << 16) & 0xFFFF0000) | (ctmp2 & 0xFFFF); \
-	ctmp3 = ((col[2][1] - col[0][1]) << 7) / dy02; \
-	ctmp4 = ((col[2][2] - col[0][2]) << 7) / dy02; \
-	deltgby = ((ctmp3 << 16) & 0xFFFF0000) | (ctmp4 & 0xFFFF); \
-	gbstart = \
-	(((int)((ydiff * ctmp3) + (col[0][1] << 7)) << 16) & 0x7FFF0000) \
-	| ((int)((ydiff * ctmp4) + (col[0][2] << 7)) & 0x7FFF); \
-	arstart = \
-	(((int)((ydiff * ctmp) + (col[0][3] << 7)) << 16) & 0x7FFF0000) \
-	| ((int)((ydiff * ctmp2) + (col[0][0] << 7)) & 0x7FFF); \
-	if (tmp) { \
-	int ax, rx, gx, bx; \
-	ax = ((col[1][3] << 7) - (dy01 * ctmp + (col[0][3] << 7))) / tmp; \
-	rx = ((col[1][0] << 7) - (dy01 * ctmp2 + (col[0][0] << 7))) / tmp; \
-	gx = ((col[1][1] << 7) - (dy01 * ctmp3 + (col[0][1] << 7))) / tmp; \
-	bx = ((col[1][2] << 7) - (dy01 * ctmp4 + (col[0][2] << 7))) / tmp; \
-	deltarx = ((ax << 16) & 0xFFFF0000) | (rx & 0xFFFF); \
-	deltgbx = ((gx << 16) & 0xFFFF0000) | (bx & 0xFFFF); \
-	} else { \
-	deltgbx = deltarx = 0; \
-	} \
-} while (0)
-
-#define SET_TEX_VERT() \
-do { \
-        t = ((s3vTextureObjectPtr) \
-                ctx->Texture.Unit[0]._Current->DriverData); \
-        deltwx = deltwy = wstart = deltdx = deltdy = dstart = 0; \
-\
-        u0 = (v[idx[0]].attrib[FRAG_ATTRIB_TEX0][0] \
-                * (GLfloat)(t->image[0].image->Width) * 256.0); \
-        u1 = (v[idx[1]].attrib[FRAG_ATTRIB_TEX0][0] \
-                * (GLfloat)(t->globj->Image[0][0]->Width) * 256.0); \
-        u2 = (v[idx[2]].attrib[FRAG_ATTRIB_TEX0][0] \
-                * (GLfloat)(t->globj->Image[0][0]->Width) * 256.0); \
-        v0 = (v[idx[0]].attrib[FRAG_ATTRIB_TEX0][1] \
-                * (GLfloat)(t->globj->Image[0][0]->Height) * 256.0); \
-        v1 = (v[idx[1]].attrib[FRAG_ATTRIB_TEX0][1] \
-                * (GLfloat)(t->globj->Image[0][0]->Height) * 256.0); \
-        v2 = (v[idx[2]].attrib[FRAG_ATTRIB_TEX0][1] \
-                * (GLfloat)(t->globj->Image[0][0]->Height) * 256.0); \
-\
-        w0 = (v[idx[0]].attrib[FRAG_ATTRIB_WPOS][3]); \
-        w1 = (v[idx[1]].attrib[FRAG_ATTRIB_WPOS][3]); \
-        w2 = (v[idx[2]].attrib[FRAG_ATTRIB_WPOS][3]); \
-} while (0)
-
-#define SET_BASEUV() \
-do { \
-        if (u0 < u1) { \
-                if (u0 < u2) { \
-                        baseu = u0; \
-                } else { \
-                        baseu = u2; \
-                } \
-        } else { \
-                if (u1 < u2) { \
-                        baseu = u1; \
-                } else { \
-                        baseu = u2; \
-                } \
-        } \
-\
-        if (v0 < v1) { \
-                if (v0 < v2) { \
-                        basev = v0; \
-                } else { \
-                        basev = v2; \
-                } \
-        } else { \
-                if (v1 < v2) { \
-                        basev = v1; \
-                } else { \
-                        basev = v2; \
-                } \
-        } \
-} while (0)
-
-
-#define SET_RW() \
-do { \
-        /* GLfloat minW; \
-\
-        if (w0 < w1) { \
-                if (w0 < w2) { \
-                        minW = w0; \
-                } else { \
-                        minW = w2; \
-                } \
-        } else { \
-                if (w1 < w2) { \
-                        minW = w1; \
-                } else { \
-                        minW = w2; \
-                } \
-        } */ \
-\
-        rw0 = (512.0 * w0); \
-        rw1 = (512.0 * w1); \
-        rw2 = (512.0 * w2); \
-} while (0)
-
-#define SET_D() \
-do { \
-        GLfloat sxy, suv; \
-        int lev; \
-\
-        suv = (v[idx[0]].attrib[FRAG_ATTRIB_TEX0][0] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][0]) * \
-                (v[idx[1]].attrib[FRAG_ATTRIB_TEX0][1] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][1]) - \
-                (v[idx[1]].attrib[FRAG_ATTRIB_TEX0][0] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][0]) * \
-                (v[idx[0]].attrib[FRAG_ATTRIB_TEX0][1] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][2]); \
-\
-        sxy = (v[idx[0]].attrib[FRAG_ATTRIB_TEX0][0] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][0]) * \
-                (v[idx[1]].attrib[FRAG_ATTRIB_TEX0][1] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][1]) - \
-                (v[idx[1]].attrib[FRAG_ATTRIB_TEX0][0] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][0]) * \
-                (v[idx[0]].attrib[FRAG_ATTRIB_TEX0][1] - \
-                v[idx[2]].attrib[FRAG_ATTRIB_TEX0][2]); \
-\
-	if (sxy < 0) sxy *= -1.0; \
-	if (suv < 0) suv *= -1.0; \
-\
-	lev = *(int*)&suv - *(int *)&sxy; \
-	if (lev < 0) \
-		lev = 0; \
-	else \
-		lev >>=23; \
-	dstart = (lev << 27); \
-} while (0)
-
-#define SET_UVWD() \
-do { \
-        SET_BASEUV(); \
-        SET_RW(); \
-        SET_D(); \
-        ru0 = (((u0 - baseu) * rw0)); \
-        ru1 = (((u1 - baseu) * rw1)); \
-        ru2 = (((u2 - baseu) * rw2)); \
-        rv0 = (((v0 - basev) * rw0)); \
-        rv1 = (((v1 - basev) * rw1)); \
-        rv2 = (((v2 - basev) * rw2)); \
-\
-        while (baseu < 0) { baseu += (t->globj->Image[0][0]->Width << 8); } \
-        while (basev < 0) { basev += (t->globj->Image[0][0]->Height << 8); } \
-\
-        if (!(baseu & 0xFF)) \
-                { baseu = (baseu >> 8); } \
-        else \
-                { baseu = (baseu >> 8) + 1; } \
-\
-        if ((basev & 0x80) || !(basev & 0xFF)) \
-                { basev = (basev >> 8); } \
-        else \
-                { basev = (basev >> 8) - 1; } \
-\
-        rbaseu = (baseu) << (16 - t->globj->Image[0][0]->WidthLog2); \
-        rbasev = (basev) << (16 - t->globj->Image[0][0]->WidthLog2); \
-        deltuy = (((ru2 - ru0) / dy02)); \
-        deltvy = (((rv2 - rv0) / dy02)); \
-        rw0 *= (1024.0 * 512.0); \
-        rw1 *= (1024.0 * 512.0); \
-        rw2 *= (1024.0 * 512.0); \
-        deltwy = ((rw2 - rw0) / dy02); \
-        if (tmp) { \
-                deltux = ((ru1 - (dy01 * deltuy + ru0)) / tmp); \
-                deltvx = ((rv1 - (dy01 * deltvy + rv0)) / tmp); \
-                deltwx = ((rw1 - (dy01 * deltwy + rw0)) / tmp); \
-        } else { deltux = deltvx = deltwx = 0; } \
-                ustart = (deltuy * ydiff) + (ru0); \
-                vstart = (deltvy * ydiff) + (rv0); \
-                wstart = (deltwy * ydiff) + (rw0); \
-} while (0)
-
-#define SEND_UVWD() \
-do { \
-        DMAOUT((rbasev & 0xFFFF)); \
-        DMAOUT((0xa0000000 | (rbaseu & 0xFFFF))); \
-        DMAOUT(deltwx); \
-        DMAOUT(deltwy); \
-        DMAOUT(wstart); \
-        DMAOUT(deltdx); \
-        DMAOUT(deltvx); \
-        DMAOUT(deltux); \
-        DMAOUT(deltdy); \
-        DMAOUT(deltvy); \
-        DMAOUT(deltuy); \
-        DMAOUT(dstart); \
-        DMAOUT(vstart); \
-        DMAOUT(ustart); \
-} while (0)
-
-#define SEND_VERT() \
-do { \
-	DMAOUT(deltzx); \
-	DMAOUT(deltzy); \
-	DMAOUT(zstart); \
-	DMAOUT(delt12); \
-	DMAOUT(end12); \
-	DMAOUT(delt01); \
-	DMAOUT(end01); \
-	DMAOUT(delt02); \
-	DMAOUT(start02); \
-	DMAOUT(ystart); \
-	DMAOUT(y01y12); \
-} while (0)
-
-#define SEND_COL() \
-do { \
-	DMAOUT(deltgbx); \
-	DMAOUT(deltarx); \
-	DMAOUT(deltgby); \
-	DMAOUT(deltary); \
-	DMAOUT(gbstart); \
-	DMAOUT(arstart); \
-} while (0)
-
-/**** MACROS end ****/
-
-
-
-
-static void TAG(s3v_point)( s3vContextPtr vmesa, 
-			     const s3vVertex *_v0 )
-{
-}
-
-static void TAG(s3v_line)( s3vContextPtr vmesa, 
-			     const s3vVertex *_v0,
-			     const s3vVertex *_v1 )
-{
-	GLcontext *ctx = vmesa->glCtx;
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-
-	LINE_VERT_VARS;
-#if (IND & S3V_RAST_FLAT_BIT)
-	LINE_FLAT_VARS;
-#else
-	LINE_GOURAUD_VARS;
-#endif
-#if (IND & S3V_RAST_CULL_BIT)
-	GLfloat cull;
-        (void) cull;
-#endif
-	LINE_VERT_VARS_VOIDS;
-#if (IND & S3V_RAST_FLAT_BIT)
-	LINE_FLAT_VARS_VOIDS;
-#else
-	LINE_GOURAUD_VARS_VOIDS;
-#endif
-
-	DEBUG(("*** s3v_line: "));
-#if (IND & S3V_RAST_CULL_BIT)
-	DEBUG(("cull "));
-#endif
-#if (IND & S3V_RAST_FLAT_BIT)
-        DEBUG(("flat "));
-#endif
-
-	DEBUG(("***\n"));
-
-#if 0
-	s3v_print_vertex(ctx, _v0);
-	s3v_print_vertex(ctx, _v1);
-#endif
-
-	s3v_translate_vertex( ctx, _v0, &v[0] );
-	s3v_translate_vertex( ctx, _v1, &v[1] );
-
-#if (IND & S3V_RAST_CULL_BIT)
-	/* FIXME: should we cull lines too? */
-#endif
-	(void)v; /* v[0]; v[1]; */
-
-	SORT_LINE_VERT();
-	SET_LINE_VERT();
-
-	SET_LINE_XY();
-	SET_LINE_DIR();
-	SET_LINE_Z();
-
-#if (IND & S3V_RAST_FLAT_BIT)
-	SET_LINE_FLAT_COL();
-#else
-	SET_LINE_GOURAUD_COL();
-#endif
-
-	DMAOUT_CHECK(3DLINE_GBD, 15);
-		SEND_LINE_COL();
-		DMAOUT(0);
-		SEND_LINE_VERT();
-	DMAFINISH();
-}
-
-static void TAG(s3v_triangle)( s3vContextPtr vmesa,
-				 const s3vVertex *_v0,
-				 const s3vVertex *_v1, 
-				 const s3vVertex *_v2 )
-{
-	GLcontext *ctx = vmesa->glCtx;
-	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-
-	VERT_VARS;
-#if (IND & S3v_RAST_FLAT_BIT)
-	FLAT_VARS;
-#else
-	GOURAUD_VARS;
-#endif
-#if (IND & S3V_RAST_TEX_BIT)
-	TEX_VARS;
-#endif
-#if (IND & S3V_RAST_CULL_BIT)
-	GLfloat cull;
-#endif
-	VERT_VARS_VOIDS;
-#if (IND & S3v_RAST_FLAT_BIT)
-	FLAT_VARS_VOIDS;
-#else
-	GOURAUD_VARS_VOIDS;
-#endif
-#if (IND & S3V_RAST_TEX_BIT)
-	TEX_VARS_VOIDS;
-#endif
-
-	DEBUG(("*** s3v_triangle: "));
-#if (IND & S3V_RAST_CULL_BIT)
-	DEBUG(("cull "));
-#endif
-#if (IND & S3V_RAST_FLAT_BIT)
-        DEBUG(("flat "));
-#endif
-#if (IND & S3V_RAST_TEX_BIT)
-        DEBUG(("tex "));
-#endif
-
-DEBUG(("***\n"));
-
-#if 0
-	s3v_print_vertex(ctx, _v0);
-	s3v_print_vertex(ctx, _v1);
-	s3v_print_vertex(ctx, _v2);
-#endif
-
-	s3v_translate_vertex( ctx, _v0, &v[0] );
-	s3v_translate_vertex( ctx, _v1, &v[1] );
-	s3v_translate_vertex( ctx, _v2, &v[2] );
-
-#if (IND & S3V_RAST_CULL_BIT)
-	cull = vmesa->backface_sign *
-		((v[1].attrib[FRAG_ATTRIB_WPOS][0] - v[0].attrib[FRAG_ATTRIB_WPOS][0]) * (v[0].attrib[FRAG_ATTRIB_WPOS][1] - v[2].attrib[FRAG_ATTRIB_WPOS][1]) +
-		 (v[1].attrib[FRAG_ATTRIB_WPOS][1] - v[0].attrib[FRAG_ATTRIB_WPOS][1]) * (v[2].attrib[FRAG_ATTRIB_WPOS][0] - v[0].attrib[FRAG_ATTRIB_WPOS][0]));
-
-	if (cull < vmesa->cull_zero /* -0.02f */) return;
-#endif
-
-	(void)v; /* v[0]; v[1]; v[2]; */
-
-	SORT_VERT();
-	SET_VERT();
-
-	if (dy02 == 0) return;
-
-	SET_XY();
-	SET_DIR();
-	SET_Z();
-	
-#if (IND & S3V_RAST_TEX_BIT)
-	SET_TEX_VERT();
-	SET_UVWD();
-#endif
-
-#if (IND & S3V_RAST_FLAT_BIT)
-	SET_FLAT_COL();
-#else
-	SET_GOURAUD_COL();
-#endif
-
-#if (IND & S3V_RAST_TEX_BIT)
-        DMAOUT_CHECK(3DTRI_BASEV, 31);
-		SEND_UVWD();
-		SEND_COL();
-		SEND_VERT();
-	DMAFINISH();
-#else
-	DMAOUT_CHECK(3DTRI_GBX, 17);
-		SEND_COL();
-		SEND_VERT();
-	DMAFINISH();
-#endif
-}
-
-static void TAG(s3v_quad)( s3vContextPtr vmesa,
-			    const s3vVertex *_v0,
-			    const s3vVertex *_v1,
-			    const s3vVertex *_v2,
-			    const s3vVertex *_v3 )
-{
-	GLcontext *ctx = vmesa->glCtx;
-        __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-
-	SWvertex temp_v[4];
-        VERT_VARS;
-#if (IND & S3v_RAST_FLAT_BIT)
-        FLAT_VARS;
-#else
-        GOURAUD_VARS;
-#endif
-#if (IND & S3V_RAST_TEX_BIT)
-        TEX_VARS;
-#endif
-#if (IND & S3V_RAST_CULL_BIT)
-        GLfloat cull;
-#endif
-        VERT_VARS_VOIDS;
-#if (IND & S3v_RAST_FLAT_BIT)
-        FLAT_VARS_VOIDS;
-#else
-        GOURAUD_VARS_VOIDS;
-#endif
-#if (IND & S3V_RAST_TEX_BIT)
-        TEX_VARS_VOIDS;
-#endif
-
-	DEBUG(("*** s3v_quad: "));
-#if (IND & S3V_RAST_CULL_BIT)
-        DEBUG(("cull "));
-		/* printf(""); */ /* speed trick */
-#endif
-#if (IND & S3V_RAST_FLAT_BIT)
-        DEBUG(("flat "));
-#endif
-#if (IND & S3V_RAST_TEX_BIT)
-        DEBUG(("tex "));
-#endif
-
-	DEBUG(("***\n"));
-
-#if 0
-	s3v_print_vertex(ctx, _v0);
-	s3v_print_vertex(ctx, _v1);
-	s3v_print_vertex(ctx, _v2);
-	s3v_print_vertex(ctx, _v3);
-#endif
-	s3v_translate_vertex( ctx, _v0, &temp_v[0] );
-	s3v_translate_vertex( ctx, _v1, &temp_v[1] );
-	s3v_translate_vertex( ctx, _v2, &temp_v[2] );
-	s3v_translate_vertex( ctx, _v3, &temp_v[3] );
-
-	/* FIRST TRI (0,1,2) */
-
-	/* ROMEO */
-	/* printf(""); */ /* speed trick (a) [turn on if (a) is return]*/
-
-	v[0] = temp_v[0];
-	v[1] = temp_v[1];
-	v[2] = temp_v[2];
-
-#if (IND & S3V_RAST_CULL_BIT)
-	cull = vmesa->backface_sign *
-		((v[1].attrib[FRAG_ATTRIB_WPOS][0] - v[0].attrib[FRAG_ATTRIB_WPOS][0]) * (v[0].attrib[FRAG_ATTRIB_WPOS][1] - v[2].attrib[FRAG_ATTRIB_WPOS][1]) +
-		 (v[1].attrib[FRAG_ATTRIB_WPOS][1] - v[0].attrib[FRAG_ATTRIB_WPOS][1]) * (v[2].attrib[FRAG_ATTRIB_WPOS][0] - v[0].attrib[FRAG_ATTRIB_WPOS][0]));
-
-	if (cull < vmesa->cull_zero /* -0.02f */) goto second; /* return; */ /* (a) */
-#endif
-	
-#if 0
-        v[0] = temp_v[0];
-	v[1] = temp_v[1];
-	v[2] = temp_v[2];
-#else
-        (void) v;
-#endif
-	SORT_VERT();
-	SET_VERT();
-
-	if (dy02 == 0) goto second;
-
-        SET_XY();
-        SET_DIR();
-        SET_Z();
-
-#if (IND & S3V_RAST_TEX_BIT)
-	SET_TEX_VERT();
-	SET_UVWD();
-#endif
-
-#if (IND & S3V_RAST_FLAT_BIT)
-	SET_FLAT_COL();
-#else
-	SET_GOURAUD_COL();
-#endif
-
-#if (IND & S3V_RAST_TEX_BIT)
-	DMAOUT_CHECK(3DTRI_BASEV, 31);
-		SEND_UVWD();
-		SEND_COL();
-		SEND_VERT();
-	DMAFINISH();
-#else
-	DMAOUT_CHECK(3DTRI_GBX, 17);
-		SEND_COL();
-		SEND_VERT();
-	DMAFINISH();
-#endif
-
-	/* SECOND TRI (0,2,3) */
-
-second:
-	v[0] = temp_v[0];
-	v[1] = temp_v[2];
-	v[2] = temp_v[3];
-
-#if (IND & S3V_RAST_CULL_BIT)
-	cull = vmesa->backface_sign *
-		((v[1].attrib[FRAG_ATTRIB_WPOS][0] - v[0].attrib[FRAG_ATTRIB_WPOS][0]) * (v[0].attrib[FRAG_ATTRIB_WPOS][1] - v[2].attrib[FRAG_ATTRIB_WPOS][1]) +
-		 (v[1].attrib[FRAG_ATTRIB_WPOS][1] - v[0].attrib[FRAG_ATTRIB_WPOS][1]) * (v[2].attrib[FRAG_ATTRIB_WPOS][0] - v[0].attrib[FRAG_ATTRIB_WPOS][0]));
-		 
-	if (cull < /* -0.02f */ vmesa->cull_zero) return;
-#endif
-
-/* second: */
-
-	/* ROMEO */
-	/* printf(""); */ /* speed trick */
-
-	v[0] = temp_v[0];
-	v[1] = temp_v[2];
-	v[2] = temp_v[3];
-
-	SORT_VERT();
-	SET_VERT();
-
-	if (dy02 == 0) return;
-
-	SET_XY();
-	SET_DIR();
-	SET_Z();
-
-#if (IND & S3V_RAST_TEX_BIT)
-	SET_TEX_VERT();
-	SET_UVWD();
-#endif
-
-#if (IND & S3V_RAST_FLAT_BIT)
-	SET_FLAT_COL();
-#else
-	SET_GOURAUD_COL();
-#endif
-
-#if (IND & S3V_RAST_TEX_BIT)
-	DMAOUT_CHECK(3DTRI_BASEV, 31);
-		SEND_UVWD();
-		SEND_COL();
-		SEND_VERT();
-	DMAFINISH();
-#else
-	DMAOUT_CHECK(3DTRI_GBX, 17);
-		SEND_COL();
-		SEND_VERT();
-	DMAFINISH();
-#endif
-}
-
-static void TAG(s3v_init)(void)
-{
-	s3v_point_tab[IND]	= TAG(s3v_point);
-	s3v_line_tab[IND]	= TAG(s3v_line);
-	s3v_tri_tab[IND]	= TAG(s3v_triangle);
-	s3v_quad_tab[IND]	= TAG(s3v_quad);
-}
-
-#undef IND
-#undef TAG
diff --git a/src/mesa/drivers/dri/s3v/s3v_vb.c b/src/mesa/drivers/dri/s3v/s3v_vb.c
deleted file mode 100644
index 00e375c6c4..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_vb.c
+++ /dev/null
@@ -1,339 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "main/glheader.h"
-#include "main/mtypes.h"
-#include "main/macros.h"
-#include "main/colormac.h"
-
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/t_context.h"
-#include "tnl/tnl.h"
-
-#include "s3v_context.h"
-#include "s3v_vb.h"
-#include "s3v_tris.h"
-
-#define S3V_XYZW_BIT       0x1
-#define S3V_RGBA_BIT       0x2
-#define S3V_TEX0_BIT       0x4
-#define S3V_PTEX_BIT       0x8
-#define S3V_FOG_BIT        0x10
-#define S3V_MAX_SETUP      0x20
-
-static struct {
-   void                (*emit)( GLcontext *, GLuint, GLuint, void *, GLuint );
-   tnl_interp_func	interp;
-   tnl_copy_pv_func	copy_pv;
-   GLboolean           (*check_tex_sizes)( GLcontext *ctx );
-   GLuint               vertex_size;
-   GLuint               vertex_stride_shift;
-   GLuint               vertex_format;
-} setup_tab[S3V_MAX_SETUP];
-
-
-/* Only one vertex format, atm, so no need to give them names:
- */
-#define TINY_VERTEX_FORMAT      1
-#define NOTEX_VERTEX_FORMAT     0
-#define TEX0_VERTEX_FORMAT      0
-#define TEX1_VERTEX_FORMAT      0
-#define PROJ_TEX1_VERTEX_FORMAT 0
-#define TEX2_VERTEX_FORMAT      0
-#define TEX3_VERTEX_FORMAT      0
-#define PROJ_TEX3_VERTEX_FORMAT 0
-
-#define DO_XYZW (IND & S3V_XYZW_BIT)
-#define DO_RGBA (IND & S3V_RGBA_BIT)
-#define DO_SPEC 0
-#define DO_FOG  (IND & S3V_FOG_BIT)
-#define DO_TEX0 (IND & S3V_TEX0_BIT)
-#define DO_TEX1 0
-#define DO_TEX2 0
-#define DO_TEX3 0
-#define DO_PTEX (IND & S3V_PTEX_BIT)
-			       
-#define VERTEX s3vVertex
-#define LOCALVARS /* s3vContextPtr vmesa = S3V_CONTEXT(ctx); */
-#define GET_VIEWPORT_MAT() 0 /* vmesa->hw_viewport */
-#define GET_TEXSOURCE(n)  n
-#define GET_VERTEX_FORMAT() 0
-#define GET_VERTEX_SIZE() S3V_CONTEXT(ctx)->vertex_size * sizeof(GLuint)
-#define GET_VERTEX_STORE() S3V_CONTEXT(ctx)->verts
-#define GET_VERTEX_STRIDE_SHIFT() S3V_CONTEXT(ctx)->vertex_stride_shift
-#define INVALIDATE_STORED_VERTICES()
-#define GET_UBYTE_COLOR_STORE() &S3V_CONTEXT(ctx)->UbyteColor
-#define GET_UBYTE_SPEC_COLOR_STORE() &S3V_CONTEXT(ctx)->UbyteSecondaryColor
-
-#define HAVE_HW_VIEWPORT    1	/* FIXME */
-#define HAVE_HW_DIVIDE      1
-#define HAVE_RGBA_COLOR     0 	/* we're BGRA */
-#define HAVE_TINY_VERTICES  1
-#define HAVE_NOTEX_VERTICES 0
-#define HAVE_TEX0_VERTICES  0
-#define HAVE_TEX1_VERTICES  0
-#define HAVE_TEX2_VERTICES  0
-#define HAVE_TEX3_VERTICES  0
-#define HAVE_PTEX_VERTICES  1
-
-/*
-#define SUBPIXEL_X -.5
-#define SUBPIXEL_Y -.5
-#define UNVIEWPORT_VARS  GLfloat h = S3V_CONTEXT(ctx)->driDrawable->h
-#define UNVIEWPORT_X(x)  x - SUBPIXEL_X
-#define UNVIEWPORT_Y(y)  - y + h + SUBPIXEL_Y
-#define UNVIEWPORT_Z(z)  z / vmesa->depth_scale
-*/
-
-#define PTEX_FALLBACK()		/* never needed */
-
-#define IMPORT_QUALIFIER
-#define IMPORT_FLOAT_COLORS s3v_import_float_colors
-#define IMPORT_FLOAT_SPEC_COLORS s3v_import_float_spec_colors
-
-#define INTERP_VERTEX setup_tab[S3V_CONTEXT(ctx)->SetupIndex].interp
-#define COPY_PV_VERTEX setup_tab[S3V_CONTEXT(ctx)->SetupIndex].copy_pv
-
-
-
-/***********************************************************************
- *         Generate  pv-copying and translation functions              *
- ***********************************************************************/
-
-#define TAG(x) s3v_##x
-#include "tnl_dd/t_dd_vb.c"
-
-/***********************************************************************
- *             Generate vertex emit and interp functions               *
- ***********************************************************************/
-
-
-#define IND (S3V_XYZW_BIT|S3V_RGBA_BIT)
-#define TAG(x) x##_wg
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_XYZW_BIT|S3V_RGBA_BIT|S3V_TEX0_BIT)
-#define TAG(x) x##_wgt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_XYZW_BIT|S3V_RGBA_BIT|S3V_TEX0_BIT|S3V_PTEX_BIT)
-#define TAG(x) x##_wgpt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_TEX0_BIT)
-#define TAG(x) x##_t0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_RGBA_BIT)
-#define TAG(x) x##_g
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_RGBA_BIT|S3V_TEX0_BIT)
-#define TAG(x) x##_gt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_XYZW_BIT|S3V_RGBA_BIT|S3V_FOG_BIT)
-#define TAG(x) x##_wgf
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_XYZW_BIT|S3V_RGBA_BIT|S3V_FOG_BIT|S3V_TEX0_BIT)
-#define TAG(x) x##_wgft0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_XYZW_BIT|S3V_RGBA_BIT|S3V_FOG_BIT|S3V_TEX0_BIT|S3V_PTEX_BIT)
-#define TAG(x) x##_wgfpt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_FOG_BIT)
-#define TAG(x) x##_f
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_RGBA_BIT | S3V_FOG_BIT)
-#define TAG(x) x##_gf
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (S3V_RGBA_BIT | S3V_FOG_BIT | S3V_TEX0_BIT)
-#define TAG(x) x##_gft0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-static void init_setup_tab( void )
-{
-   init_wg();		/* pos + col */
-   init_wgt0();		/* pos + col + tex0 */
-   init_wgpt0();	/* pos + col + p-tex0 (?) */
-   init_t0();		/* tex0 */
-   init_g();		/* col */
-   init_gt0();		/* col + tex */
-   init_wgf();
-   init_wgft0();
-   init_wgfpt0();
-   init_f();
-   init_gf();
-   init_gft0();
-}
-
-
-#if 0
-void s3vPrintSetupFlags(char *msg, GLuint flags )
-{
-   fprintf(stderr, "%s(%x): %s%s%s%s%s%s\n",
-	   msg,
-	   (int)flags,
-	   (flags & S3V_XYZW_BIT)      ? " xyzw," : "", 
-	   (flags & S3V_RGBA_BIT)     ? " rgba," : "",
-	   (flags & S3V_SPEC_BIT)     ? " spec," : "",
-	   (flags & S3V_FOG_BIT)      ? " fog," : "",
-	   (flags & S3V_TEX0_BIT)     ? " tex-0," : "",
-	   (flags & S3V_TEX1_BIT)     ? " tex-1," : "");
-}
-#endif
-
-
-void s3vCheckTexSizes( GLcontext *ctx )
-{
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-
-   if (!setup_tab[vmesa->SetupIndex].check_tex_sizes(ctx)) {
-
-      vmesa->SetupIndex |= (S3V_PTEX_BIT|S3V_RGBA_BIT);
-
-      if (1 || !(ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED))) {
-         tnl->Driver.Render.Interp = setup_tab[vmesa->SetupIndex].interp;
-         tnl->Driver.Render.CopyPV = setup_tab[vmesa->SetupIndex].copy_pv;
-      }
-   }
-}
-
-void s3vBuildVertices( GLcontext *ctx, 
-			 GLuint start, 
-			 GLuint count,
-			 GLuint newinputs )
-{
-	s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-	GLubyte *v = ((GLubyte *)vmesa->verts +
-		(start<<vmesa->vertex_stride_shift));
-	GLuint stride = 1<<vmesa->vertex_stride_shift;
-
-	DEBUG(("*** s3vBuildVertices ***\n"));
-	DEBUG(("vmesa->SetupNewInputs = 0x%x\n", vmesa->SetupNewInputs));
-	DEBUG(("vmesa->SetupIndex = 0x%x\n", vmesa->SetupIndex));
-
-#if 1
-	setup_tab[vmesa->SetupIndex].emit( ctx, start, count, v, stride );
-#else
-	newinputs |= vmesa->SetupNewInputs;
-	vmesa->SetupNewInputs = 0;
-
-	DEBUG(("newinputs is 0x%x\n", newinputs));
-
-	if (!newinputs) {
-		DEBUG(("!newinputs\n"));
-		return;
-	}
-
-	if (newinputs & VERT_CLIP) {
-	setup_tab[vmesa->SetupIndex].emit( ctx, start, count, v, stride );
-	DEBUG(("newinputs & VERT_CLIP\n"));
-	return;
-	} /* else { */
-/*      GLuint ind = 0; */
-
-	if (newinputs & VERT_RGBA) {
-		DEBUG(("newinputs & VERT_RGBA\n"));
-		ind |= S3V_RGBA_BIT;
-	}
- 
-	if (newinputs & VERT_TEX0) {
-		DEBUG(("newinputs & VERT_TEX0\n"));
-		ind |= S3V_TEX0_BIT;
-	}
-
-    if (newinputs & VERT_FOG_COORD)
-        ind |= S3V_FOG_BIT;
-
-	if (vmesa->SetupIndex & S3V_PTEX_BIT)
-		ind = ~0;
-
-    ind &= vmesa->SetupIndex;
-
-	DEBUG(("vmesa->SetupIndex = 0x%x\n", vmesa->SetupIndex));
-	DEBUG(("ind = 0x%x\n", ind));
-	DEBUG(("ind & vmesa->SetupIndex = 0x%x\n", (ind & vmesa->SetupIndex)));
-
-	if (ind) {
-		setup_tab[ind].emit( ctx, start, count, v, stride );   
-	}
-#endif
-}
-
-void s3vChooseVertexState( GLcontext *ctx )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT( ctx );
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-   GLuint ind = S3V_XYZW_BIT | S3V_RGBA_BIT;
-
-   /* FIXME: will segv in tnl_dd/t_dd_vbtmp.h (line 196) on some demos */
-/*
-   if (ctx->Fog.Enabled)
-      ind |= S3V_FOG_BIT;
-*/
-
-
-   if (ctx->Texture.Unit[0]._ReallyEnabled) {
-      _tnl_need_projected_coords( ctx, GL_FALSE );
-      ind |= S3V_TEX0_BIT;
-   } else {
-      _tnl_need_projected_coords( ctx, GL_TRUE );
-   }
-
-   vmesa->SetupIndex = ind;
-
-   if (ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED)) {
-      tnl->Driver.Render.Interp = s3v_interp_extras;
-      tnl->Driver.Render.CopyPV = s3v_copy_pv_extras;
-   } else {
-      tnl->Driver.Render.Interp = setup_tab[ind].interp;
-      tnl->Driver.Render.CopyPV = setup_tab[ind].copy_pv;
-   }
-}
-
-
-void s3vInitVB( GLcontext *ctx )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   GLuint size = TNL_CONTEXT(ctx)->vb.Size;
-
-   vmesa->verts = (char *)ALIGN_MALLOC(size * 4 * 16, 32);
-
-   {
-      static int firsttime = 1;
-      if (firsttime) {
-	 init_setup_tab();
-	 firsttime = 0;
-	 vmesa->vertex_stride_shift = 6 /* 4 */; /* FIXME - only one vertex setup */
-      }
-   }
-}
-
-
-void s3vFreeVB( GLcontext *ctx )
-{
-   s3vContextPtr vmesa = S3V_CONTEXT(ctx);
-   if (vmesa->verts) {
-      ALIGN_FREE(vmesa->verts);
-      vmesa->verts = 0;
-   }
-
-   if (vmesa->UbyteSecondaryColor.Ptr) {
-      ALIGN_FREE((void *)vmesa->UbyteSecondaryColor.Ptr);
-      vmesa->UbyteSecondaryColor.Ptr = 0;
-   }
-
-   if (vmesa->UbyteColor.Ptr) {
-      ALIGN_FREE((void *)vmesa->UbyteColor.Ptr);
-      vmesa->UbyteColor.Ptr = 0;
-   }
-}
diff --git a/src/mesa/drivers/dri/s3v/s3v_vb.h b/src/mesa/drivers/dri/s3v/s3v_vb.h
deleted file mode 100644
index 0fd5437380..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_vb.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#ifndef S3VVB_INC
-#define S3VVB_INC
-
-#include "main/mtypes.h"
-#include "swrast/swrast.h"
-
-#define _S3V_NEW_VERTEX (_NEW_TEXTURE |		\
-			   _DD_NEW_TRI_UNFILLED |	\
-			   _DD_NEW_TRI_LIGHT_TWOSIDE)
-
-
-extern void s3vChooseVertexState( GLcontext *ctx );
-extern void s3vCheckTexSizes( GLcontext *ctx );
-extern void s3vBuildVertices( GLcontext *ctx, 
-				GLuint start, 
-				GLuint count,
-				GLuint newinputs );
-
-
-extern void s3v_import_float_colors( GLcontext *ctx );
-extern void s3v_import_float_spec_colors( GLcontext *ctx );
-
-extern void s3v_translate_vertex( GLcontext *ctx, 
-				    const s3vVertex *src, 
-				    SWvertex *dst );
-
-extern void s3vInitVB( GLcontext *ctx );
-extern void s3vFreeVB( GLcontext *ctx );
-
-extern void s3v_print_vertex( GLcontext *ctx, const s3vVertex *v );
-#if 0
-extern void s3vPrintSetupFlags(char *msg, GLuint flags );
-#endif
-
-#endif
diff --git a/src/mesa/drivers/dri/s3v/s3v_xmesa.c b/src/mesa/drivers/dri/s3v/s3v_xmesa.c
deleted file mode 100644
index 85f1481769..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_xmesa.c
+++ /dev/null
@@ -1,341 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#include "s3v_context.h"
-#include "s3v_vb.h"
-#include "s3v_dri.h"
-#include "main/context.h"
-#include "main/matrix.h"
-#include "main/framebuffer.h"
-#include "main/renderbuffer.h"
-#include "main/viewport.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-#include "vbo/vbo.h"
-
-/* #define DEBUG(str) printf str */
-
-static const __DRIconfig **
-s3vInitScreen(__DRIscreen *sPriv)
-{
-    sPriv->private = (void *) s3vCreateScreen( sPriv );
-
-    if (!sPriv->private) {
-	s3vDestroyScreen( sPriv );
-	return GL_FALSE;
-    }
-
-   return NULL;
-}
-
-static void 
-s3vDestroyContext(__DRIcontextPrivate *driContextPriv)
-{
-    s3vContextPtr vmesa = (s3vContextPtr)driContextPriv->driverPrivate;
-
-    if (vmesa) {
-      _swsetup_DestroyContext( vmesa->glCtx );
-      _tnl_DestroyContext( vmesa->glCtx );
-      _vbo_DestroyContext( vmesa->glCtx );
-      _swrast_DestroyContext( vmesa->glCtx );
-
-      s3vFreeVB( vmesa->glCtx );
-
-      /* free the Mesa context */
-      vmesa->glCtx->DriverCtx = NULL;
-      _mesa_destroy_context(vmesa->glCtx);
-
-      _mesa_free(vmesa);
-      driContextPriv->driverPrivate = NULL;
-    }
-}
-
-
-static GLboolean
-s3vCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                   __DRIdrawablePrivate *driDrawPriv,
-                   const __GLcontextModes *mesaVis,
-                   GLboolean isPixmap )
-{
-   s3vScreenPtr screen = (s3vScreenPtr) driScrnPriv->private;
-
-   if (isPixmap) {
-      return GL_FALSE; /* not implemented */
-   }
-   else {
-      struct gl_framebuffer *fb = _mesa_create_framebuffer(mesaVis);
-
-      {
-         driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA, NULL, screen->cpp,
-                                 screen->frontOffset, screen->frontPitch,
-                                 driDrawPriv);
-         s3vSetSpanFunctions(frontRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
-      }
-
-      if (mesaVis->doubleBufferMode) {
-         driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA, NULL, screen->cpp,
-                                 screen->backOffset, screen->backPitch,
-                                 driDrawPriv);
-         s3vSetSpanFunctions(backRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &backRb->Base);
-         backRb->backBuffer = GL_TRUE;
-      }
-
-      if (mesaVis->depthBits == 16) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16, NULL, screen->cpp,
-                                 screen->depthOffset, screen->depthPitch,
-                                 driDrawPriv);
-         s3vSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-      else if (mesaVis->depthBits == 24) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24, NULL, screen->cpp,
-                                 screen->depthOffset, screen->depthPitch,
-                                 driDrawPriv);
-         s3vSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-
-      /* no h/w stencil yet?
-      if (mesaVis->stencilBits > 0) {
-         driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT, NULL,
-                                 screen->cpp, screen->depthOffset,
-                                 screen->depthPitch, driDrawPriv);
-         s3vSetSpanFunctions(stencilRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &stencilRb->Base);
-      }
-      */
-
-      _mesa_add_soft_renderbuffers(fb,
-                                   GL_FALSE, /* color */
-                                   GL_FALSE, /* depth */
-                                   mesaVis->stencilBits > 0,
-                                   mesaVis->accumRedBits > 0,
-                                   GL_FALSE, /* alpha */
-                                   GL_FALSE /* aux */);
-      driDrawPriv->driverPrivate = (void *) fb;
-
-      return (driDrawPriv->driverPrivate != NULL);
-   }
-}
-
-
-static void
-s3vDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
-{
-   _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
-}
-
-static void
-s3vSwapBuffers(__DRIdrawablePrivate *drawablePrivate)
-{
-   __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePrivate;
-   __DRIscreenPrivate *sPriv;
-   GLcontext *ctx;
-   s3vContextPtr vmesa;
-   s3vScreenPtr s3vscrn;
-   
-   vmesa = (s3vContextPtr) dPriv->driContextPriv->driverPrivate;
-   sPriv = vmesa->driScreen;
-   s3vscrn = vmesa->s3vScreen;
-   ctx = vmesa->glCtx;
-
-   DEBUG(("*** s3vSwapBuffers ***\n"));
-
-/* DMAFLUSH(); */
-
-   _mesa_notifySwapBuffers( ctx );
-
-   vmesa = (s3vContextPtr) dPriv->driContextPriv->driverPrivate;
-/*    driScrnPriv = vmesa->driScreen; */
-
-/*    if (vmesa->EnabledFlags & S3V_BACK_BUFFER) */
-
-/*	_mesa_notifySwapBuffers( ctx );  */
-#if 1
-{	
-	int x0, y0, x1, y1;
-/*	
-	int nRect = dPriv->numClipRects;
-	XF86DRIClipRectPtr pRect = dPriv->pClipRects;
-
-	__DRIscreenPrivate *driScrnPriv = vmesa->driScreen;
-*/
-
-/*	
-	DEBUG(("s3vSwapBuffers: S3V_BACK_BUFFER = 1 - nClip = %i\n", nRect));
-*/
-/*	vmesa->drawOffset=vmesa->s3vScreen->backOffset; */
-
-	x0 = dPriv->x;
-	y0 = dPriv->y;
-
-	x1 = x0 + dPriv->w - 1;
-	y1 = y0 + dPriv->h - 1;
-
-	DMAOUT_CHECK(BITBLT_SRC_BASE, 15);
-		DMAOUT(vmesa->s3vScreen->backOffset);
-		DMAOUT(0); /* 0xc0000000 */
-		DMAOUT( ((x0 << 16) | x1) );
-		DMAOUT( ((y0 << 16) | y1) );
-		DMAOUT( (vmesa->DestStride << 16) | vmesa->SrcStride );
-		DMAOUT( (~(0)) );
-		DMAOUT( (~(0)) );
-		DMAOUT(0);
-		DMAOUT(0);
-       /* FIXME */
-		DMAOUT(0);
-		DMAOUT(0);
-		DMAOUT( (0x01 | /* Autoexecute */
-			 0x02 | /* clip */
-			 0x04 | /* 16 bit */
-			 0x20 | /* draw */
-			0x400 | /* word alignment (bit 10=1) */
-			(0x2 << 11) | /*  offset = 1 byte */
-			(0xCC << 17) |	/* rop #204 */
-			(0x3 << 25)) ); /* l-r, t-b */
-		DMAOUT(vmesa->ScissorWH);
-		DMAOUT( /* 0 */ vmesa->SrcXY );
-		DMAOUT( (dPriv->x << 16) | dPriv->y );
-	DMAFINISH();
-
-	DMAFLUSH();
-
-	vmesa->restore_primitive = -1;
-
-}
-#endif
-}
-
-static GLboolean 
-s3vMakeCurrent(__DRIcontextPrivate *driContextPriv,
-		 __DRIdrawablePrivate *driDrawPriv,
-		 __DRIdrawablePrivate *driReadPriv)
-{
-	int x1,x2,y1,y2;
-	int cx, cy, cw, ch;
-	unsigned int src_stride, dest_stride;
-	int cl;
-
-	s3vContextPtr vmesa;
-	__DRIdrawablePrivate *dPriv = driDrawPriv;
-	vmesa = (s3vContextPtr) dPriv->driContextPriv->driverPrivate;
-	
-	DEBUG(("s3vMakeCurrent\n"));
-
-	DEBUG(("dPriv->x=%i y=%i w=%i h=%i\n", dPriv->x, dPriv->y,
-		dPriv->w, dPriv->h));
-
-	if (driContextPriv) {
-	GET_CURRENT_CONTEXT(ctx);
-	s3vContextPtr oldVirgeCtx = ctx ? S3V_CONTEXT(ctx) : NULL;
-	s3vContextPtr newVirgeCtx = (s3vContextPtr) driContextPriv->driverPrivate;
-
-	if ( newVirgeCtx != oldVirgeCtx ) {
-
-		newVirgeCtx->dirty = ~0;
-		cl = 1;
-		DEBUG(("newVirgeCtx != oldVirgeCtx\n"));
-/*		s3vUpdateClipping(newVirgeCtx->glCtx ); */
-	}
-
-	if (newVirgeCtx->driDrawable != driDrawPriv) {
-	    newVirgeCtx->driDrawable = driDrawPriv;
-		DEBUG(("driDrawable != driDrawPriv\n"));
-		s3vUpdateWindow ( newVirgeCtx->glCtx );
-		s3vUpdateViewportOffset( newVirgeCtx->glCtx );
-/*		s3vUpdateClipping(newVirgeCtx->glCtx ); */
-	}
-/*
-	s3vUpdateWindow ( newVirgeCtx->glCtx );
-	s3vUpdateViewportOffset( newVirgeCtx->glCtx );
-*/
-
-/*
-	_mesa_make_current( newVirgeCtx->glCtx,
-                          (GLframebuffer *) driDrawPriv->driverPrivate,
-                          (GLframebuffer *) driReadPriv->driverPrivate );
-
-	_mesa_set_viewport(newVirgeCtx->glCtx, 0, 0,
-	                  newVirgeCtx->driDrawable->w,
-			  newVirgeCtx->driDrawable->h);
-*/
-
-#if 0
-	newVirgeCtx->Window &= ~W_GIDMask;
-	newVirgeCtx->Window |= (driDrawPriv->index << 5);
-	CHECK_DMA_BUFFER(newVirgeCtx,1);
-	WRITE(newVirgeCtx->buf, S3VWindow, newVirgeCtx->Window);
-#endif
-
-	newVirgeCtx->new_state |= S3V_NEW_WINDOW; /* FIXME */
-
-	_mesa_make_current( newVirgeCtx->glCtx, 
-                            (GLframebuffer *) driDrawPriv->driverPrivate,
-                            (GLframebuffer *) driReadPriv->driverPrivate );
-
-	if (!newVirgeCtx->glCtx->Viewport.Width) {
-	    _mesa_set_viewport(newVirgeCtx->glCtx, 0, 0, 
-					driDrawPriv->w, driDrawPriv->h);
-
-/*		s3vUpdateClipping(newVirgeCtx->glCtx ); */
-	}
-
-/*
-	if (cl) {
-		s3vUpdateClipping(newVirgeCtx->glCtx );
-		cl =0;
-	}
-*/
-
-	newVirgeCtx->new_state |= S3V_NEW_CLIP;
-
-        if (1) {
-           cx = dPriv->x;
-           cw = dPriv->w;
-           cy = dPriv->y;
-           ch = dPriv->h;
-        }
-        
-        x1 = y1 = 0;
-        x2 = cw-1;
-        y2 = ch-1;
-
-        /*  src_stride = vmesa->s3vScreen->w * vmesa->s3vScreen->cpp; 
-            dest_stride = ((x2+31)&~31) * vmesa->s3vScreen->cpp; */
-        src_stride = vmesa->driScreen->fbWidth * 2;
-        dest_stride = ((x2+31)&~31) * 2;
-    } else {
-       _mesa_make_current( NULL, NULL, NULL );
-    }
-
-    return GL_TRUE;
-}
-
-
-static GLboolean 
-s3vUnbindContext( __DRIcontextPrivate *driContextPriv )
-{
-   return GL_TRUE;
-}
-
-const struct __DriverAPIRec driDriverAPI = {
-   .InitScreen	   = s3vInitScreen,
-   .DestroyScreen  = s3vDestroyScreen,
-   .CreateContext  = s3vCreateContext,
-   .DestroyContext = s3vDestroyContext,
-   .CreateBuffer   = s3vCreateBuffer,
-   .DestroyBuffer  = s3vDestroyBuffer,
-   .SwapBuffers	   = s3vSwapBuffers,
-   .MakeCurrent	   = s3vMakeCurrent,
-   .UnbindContext  = s3vUnbindContext,
-};
diff --git a/src/mesa/drivers/dri/s3v/s3virgetri.h b/src/mesa/drivers/dri/s3v/s3virgetri.h
deleted file mode 100644
index 5519cfd741..0000000000
--- a/src/mesa/drivers/dri/s3v/s3virgetri.h
+++ /dev/null
@@ -1,383 +0,0 @@
-/*
- * Author: Max Lingua <sunmax@libero.it>
- */
-
-#define LOCAL_VARS \
-	int vert0, vert1, vert2; \
-	GLfloat y0, y1, y2, ydiff; \
-	int iy0, iy1, iy2; \
-	int x0, x1, x2, z0, z1, z2; \
-	int dy01, dy02, dy12, dx01, dx02, dx12; \
-	int delt02, delt01, delt12, end01, end12, start02; \
-	int zstart, arstart, gbstart; \
-	int deltzy, deltzx, deltarx, deltgbx, deltary, deltgby; \
-	GLubyte (*colours)[4]; \
-	GLubyte (*scolours)[4]; \
-	static int tp = 0; \
-	int tmp, lr
-
-#define LOCAL_TEX_VARS \
-	int u0, u1, u2; \
-	GLfloat ru0, ru1, ru2; \
-	int v0, v1, v2; \
-	GLfloat rv0, rv1, rv2; \
-	GLfloat w0, w1, w2; \
-	GLfloat rw0, rw1, rw2; \
-	int baseu, basev; \
-	int d0, d1, d2; \
-	int deltdx, deltvx, deltux, deltdy, deltvy, deltuy; \
-	int deltwx, deltwy; \
-	int rbaseu, rbasev; \
-	int dstart, ustart, wstart, vstart; \
-	static int stmp = 0; \
-	s3virgeTextureObject_t *t
-
-#define CULL_BACKFACE() \
-	do { \
-		GLfloat *w0 = VB->Win.data[e0]; \
-		GLfloat *w1 = VB->Win.data[e1]; \
-		GLfloat *w2 = VB->Win.data[e2]; \
-		float cull; \
-		cull = ctx->backface_sign * ((w1[0] - w0[0]) * (w0[1] - w2[1]) + \
-			(w1[1] - w0[1]) * (w2[0] - w0[0])); \
-		if (cull < 0) \
-			return; \
-	} while (0)
-	
-#define SORT_VERTICES() \
-	do { \
-		y0 = VB->Win.data[e0][1]; \
-		y1 = VB->Win.data[e1][1]; \
-		y2 = VB->Win.data[e2][1]; \
-		if (y1 > y0) { \
-			if (y2 > y0) {  \
-				vert0 = e0; \
-				if (y1 > y2) { vert2 = e1; vert1 = e2; } else { vert2 = e2; vert1 = e1; } \
-			} else { vert0 = e2; vert1 = e0; vert2 = e1; } \
-		} else { \
-			if (y2 > y0) { vert0 = e1; vert1 = e0; vert2 = e2; } else { \
-				vert2 = e0; \
-				if (y2 > y1) { vert0 = e1; vert1 = e2; } else { vert0 = e2; vert1 = e1; } \
-			} \
-		} \
-	} while (0)
-
-#define SET_VARIABLES() \
-	do { \
-		iy0 = y0 = ((s3virgeDB->height - (VB->Win.data[vert0][1]))); \
-		iy1 = y1 = ((s3virgeDB->height - (VB->Win.data[vert1][1]))); \
-		iy2 = y2 = ((s3virgeDB->height - (VB->Win.data[vert2][1]))); \
-		if (iy0 == iy2) { return; } \
-		ydiff = y0 - (float)iy0; \
-		x0 = ((VB->Win.data[vert0][0]) * 1024.0 * 1024.0); \
-		x1 = ((VB->Win.data[vert1][0]) * 1024.0 * 1024.0); \
-		x2 = ((VB->Win.data[vert2][0]) * 1024.0 * 1024.0); \
-		z0 = (VB->Win.data[vert0][2] * 1024.0 * 32.0); \
-		z1 = (VB->Win.data[vert1][2] * 1024.0 * 32.0); \
-		z2 = (VB->Win.data[vert2][2] * 1024.0 * 32.0); \
-		dx12 = x2 - x1; \
-		dy12 = iy1 - iy2; \
-		dx01 = x1 - x0; \
-		dy01 = iy0 - iy1; \
-		dx02 = x2 - x0; \
-		dy02 = iy0 - iy2; \
-		delt12 = delt02 = delt01 = 0; \
-	} while (0)
-	
-#define SET_TEX_VARIABLES() \
-	do { \
-		t = ((s3virgeTextureObject_t *)ctx->Texture.Unit[0].Current->DriverData); \
-		deltwx = deltwy = wstart = deltdx = deltdy = dstart = 0; \
-		u0 = (VB->TexCoordPtr[0]->data[vert0][0] * (GLfloat)(t->tObj->Image[0]->Width) * 256.0); \
-		u1 = (VB->TexCoordPtr[0]->data[vert1][0] * (GLfloat)(t->tObj->Image[0]->Width) * 256.0); \
-		u2 = (VB->TexCoordPtr[0]->data[vert2][0] * (GLfloat)(t->tObj->Image[0]->Width) * 256.0); \
-		v0 = (VB->TexCoordPtr[0]->data[vert0][1] * (GLfloat)(t->tObj->Image[0]->Height) * 256.0); \
-		v1 = (VB->TexCoordPtr[0]->data[vert1][1] * (GLfloat)(t->tObj->Image[0]->Height) * 256.0); \
-		v2 = (VB->TexCoordPtr[0]->data[vert2][1] * (GLfloat)(t->tObj->Image[0]->Height) * 256.0); \
-		w0 = (VB->Win.data[vert0][3]); \
-		w1 = (VB->Win.data[vert1][3]); \
-		w2 = (VB->Win.data[vert2][3]); \
-	} while (0)
-
-#define FLATSHADE_COLORS() \
-	do { \
-		GLubyte *col = &(colours[pv][0]); \
-		deltarx = deltary = deltgbx = deltgby = 0; \
-		gbstart = (((col[1]) << 23) | ((col[2]) << 7)); \
-		arstart = (((col[3]) << 23) | ((col[0]) << 7)); \
-	} while (0)
-
-#define GOURAUD_COLORS() \
-	do { \
- 		int ctmp, ctmp2, ctmp3, ctmp4; \
-		GLubyte *col0, *col1, *col2; \
-		col0 = &(colours[vert0][0]); \
-		col1 = &(colours[vert1][0]); \
-		col2 = &(colours[vert2][0]); \
-		ctmp = ((col2[3] - col0[3]) << 7) / dy02; \
-		ctmp2 = ((col2[0] - col0[0]) << 7) / dy02; \
-		deltary = ((ctmp << 16) & 0xFFFF0000) | (ctmp2 & 0xFFFF); \
-		ctmp3 = ((col2[1] - col0[1]) << 7) / dy02; \
-		ctmp4 = ((col2[2] - col0[2]) << 7) / dy02; \
-		deltgby = ((ctmp3 << 16) & 0xFFFF0000) | (ctmp4 & 0xFFFF); \
-		gbstart = (((int)((ydiff * ctmp3) + (col0[1] << 7)) << 16) & 0x7FFF0000) | \
-          		  ((int)((ydiff * ctmp4) + (col0[2] << 7)) & 0x7FFF); \
-		arstart = (((int)((ydiff * ctmp) + (col0[3] << 7)) << 16) & 0x7FFF0000) | \
-          		  ((int)((ydiff * ctmp2) + (col0[0] << 7)) & 0x7FFF); \
-		if (tmp) { \
-			int ax, rx, gx, bx; \
-			ax = ((col1[3] << 7) - (dy01 * ctmp + (col0[3] << 7))) / tmp; \
-			rx = ((col1[0] << 7) - (dy01 * ctmp2 + (col0[0] << 7))) / tmp; \
-			gx = ((col1[1] << 7) - (dy01 * ctmp3 + (col0[1] << 7))) / tmp; \
-			bx = ((col1[2] << 7) - (dy01 * ctmp4 + (col0[2] << 7))) / tmp; \
-			deltarx = ((ax << 16) & 0xFFFF0000) | (rx & 0xFFFF); \
-			deltgbx = ((gx << 16) & 0xFFFF0000) | (bx & 0xFFFF); \
-		} else { \
-			deltgbx = deltarx = 0; \
- 		} \
- 	} while (0)
-
-#define SET_XY() \
-	do { \
-		delt02 = dx02 / dy02; \
-		if (dy12) delt12 = dx12 / dy12; \
-		if (dy01) delt01 = dx01 / dy01; \
-		start02 = (ydiff * delt02) + x0; \
-		end01 = (ydiff * delt01) + x0; \
-		end12 = ((y1 - (GLfloat)iy1) * delt12) + x1; \
-	} while (0)
-
-#define SET_DIR() \
-	do { \
-		tmp = x1 - (dy01 * delt02 + x0); \
-		if (tmp > 0) { \
-			lr = 0x80000000; \
-		} else { \
-			tmp *= -1; \
-			lr = 0; \
-		} \
-		tmp >>= 20; \
-	} while (0)
-
-#define SET_Z() \
-	do { \
-		deltzy = (z2 - z0) / dy02; \
-		if (tmp) { \
-			deltzx = (z1 - (dy01 * deltzy + z0)) / tmp; \
-		} else { deltzx = 0; } \
-		zstart = (deltzy * ydiff) + z0; \
-	} while (0)
-
-#define SET_BASEUV() \
-	do { \
-		if (u0 < u1) { \
-			if (u0 < u2) { \
-				baseu = u0; \
-			} else { \
-				baseu = u2; \
-			} \
-		} else { \
-			if (u1 < u2) { \
-				baseu = u1; \
-			} else { \
-				baseu = u2; \
-			} \
-		} \
-		if (v0 < v1) { \
-			if (v0 < v2) { \
-				basev = v0; \
-			} else { \
-				basev = v2; \
-			} \
-		} else { \
-			if (v1 < v2) { \
-				basev = v1; \
-			} else { \
-				basev = v2; \
-			} \
-		} \
-	} while (0)
-
-#define SET_RW() \
-	do { \
-		/* GLfloat minW; \
-		if (w0 < w1) { \
-			if (w0 < w2) { \
-				minW = w0; \
-			} else { \
-				minW = w2; \
-			} \
-		} else { \
-			if (w1 < w2) { \
-				minW = w1; \
-			} else { \
-				minW = w2; \
-			} \
-		} */ \
-		rw0 = (512.0 * w0); \
-		rw1 = (512.0 * w1); \
-		rw2 = (512.0 * w2); \
-	} while (0)
-
-
-#define SET_D() \
-	do { \
-		GLfloat sxy, suv; \
-		int lev; \
-		suv = (VB->TexCoordPtr[0]->data[vert0][0] - \
-		       VB->TexCoordPtr[0]->data[vert2][0]) * \
-		      (VB->TexCoordPtr[0]->data[vert1][1] - \
-		       VB->TexCoordPtr[0]->data[vert2][1]) - \
-		      (VB->TexCoordPtr[0]->data[vert1][0] - \
-		       VB->TexCoordPtr[0]->data[vert2][0]) * \
-		      (VB->TexCoordPtr[0]->data[vert0][1] - \
-		       VB->TexCoordPtr[0]->data[vert2][2]); \
-		sxy = (VB->Win.data[vert0][0] - \
-		       VB->Win.data[vert2][0]) * \
-		      (VB->Win.data[vert1][1] - \
-		       VB->Win.data[vert2][1]) - \
-		      (VB->Win.data[vert1][0] - \
-		       VB->Win.data[vert2][0]) * \
-		      (VB->Win.data[vert0][1] - \
-		       VB->Win.data[vert2][2]); \
-		if (sxy < 0) sxy *= -1.0; \
-		if (suv < 0) suv *= -1.0; \
-		lev = *(int*)&suv - *(int *)&sxy; \
-		if (lev < 0) \
-			lev = 0; \
-		else \
-			lev >>=23; \
-		dstart = (lev << 27); \
-	} while (0)
-		
-
-
-#define SET_UVWD() \
-	do { \
-		SET_BASEUV(); \
-		SET_RW(); \
-		SET_D(); \
-		ru0 = (((u0 - baseu) * rw0)); \
-		ru1 = (((u1 - baseu) * rw1)); \
-		ru2 = (((u2 - baseu) * rw2)); \
-		rv0 = (((v0 - basev) * rw0)); \
-		rv1 = (((v1 - basev) * rw1)); \
-		rv2 = (((v2 - basev) * rw2)); \
-		while (baseu < 0) { baseu += (t->tObj->Image[0]->Width << 8); } \
-		while (basev < 0) { basev += (t->tObj->Image[0]->Height << 8); } \
-		if (!(baseu & 0xFF)) { baseu = (baseu >> 8); } else { baseu = (baseu >> 8) + 1; } \
-		if ((basev & 0x80) || !(basev & 0xFF)) { basev = (basev >> 8); } else { basev = (basev >> 8) - 1; } \
-		rbaseu = (baseu) << (16 - t->widthLog2); \
-		rbasev = (basev) << (16 - t->widthLog2); \
-		deltuy = (((ru2 - ru0) / dy02)); \
-		deltvy = (((rv2 - rv0) / dy02)); \
-		rw0 *= (1024.0 * 512.0); \
-		rw1 *= (1024.0 * 512.0); \
-		rw2 *= (1024.0 * 512.0); \
-		deltwy = ((rw2 - rw0) / dy02); \
-		if (tmp) { \
-			deltux = ((ru1 - (dy01 * deltuy + ru0)) / tmp); \
-			deltvx = ((rv1 - (dy01 * deltvy + rv0)) / tmp); \
-			deltwx = ((rw1 - (dy01 * deltwy + rw0)) / tmp); \
-		} else { deltux = deltvx = deltwx = 0; } \
-		ustart = (deltuy * ydiff) + (ru0); \
-		vstart = (deltvy * ydiff) + (rv0); \
-		wstart = (deltwy * ydiff) + (rw0); \
-	} while (0)
-
-
-#define SEND_COLORS() \
-	do { \
-		WAITFIFOEMPTY(6); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_GBX), deltgbx); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_ARX), deltarx); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_GBY), deltgby); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_ARY), deltary); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_GS_BS), gbstart); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_AS_RS), arstart); \
-	} while (0)
-
-#define SEND_VERTICES() \
-	do { \
-		WAITFIFOEMPTY(6); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_ZSTART), zstart); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_ZXD), deltzx); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_ZYD), deltzy); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TXDELTA12), delt12); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TXEND12), end12); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TXDELTA01), delt01); \
-		WAITFIFOEMPTY(5); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TXEND01), end01); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TXDELTA02), delt02); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TXSTART02), start02); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TYS), iy0); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_TY01_Y12), \
-				((((iy0 - iy1) & 0x7FF) << 16) | \
-				((iy1 - iy2) & 0x7FF) | lr)); \
-	} while (0)
-
-#define SEND_UVWD() \
-	do { \
-		WAITFIFOEMPTY(7); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_BASEV), (rbasev & 0xFFFF)); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_BASEU), (0xa0000000 | (rbaseu & 0xFFFF))); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_WXD), deltwx); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_WYD), deltwy); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_WSTART), wstart); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_DXD), deltdx); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_VXD), deltvx); \
-		WAITFIFOEMPTY(7); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_UXD), deltux); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_DYD), deltdy); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_VYD), deltvy); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_UYD), deltuy); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_DSTART), dstart); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_VSTART), vstart); \
-		OUTREG( (S3VIRGE_3DTRI_REG | S3VIRGE_3DTRI_USTART), ustart); \
-	} while (0)
-
-#define DMA_SEND_UVWD() \
-	do { \
-		DMAOUT((rbasev & 0xFFFF)); \
-		DMAOUT((0xa0000000 | (rbaseu & 0xFFFF))); \
-		DMAOUT(deltwx); \
-		DMAOUT(deltwy); \
-		DMAOUT(wstart); \
-		DMAOUT(deltdx); \
-		DMAOUT(deltvx); \
-		DMAOUT(deltux); \
-		DMAOUT(deltdy); \
-		DMAOUT(deltvy); \
-		DMAOUT(deltuy); \
-		DMAOUT(dstart); \
-		DMAOUT(vstart); \
-		DMAOUT(ustart); \
-	} while (0)
-
-
-#define DMA_SEND_COLORS() \
-	do { \
-		DMAOUT(deltgbx); \
-		DMAOUT(deltarx); \
-		DMAOUT(deltgby); \
-		DMAOUT(deltary); \
-		DMAOUT(gbstart); \
-		DMAOUT(arstart); \
-	} while (0)
-
-#define DMA_SEND_VERTICES() \
-	do { \
-		DMAOUT(deltzx); \
-		DMAOUT(deltzy); \
-		DMAOUT(zstart); \
-		DMAOUT(delt12); \
-		DMAOUT(end12); \
-		DMAOUT(delt01); \
-		DMAOUT(end01); \
-		DMAOUT(delt02); \
-		DMAOUT(start02); \
-		DMAOUT(iy0); \
-		DMAOUT(((((iy0 - iy1) & 0x7FF) << 16) | \
-			((iy1 - iy2) & 0x7FF) | lr)); \
-	} while (0)
-
diff --git a/src/mesa/drivers/dri/savage/Makefile b/src/mesa/drivers/dri/savage/Makefile
index 018482f66b..2e5c40802c 100644
--- a/src/mesa/drivers/dri/savage/Makefile
+++ b/src/mesa/drivers/dri/savage/Makefile
@@ -27,4 +27,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/sis/Makefile b/src/mesa/drivers/dri/sis/Makefile
index d2354e6776..ad009fc239 100644
--- a/src/mesa/drivers/dri/sis/Makefile
+++ b/src/mesa/drivers/dri/sis/Makefile
@@ -34,4 +34,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/swrast/Makefile b/src/mesa/drivers/dri/swrast/Makefile
index 5f3a4f2191..771169c1ff 100644
--- a/src/mesa/drivers/dri/swrast/Makefile
+++ b/src/mesa/drivers/dri/swrast/Makefile
@@ -21,4 +21,3 @@ SWRAST_COMMON_SOURCES = \
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c
index a858af30c1..f4947daa06 100644
--- a/src/mesa/drivers/dri/swrast/swrast.c
+++ b/src/mesa/drivers/dri/swrast/swrast.c
@@ -43,6 +43,7 @@
 #include "tnl/t_pipeline.h"
 #include "vbo/vbo.h"
 #include "drivers/common/driverfuncs.h"
+#include "drivers/common/meta.h"
 #include "utils.h"
 
 #include "swrast_priv.h"
@@ -62,6 +63,7 @@
 #define need_GL_SGI_color_table
 
 /* sw extensions not associated with some GL version */
+#define need_GL_ARB_draw_elements_base_vertex
 #define need_GL_ARB_shader_objects
 #define need_GL_ARB_vertex_array_object
 #define need_GL_ARB_vertex_program
@@ -95,6 +97,8 @@ const struct dri_extension card_extensions[] =
     { "GL_EXT_histogram",		GL_EXT_histogram_functions },
     { "GL_SGI_color_table",		GL_SGI_color_table_functions },
 
+    { "GL_ARB_depth_clamp",		NULL },
+    { "GL_ARB_draw_elements_base_vertex", GL_ARB_draw_elements_base_vertex_functions },
     { "GL_ARB_shader_objects",		GL_ARB_shader_objects_functions },
     { "GL_ARB_vertex_array_object",	GL_ARB_vertex_array_object_functions },
     { "GL_ARB_vertex_program",		GL_ARB_vertex_program_functions },
@@ -109,8 +113,10 @@ const struct dri_extension card_extensions[] =
     { "GL_EXT_paletted_texture",	GL_EXT_paletted_texture_functions },
     { "GL_EXT_stencil_two_side",	GL_EXT_stencil_two_side_functions },
     { "GL_MESA_resize_buffers",		GL_MESA_resize_buffers_functions },
+    { "GL_NV_depth_clamp",		NULL },
     { "GL_NV_vertex_program",		GL_NV_vertex_program_functions },
     { "GL_NV_fragment_program",		GL_NV_fragment_program_functions },
+    { "GL_NV_fragment_program_option",	NULL },
     { NULL,				NULL }
 };
 
@@ -645,6 +651,8 @@ driCreateNewContext(__DRIscreen *screen, const __DRIconfig *config,
     _mesa_enable_2_0_extensions(mesaCtx);
     _mesa_enable_2_1_extensions(mesaCtx);
 
+    _mesa_meta_init(mesaCtx);
+
     return ctx;
 }
 
@@ -656,6 +664,7 @@ driDestroyContext(__DRIcontext *ctx)
 
     if (ctx) {
 	mesaCtx = &ctx->Base;
+        _mesa_meta_free(mesaCtx);
 	_swsetup_DestroyContext( mesaCtx );
 	_swrast_DestroyContext( mesaCtx );
 	_tnl_DestroyContext( mesaCtx );
diff --git a/src/mesa/drivers/dri/tdfx/Makefile b/src/mesa/drivers/dri/tdfx/Makefile
index 092c580fea..b9f25db4fe 100644
--- a/src/mesa/drivers/dri/tdfx/Makefile
+++ b/src/mesa/drivers/dri/tdfx/Makefile
@@ -32,6 +32,4 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
-
 
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
index 1f7257eaea..f6a48b3ae1 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
@@ -176,6 +176,54 @@ logbase2(int n)
 }
 
 
+static void
+tdfxGenerateMipmap(GLcontext *ctx, GLenum target,
+                   struct gl_texture_object *texObj)
+{
+   GLint mipWidth, mipHeight;
+   tdfxMipMapLevel *mip;
+   struct gl_texture_image *mipImage; /* the new/next image */
+   struct gl_texture_image *texImage;
+   const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target);
+   GLint level = texObj->BaseLevel;
+   GLsizei width, height, texelBytes;
+   const tdfxMipMapLevel *mml;
+
+   texImage = _mesa_get_tex_image(ctx, texObj, target, level);
+   assert(!texImage->IsCompressed);
+
+   mml = TDFX_TEXIMAGE_DATA(texImage);
+
+   width = texImage->Width;
+   height = texImage->Height;
+   while (level < texObj->MaxLevel && level < maxLevels - 1) {
+      mipWidth = width / 2;
+      if (!mipWidth) {
+         mipWidth = 1;
+      }
+      mipHeight = height / 2;
+      if (!mipHeight) {
+         mipHeight = 1;
+      }
+      if ((mipWidth == width) && (mipHeight == height)) {
+         break;
+      }
+      ++level;
+      mipImage = _mesa_select_tex_image(ctx, texObj, target, level);
+      mip = TDFX_TEXIMAGE_DATA(mipImage);
+      _mesa_halve2x2_teximage2d(ctx,
+                                texImage,
+                                texelBytes,
+                                mml->width, mml->height,
+                                texImage->Data, mipImage->Data);
+      texImage = mipImage;
+      mml = mip;
+      width = mipWidth;
+      height = mipHeight;
+   }
+}
+
+
 /*
  * Compute various texture image parameters.
  * Input:  w, h - source texture width and height
@@ -1397,45 +1445,6 @@ tdfxTexImage2D(GLcontext *ctx, GLenum target, GLint level,
                                           width, height, 1,
                                           format, type, pixels, packing);
        }
-
-      /* GL_SGIS_generate_mipmap */
-      if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-         GLint mipWidth, mipHeight;
-         tdfxMipMapLevel *mip;
-         struct gl_texture_image *mipImage;
-         const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target);
-   
-         assert(!texImage->IsCompressed);
-   
-         while (level < texObj->MaxLevel && level < maxLevels - 1) {
-            mipWidth = width / 2;
-            if (!mipWidth) {
-               mipWidth = 1;
-            }
-            mipHeight = height / 2;
-            if (!mipHeight) {
-               mipHeight = 1;
-            }
-            if ((mipWidth == width) && (mipHeight == height)) {
-               break;
-            }
-            _mesa_TexImage2D(target, ++level, internalFormat,
-                             mipWidth, mipHeight, border,
-                             format, type,
-                             NULL);
-            mipImage = _mesa_select_tex_image(ctx, texObj, target, level);
-            mip = TDFX_TEXIMAGE_DATA(mipImage);
-            _mesa_halve2x2_teximage2d(ctx,
-                                      texImage,
-                                      texelBytes,
-                                      mml->width, mml->height,
-                                      texImage->Data, mipImage->Data);
-            texImage = mipImage;
-            mml = mip;
-            width = mipWidth;
-            height = mipHeight;
-         }
-      }
     }
 
     RevalidateTexture(ctx, texObj);
@@ -1507,44 +1516,6 @@ tdfxTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
                                         format, type, pixels, packing);
     }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      GLint mipWidth, mipHeight;
-      tdfxMipMapLevel *mip;
-      struct gl_texture_image *mipImage;
-      const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target);
-
-      assert(!texImage->IsCompressed);
-
-      width = texImage->Width;
-      height = texImage->Height;
-      while (level < texObj->MaxLevel && level < maxLevels - 1) {
-         mipWidth = width / 2;
-         if (!mipWidth) {
-            mipWidth = 1;
-         }
-         mipHeight = height / 2;
-         if (!mipHeight) {
-            mipHeight = 1;
-         }
-         if ((mipWidth == width) && (mipHeight == height)) {
-            break;
-         }
-         ++level;
-         mipImage = _mesa_select_tex_image(ctx, texObj, target, level);
-         mip = TDFX_TEXIMAGE_DATA(mipImage);
-         _mesa_halve2x2_teximage2d(ctx,
-                                   texImage,
-                                   texelBytes,
-                                   mml->width, mml->height,
-                                   texImage->Data, mipImage->Data);
-         texImage = mipImage;
-         mml = mip;
-         width = mipWidth;
-         height = mipHeight;
-      }
-   }
-
     ti->reloadImages = GL_TRUE; /* signal the image needs to be reloaded */
     fxMesa->new_state |= TDFX_NEW_TEXTURE;  /* XXX this might be a bit much */
 }
@@ -1703,11 +1674,6 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target,
        MEMCPY(texImage->Data, data, texImage->CompressedSize);
     }
 
-    /* GL_SGIS_generate_mipmap */
-    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-       assert(!texImage->IsCompressed);
-    }
-
     RevalidateTexture(ctx, texObj);
 
     ti->reloadImages = GL_TRUE;
@@ -1770,11 +1736,6 @@ tdfxCompressedTexSubImage2D( GLcontext *ctx, GLenum target,
                                 texImage->Data);
     }
 
-    /* GL_SGIS_generate_mipmap */
-    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-       assert(!texImage->IsCompressed);
-    }
-
     RevalidateTexture(ctx, texObj);
 
     ti->reloadImages = GL_TRUE;
@@ -1914,4 +1875,5 @@ void tdfxInitTextureFuncs( struct dd_function_table *functions )
    functions->CompressedTexImage2D	= tdfxCompressedTexImage2D;
    functions->CompressedTexSubImage2D	= tdfxCompressedTexSubImage2D;
    functions->UpdateTexturePalette      = tdfxUpdateTexturePalette;
+   functions->GenerateMipmap            = tdfxGenerateMipmap;
 }
diff --git a/src/mesa/drivers/dri/trident/Makefile b/src/mesa/drivers/dri/trident/Makefile
deleted file mode 100644
index 9ee24c504c..0000000000
--- a/src/mesa/drivers/dri/trident/Makefile
+++ /dev/null
@@ -1,26 +0,0 @@
-# src/mesa/drivers/dri/trident/Makefile
-
-TOP = ../../../../..
-include $(TOP)/configs/current
-
-LIBNAME = trident_dri.so
-
-# Not yet
-# MINIGLX_SOURCES = server/trident_dri.c 
-
-DRIVER_SOURCES = \
-	trident_context.c \
-	trident_state.c \
-	trident_vb.c \
-	trident_dd.c \
-	trident_tris.c
-
-C_SOURCES = \
-	$(COMMON_SOURCES) \
-	$(DRIVER_SOURCES)
-
-ASM_SOURCES = 
-
-include ../Makefile.template
-
-symlinks:
diff --git a/src/mesa/drivers/dri/trident/trident_context.c b/src/mesa/drivers/dri/trident/trident_context.c
deleted file mode 100644
index b5126b07ea..0000000000
--- a/src/mesa/drivers/dri/trident/trident_context.c
+++ /dev/null
@@ -1,465 +0,0 @@
-/*
- * Copyright 2002 by Alan Hourihane, Sychdyn, North Wales, UK.
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Alan Hourihane not be used in
- * advertising or publicity pertaining to distribution of the software without
- * specific, written prior permission.  Alan Hourihane makes no representations
- * about the suitability of this software for any purpose.  It is provided
- * "as is" without express or implied warranty.
- *
- * ALAN HOURIHANE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
- * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
- * EVENT SHALL ALAN HOURIHANE BE LIABLE FOR ANY SPECIAL, INDIRECT OR
- * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
- * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
- *
- * Authors:  Alan Hourihane, <alanh@fairlite.demon.co.uk>
- *
- * Trident CyberBladeXP driver.
- *
- */
-#include "trident_dri.h"
-#include "trident_context.h"
-#include "trident_lock.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "vbo/vbo.h"
-
-#include "tnl/tnl.h"
-#include "tnl/t_pipeline.h"
-
-#include "main/context.h"
-#include "main/simple_list.h"
-#include "main/matrix.h"
-#include "main/extensions.h"
-#include "main/framebuffer.h"
-#include "main/renderbuffer.h"
-#include "main/viewport.h"
-#if defined(USE_X86_ASM)
-#include "x86/common_x86_asm.h"
-#endif
-#include "main/simple_list.h"
-#include "main/mm.h"
-#include "drirenderbuffer.h"
-
-#include "drivers/common/driverfuncs.h"
-#include "dri_util.h"
-#include "utils.h"
-
-static const struct tnl_pipeline_stage *trident_pipeline[] = {
-   &_tnl_vertex_transform_stage, 
-   &_tnl_normal_transform_stage, 
-   &_tnl_lighting_stage,
-   &_tnl_texgen_stage, 
-   &_tnl_texture_transform_stage, 
-   &_tnl_render_stage,		
-   0,
-};
-
-
-static GLboolean
-tridentCreateContext( const __GLcontextModes *glVisual,
-                      __DRIcontextPrivate *driContextPriv,
-                      void *sharedContextPrivate)
-{
-   GLcontext *ctx, *shareCtx;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   tridentContextPtr tmesa;
-   tridentScreenPtr tridentscrn;
-   struct dd_function_table functions;
-#if 0
-   drm_trident_sarea_t *saPriv=(drm_trident_sarea_t *)(((char*)sPriv->pSAREA)+
-						 sizeof(XF86DRISAREARec));
-#endif
-
-   tmesa = (tridentContextPtr) CALLOC( sizeof(*tmesa) );
-   if ( !tmesa ) return GL_FALSE;
-
-   /* Allocate the Mesa context */
-   if (sharedContextPrivate)
-      shareCtx = ((tridentContextPtr) sharedContextPrivate)->glCtx;
-   else
-      shareCtx = NULL;
-
-   _mesa_init_driver_functions(&functions);
-
-   tmesa->glCtx =
-      _mesa_create_context(glVisual, shareCtx, &functions, (void *)tmesa);
-
-   if (!tmesa->glCtx) {
-      FREE(tmesa);
-      return GL_FALSE;
-   }
-
-   tmesa->driContext = driContextPriv;
-   tmesa->driScreen = sPriv;
-   tmesa->driDrawable = NULL; /* Set by XMesaMakeCurrent */
-
-   tmesa->hHWContext = driContextPriv->hHWContext;
-   tmesa->driHwLock = (drmLock *)&sPriv->pSAREA->lock;
-   tmesa->driFd = sPriv->fd;
-#if 0
-   tmesa->sarea = saPriv;
-#endif
-
-   tridentscrn = tmesa->tridentScreen = (tridentScreenPtr)(sPriv->private);
-
-   ctx = tmesa->glCtx;
-
-   ctx->Const.MaxTextureLevels = 13;  /* 4K by 4K?  Is that right? */
-   ctx->Const.MaxTextureUnits = 1; /* Permedia 3 */
-
-   ctx->Const.MinLineWidth = 0.0;
-   ctx->Const.MaxLineWidth = 255.0;
-
-   ctx->Const.MinLineWidthAA = 0.0;
-   ctx->Const.MaxLineWidthAA = 65536.0;
-
-   ctx->Const.MinPointSize = 0.0;
-   ctx->Const.MaxPointSize = 255.0;
-
-   ctx->Const.MinPointSizeAA = 0.5; /* 4x4 quality mode */
-   ctx->Const.MaxPointSizeAA = 16.0; 
-   ctx->Const.PointSizeGranularity = 0.25;
-
-   ctx->Const.MaxDrawBuffers = 1;
-
-#if 0
-   tmesa->texHeap = mmInit( 0, tmesa->tridentScreen->textureSize );
-
-   make_empty_list(&tmesa->TexObjList);
-   make_empty_list(&tmesa->SwappedOut);
-
-   tmesa->CurrentTexObj[0] = 0;
-   tmesa->CurrentTexObj[1] = 0; /* Permedia 3, second texture */
-
-   tmesa->RenderIndex = ~0;
-#endif
-
-   /* Initialize the software rasterizer and helper modules.
-    */
-   _swrast_CreateContext( ctx );
-   _vbo_CreateContext( ctx );
-   _tnl_CreateContext( ctx );
-   _swsetup_CreateContext( ctx );
-
-   /* Install the customized pipeline:
-    */
-   _tnl_destroy_pipeline( ctx );
-   _tnl_install_pipeline( ctx, trident_pipeline );
-
-   /* Configure swrast to match hardware characteristics:
-    */
-   _swrast_allow_pixel_fog( ctx, GL_FALSE );
-   _swrast_allow_vertex_fog( ctx, GL_TRUE );
-
-   tridentInitVB( ctx );
-   tridentDDInitExtensions( ctx );
-   tridentDDInitDriverFuncs( ctx );
-   tridentDDInitStateFuncs( ctx );
-#if 0
-   tridentDDInitSpanFuncs( ctx );
-   tridentDDInitTextureFuncs( ctx );
-#endif
-   tridentDDInitTriFuncs( ctx );
-   tridentDDInitState( tmesa );
-
-   driContextPriv->driverPrivate = (void *)tmesa;
-
-   UNLOCK_HARDWARE(tmesa);
-
-   return GL_TRUE;
-}
-
-static void 
-tridentDestroyContext(__DRIcontextPrivate *driContextPriv)
-{
-    tridentContextPtr tmesa = (tridentContextPtr)driContextPriv->driverPrivate;
-
-    if (tmesa) {
-      _swsetup_DestroyContext( tmesa->glCtx );
-      _tnl_DestroyContext( tmesa->glCtx );
-      _vbo_DestroyContext( tmesa->glCtx );
-      _swrast_DestroyContext( tmesa->glCtx );
-
-      /* free the Mesa context */
-      tmesa->glCtx->DriverCtx = NULL;
-      _mesa_destroy_context(tmesa->glCtx);
-
-      _mesa_free(tmesa);
-      driContextPriv->driverPrivate = NULL;
-    }
-}
-
-
-static GLboolean
-tridentCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                   __DRIdrawablePrivate *driDrawPriv,
-                   const __GLcontextModes *mesaVis,
-                   GLboolean isPixmap )
-{
-   tridentScreenPtr screen = (tridentScreenPtr) driScrnPriv->private;
-
-   if (isPixmap) {
-      return GL_FALSE; /* not implemented */
-   }
-   else {
-      struct gl_framebuffer *fb = _mesa_create_framebuffer(mesaVis);
-
-      {
-         driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA, NULL, screen->cpp,
-                                 screen->frontOffset, screen->frontPitch,
-                                 driDrawPriv);
-         /*
-         tridentSetSpanFunctions(frontRb, mesaVis);
-         */
-         _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
-      }
-
-      if (mesaVis->doubleBufferMode) {
-         driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA, NULL, screen->cpp,
-                                 screen->backOffset, screen->backPitch,
-                                 driDrawPriv);
-         /*
-         tridentSetSpanFunctions(backRb, mesaVis);
-         */
-         _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &backRb->Base);
-      }
-
-      if (mesaVis->depthBits == 16) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16, NULL, screen->cpp,
-                                 screen->depthOffset, screen->depthPitch,
-                                 driDrawPriv);
-         /*
-         tridentSetSpanFunctions(depthRb, mesaVis);
-         */
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-      else if (mesaVis->depthBits == 24) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24, NULL, screen->cpp,
-                                 screen->depthOffset, screen->depthPitch,
-                                 driDrawPriv);
-         /*
-         tridentSetSpanFunctions(depthRb, mesaVis);
-         */
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-
-      /* no h/w stencil?
-      if (mesaVis->stencilBits > 0 && !swStencil) {
-         driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT);
-         tridentSetSpanFunctions(stencilRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &stencilRb->Base);
-      }
-      */
-
-      _mesa_add_soft_renderbuffers(fb,
-                                   GL_FALSE, /* color */
-                                   GL_FALSE, /* depth */
-                                   mesaVis->stencilBits > 0,
-                                   mesaVis->accumRedBits > 0,
-                                   GL_FALSE, /* alpha */
-                                   GL_FALSE /* aux */);
-      driDrawPriv->driverPrivate = (void *) fb;
-
-      return (driDrawPriv->driverPrivate != NULL);
-   }
-}
-
-
-static void
-tridentDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
-{
-   _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
-}
-
-static void
-tridentSwapBuffers(__DRIdrawablePrivate *drawablePrivate)
-{
-   __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePrivate;
-
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      tridentContextPtr tmesa;
-      GLcontext *ctx;
-      tmesa = (tridentContextPtr) dPriv->driContextPriv->driverPrivate;
-      ctx = tmesa->glCtx;
-      if (ctx->Visual.doubleBufferMode) {
-         _mesa_notifySwapBuffers( ctx );  /* flush pending rendering comands */
-         tridentCopyBuffer( dPriv );
-      }
-   }
-   else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      _mesa_problem(NULL, "tridentSwapBuffers: drawable has no context!\n");
-   }
-}
-
-static GLboolean 
-tridentMakeCurrent(__DRIcontextPrivate *driContextPriv,
-		 __DRIdrawablePrivate *driDrawPriv,
-		 __DRIdrawablePrivate *driReadPriv)
-{
-    if (driContextPriv) {
-	GET_CURRENT_CONTEXT(ctx);
-	tridentContextPtr oldCtx = ctx ? TRIDENT_CONTEXT(ctx) : NULL;
-	tridentContextPtr newCtx = (tridentContextPtr) driContextPriv->driverPrivate;
-
-	if ( newCtx != oldCtx ) {
-	    newCtx->dirty = ~0;
-	}
-
-	if (newCtx->driDrawable != driDrawPriv) {
-	    newCtx->driDrawable = driDrawPriv;
-#if 0
-	    tridentUpdateWindow ( newCtx->glCtx );
-	    tridentUpdateViewportOffset( newCtx->glCtx );
-#endif
-	}
-
-   newCtx->drawOffset = newCtx->tridentScreen->backOffset;
-   newCtx->drawPitch = newCtx->tridentScreen->backPitch;
-
-	_mesa_make_current( newCtx->glCtx, 
-                            (GLframebuffer *) driDrawPriv->driverPrivate,
-                            (GLframebuffer *) driReadPriv->driverPrivate );
-
-	if (!newCtx->glCtx->Viewport.Width) {
-	    _mesa_set_viewport(newCtx->glCtx, 0, 0, 
-					driDrawPriv->w, driDrawPriv->h);
-	}
-    } else {
-	_mesa_make_current( NULL, NULL, NULL );
-    }
-    return GL_TRUE;
-}
-
-
-static GLboolean 
-tridentUnbindContext( __DRIcontextPrivate *driContextPriv )
-{
-   return GL_TRUE;
-}
-
-
-static tridentScreenPtr
-tridentCreateScreen( __DRIscreenPrivate *sPriv )
-{
-   TRIDENTDRIPtr tDRIPriv = (TRIDENTDRIPtr)sPriv->pDevPriv;
-   tridentScreenPtr tridentScreen;
-
-   if (sPriv->devPrivSize != sizeof(TRIDENTDRIRec)) {
-      fprintf(stderr,"\nERROR!  sizeof(TRIDENTDRIRec) does not match passed size from device driver\n");
-      return GL_FALSE;
-   }
-
-    /* Allocate the private area */
-    tridentScreen = (tridentScreenPtr) CALLOC( sizeof(*tridentScreen) );
-    if ( !tridentScreen ) return NULL;
-
-   tridentScreen->driScreen = sPriv;
-
-   tridentScreen->frontOffset = tDRIPriv->frontOffset;
-   tridentScreen->backOffset = tDRIPriv->backOffset;
-   tridentScreen->depthOffset = tDRIPriv->depthOffset;
-   tridentScreen->frontPitch = tDRIPriv->frontPitch;
-   tridentScreen->backPitch = tDRIPriv->backPitch;
-   tridentScreen->depthPitch = tDRIPriv->depthPitch;
-   tridentScreen->width = tDRIPriv->width;
-   tridentScreen->height = tDRIPriv->height;
-
-printf("%d %d\n",tridentScreen->width,tridentScreen->height);
-printf("%d %d\n",tridentScreen->frontPitch,tridentScreen->backPitch);
-printf("offset 0x%x 0x%x\n",tridentScreen->backOffset,tridentScreen->depthOffset);
-
-   tridentScreen->mmio.handle = tDRIPriv->regs;
-   tridentScreen->mmio.size = 0x20000;
-    
-   if (drmMap(sPriv->fd,
-	 	tridentScreen->mmio.handle, tridentScreen->mmio.size,
-		(drmAddressPtr)&tridentScreen->mmio.map)) {
-	    FREE(tridentScreen);
-	    return GL_FALSE;
-    }
-printf("MAPPED at %p\n", tridentScreen->mmio.map);
-
-   return tridentScreen;
-}
-
-/* Destroy the device specific screen private data struct.
- */
-static void
-tridentDestroyScreen( __DRIscreenPrivate *sPriv )
-{
-    tridentScreenPtr tridentScreen = (tridentScreenPtr)sPriv->private;
-
-    FREE(tridentScreen);
-}
-
-static GLboolean 
-tridentInitDriver(__DRIscreenPrivate *sPriv)
-{
-    sPriv->private = (void *) tridentCreateScreen( sPriv );
-
-    if (!sPriv->private) {
-	tridentDestroyScreen( sPriv );
-	return GL_FALSE;
-    }
-
-    return GL_TRUE;
-}
-
-/**
- * This is the driver specific part of the createNewScreen entry point.
- * 
- * \todo maybe fold this into intelInitDriver
- *
- * \return the __GLcontextModes supported by this driver
- */
-const __DRIconfig **tridentInitScreen(__DRIscreenPrivate *psp)
-{
-   static const __DRIversion ddx_expected = { 4, 0, 0 };
-   static const __DRIversion dri_expected = { 3, 1, 0 };
-   static const __DRIversion drm_expected = { 1, 0, 0 };
-   
-   if ( ! driCheckDriDdxDrmVersions2( "Trident",
-				      &psp->dri_version, & dri_expected,
-				      &psp->ddx_version, & ddx_expected,
-				      &psp->drm_version, & drm_expected ) )
-      return NULL;
-
-   if (!tridentInitDriver(psp))
-	return NULL;
-
-    /* Wait... what?  This driver doesn't report any modes... */
-#if 0
-   TRIDENTDRIPtr dri_priv = (TRIDENTDRIPtr) psp->pDevPriv;
-   *driver_modes = tridentFillInModes( dri_priv->bytesPerPixel * 8,
-				       GL_TRUE );
-#endif
-
-   return NULL;
-}
-
-const struct __DriverAPIRec driDriverAPI = {
-   tridentInitScreen,
-   tridentDestroyScreen,
-   tridentCreateContext,
-   tridentDestroyContext,
-   tridentCreateBuffer,
-   tridentDestroyBuffer,
-   tridentSwapBuffers,
-   tridentMakeCurrent,
-   tridentUnbindContext,
-};
diff --git a/src/mesa/drivers/dri/trident/trident_context.h b/src/mesa/drivers/dri/trident/trident_context.h
deleted file mode 100644
index fbbb4a96e7..0000000000
--- a/src/mesa/drivers/dri/trident/trident_context.h
+++ /dev/null
@@ -1,242 +0,0 @@
-/*
- * Copyright 2002 by Alan Hourihane, Sychdyn, North Wales, UK.
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Alan Hourihane not be used in
- * advertising or publicity pertaining to distribution of the software without
- * specific, written prior permission.  Alan Hourihane makes no representations
- * about the suitability of this software for any purpose.  It is provided
- * "as is" without express or implied warranty.
- *
- * ALAN HOURIHANE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
- * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
- * EVENT SHALL ALAN HOURIHANE BE LIABLE FOR ANY SPECIAL, INDIRECT OR
- * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
- * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
- *
- * Authors:  Alan Hourihane, <alanh@fairlite.demon.co.uk>
- *
- * Trident CyberBladeXP driver.
- *
- */
-#ifndef _TRIDENT_CONTEXT_H_
-#define _TRIDENT_CONTEXT_H_
-
-#include "dri_util.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-#include "drm.h"
-#include "main/mm.h"
-
-#define SUBPIXEL_X (0.0F)
-#define SUBPIXEL_Y (0.125F)
-
-#define _TRIDENT_NEW_VERTEX (_NEW_TEXTURE |		\
-			   _DD_NEW_TRI_UNFILLED |	\
-			   _DD_NEW_TRI_LIGHT_TWOSIDE)
-
-#define TRIDENT_FALLBACK_TEXTURE	0x01
-#define TRIDENT_FALLBACK_DRAW_BUFFER	0x02
-
-#define TRIDENT_NEW_CLIP		0x01
-
-#define TRIDENT_UPLOAD_COMMAND_D	0x00000001
-#define TRIDENT_UPLOAD_CONTEXT		0x04000000
-#define TRIDENT_UPLOAD_CLIPRECTS	0x80000000
-
-#define TAG(x) trident##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-/* these require that base be dword-aligned */
-static INLINE void MMIO_OUT32(unsigned char *base, unsigned int offset,
-                              unsigned int val)
-{
-    unsigned int *addr = (unsigned int *)(base + offset);
-    *addr = val;
-}
-
-static INLINE unsigned int MMIO_IN32(unsigned char *base, unsigned int offset)
-{
-    unsigned int *addr = (unsigned int *)(base + offset);
-    return *addr;
-}
-
-#define MMIO_OUT8(base, offset, val) *((base) + (offset)) = (val)
-#define MMIO_IN8(base, offset) *((base) + (offset))
-
-struct trident_context;
-typedef struct trident_context tridentContextRec;
-typedef struct trident_context *tridentContextPtr;
-
-typedef void (*trident_quad_func)( tridentContextPtr, 
-				 const tridentVertex *, 
-				 const tridentVertex *,
-				 const tridentVertex *,
-				 const tridentVertex * );
-typedef void (*trident_tri_func)( tridentContextPtr, 
-				const tridentVertex *, 
-				const tridentVertex *,
-				const tridentVertex * );
-typedef void (*trident_line_func)( tridentContextPtr, 
-				 const tridentVertex *, 
-				 const tridentVertex * );
-typedef void (*trident_point_func)( tridentContextPtr, 
-				  const tridentVertex * );
-
-typedef struct {
-   drm_handle_t handle;			/* Handle to the DRM region */
-   drmSize size;			/* Size of the DRM region */
-   unsigned char *map;			/* Mapping of the DRM region */
-} tridentRegionRec, *tridentRegionPtr;
-
-typedef struct {
-    __DRIscreenPrivate *driScreen; /* Back pointer to DRI screen */
-
-    drmBufMapPtr buffers;
-
-    unsigned int frontOffset;
-    unsigned int frontPitch;
-    unsigned int backOffset;
-    unsigned int backPitch;
-    unsigned int depthOffset;
-    unsigned int depthPitch;
-    unsigned int width;
-    unsigned int height;
-    unsigned int cpp;
-
-#if 0
-    unsigned int sarea_priv_offset;
-#endif
-
-    tridentRegionRec mmio;
-} tridentScreenRec, *tridentScreenPtr;
-
-/**
- * tridentRenderbuffer, derived from Mesa's gl_renderbuffer
- */
-typedef struct {
-   struct gl_renderbuffer Base;
-   /* XXX per-window info should go here */
-   int foo, bar;
-} tridentRenderbuffer;
-
-
-struct trident_context {
-	GLcontext 		*glCtx;		/* Mesa context */
-
-	__DRIcontextPrivate	*driContext;
-	__DRIscreenPrivate	*driScreen;
-	__DRIdrawablePrivate	*driDrawable;
-
-	GLuint 			new_gl_state;
-	GLuint 			new_state;
-	GLuint 			dirty;
-
-#if 0
-  	drm_trident_sarea_t	*sarea; 
-#endif
-
-        /* Temporaries for translating away float colors:
-	 */
-        struct gl_client_array UbyteColor;
-        struct gl_client_array UbyteSecondaryColor;
-
-   	/* Mirrors of some DRI state
-    	 */
-	int lastStamp;		        /* mirror driDrawable->lastStamp */
-
-   	drm_context_t hHWContext;
-   	drmLock *driHwLock;
-   	int driFd;
-
-   	tridentScreenPtr tridentScreen;	/* Screen private DRI data */
-
-  	/* Visual, drawable, cliprect and scissor information
-    	 */
-   	GLenum DrawBuffer;
-   	GLint drawOffset, drawPitch;
-   	GLint drawX, drawY;             /* origin of drawable in draw buffer */
-   	GLint readOffset, readPitch;
-
-   	GLuint numClipRects;		/* Cliprects for the draw buffer */
-   	drm_clip_rect_t *pClipRects;
-
-   	GLint scissor;
-   	drm_clip_rect_t ScissorRect;	/* Current software scissor */
-
-   	GLuint Fallback;
-	GLuint RenderIndex;
-	GLuint SetupNewInputs;
-	GLuint SetupIndex;
-	GLfloat hw_viewport[16];
-	GLfloat	depth_scale;
-	GLuint vertex_format;
-	GLuint vertex_size;
-	GLuint vertex_stride_shift;
-	GLubyte *verts;
-
-	GLint tmu_source[2];
-
-	GLuint hw_primitive;
-	GLenum render_primitive;
-
-   	trident_point_func    draw_point;
-   	trident_line_func     draw_line;
-   	trident_tri_func      draw_tri;
-   	trident_quad_func     draw_quad;
-
-#if 0
-   	gammaTextureObjectPtr CurrentTexObj[2];
-   	struct gamma_texture_object_t TexObjList;
-   	struct gamma_texture_object_t SwappedOut; 
-	GLenum TexEnvImageFmt[2];
-
-	struct mem_block *texHeap;
-
-   	int lastSwap;
-   	int texAge;
-   	int ctxAge;
-   	int dirtyAge;
-        int lastStamp;
-#endif
-
-	/* Chip state */
-	
-	int	commandD;
-
-	/* Context State */
-	
-	int	ClearColor;
-};
-
-void tridentDDInitExtensions( GLcontext *ctx );
-void tridentDDInitDriverFuncs( GLcontext *ctx );
-void tridentDDInitSpanFuncs( GLcontext *ctx );
-void tridentDDInitState( tridentContextPtr tmesa );
-void tridentInitHW( tridentContextPtr tmesa );
-void tridentDDInitStateFuncs( GLcontext *ctx );
-void tridentDDInitTextureFuncs( GLcontext *ctx );
-void tridentDDInitTriFuncs( GLcontext *ctx );
-
-extern void tridentBuildVertices( GLcontext *ctx, 
-				GLuint start, 
-				GLuint count,
-				GLuint newinputs );
-extern void tridentInitVB( GLcontext *ctx );
-extern void tridentCopyBuffer( const __DRIdrawablePrivate *dPriv );
-extern void tridentFallback( tridentContextPtr tmesa, GLuint bit,
-                             GLboolean mode );
-extern void tridentCheckTexSizes( GLcontext *ctx );
-extern void tridentChooseVertexState( GLcontext *ctx );
-extern void tridentDDUpdateHWState( GLcontext *ctx );
-extern void tridentUploadHwStateLocked( tridentContextPtr tmesa );
-
-#define TRIDENT_CONTEXT(ctx)		((tridentContextPtr)(ctx->DriverCtx))
-
-#endif /* _TRIDENT_CONTEXT_H_ */
diff --git a/src/mesa/drivers/dri/trident/trident_dd.c b/src/mesa/drivers/dri/trident/trident_dd.c
deleted file mode 100644
index faa40c36a2..0000000000
--- a/src/mesa/drivers/dri/trident/trident_dd.c
+++ /dev/null
@@ -1,110 +0,0 @@
-/*
- * Copyright 2002 by Alan Hourihane, Sychdyn, North Wales, UK.
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Alan Hourihane not be used in
- * advertising or publicity pertaining to distribution of the software without
- * specific, written prior permission.  Alan Hourihane makes no representations
- * about the suitability of this software for any purpose.  It is provided
- * "as is" without express or implied warranty.
- *
- * ALAN HOURIHANE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
- * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
- * EVENT SHALL ALAN HOURIHANE BE LIABLE FOR ANY SPECIAL, INDIRECT OR
- * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
- * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
- *
- * Authors:  Alan Hourihane, <alanh@fairlite.demon.co.uk>
- *
- * Trident CyberBladeXP driver.
- *
- */
-#include "trident_context.h"
-#include "trident_lock.h"
-#if defined(USE_X86_ASM)
-#include "x86/common_x86_asm.h"
-#endif
-
-#include "swrast/swrast.h"
-#include "main/context.h"
-#include "main/framebuffer.h"
-
-#define TRIDENT_DATE	"20041223"
-
-/* Return the width and height of the current color buffer.
- */
-static void tridentDDGetBufferSize( GLframebuffer *framebuffer,
-				 GLuint *width, GLuint *height )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-
-   LOCK_HARDWARE(tmesa);
-   *width  = tmesa->driDrawable->w;
-   *height = tmesa->driDrawable->h;
-   UNLOCK_HARDWARE(tmesa);
-}
-
-
-/* Return various strings for glGetString().
- */
-static const GLubyte *tridentDDGetString( GLcontext *ctx, GLenum name )
-{
-   static char buffer[128];
-
-   switch ( name ) {
-   case GL_VENDOR:
-      return (GLubyte *)"Alan Hourihane";
-
-   case GL_RENDERER:
-      sprintf( buffer, "Mesa DRI Trident " TRIDENT_DATE );
-
-      /* Append any CPU-specific information.
-       */
-#ifdef USE_X86_ASM
-      if ( _mesa_x86_cpu_features ) {
-	 strncat( buffer, " x86", 4 );
-      }
-#ifdef USE_MMX_ASM
-      if ( cpu_has_mmx ) {
-	 strncat( buffer, "/MMX", 4 );
-      }
-#endif
-#ifdef USE_3DNOW_ASM
-      if ( cpu_has_3dnow ) {
-	 strncat( buffer, "/3DNow!", 7 );
-      }
-#endif
-#ifdef USE_SSE_ASM
-      if ( cpu_has_xmm ) {
-	 strncat( buffer, "/SSE", 4 );
-      }
-#endif
-#endif
-      return (GLubyte *)buffer;
-
-   default:
-      return NULL;
-   }
-}
-
-/* Enable the extensions supported by this driver.
- */
-void tridentDDInitExtensions( GLcontext *ctx )
-{
-   /* None... */
-}
-
-/* Initialize the driver's misc functions.
- */
-void tridentDDInitDriverFuncs( GLcontext *ctx )
-{
-   ctx->Driver.GetBufferSize = tridentDDGetBufferSize;
-   ctx->Driver.GetString = tridentDDGetString;
-   ctx->Driver.Error = NULL;
-}
diff --git a/src/mesa/drivers/dri/trident/trident_dri.h b/src/mesa/drivers/dri/trident/trident_dri.h
deleted file mode 100644
index c1ce3c4682..0000000000
--- a/src/mesa/drivers/dri/trident/trident_dri.h
+++ /dev/null
@@ -1,28 +0,0 @@
-#ifndef _TRIDENT_DRI_
-#define _TRIDENT_DRI_
-
-#include "xf86drm.h"
-
-typedef struct {
-  drm_handle_t regs;
-  drmSize regsSize;
-  drmAddress regsMap;
-  int deviceID;
-  int width;
-  int height;
-  int mem;
-  int frontOffset;
-  int frontPitch;
-  int backOffset;
-  int backPitch;
-  int depthOffset;
-  int depthPitch;
-  int cpp;
-#if 0
-  int textureOffset;
-  int textureSize;
-#endif
-  unsigned int sarea_priv_offset;
-} TRIDENTDRIRec, *TRIDENTDRIPtr;
-
-#endif
diff --git a/src/mesa/drivers/dri/trident/trident_lock.h b/src/mesa/drivers/dri/trident/trident_lock.h
deleted file mode 100644
index ee0819f5ca..0000000000
--- a/src/mesa/drivers/dri/trident/trident_lock.h
+++ /dev/null
@@ -1,11 +0,0 @@
-/* XXX tridentGetLock doesn't exist... */
-
-#define LOCK_HARDWARE(tmesa) \
-    do { \
-        char __ret = 0; \
-        DRM_CAS(tmesa->driHwLock, tmesa->hHWContext, \
-            DRM_LOCK_HELD | tmesa->hHWContext, __ret); \
-    } while (0)
-
-#define UNLOCK_HARDWARE(tmesa) \
-    DRM_UNLOCK(tmesa->driFd, tmesa->driHwLock, tmesa->hHWContext)
diff --git a/src/mesa/drivers/dri/trident/trident_state.c b/src/mesa/drivers/dri/trident/trident_state.c
deleted file mode 100644
index e68d3a73c6..0000000000
--- a/src/mesa/drivers/dri/trident/trident_state.c
+++ /dev/null
@@ -1,548 +0,0 @@
-/*
- * Copyright 2002 by Alan Hourihane, Sychdyn, North Wales, UK.
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Alan Hourihane not be used in
- * advertising or publicity pertaining to distribution of the software without
- * specific, written prior permission.  Alan Hourihane makes no representations
- * about the suitability of this software for any purpose.  It is provided
- * "as is" without express or implied warranty.
- *
- * ALAN HOURIHANE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
- * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
- * EVENT SHALL ALAN HOURIHANE BE LIABLE FOR ANY SPECIAL, INDIRECT OR
- * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
- * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
- *
- * Authors:  Alan Hourihane, <alanh@fairlite.demon.co.uk>
- *
- * Trident CyberBladeXP driver.
- *
- */
-#include "trident_context.h"
-#include "trident_lock.h"
-#include "vbo/vbo.h"
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-#include "main/framebuffer.h"
-
-#define TRIDENTPACKCOLOR332(r, g, b)					\
-   (((r) & 0xe0) | (((g) & 0xe0) >> 3) | (((b) & 0xc0) >> 6))
-
-#define TRIDENTPACKCOLOR1555(r, g, b, a)					\
-   ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) |	\
-    ((a) ? 0x8000 : 0))
-
-#define TRIDENTPACKCOLOR565(r, g, b)					\
-   ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define TRIDENTPACKCOLOR888(r, g, b)					\
-   (((r) << 16) | ((g) << 8) | (b))
-
-#define TRIDENTPACKCOLOR8888(r, g, b, a)					\
-   (((a) << 24) | ((r) << 16) | ((g) << 8) | (b))
-
-#define TRIDENTPACKCOLOR4444(r, g, b, a)					\
-   ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-static INLINE GLuint tridentPackColor( GLuint cpp,
-                                       GLubyte r, GLubyte g,
-                                       GLubyte b, GLubyte a )
-{
-   switch ( cpp ) {
-   case 2:
-      return TRIDENTPACKCOLOR565( r, g, b );
-   case 4:
-      return TRIDENTPACKCOLOR8888( r, g, b, a );
-   default:
-      return 0;
-   }
-}
-
-void tridentUploadHwStateLocked( tridentContextPtr tmesa )
-{
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-#if 0
-   ATISAREAPrivPtr sarea = tmesa->sarea;
-   trident_context_regs_t *regs = &(sarea->ContextState);
-#endif
-
-   if ( tmesa->dirty & TRIDENT_UPLOAD_COMMAND_D ) {
-      MMIO_OUT32(MMIO, 0x00281C, tmesa->commandD );
-      tmesa->dirty &= ~TRIDENT_UPLOAD_COMMAND_D;
-   }
-
-   if ( tmesa->dirty & TRIDENT_UPLOAD_CLIPRECTS ) {
-      /* XXX FIX ME ! */
-      MMIO_OUT32(MMIO, 0x002C80 , 0x20008000 | tmesa->tridentScreen->height );    
-      MMIO_OUT32(MMIO, 0x002C84 , 0x20000000 | tmesa->tridentScreen->width );    
-      tmesa->dirty &= ~TRIDENT_UPLOAD_CLIPRECTS;
-   }
-
-   tmesa->dirty = 0;
-}
-
-/* Copy the back color buffer to the front color buffer.
- */
-void tridentCopyBuffer( const __DRIdrawablePrivate *dPriv )
-{
-   unsigned char *MMIO;
-   tridentContextPtr tmesa;
-   GLint nbox, i;
-   int busy;
-   drm_clip_rect_t *pbox;
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   tmesa = (tridentContextPtr) dPriv->driContextPriv->driverPrivate;
-   MMIO = tmesa->tridentScreen->mmio.map;
-
-   LOCK_HARDWARE( tmesa );
-
-   /* use front buffer cliprects */
-   nbox = dPriv->numClipRects;
-   pbox = dPriv->pClipRects;
-
-   for ( i = 0 ; i < nbox ; i++ ) {
-#if 0
-      GLint nr = MIN2( i + MACH64_NR_SAREA_CLIPRECTS , nbox );
-      drm_clip_rect_t *b = tmesa->sarea->boxes;
-      GLint n = 0;
-
-      for ( ; i < nr ; i++ ) {
-	 *b++ = pbox[i];
-	 n++;
-      }
-      tmesa->sarea->nbox = n;
-#endif
-
-    MMIO_OUT32(MMIO, 0x2150, tmesa->tridentScreen->frontPitch << 20 | tmesa->tridentScreen->frontOffset>>4);
-    MMIO_OUT32(MMIO, 0x2154, tmesa->tridentScreen->backPitch << 20 | tmesa->tridentScreen->backOffset>>4);
-    MMIO_OUT8(MMIO, 0x2127, 0xCC); /* Copy Rop */
-    MMIO_OUT32(MMIO, 0x2128, 0x4); /* scr2scr */
-    MMIO_OUT32(MMIO, 0x2138, (pbox->x1 << 16) | pbox->y1);
-    MMIO_OUT32(MMIO, 0x213C, (pbox->x1 << 16) | pbox->y1);
-    MMIO_OUT32(MMIO, 0x2140, (pbox->x2 - pbox->x1) << 16 | (pbox->y2 - pbox->y1) );
-    MMIO_OUT8(MMIO, 0x2124, 0x01); /* BLT */
-#define GE_BUSY 0x80
-    for (;;) {
-	busy = MMIO_IN8(MMIO, 0x2120);
-	if ( !(busy & GE_BUSY) )
-		break;
-    }
-   }
-
-   UNLOCK_HARDWARE( tmesa );
-
-#if 0
-   tmesa->dirty |= (MACH64_UPLOAD_CONTEXT |
-		    MACH64_UPLOAD_MISC |
-		    MACH64_UPLOAD_CLIPRECTS);
-#endif
-}
-
-
-static void tridentDDClear( GLcontext *ctx, GLbitfield mask )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-   int busy;
-   GLuint flags = 0;
-   GLint i;
-   GLint cx, cy, cw, ch;
-
-#define DRM_TRIDENT_FRONT	0x01
-#define DRM_TRIDENT_BACK	0x02
-#define DRM_TRIDENT_DEPTH	0x04
-
-   if ( tmesa->new_state )
-      tridentDDUpdateHWState( ctx );
-
-   if ( mask & BUFFER_BIT_FRONT_LEFT ) {
-      flags |= DRM_TRIDENT_FRONT;
-      mask &= ~BUFFER_BIT_FRONT_LEFT;
-   }
-
-   if ( mask & BUFFER_BIT_BACK_LEFT ) {
-      flags |= DRM_TRIDENT_BACK;
-      mask &= ~BUFFER_BIT_BACK_LEFT;
-   }
-
-   if ( ( mask & BUFFER_BIT_DEPTH ) && ctx->Depth.Mask ) {
-      flags |= DRM_TRIDENT_DEPTH;
-      mask &= ~BUFFER_BIT_DEPTH;
-   }
-
-   LOCK_HARDWARE(tmesa);
-
-   /* get region after locking: */
-   cx = ctx->DrawBuffer->_Xmin;
-   cy = ctx->DrawBuffer->_Ymin;
-   cw = ctx->DrawBuffer->_Xmax - cx;
-   ch = ctx->DrawBuffer->_Ymax - cy;
-
-   if ( flags ) {
-   
-      cx += tmesa->drawX;
-      cy += tmesa->drawY;
-   
-      /* HACK!!!
-       */
-      if ( tmesa->dirty & ~TRIDENT_UPLOAD_CLIPRECTS ) {
-         tridentUploadHwStateLocked( tmesa );
-      }
-   
-      for ( i = 0 ; i < tmesa->numClipRects ; i++ ) {
-#if 0
-         int nr = MIN2( i + TRIDENT_NR_SAREA_CLIPRECTS, tmesa->numClipRects );
-         drm_clip_rect_t *box = tmesa->pClipRects;
-         drm_clip_rect_t *b = tmesa->sarea->boxes;
-         GLint n = 0;
-   
-         if ( !all ) {
-	    for ( ; i < nr ; i++ ) {
-	       GLint x = box[i].x1;
-	       GLint y = box[i].y1;
-	       GLint w = box[i].x2 - x;
-	       GLint h = box[i].y2 - y;
-	       
-	       if ( x < cx ) w -= cx - x, x = cx;
-	       if ( y < cy ) h -= cy - y, y = cy;
-	       if ( x + w > cx + cw ) w = cx + cw - x;
-	       if ( y + h > cy + ch ) h = cy + ch - y;
-	       if ( w <= 0 ) continue;
-	       if ( h <= 0 ) continue;
-	       
-	       b->x1 = x;
-	       b->y1 = y;
-	       b->x2 = x + w;
-	       b->y2 = y + h;
-	       b++;
-	       n++;
-	    }
-         } else {
-	    for ( ; i < nr ; i++ ) {
-	       *b++ = box[i];
-	       n++;
-	    }
-         }
-   
-         tmesa->sarea->nbox = n;
-#endif
-   
-if (flags & DRM_TRIDENT_BACK) {
-    MMIO_OUT32(MMIO, 0x2150, tmesa->tridentScreen->backPitch << 20 | tmesa->tridentScreen->backOffset>>4);
-         MMIO_OUT8(MMIO, 0x2127, 0xF0); /* Pat Rop */
-         MMIO_OUT32(MMIO, 0x2158, tmesa->ClearColor);
-         MMIO_OUT32(MMIO, 0x2128, 0x4000); /* solidfill */
-         MMIO_OUT32(MMIO, 0x2138, cx << 16 | cy);
-         MMIO_OUT32(MMIO, 0x2140, cw << 16 | ch);
-         MMIO_OUT8(MMIO, 0x2124, 0x01); /* BLT */
-#define GE_BUSY 0x80
-	 for (;;) {
-		busy = MMIO_IN8(MMIO, 0x2120);
-		if ( !(busy & GE_BUSY) )
-			break;
-	 }
-}
-if (flags & DRM_TRIDENT_DEPTH) {
-    MMIO_OUT32(MMIO, 0x2150, tmesa->tridentScreen->depthPitch << 20 | tmesa->tridentScreen->depthOffset>>4);
-         MMIO_OUT8(MMIO, 0x2127, 0xF0); /* Pat Rop */
-         MMIO_OUT32(MMIO, 0x2158, tmesa->ClearColor);
-         MMIO_OUT32(MMIO, 0x2128, 0x4000); /* solidfill */
-         MMIO_OUT32(MMIO, 0x2138, cx << 16 | cy);
-         MMIO_OUT32(MMIO, 0x2140, cw << 16 | ch);
-         MMIO_OUT8(MMIO, 0x2124, 0x01); /* BLT */
-#define GE_BUSY 0x80
-	 for (;;) {
-		busy = MMIO_IN8(MMIO, 0x2120);
-		if ( !(busy & GE_BUSY) )
-			break;
-	 }
-}
-    MMIO_OUT32(MMIO, 0x2150, tmesa->tridentScreen->frontPitch << 20 | tmesa->tridentScreen->frontOffset>>4);
-if (flags & DRM_TRIDENT_FRONT) {
-         MMIO_OUT8(MMIO, 0x2127, 0xF0); /* Pat Rop */
-         MMIO_OUT32(MMIO, 0x2158, tmesa->ClearColor);
-         MMIO_OUT32(MMIO, 0x2128, 0x4000); /* solidfill */
-         MMIO_OUT32(MMIO, 0x2138, cx << 16 | cy);
-         MMIO_OUT32(MMIO, 0x2140, cw << 16 | ch);
-         MMIO_OUT8(MMIO, 0x2124, 0x01); /* BLT */
-#define GE_BUSY 0x80
-	 for (;;) {
-		busy = MMIO_IN8(MMIO, 0x2120);
-		if ( !(busy & GE_BUSY) )
-			break;
-	 }
-}
-   
-      }
-   
-#if 0
-      tmesa->dirty |= (TRIDENT_UPLOAD_CONTEXT |
-   		       TRIDENT_UPLOAD_MISC |
-   		       TRIDENT_UPLOAD_CLIPRECTS);
-#endif
-   }
-
-   UNLOCK_HARDWARE(tmesa);
-
-   if ( mask )
-      _swrast_Clear( ctx, mask );
-}
-
-static void tridentDDShadeModel( GLcontext *ctx, GLenum mode )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   GLuint s = tmesa->commandD;
-
-#define TRIDENT_FLAT_SHADE 			0x000000E0
-#define TRIDENT_FLAT_SHADE_VERTEX_C		0x00000060
-#define TRIDENT_FLAT_SHADE_GOURAUD		0x00000080
-
-   s &= ~TRIDENT_FLAT_SHADE;
-
-   switch ( mode ) {
-   case GL_FLAT:
-      s |= TRIDENT_FLAT_SHADE_VERTEX_C;
-      break;
-   case GL_SMOOTH:
-      s |= TRIDENT_FLAT_SHADE_GOURAUD;
-      break;
-   default:
-      return;
-   }
-
-   if ( tmesa->commandD != s ) {
-      tmesa->commandD = s;
-
-      tmesa->dirty |= TRIDENT_UPLOAD_COMMAND_D;
-   }
-}
-
-static void
-tridentCalcViewport( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   const GLfloat *v = ctx->Viewport._WindowMap.m;
-   GLfloat *m = tmesa->hw_viewport;
-
-   /* See also trident_translate_vertex.
-    */
-   m[MAT_SX] =   v[MAT_SX];
-   m[MAT_TX] =   v[MAT_TX] + tmesa->drawX + SUBPIXEL_X;
-   m[MAT_SY] = - v[MAT_SY];
-   m[MAT_TY] = - v[MAT_TY] + tmesa->driDrawable->h + tmesa->drawY + SUBPIXEL_Y;
-#if 0
-   m[MAT_SZ] =   v[MAT_SZ] * tmesa->depth_scale;
-   m[MAT_TZ] =   v[MAT_TZ] * tmesa->depth_scale;
-#else
-   m[MAT_SZ] =   v[MAT_SZ];
-   m[MAT_TZ] =   v[MAT_TZ];
-#endif
-
-   tmesa->SetupNewInputs = ~0;
-}
-
-static void tridentDDViewport( GLcontext *ctx,
-			  GLint x, GLint y,
-			  GLsizei width, GLsizei height )
-{
-   tridentCalcViewport( ctx );
-}
-
-static void tridentDDDepthRange( GLcontext *ctx,
-			    GLclampd nearval, GLclampd farval )
-{
-   tridentCalcViewport( ctx );
-}
-
-static void
-tridentSetCliprects( tridentContextPtr tmesa, GLenum mode )
-{
-   __DRIdrawablePrivate *dPriv = tmesa->driDrawable;
-
-   switch ( mode ) {
-   case GL_FRONT_LEFT:
-      if (dPriv->numClipRects == 0) {
-	 static drm_clip_rect_t zeroareacliprect = {0,0,0,0};
-	 tmesa->numClipRects = 1;
-	 tmesa->pClipRects = &zeroareacliprect;
-      } else {
-	 tmesa->numClipRects = dPriv->numClipRects;
-	 tmesa->pClipRects = (drm_clip_rect_t *)dPriv->pClipRects;
-      }
-      tmesa->drawX = dPriv->x;
-      tmesa->drawY = dPriv->y;
-      break;
-   case GL_BACK_LEFT:
-      if ( dPriv->numBackClipRects == 0 ) {
-	  if (dPriv->numClipRects == 0) {
-	     static drm_clip_rect_t zeroareacliprect = {0,0,0,0};
-	     tmesa->numClipRects = 1;
-	     tmesa->pClipRects = &zeroareacliprect;
-	  } else {
-	     tmesa->numClipRects = dPriv->numClipRects;
-	     tmesa->pClipRects = (drm_clip_rect_t *)dPriv->pClipRects;
-	     tmesa->drawX = dPriv->x;
-	     tmesa->drawY = dPriv->y;
-	  }
-      }
-      else {
-	 tmesa->numClipRects = dPriv->numBackClipRects;
-	 tmesa->pClipRects = (drm_clip_rect_t *)dPriv->pBackClipRects;
-	 tmesa->drawX = dPriv->backX;
-	 tmesa->drawY = dPriv->backY;
-      }
-      break;
-   default:
-      return;
-   }
-
-#if 0
-   tmesa->dirty |= TRIDENT_UPLOAD_CLIPRECTS;
-#endif
-}
-
-#if 0
-static GLboolean tridentDDSetDrawBuffer( GLcontext *ctx, GLenum mode )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   int found = GL_TRUE;
-
-   if ( tmesa->DrawBuffer != mode ) {
-      tmesa->DrawBuffer = mode;
-      
-      switch ( mode ) {
-      case GL_FRONT_LEFT:
-	 tridentFallback( tmesa, TRIDENT_FALLBACK_DRAW_BUFFER, GL_FALSE );
-	 tmesa->drawOffset = tmesa->tridentScreen->frontOffset;
-	 tmesa->drawPitch  = tmesa->tridentScreen->frontPitch;
-	 tridentSetCliprects( tmesa, GL_FRONT_LEFT );
-	 break;
-      case GL_BACK_LEFT:
-	 tridentFallback( tmesa, TRIDENT_FALLBACK_DRAW_BUFFER, GL_FALSE );
-	 tmesa->drawOffset = tmesa->tridentScreen->backOffset;
-	 tmesa->drawPitch  = tmesa->tridentScreen->backPitch;
-	 tridentSetCliprects( tmesa, GL_BACK_LEFT );
-	 break;
-      default:
-	 tridentFallback( tmesa, TRIDENT_FALLBACK_DRAW_BUFFER, GL_TRUE );
-	 found = GL_FALSE;
-	 break;
-      }
-
-#if 0
-      tmesa->setup.dst_off_pitch = (((tmesa->drawPitch/8) << 22) |
-				    (tmesa->drawOffset >> 3));
-
-      tmesa->dirty |= MACH64_UPLOAD_DST_OFF_PITCH | MACH64_UPLOAD_CONTEXT;
-#endif
-      
-   }
-
-   return found;
-}
-
-static void tridentDDClearColor( GLcontext *ctx,
-				const GLchan color[4] )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-
-   tmesa->ClearColor = tridentPackColor( tmesa->tridentScreen->cpp,
-					color[0], color[1], 
-					color[2], color[3] );
-}
-#endif
-
-static void
-tridentDDUpdateState( GLcontext *ctx, GLuint new_state )
-{
-   _swrast_InvalidateState( ctx, new_state );
-   _swsetup_InvalidateState( ctx, new_state );
-   _vbo_InvalidateState( ctx, new_state );
-   _tnl_InvalidateState( ctx, new_state );
-   TRIDENT_CONTEXT(ctx)->new_gl_state |= new_state;
-}
-
-
-/* Initialize the context's hardware state.
- */
-void tridentDDInitState( tridentContextPtr tmesa )
-{
-   tmesa->new_state = 0;
-
-   switch ( tmesa->glCtx->Visual.depthBits ) {
-   case 16:
-      tmesa->depth_scale = 1.0 / (GLfloat)0xffff;
-      break;
-   case 24:
-      tmesa->depth_scale = 1.0 / (GLfloat)0xffffff;
-      break;
-   }
-}
-
-void tridentDDUpdateHWState( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   int new_state = tmesa->new_state;
-
-   if ( new_state )
-   {
-      tmesa->new_state = 0;
-
-#if 0
-      /* Update the various parts of the context's state.
-       */
-      if ( new_state & GAMMA_NEW_ALPHA )
-	 tridentUpdateAlphaMode( ctx );
-
-      if ( new_state & GAMMA_NEW_DEPTH )
-	 tridentUpdateZMode( ctx );
-
-      if ( new_state & GAMMA_NEW_FOG )
-	 gammaUpdateFogAttrib( ctx );
-
-      if ( new_state & GAMMA_NEW_CLIP )
-	 gammaUpdateClipping( ctx );
-
-      if ( new_state & GAMMA_NEW_POLYGON )
-	 gammaUpdatePolygon( ctx );
-
-      if ( new_state & GAMMA_NEW_CULL )
-	 gammaUpdateCull( ctx );
-
-      if ( new_state & GAMMA_NEW_MASKS )
-	 gammaUpdateMasks( ctx );
-
-      if ( new_state & GAMMA_NEW_STIPPLE )
-	 gammaUpdateStipple( ctx );
-#endif
-   }
-
-   /* HACK ! */
-
-#if 0
-   gammaEmitHwState( tmesa );
-#endif
-}
-
-/* Initialize the driver's state functions.
- */
-void tridentDDInitStateFuncs( GLcontext *ctx )
-{
-   ctx->Driver.UpdateState		= tridentDDUpdateState;
-
-   ctx->Driver.Clear			= tridentDDClear;
-   ctx->Driver.DepthRange		= tridentDDDepthRange;
-   ctx->Driver.ShadeModel		= tridentDDShadeModel;
-   ctx->Driver.Viewport			= tridentDDViewport;
-}
diff --git a/src/mesa/drivers/dri/trident/trident_tris.c b/src/mesa/drivers/dri/trident/trident_tris.c
deleted file mode 100644
index ee85ab482c..0000000000
--- a/src/mesa/drivers/dri/trident/trident_tris.c
+++ /dev/null
@@ -1,1125 +0,0 @@
-/*
- * Copyright 2002 by Alan Hourihane, Sychdyn, North Wales, UK.
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Alan Hourihane not be used in
- * advertising or publicity pertaining to distribution of the software without
- * specific, written prior permission.  Alan Hourihane makes no representations
- * about the suitability of this software for any purpose.  It is provided
- * "as is" without express or implied warranty.
- *
- * ALAN HOURIHANE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
- * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
- * EVENT SHALL ALAN HOURIHANE BE LIABLE FOR ANY SPECIAL, INDIRECT OR
- * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
- * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
- *
- * Authors:  Alan Hourihane, <alanh@fairlite.demon.co.uk>
- *
- * Trident CyberBladeXP driver.
- *
- */
-
-#include "trident_context.h"
-#include "trident_lock.h"
-#include "tnl/tnl.h"
-#include "tnl/t_context.h"
-#include "tnl/t_pipeline.h"
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-
-static int first = 1;
-
-typedef struct reg {
-	int addr;
-	int data;
-} RegData;
-
-RegData initRegData[]={
-    {0x2804,  0x19980824},
-    {0x2F70,  0x46455858},
-    {0x2F74,    0x41584998},
-    {0x2F00,    0x00000000},
-    {0x2F04,    0x80000800},
-    {0x2F08,    0x00550200},
-    {0x2F40,    0x00000001},
-    {0x2F40,    0x00000001},
-    {0x2F44,    0x00830097},
-    {0x2F48,    0x0087009F},
-    {0x2F4C,    0x00BF0003},
-    {0x2F50,    0xF00B6C1B},
-    {0x2C04,    0x00000000},
-    {0x2D00,    0x00000080},
-    {0x2D00,    0x00000000},
-    {0x2DD4,    0x00100000},
-    {0x2DD4,    0x00100010},
-    {0x2DD8,    0x00100000},
-    {0x2DD8,    0x00100010},
-    {0x2C88,    0xFFFFFFFF},
-    {0x2C94 ,   0xFFFFFFFF},
-    {0x281C,    0x00008000},
-    {0x2C80,    0x00000000},
-    {0x2C80,    0x00000000},
-    {0x2C80 ,   0x00008000},
-    {0x2C00  ,  0x00000000},
-    {0x2C04  ,  0x00000000},
-    {0x2C08  ,  0x00000000},
-    {0x2C0C  ,  0x00000000},
-    {0x2C10  ,  0x00000000},
-    {0x2C14  ,  0x00000000},
-    {0x2C18  ,  0x00000000},
-    {0x2C1C  ,  0x00000000},
-    {0x2C20  ,  0x00000000},
-    {0x2C24  ,  0x00000000},
-    {0x2C2C  ,  0x00000000},
-    {0x2C30  ,  0x00000000},
-    {0x2C34  ,  0x00000000},
-    {0x2C38  ,  0x00000000},
-    {0x2C3C  ,  0x00000000},
-    {0x2C40  ,  0x00000000},
-    {0x2C44  ,  0x00000000},
-    {0x2C48  ,  0x00000000},
-    {0x2C4C  ,  0x00000000},
-    {0x2C50  ,  0x00000000},
-    {0x2C54  ,  0x00000000},
-    {0x2C58  ,  0x00000000},
-    {0x2C5C  ,  0x00000000},
-    {0x2C60  ,  0x00000000},
-    {0x2C64  ,  0x00000000},
-    {0x2C68  ,  0x00000000},
-    {0x2C6C  ,  0x00000000},
-    {0x2C70  ,  0x00000000},
-    {0x2C74  ,  0x00000000},
-    {0x2C78  ,  0x00000000},
-    {0x2C7C  ,  0x00000000},
-    {0x2C80  ,  0x00008000},
-    {0x2C84  ,  0x00000000},
-    {0x2C88  ,  0xFFFFFFFF},
-    {0x2C8C  ,  0x00000000},
-    {0x2C90  ,  0x00000000},
-    {0x2C94  ,  0xFFFFFFFF},
-    {0x2C98  ,  0x00000000},
-    {0x2C9C  ,  0x00000000},
-    {0x2CA0  ,  0x00000000},
-    {0x2CA4   , 0x00000000},
-    {0x2CA8   , 0x00000000},
-    {0x2CAC  ,  0x00000000},
-    {0x2CB0  ,  0x00000000},
-    {0x2CB4  ,  0x00000000},
-    {0x2CB8  ,  0x00000000},
-    {0x2CBC  ,  0x00000000},
-    {0x2CC0  ,  0x00000000},
-    {0x2CC4  ,  0x00000000},
-    {0x2CC8  ,  0x00000000},
-    {0x2CCC  ,  0x00000000},
-    {0x2CD0  ,  0x00000000},
-    {0x2CD4  ,  0x00000000},
-    {0x2CD8  ,  0x00000000},
-    {0x2CDC  ,  0x00000000},
-    {0x2CE0  ,  0x00000000},
-    {0x2CE4  ,  0x00000000},
-    {0x2CE8  ,  0x00000000},
-    {0x2CEC  ,  0x00000000},
-    {0x2CF0  ,  0x00000000},
-    {0x2CF4  ,  0x00000000},
-    {0x2CF8  ,  0x00000000},
-    {0x2CFC  ,  0x00000000},
-    {0x2D00  ,  0x00000000},
-    {0x2D04  ,  0x00000000},
-    {0x2D08  ,  0x00000000},
-    {0x2D0C  ,  0x00000000},
-    {0x2D10  ,  0x00000000},
-    {0x2D14  ,  0x00000000},
-    {0x2D18  ,  0x00000000},
-    {0x2D1C  ,  0x00000000},
-    {0x2D20  ,  0x00000000},
-    {0x2D24  ,  0x00000000},
-    {0x2D28  ,  0x00000000},
-    {0x2D2C  ,  0x00000000},
-    {0x2D30  ,  0x00000000},
-    {0x2D34  ,  0x00000000},
-    {0x2D38   , 0x00000000},
-    {0x2D3C   , 0x00000000},
-    {0x2D40   , 0x00000000},
-    {0x2D44   , 0x00000000},
-    {0x2D48   , 0x00000000},
-    {0x2D4C   , 0x00000000},
-    {0x2D50   , 0x00000000},
-    {0x2D54  ,  0x00000000},
-    {0x2D58  ,  0x00000000},
-    {0x2D5C  ,  0x00000000},
-    {0x2D60  ,  0x00000000},
-    {0x2D64  ,  0x00000000},
-    {0x2D68  ,  0x00000000},
-    {0x2D6C  ,  0x00000000},
-    {0x2D70   , 0x00000000},
-    {0x2D74   , 0x00000000},
-    {0x2D78   , 0x00000000},
-    {0x2D7C   , 0x00000000},
-    {0x2D80   , 0x00000000},
-    {0x2D84   , 0x00000000},
-    {0x2D88   , 0x00000000},
-    {0x2D8C   , 0x00000000},
-    {0x2D90   , 0x00000000},
-    {0x2D94   , 0x00000000},
-    {0x2D98   , 0x00000000},
-    {0x2D9C   , 0x00000000},
-    {0x2DA0   , 0x00000000},
-    {0x2DA4   , 0x00000000},
-    {0x2DA8   , 0x00000000},
-    {0x2DAC   , 0x00000000},
-    {0x2DB0   , 0x00000000},
-    {0x2DB4   , 0x00000000},
-    {0x2DB8   , 0x00000000},
-    {0x2DBC   , 0x00000000},
-    {0x2DC0   , 0x00000000},
-    {0x2DC4   , 0x00000000},
-    {0x2DC8   , 0x00000000},
-    {0x2DCC   , 0x00000000},
-    {0x2DD0   , 0x00000000},
-    {0x2DD4   , 0x00100010},
-    {0x2DD8   , 0x00100010},
-    {0x2DDC   , 0x00000000},
-    {0x2DE0   , 0x00000000},
-    {0x2DE4   , 0x00000000},
-    {0x2DE8   , 0x00000000},
-    {0x2DEC   , 0x00000000},
-    {0x2DF0   , 0x00000000},
-    {0x2DF4   , 0x00000000},
-    {0x2DF8   , 0x00000000},
-    {0x2DFC   , 0x00000000},
-    {0x2E00   , 0x00000000},
-    {0x2E04   , 0x00000000},
-    {0x2E08   , 0x00000000},
-    {0x2E0C   , 0x00000000},
-    {0x2E10   , 0x00000000},
-    {0x2E14   , 0x00000000},
-    {0x2E18   , 0x00000000},
-    {0x2E1C   , 0x00000000},
-    {0x2E20   , 0x00000000},
-    {0x2E24   , 0x00000000},
-    {0x2E28   , 0x00000000},
-    {0x2E2C   , 0x00000000},
-    {0x2E30   , 0x00000000},
-    {0x2E34   , 0x00000000},
-    {0x2E38   , 0x00000000},
-    {0x2E3C   , 0x00000000},
-    {0x2E40   , 0x00000000},
-    {0x2E44   , 0x00000000},
-    {0x2E48   , 0x00000000},
-    {0x2E4C   , 0x00000000},
-    {0x2E50   , 0x00000000},
-    {0x2E54   , 0x00000000},
-    {0x2E58   , 0x00000000},
-    {0x2E5C   , 0x00000000},
-    {0x2E60   , 0x00000000},
-    {0x2E64   , 0x00000000},
-    {0x2E68   , 0x00000000},
-    {0x2E6C   , 0x00000000},
-    {0x2E70   , 0x00000000},
-    {0x2E74   , 0x00000000},
-    {0x2E78   , 0x00000000},
-    {0x2E7C   , 0x00000000},
-    {0x2E80   , 0x00000000},
-    {0x2E84   , 0x00000000},
-    {0x2E88   , 0x00000000},
-    {0x2E8C   , 0x00000000},
-    {0x2E90   , 0x00000000},
-    {0x2E94   , 0x00000000},
-    {0x2E98   , 0x00000000},
-    {0x2E9C   , 0x00000000},
-    {0x2EA0   , 0x00000000},
-    {0x2EA4   , 0x00000000},
-    {0x2EA8   , 0x00000000},
-    {0x2EAC   , 0x00000000},
-    {0x2EB0   , 0x00000000},
-    {0x2EB4   , 0x00000000},
-    {0x2EB8   , 0x00000000},
-    {0x2EBC   , 0x00000000},
-    {0x2EC0   , 0x00000000},
-    {0x2EC4   , 0x00000000},
-    {0x2EC8   , 0x00000000},
-    {0x2ECC   , 0x00000000},
-    {0x2ED0   , 0x00000000},
-    {0x2ED4   , 0x00000000},
-    {0x2ED8   , 0x00000000},
-    {0x2EDC   , 0x00000000},
-    {0x2EE0   , 0x00000000},
-    {0x2EE4    ,0x00000000},
-    {0x2EE8    ,0x00000000},
-    {0x2EEC   , 0x00000000},
-    {0x2EF0   , 0x00000000},
-    {0x2EF4   , 0x00000000},
-    {0x2EF8   , 0x00000000},
-    {0x2EFC   , 0x00000000},
-    /*{0x2F60   , 0x00000000},*/
-};
-
-int initRegDataNum=sizeof(initRegData)/sizeof(RegData);
-
-typedef union {
-    unsigned int i;
-    float        f;
-} dmaBufRec, *dmaBuf;
-
-void Init3D( tridentContextPtr tmesa )
-{
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-    int i;
-
-    for(i=0;i<initRegDataNum;++i)
-       MMIO_OUT32(MMIO, initRegData[i].addr, initRegData[i].data);
-}
-
-int DrawTriangle( tridentContextPtr tmesa)
-{
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-   dmaBufRec clr;
-
-printf("DRAW TRI\n");
-	Init3D(tmesa);
-
-printf("ENGINE STATUS 0x%x\n",MMIO_IN32(MMIO, 0x2800));
-    MMIO_OUT32(MMIO,  0x002800, 0x00000000 );
-#if 0
-    MMIO_OUT32(MMIO,  0x002368 , MMIO_IN32(MMIO,0x002368)|1 );
-#endif
-
-    MMIO_OUT32(MMIO, 0x002C00 , 0x00000014 );
-#if 0
-    MMIO_OUT32(MMIO, 0x002C04 , 0x0A8004C0 );
-#else
-    MMIO_OUT32(MMIO, 0x002C04 , 0x0A8000C0 );
-#endif
-
-#if 0
-    MMIO_OUT32(MMIO, 0x002C08 , 0x00000000 );
-    MMIO_OUT32(MMIO, 0x002C0C , 0xFFCCCCCC );
-    MMIO_OUT32(MMIO, 0x002C10 , 0x3F800000 );
-    MMIO_OUT32(MMIO, 0x002C14 , 0x3D0D3DCB );
-    MMIO_OUT32(MMIO, 0x002C2C , 0x70000000 );
-    MMIO_OUT32(MMIO, 0x002C24 , 0x00202C00 );
-    MMIO_OUT32(MMIO, 0x002C28 , 0xE0002500 );
-    MMIO_OUT32(MMIO, 0x002C30 , 0x00000000 );
-    MMIO_OUT32(MMIO, 0x002C34 , 0xE0000000 );
-    MMIO_OUT32(MMIO, 0x002C38 , 0x00000000 );
-#endif
-
-    MMIO_OUT32(MMIO, 0x002C50 , 0x00000000 );
-    MMIO_OUT32(MMIO, 0x002C54 , 0x0C320C80 );    
-    MMIO_OUT32(MMIO, 0x002C50 , 0x00000000 );
-    MMIO_OUT32(MMIO, 0x002C54 , 0x0C320C80 );    
-    MMIO_OUT32(MMIO, 0x002C80 , 0x20008258 );    
-    MMIO_OUT32(MMIO, 0x002C84 , 0x20000320 );    
-    MMIO_OUT32(MMIO, 0x002C94 , 0xFFFFFFFF );
-
-#if 0
-    MMIO_OUT32(MMIO, 0x002D00 , 0x00009009 );    
-    MMIO_OUT32(MMIO, 0x002D38 , 0x00000000 );
-    MMIO_OUT32(MMIO, 0x002D94 , 0x20002000 );
-    MMIO_OUT32(MMIO, 0x002D50 , 0xf0000000 );
-    MMIO_OUT32(MMIO, 0x002D80 , 0x24002000 );        
-    MMIO_OUT32(MMIO, 0x002D98 , 0x81000000 );        
-    MMIO_OUT32(MMIO, 0x002DB0 , 0x81000000 );        
-    MMIO_OUT32(MMIO, 0x002DC8 , 0x808000FF );
-    MMIO_OUT32(MMIO, 0x002DD4 , 0x02000200 );
-    MMIO_OUT32(MMIO, 0x002DD8 , 0x02000200 );
-    MMIO_OUT32(MMIO, 0x002D30 , 0x02092400 );    
-    MMIO_OUT32(MMIO, 0x002D04 , 0x00102120 );    
-    MMIO_OUT32(MMIO, 0x002D08 , 0xFFFFFFFF );
-    MMIO_OUT32(MMIO, 0x002D0C , 0xF00010D0 );    
-    MMIO_OUT32(MMIO, 0x002D10 , 0xC0000400 );
-#endif
-
-    MMIO_OUT32(MMIO, 0x002814,  0x00000000 );
-#if 0
-    MMIO_OUT32(MMIO, 0x002818 , 0x00036C20 );        
-#else
-    MMIO_OUT32(MMIO, 0x002818 , 0x00036020 );        
-#endif
-    MMIO_OUT32(MMIO, 0x00281C , 0x00098081 );	
-
-printf("first TRI\n");
-    clr.f = 5.0;
-    MMIO_OUT32(MMIO, 0x002820 , clr.i );               
-    clr.f = 595.0;
-    MMIO_OUT32(MMIO, 0x002824 , clr.i );               
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002828 , clr.i );
-    MMIO_OUT32(MMIO, 0x00282C , 0x00FF00 );        
-#if 0
-    clr.f = 0.0;
-    MMIO_OUT32(MMIO, 0x002830 , clr.i );
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002834 , clr.i );
-#endif
-
-    clr.f = 5.0;
-    MMIO_OUT32(MMIO, 0x002820 , clr.i );               
-    clr.f = 5.0;
-    MMIO_OUT32(MMIO, 0x002824 , clr.i );               
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002828 , clr.i );
-    MMIO_OUT32(MMIO, 0x00282C , 0xFF0000 );        
-#if 0
-    clr.f = 0.0;
-    MMIO_OUT32(MMIO, 0x002830 , clr.i );
-    clr.f = 0.0;
-    MMIO_OUT32(MMIO, 0x002834 , clr.i );
-#endif
-
-    clr.f = 395.0;
-printf("0x%x\n",clr.i);
-    MMIO_OUT32(MMIO, 0x002820 , clr.i );               
-    clr.f = 5.0;
-    MMIO_OUT32(MMIO, 0x002824 , clr.i );               
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002828 , clr.i );
-    MMIO_OUT32(MMIO, 0x00282C , 0xFF );        
-#if 0
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002830 , clr.i );
-    clr.f = 0.0;
-    MMIO_OUT32(MMIO, 0x002834 , clr.i );
-#endif
-
-printf("sec TRI\n");
-    MMIO_OUT32(MMIO, 0x00281C , 0x00093980 );    
-    clr.f = 395.0;
-    MMIO_OUT32(MMIO, 0x002820 , clr.i );               
-    clr.f = 595.0;
-    MMIO_OUT32(MMIO, 0x002824 , clr.i );               
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002828 , clr.i );               
-    MMIO_OUT32(MMIO, 0x00282C , 0x00FF00 );        
-#if 0
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002830 , clr.i );
-    clr.f = 1.0;
-    MMIO_OUT32(MMIO, 0x002834 , clr.i );
-#endif
-
-#if 0
-    MMIO_OUT32(MMIO,  0x002368 , MMIO_IN32(MMIO,0x002368)&0xfffffffe );
-#endif
-
-printf("fin TRI\n");
-
-    return 0;  
-}
-
-static INLINE void trident_draw_point(tridentContextPtr tmesa, 
-			     const tridentVertex *v0 )
-{
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-   (void) MMIO;
-}
-
-static INLINE void trident_draw_line( tridentContextPtr tmesa, 
-			     const tridentVertex *v0,
-			     const tridentVertex *v1 )
-{
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-   (void) MMIO;
-}
-
-static INLINE void trident_draw_triangle( tridentContextPtr tmesa,
-				 const tridentVertex *v0,
-				 const tridentVertex *v1, 
-				 const tridentVertex *v2 )
-{
-}
-
-static INLINE void trident_draw_quad( tridentContextPtr tmesa,
-			    const tridentVertex *v0,
-			    const tridentVertex *v1,
-			    const tridentVertex *v2,
-			    const tridentVertex *v3 )
-{
-   GLuint vertsize = tmesa->vertex_size;
-   GLint coloridx = (vertsize > 4) ? 4 : 3;
-   unsigned char *MMIO = tmesa->tridentScreen->mmio.map;
-   int clr;
-   float *ftmp = (float *)(&clr);
-
-   if (tmesa->dirty)
-	tridentUploadHwStateLocked( tmesa );
-#if 0
-	DrawTriangle(tmesa);
-	exit(0);
-#else
-#if 1
-	if (first) {
-	Init3D(tmesa);
-#if 0
-	DrawTriangle(tmesa);
-#endif
-	first = 0;
-	}
-#endif
-    
-    LOCK_HARDWARE( tmesa );
-
-    MMIO_OUT32(MMIO, 0x002C00 , 0x00000010 );
-    MMIO_OUT32(MMIO, 0x002C04 , 0x029C00C0 );
-
-    /* Z buffer */
-    MMIO_OUT32(MMIO, 0x002C24 , 0x00100000 /*| (tmesa->tridentScreen->depthOffset)*/ );
-    MMIO_OUT32(MMIO, 0x002C28 , 0xE0000000 | (tmesa->tridentScreen->depthPitch * 4) );
-
-    /* front buffer */
-    MMIO_OUT32(MMIO, 0x002C50 , 0x00000000 | (tmesa->drawOffset) );
-    MMIO_OUT32(MMIO, 0x002C54 , 0x0C320000 | (tmesa->drawPitch * 4) );    
-
-    /* clipper */
-    MMIO_OUT32(MMIO, 0x002C80 , 0x20008000 | tmesa->tridentScreen->height );    
-    MMIO_OUT32(MMIO, 0x002C84 , 0x20000000 | tmesa->tridentScreen->width );    
-
-    /* writemask */
-    MMIO_OUT32(MMIO, 0x002C94 , 0xFFFFFFFF );
-
-if (vertsize == 4) {
-    MMIO_OUT32(MMIO, 0x002818 , 0x0003A020 );        
-    MMIO_OUT32(MMIO, 0x00281C , 0x00098021 );	
-
-    *ftmp = v0->v.x;
-    MMIO_OUT32(MMIO, 0x002820 , clr );               
-    *ftmp = v0->v.y;
-    MMIO_OUT32(MMIO, 0x002824 , clr );               
-    *ftmp = v0->v.z;
-    MMIO_OUT32(MMIO, 0x002828 , clr );
-#if 0
-    *ftmp = v0->v.w;
-    MMIO_OUT32(MMIO, 0x00282C , clr );
-#endif
-    MMIO_OUT32(MMIO, 0x00282C , v0->ui[coloridx] );        
-
-    *ftmp = v1->v.x;
-    MMIO_OUT32(MMIO, 0x002820 , clr );               
-    *ftmp = v1->v.y;
-    MMIO_OUT32(MMIO, 0x002824 , clr );               
-    *ftmp = v1->v.z;
-    MMIO_OUT32(MMIO, 0x002828 , clr );
-#if 0
-    *ftmp = v1->v.w;
-    MMIO_OUT32(MMIO, 0x00282C , clr );
-#endif
-    MMIO_OUT32(MMIO, 0x00282C , v1->ui[coloridx] );        
-
-    *ftmp = v2->v.x;
-    MMIO_OUT32(MMIO, 0x002820 , clr );               
-    *ftmp = v2->v.y;
-    MMIO_OUT32(MMIO, 0x002824 , clr );               
-    *ftmp = v2->v.z;
-    MMIO_OUT32(MMIO, 0x002828 , clr );
-#if 0
-    *ftmp = v2->v.w;
-    MMIO_OUT32(MMIO, 0x00282C , clr );
-#endif
-    MMIO_OUT32(MMIO, 0x00282C , v2->ui[coloridx] );        
-
-    MMIO_OUT32(MMIO, 0x00281C , 0x00093020 );    
-    *ftmp = v3->v.x;
-    MMIO_OUT32(MMIO, 0x002820 , clr );               
-    *ftmp = v3->v.y;
-    MMIO_OUT32(MMIO, 0x002824 , clr );               
-    *ftmp = v3->v.z;
-    MMIO_OUT32(MMIO, 0x002828 , clr );
-#if 0
-    *ftmp = v3->v.w;
-    MMIO_OUT32(MMIO, 0x00282C , clr );
-#endif
-    MMIO_OUT32(MMIO, 0x00282C , v3->ui[coloridx] );        
-
-}
-#endif
-
-    UNLOCK_HARDWARE( tmesa );
-}
-/***********************************************************************
- *                    Rasterization fallback helpers                   *
- ***********************************************************************/
-
-
-/* This code is hit only when a mix of accelerated and unaccelerated
- * primitives are being drawn, and only for the unaccelerated
- * primitives.  
- */
-#if 0
-static void 
-trident_fallback_quad( tridentContextPtr tmesa, 
-		    const tridentVertex *v0, 
-		    const tridentVertex *v1, 
-		    const tridentVertex *v2, 
-		    const tridentVertex *v3 )
-{
-   GLcontext *ctx = tmesa->glCtx;
-   SWvertex v[4];
-   trident_translate_vertex( ctx, v0, &v[0] );
-   trident_translate_vertex( ctx, v1, &v[1] );
-   trident_translate_vertex( ctx, v2, &v[2] );
-   trident_translate_vertex( ctx, v3, &v[3] );
-   _swrast_Quad( ctx, &v[0], &v[1], &v[2], &v[3] );
-}
-#endif
-
-/* XXX hack to get the prototype defined in time... */
-void trident_translate_vertex(GLcontext *ctx, const tridentVertex *src,
-                              SWvertex *dst);
-
-static void 
-trident_fallback_tri( tridentContextPtr tmesa, 
-		    const tridentVertex *v0, 
-		    const tridentVertex *v1, 
-		    const tridentVertex *v2 )
-{
-   GLcontext *ctx = tmesa->glCtx;
-   SWvertex v[3];
-   trident_translate_vertex( ctx, v0, &v[0] );
-   trident_translate_vertex( ctx, v1, &v[1] );
-   trident_translate_vertex( ctx, v2, &v[2] );
-   _swrast_Triangle( ctx, &v[0], &v[1], &v[2] );
-}
-
-static void 
-trident_fallback_line( tridentContextPtr tmesa,
-		     const tridentVertex *v0,
-		     const tridentVertex *v1 )
-{
-   GLcontext *ctx = tmesa->glCtx;
-   SWvertex v[2];
-   trident_translate_vertex( ctx, v0, &v[0] );
-   trident_translate_vertex( ctx, v1, &v[1] );
-   _swrast_Line( ctx, &v[0], &v[1] );
-}
-
-
-static void 
-trident_fallback_point( tridentContextPtr tmesa, 
-		      const tridentVertex *v0 )
-{
-   GLcontext *ctx = tmesa->glCtx;
-   SWvertex v[1];
-   trident_translate_vertex( ctx, v0, &v[0] );
-   _swrast_Point( ctx, &v[0] );
-}
-
-/***********************************************************************
- *          Macros for t_dd_tritmp.h to draw basic primitives          *
- ***********************************************************************/
-
-#define TRI( a, b, c )				\
-do {						\
-   if (DO_FALLBACK)				\
-      tmesa->draw_tri( tmesa, a, b, c );	\
-   else						\
-      trident_draw_triangle( tmesa, a, b, c );	\
-} while (0)
-
-#define QUAD( a, b, c, d )			\
-do {						\
-   if (DO_FALLBACK) {				\
-      tmesa->draw_tri( tmesa, a, b, d );	\
-      tmesa->draw_tri( tmesa, b, c, d );	\
-   } else 					\
-      trident_draw_quad( tmesa, a, b, c, d );	\
-} while (0)
-
-#define LINE( v0, v1 )				\
-do {						\
-   if (DO_FALLBACK)				\
-      tmesa->draw_line( tmesa, v0, v1 );	\
-   else 					\
-      trident_draw_line( tmesa, v0, v1 );	\
-} while (0)
-
-#define POINT( v0 )				\
-do {						\
-   if (DO_FALLBACK)				\
-      tmesa->draw_point( tmesa, v0 );		\
-   else 					\
-      trident_draw_point( tmesa, v0 );		\
-} while (0)
-
-/***********************************************************************
- *              Build render functions from dd templates               *
- ***********************************************************************/
-
-#define TRIDENT_OFFSET_BIT 	0x01
-#define TRIDENT_TWOSIDE_BIT	0x02
-#define TRIDENT_UNFILLED_BIT	0x04
-#define TRIDENT_FALLBACK_BIT	0x08
-#define TRIDENT_MAX_TRIFUNC	0x10
-
-
-static struct {
-   tnl_points_func	points;
-   tnl_line_func	line;
-   tnl_triangle_func	triangle;
-   tnl_quad_func	quad;
-} rast_tab[TRIDENT_MAX_TRIFUNC];
-
-
-#define DO_FALLBACK (IND & TRIDENT_FALLBACK_BIT)
-#define DO_OFFSET   (IND & TRIDENT_OFFSET_BIT)
-#define DO_UNFILLED (IND & TRIDENT_UNFILLED_BIT)
-#define DO_TWOSIDE  (IND & TRIDENT_TWOSIDE_BIT)
-#define DO_FLAT      0
-#define DO_TRI       1
-#define DO_QUAD      1
-#define DO_LINE      1
-#define DO_POINTS    1
-#define DO_FULL_QUAD 1
-
-#define HAVE_RGBA         1
-#define HAVE_SPEC         1
-#define HAVE_BACK_COLORS  0
-#define HAVE_HW_FLATSHADE 1
-#define VERTEX            tridentVertex
-#define TAB               rast_tab
-
-#define DEPTH_SCALE 1.0
-#define UNFILLED_TRI unfilled_tri
-#define UNFILLED_QUAD unfilled_quad
-#define VERT_X(_v) _v->v.x
-#define VERT_Y(_v) _v->v.y
-#define VERT_Z(_v) _v->v.z
-#define AREA_IS_CCW( a ) (a > 0)
-#define GET_VERTEX(e) (tmesa->verts + (e<<tmesa->vertex_stride_shift))
-
-#define TRIDENT_COLOR( dst, src )                \
-do {						\
-   dst[0] = src[2];				\
-   dst[1] = src[1];				\
-   dst[2] = src[0];				\
-   dst[3] = src[3];				\
-} while (0)
-
-#define TRIDENT_SPEC( dst, src )			\
-do {						\
-   dst[0] = src[2];				\
-   dst[1] = src[1];				\
-   dst[2] = src[0];				\
-} while (0)
-
-#define VERT_SET_RGBA( v, c )    TRIDENT_COLOR( v->ub4[coloroffset], c )
-#define VERT_COPY_RGBA( v0, v1 ) v0->ui[coloroffset] = v1->ui[coloroffset]
-#define VERT_SAVE_RGBA( idx )    color[idx] = v[idx]->ui[coloroffset]
-#define VERT_RESTORE_RGBA( idx ) v[idx]->ui[coloroffset] = color[idx]
-
-#define VERT_SET_SPEC( v, c )    if (havespec) TRIDENT_SPEC( v->ub4[5], c )
-#define VERT_COPY_SPEC( v0, v1 ) if (havespec) COPY_3V(v0->ub4[5], v1->ub4[5])
-#define VERT_SAVE_SPEC( idx )    if (havespec) spec[idx] = v[idx]->ui[5]
-#define VERT_RESTORE_SPEC( idx ) if (havespec) v[idx]->ui[5] = spec[idx]
-
-#define LOCAL_VARS(n)						\
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);		\
-   GLuint color[n], spec[n];					\
-   GLuint coloroffset = (tmesa->vertex_size == 4 ? 3 : 4);	\
-   GLboolean havespec = (tmesa->vertex_size == 4 ? 0 : 1);	\
-   (void) color; (void) spec; (void) coloroffset; (void) havespec;
-/***********************************************************************
- *                Helpers for rendering unfilled primitives            *
- ***********************************************************************/
-#if 0
-static const GLuint hw_prim[GL_POLYGON+1] = {
-   B_PrimType_Points,
-   B_PrimType_Lines,
-   B_PrimType_Lines,
-   B_PrimType_Lines,
-   B_PrimType_Triangles,
-   B_PrimType_Triangles,
-   B_PrimType_Triangles,
-   B_PrimType_Triangles,
-   B_PrimType_Triangles,
-   B_PrimType_Triangles
-};
-#endif
-
-static void tridentResetLineStipple( GLcontext *ctx );
-#if 0
-static void tridentRasterPrimitive( GLcontext *ctx, GLuint hwprim );
-#endif
-static void tridentRenderPrimitive( GLcontext *ctx, GLenum prim );
-
-#define RASTERIZE(x) /*if (tmesa->hw_primitive != hw_prim[x]) \
-                        tridentRasterPrimitive( ctx, hw_prim[x] ) */
-#define RENDER_PRIMITIVE tmesa->render_primitive
-#define TAG(x) x
-#define IND TRIDENT_FALLBACK_BIT
-#include "tnl_dd/t_dd_unfilled.h"
-#undef IND
-
-/***********************************************************************
- *                      Generate GL render functions                   *
- ***********************************************************************/
-
-#define IND (0)
-#define TAG(x) x
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_OFFSET_BIT)
-#define TAG(x) x##_offset
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT)
-#define TAG(x) x##_twoside
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_OFFSET_BIT)
-#define TAG(x) x##_twoside_offset
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_UNFILLED_BIT)
-#define TAG(x) x##_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_OFFSET_BIT|TRIDENT_UNFILLED_BIT)
-#define TAG(x) x##_offset_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_UNFILLED_BIT)
-#define TAG(x) x##_twoside_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_OFFSET_BIT|TRIDENT_UNFILLED_BIT)
-#define TAG(x) x##_twoside_offset_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_OFFSET_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_offset_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_twoside_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_OFFSET_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_twoside_offset_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_UNFILLED_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_OFFSET_BIT|TRIDENT_UNFILLED_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_offset_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_UNFILLED_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_twoside_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (TRIDENT_TWOSIDE_BIT|TRIDENT_OFFSET_BIT|TRIDENT_UNFILLED_BIT|TRIDENT_FALLBACK_BIT)
-#define TAG(x) x##_twoside_offset_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-static void init_rast_tab( void )
-{
-   init();
-   init_offset();
-   init_twoside();
-   init_twoside_offset();
-   init_unfilled();
-   init_offset_unfilled();
-   init_twoside_unfilled();
-   init_twoside_offset_unfilled();
-   init_fallback();
-   init_offset_fallback();
-   init_twoside_fallback();
-   init_twoside_offset_fallback();
-   init_unfilled_fallback();
-   init_offset_unfilled_fallback();
-   init_twoside_unfilled_fallback();
-   init_twoside_offset_unfilled_fallback();
-}
-
-
-/**********************************************************************/
-/*               Render unclipped begin/end objects                   */
-/**********************************************************************/
-
-#define VERT(x) (tridentVertex *)(tridentverts + (x << shift))
-#define RENDER_POINTS( start, count )		\
-   for ( ; start < count ; start++) 		\
-      trident_draw_point( tmesa, VERT(start) )
-#define RENDER_LINE( v0, v1 ) \
-   trident_draw_line( tmesa, VERT(v0), VERT(v1) )
-#define RENDER_TRI( v0, v1, v2 )  \
-   trident_draw_triangle( tmesa, VERT(v0), VERT(v1), VERT(v2) )
-#define RENDER_QUAD( v0, v1, v2, v3 ) \
-   trident_draw_quad( tmesa, VERT(v0), VERT(v1), VERT(v2), VERT(v3) )
-#define INIT(x) tridentRenderPrimitive( ctx, x );
-#undef LOCAL_VARS
-#define LOCAL_VARS						\
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);		\
-   const GLuint shift = tmesa->vertex_stride_shift;		\
-   const char *tridentverts = (char *)tmesa->verts;		\
-   const GLboolean stipple = ctx->Line.StippleFlag;		\
-   const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;	\
-   (void) elt;
-#define RESET_STIPPLE	if ( stipple ) tridentResetLineStipple( ctx );
-#define RESET_OCCLUSION
-#define PRESERVE_VB_DEFS
-#define ELT(x) (x)
-#define TAG(x) trident_##x##_verts
-#include "tnl/t_vb_rendertmp.h"
-#undef ELT
-#undef TAG
-#define TAG(x) trident_##x##_elts
-#define ELT(x) elt[x]
-#include "tnl/t_vb_rendertmp.h"
-
-/**********************************************************************/
-/*                   Render clipped primitives                        */
-/**********************************************************************/
-
-static void tridentRenderClippedPoly( GLcontext *ctx, const GLuint *elts, 
-				   GLuint n )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint prim = tmesa->render_primitive;
-
-   /* Render the new vertices as an unclipped polygon. 
-    */
-   {
-      GLuint *tmp = VB->Elts;
-      VB->Elts = (GLuint *)elts;
-      tnl->Driver.Render.PrimTabElts[GL_POLYGON]( ctx, 0, n, PRIM_BEGIN|PRIM_END );
-      VB->Elts = tmp;
-   }
-
-   /* Restore the render primitive
-    */
-   if (prim != GL_POLYGON)
-      tnl->Driver.Render.PrimitiveNotify( ctx, prim );
-}
-
-static void tridentRenderClippedLine( GLcontext *ctx, GLuint ii, GLuint jj )
-{
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   tnl->Driver.Render.Line( ctx, ii, jj );
-}
-
-
-/**********************************************************************/
-/*                    Choose render functions                         */
-/**********************************************************************/
-
-#define _TRIDENT_NEW_RENDER_STATE (_DD_NEW_LINE_STIPPLE |	\
-			          _DD_NEW_LINE_SMOOTH |		\
-			          _DD_NEW_POINT_SMOOTH |	\
-			          _DD_NEW_TRI_SMOOTH |		\
-			          _DD_NEW_TRI_UNFILLED |	\
-			          _DD_NEW_TRI_LIGHT_TWOSIDE |	\
-			          _DD_NEW_TRI_OFFSET)		\
-
-
-#define POINT_FALLBACK (DD_POINT_SMOOTH)
-#define LINE_FALLBACK (DD_LINE_STIPPLE|DD_LINE_SMOOTH)
-#define TRI_FALLBACK (DD_TRI_SMOOTH)
-#define ANY_FALLBACK_FLAGS (POINT_FALLBACK|LINE_FALLBACK|TRI_FALLBACK)
-#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE|DD_TRI_OFFSET|DD_TRI_UNFILLED)
-
-
-static void tridentChooseRenderState(GLcontext *ctx)
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint flags = ctx->_TriangleCaps;
-   GLuint index = 0;
-
-   if (flags & (ANY_RASTER_FLAGS|ANY_FALLBACK_FLAGS)) {
-      tmesa->draw_point = trident_draw_point;
-      tmesa->draw_line = trident_draw_line;
-      tmesa->draw_tri = trident_draw_triangle;
-
-      if (flags & ANY_RASTER_FLAGS) {
-	 if (flags & DD_TRI_LIGHT_TWOSIDE) index |= TRIDENT_TWOSIDE_BIT;
-	 if (flags & DD_TRI_OFFSET)        index |= TRIDENT_OFFSET_BIT;
-	 if (flags & DD_TRI_UNFILLED)      index |= TRIDENT_UNFILLED_BIT;
-      }
-
-      /* Hook in fallbacks for specific primitives.
-       */
-      if (flags & (POINT_FALLBACK|LINE_FALLBACK|TRI_FALLBACK)) {
-	 if (flags & POINT_FALLBACK) tmesa->draw_point = trident_fallback_point;
-	 if (flags & LINE_FALLBACK)  tmesa->draw_line = trident_fallback_line;
-	 if (flags & TRI_FALLBACK)   tmesa->draw_tri = trident_fallback_tri;
-	 index |= TRIDENT_FALLBACK_BIT;
-      }
-   }
-
-   if (tmesa->RenderIndex != index) {
-      tmesa->RenderIndex = index;
-
-      tnl->Driver.Render.Points = rast_tab[index].points;
-      tnl->Driver.Render.Line = rast_tab[index].line;
-      tnl->Driver.Render.Triangle = rast_tab[index].triangle;
-      tnl->Driver.Render.Quad = rast_tab[index].quad;
-         
-      if (tmesa->RenderIndex == 0) {
-         tnl->Driver.Render.PrimTabVerts = trident_render_tab_verts;
-         tnl->Driver.Render.PrimTabElts = trident_render_tab_elts;
-      } else {
-         tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
-         tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
-      }
-      tnl->Driver.Render.ClippedLine = tridentRenderClippedLine;
-      tnl->Driver.Render.ClippedPolygon = tridentRenderClippedPoly;
-   }
-}
-
-
-/**********************************************************************/
-/*                 High level hooks for t_vb_render.c                 */
-/**********************************************************************/
-
-
-
-/* Determine the rasterized primitive when not drawing unfilled 
- * polygons.
- *
- * Used only for the default render stage which always decomposes
- * primitives to trianges/lines/points.  For the accelerated stage,
- * which renders strips as strips, the equivalent calculations are
- * performed in tridentrender.c.
- */
-#if 0
-static void tridentRasterPrimitive( GLcontext *ctx, GLuint hwprim )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   if (tmesa->hw_primitive != hwprim)
-      tmesa->hw_primitive = hwprim;
-}
-#endif
-
-static void tridentRenderPrimitive( GLcontext *ctx, GLenum prim )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   tmesa->render_primitive = prim;
-}
-
-static void tridentRunPipeline( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-
-   if ( tmesa->new_state )
-      tridentDDUpdateHWState( ctx );
-
-   if (tmesa->new_gl_state) {
-#if 0
-      if (tmesa->new_gl_state & _NEW_TEXTURE)
-	 tridentUpdateTextureState( ctx );
-#endif
-
-   if (!tmesa->Fallback) {
-      if (tmesa->new_gl_state & _TRIDENT_NEW_VERTEX)
-	 tridentChooseVertexState( ctx );
-      
-      if (tmesa->new_gl_state & _TRIDENT_NEW_RENDER_STATE)
-	 tridentChooseRenderState( ctx );
-   }
-      
-      tmesa->new_gl_state = 0;
-   }
-
-   _tnl_run_pipeline( ctx );
-}
-
-static void tridentRenderStart( GLcontext *ctx )
-{
-   /* Check for projective texturing.  Make sure all texcoord
-    * pointers point to something.  (fix in mesa?)  
-    */
-   tridentCheckTexSizes( ctx );
-}
-
-static void tridentRenderFinish( GLcontext *ctx )
-{
-   if (0)
-      _swrast_flush( ctx );	/* never needed */
-}
-
-static void tridentResetLineStipple( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   (void) tmesa;
-
-   /* Reset the hardware stipple counter.
-    */
-}
-
-
-/**********************************************************************/
-/*           Transition to/from hardware rasterization.               */
-/**********************************************************************/
-
-
-void tridentFallback( tridentContextPtr tmesa, GLuint bit, GLboolean mode )
-{
-   GLcontext *ctx = tmesa->glCtx;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint oldfallback = tmesa->Fallback;
-
-      _tnl_need_projected_coords( ctx, GL_FALSE );
-
-   if (mode) {
-      tmesa->Fallback |= bit;
-      if (oldfallback == 0) {
-	 _swsetup_Wakeup( ctx );
-	 tmesa->RenderIndex = ~0;
-      }
-   }
-   else {
-      tmesa->Fallback &= ~bit;
-      if (oldfallback == bit) {
-	 _swrast_flush( ctx );
-	 tnl->Driver.Render.Start = tridentRenderStart;
-	 tnl->Driver.Render.PrimitiveNotify = tridentRenderPrimitive;
-	 tnl->Driver.Render.Finish = tridentRenderFinish;
-	 tnl->Driver.Render.BuildVertices = tridentBuildVertices;
-         tnl->Driver.Render.ResetLineStipple = tridentResetLineStipple;
-	 tmesa->new_gl_state |= (_TRIDENT_NEW_RENDER_STATE|
-				 _TRIDENT_NEW_VERTEX);
-      }
-   }
-}
-
-
-/**********************************************************************/
-/*                            Initialization.                         */
-/**********************************************************************/
-
-
-void tridentDDInitTriFuncs( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   static int firsttime = 1;
-
-   if (firsttime) {
-      init_rast_tab();
-      firsttime = 0;
-   }
-   
-   tmesa->RenderIndex = ~0;
-
-   tnl->Driver.RunPipeline = tridentRunPipeline;
-   tnl->Driver.Render.Start = tridentRenderStart;
-   tnl->Driver.Render.Finish = tridentRenderFinish; 
-   tnl->Driver.Render.PrimitiveNotify = tridentRenderPrimitive;
-   tnl->Driver.Render.ResetLineStipple = tridentResetLineStipple;
-   tnl->Driver.Render.BuildVertices = tridentBuildVertices;
-}
diff --git a/src/mesa/drivers/dri/trident/trident_vb.c b/src/mesa/drivers/dri/trident/trident_vb.c
deleted file mode 100644
index 055a914595..0000000000
--- a/src/mesa/drivers/dri/trident/trident_vb.c
+++ /dev/null
@@ -1,434 +0,0 @@
-/*
- * Copyright 2002 by Alan Hourihane, Sychdyn, North Wales, UK.
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Alan Hourihane not be used in
- * advertising or publicity pertaining to distribution of the software without
- * specific, written prior permission.  Alan Hourihane makes no representations
- * about the suitability of this software for any purpose.  It is provided
- * "as is" without express or implied warranty.
- *
- * ALAN HOURIHANE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
- * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
- * EVENT SHALL ALAN HOURIHANE BE LIABLE FOR ANY SPECIAL, INDIRECT OR
- * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
- * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
- *
- * Authors:  Alan Hourihane, <alanh@fairlite.demon.co.uk>
- *
- * Trident CyberBladeXP driver.
- *
- */
-#include "main/glheader.h"
-#include "main/mtypes.h"
-#include "main/macros.h"
-#include "main/colormac.h"
-
-#include "swrast_setup/swrast_setup.h"
-#include "swrast/swrast.h"
-#include "tnl/t_context.h"
-#include "tnl/tnl.h"
-
-#include "trident_context.h"
-
-#define TRIDENT_TEX1_BIT       0x1
-#define TRIDENT_TEX0_BIT       0x2
-#define TRIDENT_RGBA_BIT       0x4
-#define TRIDENT_SPEC_BIT       0x8
-#define TRIDENT_FOG_BIT        0x10
-#define TRIDENT_XYZW_BIT       0x20
-#define TRIDENT_PTEX_BIT       0x40
-#define TRIDENT_MAX_SETUP      0x80
-
-static struct {
-   void                (*emit)( GLcontext *, GLuint, GLuint, void *, GLuint );
-   tnl_interp_func	interp;
-   tnl_copy_pv_func     copy_pv;
-   GLboolean           (*check_tex_sizes)( GLcontext *ctx );
-   GLuint               vertex_size;
-   GLuint               vertex_stride_shift;
-   GLuint               vertex_format;
-} setup_tab[TRIDENT_MAX_SETUP];
-
-#define TINY_VERTEX_FORMAT      1
-#define NOTEX_VERTEX_FORMAT     2
-#define TEX0_VERTEX_FORMAT      3
-#define TEX1_VERTEX_FORMAT      4
-#define PROJ_TEX1_VERTEX_FORMAT 5
-#define TEX2_VERTEX_FORMAT      6
-#define TEX3_VERTEX_FORMAT      7
-#define PROJ_TEX3_VERTEX_FORMAT 8
-
-#define DO_XYZW (IND & TRIDENT_XYZW_BIT)
-#define DO_RGBA (IND & TRIDENT_RGBA_BIT)
-#define DO_SPEC (IND & TRIDENT_SPEC_BIT)
-#define DO_FOG  (IND & TRIDENT_FOG_BIT)
-#define DO_TEX0 (IND & TRIDENT_TEX0_BIT)
-#define DO_TEX1 (IND & TRIDENT_TEX1_BIT)
-#define DO_TEX2 0
-#define DO_TEX3 0
-#define DO_PTEX (IND & TRIDENT_PTEX_BIT)
-
-#define VERTEX tridentVertex
-#define VERTEX_COLOR trident_color_t
-#define LOCALVARS tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-#define GET_VIEWPORT_MAT() 		tmesa->hw_viewport
-#define GET_TEXSOURCE(n)  		tmesa->tmu_source[n]
-#define GET_VERTEX_FORMAT() 		tmesa->vertex_format
-#define GET_VERTEX_SIZE()               tmesa->vertex_size
-#define GET_VERTEX_STORE()  		tmesa->verts
-#define GET_VERTEX_STRIDE_SHIFT() 	tmesa->vertex_stride_shift
-#define GET_UBYTE_COLOR_STORE() 	&tmesa->UbyteColor
-#define GET_UBYTE_SPEC_COLOR_STORE() 	&tmesa->UbyteSecondaryColor
-			       
-#define HAVE_HW_VIEWPORT    0
-#define HAVE_HW_DIVIDE      0
-#define HAVE_RGBA_COLOR     0 
-#define HAVE_TINY_VERTICES  1
-#define HAVE_NOTEX_VERTICES 1
-#define HAVE_TEX0_VERTICES  1
-#define HAVE_TEX1_VERTICES  1
-#define HAVE_TEX2_VERTICES  0
-#define HAVE_TEX3_VERTICES  0
-#define HAVE_PTEX_VERTICES  0
-
-#define UNVIEWPORT_VARS                                 \
-   const GLfloat dx = - tmesa->drawX - SUBPIXEL_X;	\
-   const GLfloat dy = (tmesa->driDrawable->h + 	        \
-		       tmesa->drawY + SUBPIXEL_Y);	\
-   const GLfloat sz = 1.0 / tmesa->depth_scale
-
-#define UNVIEWPORT_X(x)    x  +  dx;
-#define UNVIEWPORT_Y(y)  - y  +  dy;
-#define UNVIEWPORT_Z(z)    z  *  sz;
-
-#define PTEX_FALLBACK()	 tridentFallback(TRIDENT_CONTEXT(ctx), TRIDENT_FALLBACK_TEXTURE, 1)
-
-#define IMPORT_FLOAT_COLORS trident_import_float_colors
-#define IMPORT_FLOAT_SPEC_COLORS trident_import_float_spec_colors
-
-#define INTERP_VERTEX setup_tab[tmesa->SetupIndex].interp
-#define COPY_PV_VERTEX setup_tab[tmesa->SetupIndex].copy_pv
-
-/***********************************************************************
- *         Generate  pv-copying and translation functions              *
- ***********************************************************************/
-
-#define TAG(x) trident_##x
-#include "tnl_dd/t_dd_vb.c"
-
-/***********************************************************************
- *             Generate vertex emit and interp functions               *
- ***********************************************************************/
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT)
-#define TAG(x) x##_wg
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT)
-#define TAG(x) x##_wgs
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_wgt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_wgt0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_TEX0_BIT|TRIDENT_PTEX_BIT)
-#define TAG(x) x##_wgpt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_wgst0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_wgst0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT|TRIDENT_PTEX_BIT)
-#define TAG(x) x##_wgspt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT)
-#define TAG(x) x##_wgf
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT)
-#define TAG(x) x##_wgfs
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_wgft0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_wgft0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT|TRIDENT_PTEX_BIT)
-#define TAG(x) x##_wgfpt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_wgfst0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_wgfst0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT|TRIDENT_PTEX_BIT)
-#define TAG(x) x##_wgfspt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_TEX0_BIT)
-#define TAG(x) x##_t0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_t0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_FOG_BIT)
-#define TAG(x) x##_f
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_ft0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_ft0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT)
-#define TAG(x) x##_g
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT)
-#define TAG(x) x##_gs
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_gt0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_gt0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_gst0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_gst0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT)
-#define TAG(x) x##_gf
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT)
-#define TAG(x) x##_gfs
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_gft0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_gft0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT)
-#define TAG(x) x##_gfst0
-#include "tnl_dd/t_dd_vbtmp.h"
-
-#define IND (TRIDENT_RGBA_BIT|TRIDENT_FOG_BIT|TRIDENT_SPEC_BIT|TRIDENT_TEX0_BIT|TRIDENT_TEX1_BIT)
-#define TAG(x) x##_gfst0t1
-#include "tnl_dd/t_dd_vbtmp.h"
-
-static void init_setup_tab( void )
-{
-   init_wg();
-   init_wgs();
-   init_wgt0();
-   init_wgt0t1();
-   init_wgpt0();
-   init_wgst0();
-   init_wgst0t1();
-   init_wgspt0();
-   init_wgf();
-   init_wgfs();
-   init_wgft0();
-   init_wgft0t1();
-   init_wgfpt0();
-   init_wgfst0();
-   init_wgfst0t1();
-   init_wgfspt0();
-   init_t0();
-   init_t0t1();
-   init_f();
-   init_ft0();
-   init_ft0t1();
-   init_g();
-   init_gs();
-   init_gt0();
-   init_gt0t1();
-   init_gst0();
-   init_gst0t1();
-   init_gf();
-   init_gfs();
-   init_gft0();
-   init_gft0t1();
-   init_gfst0();
-   init_gfst0t1();
-}
-
-void tridentBuildVertices( GLcontext *ctx, 
-			 GLuint start, 
-			 GLuint count,
-			 GLuint newinputs )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT( ctx );
-   GLubyte *v = ((GLubyte *)tmesa->verts + (start<<tmesa->vertex_stride_shift));
-   GLuint stride = 1<<tmesa->vertex_stride_shift;
-
-   newinputs |= tmesa->SetupNewInputs;
-   tmesa->SetupNewInputs = 0;
-
-   if (!newinputs)
-      return;
-
-   if (newinputs & VERT_BIT_POS) {
-      setup_tab[tmesa->SetupIndex].emit( ctx, start, count, v, stride );   
-   } else {
-      GLuint ind = 0;
-
-      if (newinputs & VERT_BIT_COLOR0)
-	 ind |= TRIDENT_RGBA_BIT;
-
-      if (newinputs & VERT_BIT_COLOR1)
-	 ind |= TRIDENT_SPEC_BIT;
-
-      if (newinputs & VERT_BIT_TEX0)
-	 ind |= TRIDENT_TEX0_BIT;
-
-      if (newinputs & VERT_BIT_TEX1)
-	 ind |= TRIDENT_TEX1_BIT;
-
-      if (newinputs & VERT_BIT_FOG)
-	 ind |= TRIDENT_FOG_BIT;
-
-      if (tmesa->SetupIndex & TRIDENT_PTEX_BIT)
-	 ind = ~0;
-
-      ind &= tmesa->SetupIndex;
-
-      if (ind) {
-	 setup_tab[ind].emit( ctx, start, count, v, stride );   
-      }
-   }
-}
-
-void tridentCheckTexSizes( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT( ctx );
-
-   if (!setup_tab[tmesa->SetupIndex].check_tex_sizes(ctx)) {
-      TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-      /* Invalidate stored verts
-       */
-      tmesa->SetupNewInputs = ~0;
-      tmesa->SetupIndex |= TRIDENT_PTEX_BIT;
-
-      if (!tmesa->Fallback &&
-	  !(ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED))) {
-	 tnl->Driver.Render.Interp = setup_tab[tmesa->SetupIndex].interp;
-	 tnl->Driver.Render.CopyPV = setup_tab[tmesa->SetupIndex].copy_pv;
-      }
-   }
-}
-
-void tridentChooseVertexState( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT( ctx );
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint ind = TRIDENT_XYZW_BIT|TRIDENT_RGBA_BIT;
-
-   if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR)
-      ind |= TRIDENT_SPEC_BIT;
-
-   if (ctx->Fog.Enabled)
-      ind |= TRIDENT_FOG_BIT;
-
-   if (ctx->Texture.Unit[0]._ReallyEnabled) {
-      ind |= TRIDENT_TEX0_BIT;
-      if (ctx->Texture.Unit[1]._ReallyEnabled) {
-	 ind |= TRIDENT_TEX1_BIT;
-      }
-   }
-
-   tmesa->SetupIndex = ind;
-
-   if (ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED)) {
-      tnl->Driver.Render.Interp = trident_interp_extras;
-      tnl->Driver.Render.CopyPV = trident_copy_pv_extras;
-   } else {
-      tnl->Driver.Render.Interp = setup_tab[ind].interp;
-      tnl->Driver.Render.CopyPV = setup_tab[ind].copy_pv;
-   }
-
-   if (setup_tab[ind].vertex_format != tmesa->vertex_format) {
-      tmesa->vertex_format = setup_tab[ind].vertex_format;
-      tmesa->vertex_size = setup_tab[ind].vertex_size;
-      tmesa->vertex_stride_shift = setup_tab[ind].vertex_stride_shift;
-   }
-}
-
-void tridentInitVB( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-   GLuint size = TNL_CONTEXT(ctx)->vb.Size;
-
-   tmesa->verts = (GLubyte *)ALIGN_MALLOC( size * 16 * 4, 32 );
-
-   {
-      static int firsttime = 1;
-      if (firsttime) {
-	 init_setup_tab();
-	 firsttime = 0;
-      }
-   }
-}
-
-void tridentFreeVB( GLcontext *ctx )
-{
-   tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
-
-   if (tmesa->verts) {
-      ALIGN_FREE(tmesa->verts);
-      tmesa->verts = 0;
-   }
-
-   if (tmesa->UbyteSecondaryColor.Ptr) {
-      ALIGN_FREE((void *)tmesa->UbyteSecondaryColor.Ptr);
-      tmesa->UbyteSecondaryColor.Ptr = 0;
-   }
-
-   if (tmesa->UbyteColor.Ptr) {
-      ALIGN_FREE((void *)tmesa->UbyteColor.Ptr);
-      tmesa->UbyteColor.Ptr = 0;
-   }
-}
diff --git a/src/mesa/drivers/dri/unichrome/Makefile b/src/mesa/drivers/dri/unichrome/Makefile
index 5fe00c1bd1..344d34fce3 100644
--- a/src/mesa/drivers/dri/unichrome/Makefile
+++ b/src/mesa/drivers/dri/unichrome/Makefile
@@ -29,4 +29,3 @@ ASM_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/unichrome/via_tex.c b/src/mesa/drivers/dri/unichrome/via_tex.c
index d2010f0907..54073e7691 100644
--- a/src/mesa/drivers/dri/unichrome/via_tex.c
+++ b/src/mesa/drivers/dri/unichrome/via_tex.c
@@ -818,11 +818,6 @@ static void viaTexImage(GLcontext *ctx,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
diff --git a/src/mesa/drivers/osmesa/osmesa.c b/src/mesa/drivers/osmesa/osmesa.c
index 904659e345..692657a5df 100644
--- a/src/mesa/drivers/osmesa/osmesa.c
+++ b/src/mesa/drivers/osmesa/osmesa.c
@@ -50,6 +50,7 @@
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
 #include "drivers/common/driverfuncs.h"
+#include "drivers/common/meta.h"
 #include "vbo/vbo.h"
 
 
@@ -1258,6 +1259,8 @@ OSMesaCreateContextExt( GLenum format, GLint depthBits, GLint stencilBits,
       osmesa->bInd = bind;
       osmesa->aInd = aind;
 
+      _mesa_meta_init(&osmesa->mesa);
+
       /* Initialize the software rasterizer and helper modules. */
       {
 	 GLcontext *ctx = &osmesa->mesa;
@@ -1304,6 +1307,8 @@ OSMesaDestroyContext( OSMesaContext osmesa )
       if (osmesa->rb)
          _mesa_reference_renderbuffer(&osmesa->rb, NULL);
 
+      _mesa_meta_free( &osmesa->mesa );
+
       _swsetup_DestroyContext( &osmesa->mesa );
       _tnl_DestroyContext( &osmesa->mesa );
       _vbo_DestroyContext( &osmesa->mesa );
diff --git a/src/mesa/drivers/windows/gdi/mesa.def b/src/mesa/drivers/windows/gdi/mesa.def
index bd3e5b2137..5abcd1d927 100644
--- a/src/mesa/drivers/windows/gdi/mesa.def
+++ b/src/mesa/drivers/windows/gdi/mesa.def
@@ -943,6 +943,15 @@ EXPORTS
 	_mesa_update_framebuffer_visual
 	_mesa_use_program
 	_mesa_Viewport
+	_mesa_meta_CopyColorSubTable
+	_mesa_meta_CopyColorTable
+	_mesa_meta_CopyConvolutionFilter1D
+	_mesa_meta_CopyConvolutionFilter2D
+	_mesa_meta_CopyTexImage1D
+	_mesa_meta_CopyTexImage2D
+	_mesa_meta_CopyTexSubImage1D
+	_mesa_meta_CopyTexSubImage2D
+	_mesa_meta_CopyTexSubImage3D
 	_mesa_wait_query
 	_swrast_Accum
 	_swrast_Bitmap
@@ -953,15 +962,6 @@ EXPORTS
 	_swrast_Clear
 	_swrast_choose_line
 	_swrast_choose_triangle
-	_swrast_CopyColorSubTable
-	_swrast_CopyColorTable
-	_swrast_CopyConvolutionFilter1D
-	_swrast_CopyConvolutionFilter2D
-	_swrast_copy_teximage1d
-	_swrast_copy_teximage2d
-	_swrast_copy_texsubimage1d
-	_swrast_copy_texsubimage2d
-	_swrast_copy_texsubimage3d
 	_swrast_CreateContext
 	_swrast_DestroyContext
 	_swrast_exec_fragment_program
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index e1971db693..8929b22af1 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -12,6 +12,7 @@
 #include "framebuffer.h"
 #include "renderbuffer.h"
 #include "drivers/common/driverfuncs.h"
+#include "drivers/common/meta.h"
 #include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -1515,6 +1516,8 @@ WMesaContext WMesaCreateContext(HDC hDC,
     _mesa_enable_2_0_extensions(ctx);
     _mesa_enable_2_1_extensions(ctx);
   
+    _mesa_meta_init(ctx);
+
     /* Initialize the software rasterizer and helper modules. */
     if (!_swrast_CreateContext(ctx) ||
         !_vbo_CreateContext(ctx) ||
@@ -1558,6 +1561,8 @@ void WMesaDestroyContext( WMesaContext pwc )
     DeleteObject(pwc->clearPen); 
     DeleteObject(pwc->clearBrush); 
     
+    _mesa_meta_free(ctx);
+
     _swsetup_DestroyContext(ctx);
     _tnl_DestroyContext(ctx);
     _vbo_DestroyContext(ctx);
diff --git a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
index 342a742867..7ac425a109 100644
--- a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
+++ b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
@@ -1346,6 +1346,8 @@ static void wmesa_update_state_first_time(
 	struct swrast_device_driver	*swdd = _swrast_GetDeviceDriverReference( ctx );
 	TNLcontext					*tnl = TNL_CONTEXT(ctx);
 	
+        _mesa_init_driver_functions(&ctx->Driver);
+
 	/*
 	* XXX these function pointers could be initialized just once during
 	* context creation since they don't depend on any state changes.
@@ -1362,8 +1364,6 @@ static void wmesa_update_state_first_time(
 
 	ctx->Driver.Viewport				= wmesa_viewport;
 	
-	ctx->Driver.Accum					= _swrast_Accum;
-	ctx->Driver.Bitmap					= _swrast_Bitmap;
 	ctx->Driver.Clear					= clear;
 	
 	ctx->Driver.Flush					= flush;
@@ -1371,28 +1371,6 @@ static void wmesa_update_state_first_time(
 	ctx->Driver.ClearColor				= clear_color;
 	ctx->Driver.Enable					= enable;
 	
-	ctx->Driver.CopyPixels				= _swrast_CopyPixels;
-	ctx->Driver.DrawPixels				= _swrast_DrawPixels;
-	ctx->Driver.ReadPixels				= _swrast_ReadPixels;
-	
-	ctx->Driver.ChooseTextureFormat		= _mesa_choose_tex_format;
-	ctx->Driver.TexImage1D				= _mesa_store_teximage1d;
-	ctx->Driver.TexImage2D				= _mesa_store_teximage2d;
-	ctx->Driver.TexImage3D				= _mesa_store_teximage3d;
-	ctx->Driver.TexSubImage1D			= _mesa_store_texsubimage1d;
-	ctx->Driver.TexSubImage2D			= _mesa_store_texsubimage2d;
-	ctx->Driver.TexSubImage3D			= _mesa_store_texsubimage3d;
-	ctx->Driver.TestProxyTexImage		= _mesa_test_proxy_teximage;
-	
-	ctx->Driver.CopyTexImage1D			= _swrast_copy_teximage1d;
-	ctx->Driver.CopyTexImage2D			= _swrast_copy_teximage2d;
-	ctx->Driver.CopyTexSubImage1D		= _swrast_copy_texsubimage1d;
-	ctx->Driver.CopyTexSubImage2D		= _swrast_copy_texsubimage2d;
-	ctx->Driver.CopyTexSubImage3D		= _swrast_copy_texsubimage3d;
-	ctx->Driver.CopyColorTable			= _swrast_CopyColorTable;
-	ctx->Driver.CopyColorSubTable		= _swrast_CopyColorSubTable;
-	ctx->Driver.CopyConvolutionFilter1D	= _swrast_CopyConvolutionFilter1D;
-	ctx->Driver.CopyConvolutionFilter2D	= _swrast_CopyConvolutionFilter2D;
 	
 	// Does not apply for Mesa 5.x
 	//ctx->Driver.BaseCompressedTexFormat	= _mesa_base_compressed_texformat;
diff --git a/src/mesa/drivers/windows/icd/mesa.def b/src/mesa/drivers/windows/icd/mesa.def
index 465b380a0c..25ac08a2f0 100644
--- a/src/mesa/drivers/windows/icd/mesa.def
+++ b/src/mesa/drivers/windows/icd/mesa.def
@@ -75,6 +75,15 @@ EXPORTS
 	_mesa_strcmp
 	_mesa_test_proxy_teximage
 	_mesa_Viewport
+	_mesa_meta_CopyColorSubTable
+	_mesa_meta_CopyColorTable
+	_mesa_meta_CopyConvolutionFilter1D
+	_mesa_meta_CopyConvolutionFilter2D
+	_mesa_meta_CopyTexImage1D
+	_mesa_meta_CopyTexImage2D
+	_mesa_meta_CopyTexSubImage1D
+	_mesa_meta_CopyTexSubImage2D
+	_mesa_meta_CopyTexSubImage3D
 	_swrast_Accum
 	_swrast_Bitmap
 	_swrast_CopyPixels
@@ -84,15 +93,6 @@ EXPORTS
 	_swrast_Clear
 	_swrast_choose_line
 	_swrast_choose_triangle
-	_swrast_CopyColorSubTable
-	_swrast_CopyColorTable
-	_swrast_CopyConvolutionFilter1D
-	_swrast_CopyConvolutionFilter2D
-	_swrast_copy_teximage1d
-	_swrast_copy_teximage2d
-	_swrast_copy_texsubimage1d
-	_swrast_copy_texsubimage2d
-	_swrast_copy_texsubimage3d
 	_swrast_CreateContext
 	_swrast_DestroyContext
 	_swrast_InvalidateState
diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c
index 2c7be9f182..79b058634c 100644
--- a/src/mesa/drivers/x11/xm_api.c
+++ b/src/mesa/drivers/x11/xm_api.c
@@ -1317,6 +1317,7 @@ xmesa_convert_from_x_visual_type( int visualType )
 #define need_GL_SGI_color_table
 
 /* sw extensions not associated with some GL version */
+#define need_GL_ARB_draw_elements_base_vertex
 #define need_GL_ARB_shader_objects
 #define need_GL_ARB_sync
 #define need_GL_ARB_vertex_program
@@ -1347,6 +1348,8 @@ const struct dri_extension card_extensions[] =
    { "GL_EXT_histogram",		GL_EXT_histogram_functions },
    { "GL_SGI_color_table",		GL_SGI_color_table_functions },
 
+   { "GL_ARB_depth_clamp",		NULL },
+   { "GL_ARB_draw_elements_base_vertex", GL_ARB_draw_elements_base_vertex_functions },
    { "GL_ARB_shader_objects",		GL_ARB_shader_objects_functions },
    { "GL_ARB_sync",			GL_ARB_sync_functions },
    { "GL_ARB_vertex_program",		GL_ARB_vertex_program_functions },
@@ -1358,6 +1361,7 @@ const struct dri_extension card_extensions[] =
    { "GL_EXT_gpu_program_parameters",	GL_EXT_gpu_program_parameters_functions },
    { "GL_EXT_paletted_texture",		GL_EXT_paletted_texture_functions },
    { "GL_MESA_resize_buffers",		GL_MESA_resize_buffers_functions },
+   { "GL_NV_depth_clamp",		NULL },
    { "GL_NV_vertex_program",		GL_NV_vertex_program_functions },
    { "GL_NV_fragment_program",		GL_NV_fragment_program_functions },
    { NULL,				NULL }
@@ -1644,8 +1648,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
    xmesa_register_swrast_functions( mesaCtx );
    _swsetup_Wakeup(mesaCtx);
 
-   if (TEST_META_FUNCS)
-      _mesa_meta_init(mesaCtx);
+   _mesa_meta_init(mesaCtx);
 
    return c;
 }
@@ -1661,8 +1664,7 @@ void XMesaDestroyContext( XMesaContext c )
    FXdestroyContext( XMESA_BUFFER(mesaCtx->DrawBuffer) );
 #endif
 
-   if (TEST_META_FUNCS)
-      _mesa_meta_free( mesaCtx );
+   _mesa_meta_free( mesaCtx );
 
    _swsetup_DestroyContext( mesaCtx );
    _swrast_DestroyContext( mesaCtx );
diff --git a/src/mesa/drivers/x11/xm_dd.c b/src/mesa/drivers/x11/xm_dd.c
index 465a40ac8d..e2d4aa9b2d 100644
--- a/src/mesa/drivers/x11/xm_dd.c
+++ b/src/mesa/drivers/x11/xm_dd.c
@@ -1150,10 +1150,11 @@ xmesa_init_driver_functions( XMesaVisual xmvisual,
    driver->Enable = enable;
    driver->Viewport = xmesa_viewport;
    if (TEST_META_FUNCS) {
-      driver->Clear = _mesa_meta_clear;
-      driver->CopyPixels = _mesa_meta_copy_pixels;
-      driver->BlitFramebuffer = _mesa_meta_blit_framebuffer;
-      driver->DrawPixels = _mesa_meta_draw_pixels;
+      driver->Clear = _mesa_meta_Clear;
+      driver->CopyPixels = _mesa_meta_CopyPixels;
+      driver->BlitFramebuffer = _mesa_meta_BlitFramebuffer;
+      driver->DrawPixels = _mesa_meta_DrawPixels;
+      driver->Bitmap = _mesa_meta_Bitmap;
    }
    else {
       driver->Clear = clear_buffers;
@@ -1167,8 +1168,8 @@ xmesa_init_driver_functions( XMesaVisual xmvisual,
       else if (xmvisual->undithered_pf == PF_5R6G5B) {
          driver->DrawPixels = xmesa_DrawPixels_5R6G5B;
       }
-   }
 #endif
+   }
    driver->TestProxyTexImage = test_proxy_teximage;
 #if ENABLE_EXT_texure_compression_s3tc
    driver->ChooseTextureFormat = choose_tex_format;
diff --git a/src/mesa/glapi/ARB_depth_clamp.xml b/src/mesa/glapi/ARB_depth_clamp.xml
new file mode 100644
index 0000000000..157c9a86b1
--- /dev/null
+++ b/src/mesa/glapi/ARB_depth_clamp.xml
@@ -0,0 +1,12 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+
+<category name="GL_ARB_depth_clamp" number="61">
+    <enum name="DEPTH_CLAMP" count="1"  value="0x864F">
+        <size name="Get" mode="get"/>
+    </enum>
+</category>
+
+</OpenGLAPI>
diff --git a/src/mesa/glapi/ARB_draw_elements_base_vertex.xml b/src/mesa/glapi/ARB_draw_elements_base_vertex.xml
new file mode 100644
index 0000000000..f4067f4c8d
--- /dev/null
+++ b/src/mesa/glapi/ARB_draw_elements_base_vertex.xml
@@ -0,0 +1,40 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<!-- Note: no GLX protocol info yet. -->
+
+
+<OpenGLAPI>
+
+<category name="GL_ARB_draw_elements_base_vertex" number="62">
+
+    <function name="DrawElementsBaseVertex" offset="assign">
+        <param name="mode" type="GLenum"/>
+        <param name="count" type="GLsizei"/>
+        <param name="type" type="GLenum"/>
+        <param name="indices" type="const GLvoid *"/>
+        <param name="basevertex" type="GLint"/>
+    </function>
+
+    <function name="DrawRangeElementsBaseVertex" offset="assign">
+        <param name="mode" type="GLenum"/>
+        <param name="start" type="GLuint"/>
+        <param name="end" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="type" type="GLenum"/>
+        <param name="indices" type="const GLvoid *"/>
+        <param name="basevertex" type="GLint"/>
+    </function>
+
+    <function name="MultiDrawElementsBaseVertex" offset="assign">
+        <param name="mode" type="GLenum"/>
+        <param name="count" type="const GLsizei *"/>
+        <param name="type" type="GLenum"/>
+        <param name="indices" type="const GLvoid **"/>
+        <param name="primcount" type="GLsizei"/>
+        <param name="basevertex" type="const GLint *"/>
+    </function>
+
+</category>
+
+</OpenGLAPI>
diff --git a/src/mesa/glapi/EXT_provoking_vertex.xml b/src/mesa/glapi/EXT_provoking_vertex.xml
index f528a2c7d3..71d2c72909 100644
--- a/src/mesa/glapi/EXT_provoking_vertex.xml
+++ b/src/mesa/glapi/EXT_provoking_vertex.xml
@@ -19,4 +19,17 @@
 
 </category>
 
+<category name="GL_ARB_provoking_vertex" number="64">
+
+    <enum name="FIRST_VERTEX_CONVENTION"                  value="0x8E4D"/>
+    <enum name="LAST_VERTEX_CONVENTION"                   value="0x8E4E"/>
+    <enum name="PROVOKING_VERTEX"                         value="0x8E4F"/>
+    <enum name="QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION" value="0x8E4C"/>
+
+    <function name="ProvokingVertex" alias="ProvokingVertexEXT">
+        <param name="mode" type="GLenum"/>
+    </function>
+
+</category>
+
 </OpenGLAPI>
diff --git a/src/mesa/glapi/Makefile b/src/mesa/glapi/Makefile
index 65edab7cec..22f65b74c2 100644
--- a/src/mesa/glapi/Makefile
+++ b/src/mesa/glapi/Makefile
@@ -48,6 +48,7 @@ SERVER_OUTPUTS = \
 API_XML = gl_API.xml \
 	EXT_framebuffer_object.xml \
 	ARB_copy_buffer.xml \
+	ARB_depth_clamp.xml \
 	ARB_framebuffer_object.xml \
 	ARB_map_buffer_range.xml \
 	ARB_seamless_cube_map.xml \
diff --git a/src/mesa/glapi/dispatch.h b/src/mesa/glapi/dispatch.h
index 0d32045916..efbd56fa71 100644
--- a/src/mesa/glapi/dispatch.h
+++ b/src/mesa/glapi/dispatch.h
@@ -1782,6 +1782,15 @@
 #define CALL_WaitSync(disp, parameters) (*((disp)->WaitSync)) parameters
 #define GET_WaitSync(disp) ((disp)->WaitSync)
 #define SET_WaitSync(disp, fn) ((disp)->WaitSync = fn)
+#define CALL_DrawElementsBaseVertex(disp, parameters) (*((disp)->DrawElementsBaseVertex)) parameters
+#define GET_DrawElementsBaseVertex(disp) ((disp)->DrawElementsBaseVertex)
+#define SET_DrawElementsBaseVertex(disp, fn) ((disp)->DrawElementsBaseVertex = fn)
+#define CALL_DrawRangeElementsBaseVertex(disp, parameters) (*((disp)->DrawRangeElementsBaseVertex)) parameters
+#define GET_DrawRangeElementsBaseVertex(disp) ((disp)->DrawRangeElementsBaseVertex)
+#define SET_DrawRangeElementsBaseVertex(disp, fn) ((disp)->DrawRangeElementsBaseVertex = fn)
+#define CALL_MultiDrawElementsBaseVertex(disp, parameters) (*((disp)->MultiDrawElementsBaseVertex)) parameters
+#define GET_MultiDrawElementsBaseVertex(disp) ((disp)->MultiDrawElementsBaseVertex)
+#define SET_MultiDrawElementsBaseVertex(disp, fn) ((disp)->MultiDrawElementsBaseVertex = fn)
 #define CALL_PolygonOffsetEXT(disp, parameters) (*((disp)->PolygonOffsetEXT)) parameters
 #define GET_PolygonOffsetEXT(disp) ((disp)->PolygonOffsetEXT)
 #define SET_PolygonOffsetEXT(disp, fn) ((disp)->PolygonOffsetEXT = fn)
@@ -2439,7 +2448,7 @@
 
 #else
 
-#define driDispatchRemapTable_size 384
+#define driDispatchRemapTable_size 387
 extern int driDispatchRemapTable[ driDispatchRemapTable_size ];
 
 #define AttachShader_remap_index 0
@@ -2608,224 +2617,227 @@ extern int driDispatchRemapTable[ driDispatchRemapTable_size ];
 #define GetSynciv_remap_index 163
 #define IsSync_remap_index 164
 #define WaitSync_remap_index 165
-#define PolygonOffsetEXT_remap_index 166
-#define GetPixelTexGenParameterfvSGIS_remap_index 167
-#define GetPixelTexGenParameterivSGIS_remap_index 168
-#define PixelTexGenParameterfSGIS_remap_index 169
-#define PixelTexGenParameterfvSGIS_remap_index 170
-#define PixelTexGenParameteriSGIS_remap_index 171
-#define PixelTexGenParameterivSGIS_remap_index 172
-#define SampleMaskSGIS_remap_index 173
-#define SamplePatternSGIS_remap_index 174
-#define ColorPointerEXT_remap_index 175
-#define EdgeFlagPointerEXT_remap_index 176
-#define IndexPointerEXT_remap_index 177
-#define NormalPointerEXT_remap_index 178
-#define TexCoordPointerEXT_remap_index 179
-#define VertexPointerEXT_remap_index 180
-#define PointParameterfEXT_remap_index 181
-#define PointParameterfvEXT_remap_index 182
-#define LockArraysEXT_remap_index 183
-#define UnlockArraysEXT_remap_index 184
-#define CullParameterdvEXT_remap_index 185
-#define CullParameterfvEXT_remap_index 186
-#define SecondaryColor3bEXT_remap_index 187
-#define SecondaryColor3bvEXT_remap_index 188
-#define SecondaryColor3dEXT_remap_index 189
-#define SecondaryColor3dvEXT_remap_index 190
-#define SecondaryColor3fEXT_remap_index 191
-#define SecondaryColor3fvEXT_remap_index 192
-#define SecondaryColor3iEXT_remap_index 193
-#define SecondaryColor3ivEXT_remap_index 194
-#define SecondaryColor3sEXT_remap_index 195
-#define SecondaryColor3svEXT_remap_index 196
-#define SecondaryColor3ubEXT_remap_index 197
-#define SecondaryColor3ubvEXT_remap_index 198
-#define SecondaryColor3uiEXT_remap_index 199
-#define SecondaryColor3uivEXT_remap_index 200
-#define SecondaryColor3usEXT_remap_index 201
-#define SecondaryColor3usvEXT_remap_index 202
-#define SecondaryColorPointerEXT_remap_index 203
-#define MultiDrawArraysEXT_remap_index 204
-#define MultiDrawElementsEXT_remap_index 205
-#define FogCoordPointerEXT_remap_index 206
-#define FogCoorddEXT_remap_index 207
-#define FogCoorddvEXT_remap_index 208
-#define FogCoordfEXT_remap_index 209
-#define FogCoordfvEXT_remap_index 210
-#define PixelTexGenSGIX_remap_index 211
-#define BlendFuncSeparateEXT_remap_index 212
-#define FlushVertexArrayRangeNV_remap_index 213
-#define VertexArrayRangeNV_remap_index 214
-#define CombinerInputNV_remap_index 215
-#define CombinerOutputNV_remap_index 216
-#define CombinerParameterfNV_remap_index 217
-#define CombinerParameterfvNV_remap_index 218
-#define CombinerParameteriNV_remap_index 219
-#define CombinerParameterivNV_remap_index 220
-#define FinalCombinerInputNV_remap_index 221
-#define GetCombinerInputParameterfvNV_remap_index 222
-#define GetCombinerInputParameterivNV_remap_index 223
-#define GetCombinerOutputParameterfvNV_remap_index 224
-#define GetCombinerOutputParameterivNV_remap_index 225
-#define GetFinalCombinerInputParameterfvNV_remap_index 226
-#define GetFinalCombinerInputParameterivNV_remap_index 227
-#define ResizeBuffersMESA_remap_index 228
-#define WindowPos2dMESA_remap_index 229
-#define WindowPos2dvMESA_remap_index 230
-#define WindowPos2fMESA_remap_index 231
-#define WindowPos2fvMESA_remap_index 232
-#define WindowPos2iMESA_remap_index 233
-#define WindowPos2ivMESA_remap_index 234
-#define WindowPos2sMESA_remap_index 235
-#define WindowPos2svMESA_remap_index 236
-#define WindowPos3dMESA_remap_index 237
-#define WindowPos3dvMESA_remap_index 238
-#define WindowPos3fMESA_remap_index 239
-#define WindowPos3fvMESA_remap_index 240
-#define WindowPos3iMESA_remap_index 241
-#define WindowPos3ivMESA_remap_index 242
-#define WindowPos3sMESA_remap_index 243
-#define WindowPos3svMESA_remap_index 244
-#define WindowPos4dMESA_remap_index 245
-#define WindowPos4dvMESA_remap_index 246
-#define WindowPos4fMESA_remap_index 247
-#define WindowPos4fvMESA_remap_index 248
-#define WindowPos4iMESA_remap_index 249
-#define WindowPos4ivMESA_remap_index 250
-#define WindowPos4sMESA_remap_index 251
-#define WindowPos4svMESA_remap_index 252
-#define MultiModeDrawArraysIBM_remap_index 253
-#define MultiModeDrawElementsIBM_remap_index 254
-#define DeleteFencesNV_remap_index 255
-#define FinishFenceNV_remap_index 256
-#define GenFencesNV_remap_index 257
-#define GetFenceivNV_remap_index 258
-#define IsFenceNV_remap_index 259
-#define SetFenceNV_remap_index 260
-#define TestFenceNV_remap_index 261
-#define AreProgramsResidentNV_remap_index 262
-#define BindProgramNV_remap_index 263
-#define DeleteProgramsNV_remap_index 264
-#define ExecuteProgramNV_remap_index 265
-#define GenProgramsNV_remap_index 266
-#define GetProgramParameterdvNV_remap_index 267
-#define GetProgramParameterfvNV_remap_index 268
-#define GetProgramStringNV_remap_index 269
-#define GetProgramivNV_remap_index 270
-#define GetTrackMatrixivNV_remap_index 271
-#define GetVertexAttribPointervNV_remap_index 272
-#define GetVertexAttribdvNV_remap_index 273
-#define GetVertexAttribfvNV_remap_index 274
-#define GetVertexAttribivNV_remap_index 275
-#define IsProgramNV_remap_index 276
-#define LoadProgramNV_remap_index 277
-#define ProgramParameters4dvNV_remap_index 278
-#define ProgramParameters4fvNV_remap_index 279
-#define RequestResidentProgramsNV_remap_index 280
-#define TrackMatrixNV_remap_index 281
-#define VertexAttrib1dNV_remap_index 282
-#define VertexAttrib1dvNV_remap_index 283
-#define VertexAttrib1fNV_remap_index 284
-#define VertexAttrib1fvNV_remap_index 285
-#define VertexAttrib1sNV_remap_index 286
-#define VertexAttrib1svNV_remap_index 287
-#define VertexAttrib2dNV_remap_index 288
-#define VertexAttrib2dvNV_remap_index 289
-#define VertexAttrib2fNV_remap_index 290
-#define VertexAttrib2fvNV_remap_index 291
-#define VertexAttrib2sNV_remap_index 292
-#define VertexAttrib2svNV_remap_index 293
-#define VertexAttrib3dNV_remap_index 294
-#define VertexAttrib3dvNV_remap_index 295
-#define VertexAttrib3fNV_remap_index 296
-#define VertexAttrib3fvNV_remap_index 297
-#define VertexAttrib3sNV_remap_index 298
-#define VertexAttrib3svNV_remap_index 299
-#define VertexAttrib4dNV_remap_index 300
-#define VertexAttrib4dvNV_remap_index 301
-#define VertexAttrib4fNV_remap_index 302
-#define VertexAttrib4fvNV_remap_index 303
-#define VertexAttrib4sNV_remap_index 304
-#define VertexAttrib4svNV_remap_index 305
-#define VertexAttrib4ubNV_remap_index 306
-#define VertexAttrib4ubvNV_remap_index 307
-#define VertexAttribPointerNV_remap_index 308
-#define VertexAttribs1dvNV_remap_index 309
-#define VertexAttribs1fvNV_remap_index 310
-#define VertexAttribs1svNV_remap_index 311
-#define VertexAttribs2dvNV_remap_index 312
-#define VertexAttribs2fvNV_remap_index 313
-#define VertexAttribs2svNV_remap_index 314
-#define VertexAttribs3dvNV_remap_index 315
-#define VertexAttribs3fvNV_remap_index 316
-#define VertexAttribs3svNV_remap_index 317
-#define VertexAttribs4dvNV_remap_index 318
-#define VertexAttribs4fvNV_remap_index 319
-#define VertexAttribs4svNV_remap_index 320
-#define VertexAttribs4ubvNV_remap_index 321
-#define GetTexBumpParameterfvATI_remap_index 322
-#define GetTexBumpParameterivATI_remap_index 323
-#define TexBumpParameterfvATI_remap_index 324
-#define TexBumpParameterivATI_remap_index 325
-#define AlphaFragmentOp1ATI_remap_index 326
-#define AlphaFragmentOp2ATI_remap_index 327
-#define AlphaFragmentOp3ATI_remap_index 328
-#define BeginFragmentShaderATI_remap_index 329
-#define BindFragmentShaderATI_remap_index 330
-#define ColorFragmentOp1ATI_remap_index 331
-#define ColorFragmentOp2ATI_remap_index 332
-#define ColorFragmentOp3ATI_remap_index 333
-#define DeleteFragmentShaderATI_remap_index 334
-#define EndFragmentShaderATI_remap_index 335
-#define GenFragmentShadersATI_remap_index 336
-#define PassTexCoordATI_remap_index 337
-#define SampleMapATI_remap_index 338
-#define SetFragmentShaderConstantATI_remap_index 339
-#define PointParameteriNV_remap_index 340
-#define PointParameterivNV_remap_index 341
-#define ActiveStencilFaceEXT_remap_index 342
-#define BindVertexArrayAPPLE_remap_index 343
-#define DeleteVertexArraysAPPLE_remap_index 344
-#define GenVertexArraysAPPLE_remap_index 345
-#define IsVertexArrayAPPLE_remap_index 346
-#define GetProgramNamedParameterdvNV_remap_index 347
-#define GetProgramNamedParameterfvNV_remap_index 348
-#define ProgramNamedParameter4dNV_remap_index 349
-#define ProgramNamedParameter4dvNV_remap_index 350
-#define ProgramNamedParameter4fNV_remap_index 351
-#define ProgramNamedParameter4fvNV_remap_index 352
-#define DepthBoundsEXT_remap_index 353
-#define BlendEquationSeparateEXT_remap_index 354
-#define BindFramebufferEXT_remap_index 355
-#define BindRenderbufferEXT_remap_index 356
-#define CheckFramebufferStatusEXT_remap_index 357
-#define DeleteFramebuffersEXT_remap_index 358
-#define DeleteRenderbuffersEXT_remap_index 359
-#define FramebufferRenderbufferEXT_remap_index 360
-#define FramebufferTexture1DEXT_remap_index 361
-#define FramebufferTexture2DEXT_remap_index 362
-#define FramebufferTexture3DEXT_remap_index 363
-#define GenFramebuffersEXT_remap_index 364
-#define GenRenderbuffersEXT_remap_index 365
-#define GenerateMipmapEXT_remap_index 366
-#define GetFramebufferAttachmentParameterivEXT_remap_index 367
-#define GetRenderbufferParameterivEXT_remap_index 368
-#define IsFramebufferEXT_remap_index 369
-#define IsRenderbufferEXT_remap_index 370
-#define RenderbufferStorageEXT_remap_index 371
-#define BlitFramebufferEXT_remap_index 372
-#define BufferParameteriAPPLE_remap_index 373
-#define FlushMappedBufferRangeAPPLE_remap_index 374
-#define FramebufferTextureLayerEXT_remap_index 375
-#define ProvokingVertexEXT_remap_index 376
-#define GetTexParameterPointervAPPLE_remap_index 377
-#define TextureRangeAPPLE_remap_index 378
-#define StencilFuncSeparateATI_remap_index 379
-#define ProgramEnvParameters4fvEXT_remap_index 380
-#define ProgramLocalParameters4fvEXT_remap_index 381
-#define GetQueryObjecti64vEXT_remap_index 382
-#define GetQueryObjectui64vEXT_remap_index 383
+#define DrawElementsBaseVertex_remap_index 166
+#define DrawRangeElementsBaseVertex_remap_index 167
+#define MultiDrawElementsBaseVertex_remap_index 168
+#define PolygonOffsetEXT_remap_index 169
+#define GetPixelTexGenParameterfvSGIS_remap_index 170
+#define GetPixelTexGenParameterivSGIS_remap_index 171
+#define PixelTexGenParameterfSGIS_remap_index 172
+#define PixelTexGenParameterfvSGIS_remap_index 173
+#define PixelTexGenParameteriSGIS_remap_index 174
+#define PixelTexGenParameterivSGIS_remap_index 175
+#define SampleMaskSGIS_remap_index 176
+#define SamplePatternSGIS_remap_index 177
+#define ColorPointerEXT_remap_index 178
+#define EdgeFlagPointerEXT_remap_index 179
+#define IndexPointerEXT_remap_index 180
+#define NormalPointerEXT_remap_index 181
+#define TexCoordPointerEXT_remap_index 182
+#define VertexPointerEXT_remap_index 183
+#define PointParameterfEXT_remap_index 184
+#define PointParameterfvEXT_remap_index 185
+#define LockArraysEXT_remap_index 186
+#define UnlockArraysEXT_remap_index 187
+#define CullParameterdvEXT_remap_index 188
+#define CullParameterfvEXT_remap_index 189
+#define SecondaryColor3bEXT_remap_index 190
+#define SecondaryColor3bvEXT_remap_index 191
+#define SecondaryColor3dEXT_remap_index 192
+#define SecondaryColor3dvEXT_remap_index 193
+#define SecondaryColor3fEXT_remap_index 194
+#define SecondaryColor3fvEXT_remap_index 195
+#define SecondaryColor3iEXT_remap_index 196
+#define SecondaryColor3ivEXT_remap_index 197
+#define SecondaryColor3sEXT_remap_index 198
+#define SecondaryColor3svEXT_remap_index 199
+#define SecondaryColor3ubEXT_remap_index 200
+#define SecondaryColor3ubvEXT_remap_index 201
+#define SecondaryColor3uiEXT_remap_index 202
+#define SecondaryColor3uivEXT_remap_index 203
+#define SecondaryColor3usEXT_remap_index 204
+#define SecondaryColor3usvEXT_remap_index 205
+#define SecondaryColorPointerEXT_remap_index 206
+#define MultiDrawArraysEXT_remap_index 207
+#define MultiDrawElementsEXT_remap_index 208
+#define FogCoordPointerEXT_remap_index 209
+#define FogCoorddEXT_remap_index 210
+#define FogCoorddvEXT_remap_index 211
+#define FogCoordfEXT_remap_index 212
+#define FogCoordfvEXT_remap_index 213
+#define PixelTexGenSGIX_remap_index 214
+#define BlendFuncSeparateEXT_remap_index 215
+#define FlushVertexArrayRangeNV_remap_index 216
+#define VertexArrayRangeNV_remap_index 217
+#define CombinerInputNV_remap_index 218
+#define CombinerOutputNV_remap_index 219
+#define CombinerParameterfNV_remap_index 220
+#define CombinerParameterfvNV_remap_index 221
+#define CombinerParameteriNV_remap_index 222
+#define CombinerParameterivNV_remap_index 223
+#define FinalCombinerInputNV_remap_index 224
+#define GetCombinerInputParameterfvNV_remap_index 225
+#define GetCombinerInputParameterivNV_remap_index 226
+#define GetCombinerOutputParameterfvNV_remap_index 227
+#define GetCombinerOutputParameterivNV_remap_index 228
+#define GetFinalCombinerInputParameterfvNV_remap_index 229
+#define GetFinalCombinerInputParameterivNV_remap_index 230
+#define ResizeBuffersMESA_remap_index 231
+#define WindowPos2dMESA_remap_index 232
+#define WindowPos2dvMESA_remap_index 233
+#define WindowPos2fMESA_remap_index 234
+#define WindowPos2fvMESA_remap_index 235
+#define WindowPos2iMESA_remap_index 236
+#define WindowPos2ivMESA_remap_index 237
+#define WindowPos2sMESA_remap_index 238
+#define WindowPos2svMESA_remap_index 239
+#define WindowPos3dMESA_remap_index 240
+#define WindowPos3dvMESA_remap_index 241
+#define WindowPos3fMESA_remap_index 242
+#define WindowPos3fvMESA_remap_index 243
+#define WindowPos3iMESA_remap_index 244
+#define WindowPos3ivMESA_remap_index 245
+#define WindowPos3sMESA_remap_index 246
+#define WindowPos3svMESA_remap_index 247
+#define WindowPos4dMESA_remap_index 248
+#define WindowPos4dvMESA_remap_index 249
+#define WindowPos4fMESA_remap_index 250
+#define WindowPos4fvMESA_remap_index 251
+#define WindowPos4iMESA_remap_index 252
+#define WindowPos4ivMESA_remap_index 253
+#define WindowPos4sMESA_remap_index 254
+#define WindowPos4svMESA_remap_index 255
+#define MultiModeDrawArraysIBM_remap_index 256
+#define MultiModeDrawElementsIBM_remap_index 257
+#define DeleteFencesNV_remap_index 258
+#define FinishFenceNV_remap_index 259
+#define GenFencesNV_remap_index 260
+#define GetFenceivNV_remap_index 261
+#define IsFenceNV_remap_index 262
+#define SetFenceNV_remap_index 263
+#define TestFenceNV_remap_index 264
+#define AreProgramsResidentNV_remap_index 265
+#define BindProgramNV_remap_index 266
+#define DeleteProgramsNV_remap_index 267
+#define ExecuteProgramNV_remap_index 268
+#define GenProgramsNV_remap_index 269
+#define GetProgramParameterdvNV_remap_index 270
+#define GetProgramParameterfvNV_remap_index 271
+#define GetProgramStringNV_remap_index 272
+#define GetProgramivNV_remap_index 273
+#define GetTrackMatrixivNV_remap_index 274
+#define GetVertexAttribPointervNV_remap_index 275
+#define GetVertexAttribdvNV_remap_index 276
+#define GetVertexAttribfvNV_remap_index 277
+#define GetVertexAttribivNV_remap_index 278
+#define IsProgramNV_remap_index 279
+#define LoadProgramNV_remap_index 280
+#define ProgramParameters4dvNV_remap_index 281
+#define ProgramParameters4fvNV_remap_index 282
+#define RequestResidentProgramsNV_remap_index 283
+#define TrackMatrixNV_remap_index 284
+#define VertexAttrib1dNV_remap_index 285
+#define VertexAttrib1dvNV_remap_index 286
+#define VertexAttrib1fNV_remap_index 287
+#define VertexAttrib1fvNV_remap_index 288
+#define VertexAttrib1sNV_remap_index 289
+#define VertexAttrib1svNV_remap_index 290
+#define VertexAttrib2dNV_remap_index 291
+#define VertexAttrib2dvNV_remap_index 292
+#define VertexAttrib2fNV_remap_index 293
+#define VertexAttrib2fvNV_remap_index 294
+#define VertexAttrib2sNV_remap_index 295
+#define VertexAttrib2svNV_remap_index 296
+#define VertexAttrib3dNV_remap_index 297
+#define VertexAttrib3dvNV_remap_index 298
+#define VertexAttrib3fNV_remap_index 299
+#define VertexAttrib3fvNV_remap_index 300
+#define VertexAttrib3sNV_remap_index 301
+#define VertexAttrib3svNV_remap_index 302
+#define VertexAttrib4dNV_remap_index 303
+#define VertexAttrib4dvNV_remap_index 304
+#define VertexAttrib4fNV_remap_index 305
+#define VertexAttrib4fvNV_remap_index 306
+#define VertexAttrib4sNV_remap_index 307
+#define VertexAttrib4svNV_remap_index 308
+#define VertexAttrib4ubNV_remap_index 309
+#define VertexAttrib4ubvNV_remap_index 310
+#define VertexAttribPointerNV_remap_index 311
+#define VertexAttribs1dvNV_remap_index 312
+#define VertexAttribs1fvNV_remap_index 313
+#define VertexAttribs1svNV_remap_index 314
+#define VertexAttribs2dvNV_remap_index 315
+#define VertexAttribs2fvNV_remap_index 316
+#define VertexAttribs2svNV_remap_index 317
+#define VertexAttribs3dvNV_remap_index 318
+#define VertexAttribs3fvNV_remap_index 319
+#define VertexAttribs3svNV_remap_index 320
+#define VertexAttribs4dvNV_remap_index 321
+#define VertexAttribs4fvNV_remap_index 322
+#define VertexAttribs4svNV_remap_index 323
+#define VertexAttribs4ubvNV_remap_index 324
+#define GetTexBumpParameterfvATI_remap_index 325
+#define GetTexBumpParameterivATI_remap_index 326
+#define TexBumpParameterfvATI_remap_index 327
+#define TexBumpParameterivATI_remap_index 328
+#define AlphaFragmentOp1ATI_remap_index 329
+#define AlphaFragmentOp2ATI_remap_index 330
+#define AlphaFragmentOp3ATI_remap_index 331
+#define BeginFragmentShaderATI_remap_index 332
+#define BindFragmentShaderATI_remap_index 333
+#define ColorFragmentOp1ATI_remap_index 334
+#define ColorFragmentOp2ATI_remap_index 335
+#define ColorFragmentOp3ATI_remap_index 336
+#define DeleteFragmentShaderATI_remap_index 337
+#define EndFragmentShaderATI_remap_index 338
+#define GenFragmentShadersATI_remap_index 339
+#define PassTexCoordATI_remap_index 340
+#define SampleMapATI_remap_index 341
+#define SetFragmentShaderConstantATI_remap_index 342
+#define PointParameteriNV_remap_index 343
+#define PointParameterivNV_remap_index 344
+#define ActiveStencilFaceEXT_remap_index 345
+#define BindVertexArrayAPPLE_remap_index 346
+#define DeleteVertexArraysAPPLE_remap_index 347
+#define GenVertexArraysAPPLE_remap_index 348
+#define IsVertexArrayAPPLE_remap_index 349
+#define GetProgramNamedParameterdvNV_remap_index 350
+#define GetProgramNamedParameterfvNV_remap_index 351
+#define ProgramNamedParameter4dNV_remap_index 352
+#define ProgramNamedParameter4dvNV_remap_index 353
+#define ProgramNamedParameter4fNV_remap_index 354
+#define ProgramNamedParameter4fvNV_remap_index 355
+#define DepthBoundsEXT_remap_index 356
+#define BlendEquationSeparateEXT_remap_index 357
+#define BindFramebufferEXT_remap_index 358
+#define BindRenderbufferEXT_remap_index 359
+#define CheckFramebufferStatusEXT_remap_index 360
+#define DeleteFramebuffersEXT_remap_index 361
+#define DeleteRenderbuffersEXT_remap_index 362
+#define FramebufferRenderbufferEXT_remap_index 363
+#define FramebufferTexture1DEXT_remap_index 364
+#define FramebufferTexture2DEXT_remap_index 365
+#define FramebufferTexture3DEXT_remap_index 366
+#define GenFramebuffersEXT_remap_index 367
+#define GenRenderbuffersEXT_remap_index 368
+#define GenerateMipmapEXT_remap_index 369
+#define GetFramebufferAttachmentParameterivEXT_remap_index 370
+#define GetRenderbufferParameterivEXT_remap_index 371
+#define IsFramebufferEXT_remap_index 372
+#define IsRenderbufferEXT_remap_index 373
+#define RenderbufferStorageEXT_remap_index 374
+#define BlitFramebufferEXT_remap_index 375
+#define BufferParameteriAPPLE_remap_index 376
+#define FlushMappedBufferRangeAPPLE_remap_index 377
+#define FramebufferTextureLayerEXT_remap_index 378
+#define ProvokingVertexEXT_remap_index 379
+#define GetTexParameterPointervAPPLE_remap_index 380
+#define TextureRangeAPPLE_remap_index 381
+#define StencilFuncSeparateATI_remap_index 382
+#define ProgramEnvParameters4fvEXT_remap_index 383
+#define ProgramLocalParameters4fvEXT_remap_index 384
+#define GetQueryObjecti64vEXT_remap_index 385
+#define GetQueryObjectui64vEXT_remap_index 386
 
 #define CALL_AttachShader(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLuint, GLuint)), driDispatchRemapTable[AttachShader_remap_index], parameters)
 #define GET_AttachShader(disp) GET_by_offset(disp, driDispatchRemapTable[AttachShader_remap_index])
@@ -3325,6 +3337,15 @@ extern int driDispatchRemapTable[ driDispatchRemapTable_size ];
 #define CALL_WaitSync(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLsync, GLbitfield, GLuint64)), driDispatchRemapTable[WaitSync_remap_index], parameters)
 #define GET_WaitSync(disp) GET_by_offset(disp, driDispatchRemapTable[WaitSync_remap_index])
 #define SET_WaitSync(disp, fn) SET_by_offset(disp, driDispatchRemapTable[WaitSync_remap_index], fn)
+#define CALL_DrawElementsBaseVertex(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLsizei, GLenum, const GLvoid *, GLint)), driDispatchRemapTable[DrawElementsBaseVertex_remap_index], parameters)
+#define GET_DrawElementsBaseVertex(disp) GET_by_offset(disp, driDispatchRemapTable[DrawElementsBaseVertex_remap_index])
+#define SET_DrawElementsBaseVertex(disp, fn) SET_by_offset(disp, driDispatchRemapTable[DrawElementsBaseVertex_remap_index], fn)
+#define CALL_DrawRangeElementsBaseVertex(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, GLuint, GLsizei, GLenum, const GLvoid *, GLint)), driDispatchRemapTable[DrawRangeElementsBaseVertex_remap_index], parameters)
+#define GET_DrawRangeElementsBaseVertex(disp) GET_by_offset(disp, driDispatchRemapTable[DrawRangeElementsBaseVertex_remap_index])
+#define SET_DrawRangeElementsBaseVertex(disp, fn) SET_by_offset(disp, driDispatchRemapTable[DrawRangeElementsBaseVertex_remap_index], fn)
+#define CALL_MultiDrawElementsBaseVertex(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, const GLsizei *, GLenum, const GLvoid **, GLsizei, const GLint *)), driDispatchRemapTable[MultiDrawElementsBaseVertex_remap_index], parameters)
+#define GET_MultiDrawElementsBaseVertex(disp) GET_by_offset(disp, driDispatchRemapTable[MultiDrawElementsBaseVertex_remap_index])
+#define SET_MultiDrawElementsBaseVertex(disp, fn) SET_by_offset(disp, driDispatchRemapTable[MultiDrawElementsBaseVertex_remap_index], fn)
 #define CALL_PolygonOffsetEXT(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLfloat, GLfloat)), driDispatchRemapTable[PolygonOffsetEXT_remap_index], parameters)
 #define GET_PolygonOffsetEXT(disp) GET_by_offset(disp, driDispatchRemapTable[PolygonOffsetEXT_remap_index])
 #define SET_PolygonOffsetEXT(disp, fn) SET_by_offset(disp, driDispatchRemapTable[PolygonOffsetEXT_remap_index], fn)
diff --git a/src/mesa/glapi/gl_API.xml b/src/mesa/glapi/gl_API.xml
index 1dfd92be08..da4be14707 100644
--- a/src/mesa/glapi/gl_API.xml
+++ b/src/mesa/glapi/gl_API.xml
@@ -7950,6 +7950,8 @@
 
 <xi:include href="ARB_copy_buffer.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
+<xi:include href="ARB_depth_clamp.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
+
 <xi:include href="ARB_map_buffer_range.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
 <xi:include href="ARB_vertex_array_object.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
@@ -7958,6 +7960,8 @@
 
 <xi:include href="ARB_seamless_cube_map.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
+<xi:include href="ARB_draw_elements_base_vertex.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
+
 
 <!-- Non-ARB extensions sorted by extension number. -->
 
diff --git a/src/mesa/glapi/glapioffsets.h b/src/mesa/glapi/glapioffsets.h
index 6d4ab09b07..a3807744ff 100644
--- a/src/mesa/glapi/glapioffsets.h
+++ b/src/mesa/glapi/glapioffsets.h
@@ -606,225 +606,228 @@
 #define _gloffset_GetSynciv 571
 #define _gloffset_IsSync 572
 #define _gloffset_WaitSync 573
-#define _gloffset_PolygonOffsetEXT 574
-#define _gloffset_GetPixelTexGenParameterfvSGIS 575
-#define _gloffset_GetPixelTexGenParameterivSGIS 576
-#define _gloffset_PixelTexGenParameterfSGIS 577
-#define _gloffset_PixelTexGenParameterfvSGIS 578
-#define _gloffset_PixelTexGenParameteriSGIS 579
-#define _gloffset_PixelTexGenParameterivSGIS 580
-#define _gloffset_SampleMaskSGIS 581
-#define _gloffset_SamplePatternSGIS 582
-#define _gloffset_ColorPointerEXT 583
-#define _gloffset_EdgeFlagPointerEXT 584
-#define _gloffset_IndexPointerEXT 585
-#define _gloffset_NormalPointerEXT 586
-#define _gloffset_TexCoordPointerEXT 587
-#define _gloffset_VertexPointerEXT 588
-#define _gloffset_PointParameterfEXT 589
-#define _gloffset_PointParameterfvEXT 590
-#define _gloffset_LockArraysEXT 591
-#define _gloffset_UnlockArraysEXT 592
-#define _gloffset_CullParameterdvEXT 593
-#define _gloffset_CullParameterfvEXT 594
-#define _gloffset_SecondaryColor3bEXT 595
-#define _gloffset_SecondaryColor3bvEXT 596
-#define _gloffset_SecondaryColor3dEXT 597
-#define _gloffset_SecondaryColor3dvEXT 598
-#define _gloffset_SecondaryColor3fEXT 599
-#define _gloffset_SecondaryColor3fvEXT 600
-#define _gloffset_SecondaryColor3iEXT 601
-#define _gloffset_SecondaryColor3ivEXT 602
-#define _gloffset_SecondaryColor3sEXT 603
-#define _gloffset_SecondaryColor3svEXT 604
-#define _gloffset_SecondaryColor3ubEXT 605
-#define _gloffset_SecondaryColor3ubvEXT 606
-#define _gloffset_SecondaryColor3uiEXT 607
-#define _gloffset_SecondaryColor3uivEXT 608
-#define _gloffset_SecondaryColor3usEXT 609
-#define _gloffset_SecondaryColor3usvEXT 610
-#define _gloffset_SecondaryColorPointerEXT 611
-#define _gloffset_MultiDrawArraysEXT 612
-#define _gloffset_MultiDrawElementsEXT 613
-#define _gloffset_FogCoordPointerEXT 614
-#define _gloffset_FogCoorddEXT 615
-#define _gloffset_FogCoorddvEXT 616
-#define _gloffset_FogCoordfEXT 617
-#define _gloffset_FogCoordfvEXT 618
-#define _gloffset_PixelTexGenSGIX 619
-#define _gloffset_BlendFuncSeparateEXT 620
-#define _gloffset_FlushVertexArrayRangeNV 621
-#define _gloffset_VertexArrayRangeNV 622
-#define _gloffset_CombinerInputNV 623
-#define _gloffset_CombinerOutputNV 624
-#define _gloffset_CombinerParameterfNV 625
-#define _gloffset_CombinerParameterfvNV 626
-#define _gloffset_CombinerParameteriNV 627
-#define _gloffset_CombinerParameterivNV 628
-#define _gloffset_FinalCombinerInputNV 629
-#define _gloffset_GetCombinerInputParameterfvNV 630
-#define _gloffset_GetCombinerInputParameterivNV 631
-#define _gloffset_GetCombinerOutputParameterfvNV 632
-#define _gloffset_GetCombinerOutputParameterivNV 633
-#define _gloffset_GetFinalCombinerInputParameterfvNV 634
-#define _gloffset_GetFinalCombinerInputParameterivNV 635
-#define _gloffset_ResizeBuffersMESA 636
-#define _gloffset_WindowPos2dMESA 637
-#define _gloffset_WindowPos2dvMESA 638
-#define _gloffset_WindowPos2fMESA 639
-#define _gloffset_WindowPos2fvMESA 640
-#define _gloffset_WindowPos2iMESA 641
-#define _gloffset_WindowPos2ivMESA 642
-#define _gloffset_WindowPos2sMESA 643
-#define _gloffset_WindowPos2svMESA 644
-#define _gloffset_WindowPos3dMESA 645
-#define _gloffset_WindowPos3dvMESA 646
-#define _gloffset_WindowPos3fMESA 647
-#define _gloffset_WindowPos3fvMESA 648
-#define _gloffset_WindowPos3iMESA 649
-#define _gloffset_WindowPos3ivMESA 650
-#define _gloffset_WindowPos3sMESA 651
-#define _gloffset_WindowPos3svMESA 652
-#define _gloffset_WindowPos4dMESA 653
-#define _gloffset_WindowPos4dvMESA 654
-#define _gloffset_WindowPos4fMESA 655
-#define _gloffset_WindowPos4fvMESA 656
-#define _gloffset_WindowPos4iMESA 657
-#define _gloffset_WindowPos4ivMESA 658
-#define _gloffset_WindowPos4sMESA 659
-#define _gloffset_WindowPos4svMESA 660
-#define _gloffset_MultiModeDrawArraysIBM 661
-#define _gloffset_MultiModeDrawElementsIBM 662
-#define _gloffset_DeleteFencesNV 663
-#define _gloffset_FinishFenceNV 664
-#define _gloffset_GenFencesNV 665
-#define _gloffset_GetFenceivNV 666
-#define _gloffset_IsFenceNV 667
-#define _gloffset_SetFenceNV 668
-#define _gloffset_TestFenceNV 669
-#define _gloffset_AreProgramsResidentNV 670
-#define _gloffset_BindProgramNV 671
-#define _gloffset_DeleteProgramsNV 672
-#define _gloffset_ExecuteProgramNV 673
-#define _gloffset_GenProgramsNV 674
-#define _gloffset_GetProgramParameterdvNV 675
-#define _gloffset_GetProgramParameterfvNV 676
-#define _gloffset_GetProgramStringNV 677
-#define _gloffset_GetProgramivNV 678
-#define _gloffset_GetTrackMatrixivNV 679
-#define _gloffset_GetVertexAttribPointervNV 680
-#define _gloffset_GetVertexAttribdvNV 681
-#define _gloffset_GetVertexAttribfvNV 682
-#define _gloffset_GetVertexAttribivNV 683
-#define _gloffset_IsProgramNV 684
-#define _gloffset_LoadProgramNV 685
-#define _gloffset_ProgramParameters4dvNV 686
-#define _gloffset_ProgramParameters4fvNV 687
-#define _gloffset_RequestResidentProgramsNV 688
-#define _gloffset_TrackMatrixNV 689
-#define _gloffset_VertexAttrib1dNV 690
-#define _gloffset_VertexAttrib1dvNV 691
-#define _gloffset_VertexAttrib1fNV 692
-#define _gloffset_VertexAttrib1fvNV 693
-#define _gloffset_VertexAttrib1sNV 694
-#define _gloffset_VertexAttrib1svNV 695
-#define _gloffset_VertexAttrib2dNV 696
-#define _gloffset_VertexAttrib2dvNV 697
-#define _gloffset_VertexAttrib2fNV 698
-#define _gloffset_VertexAttrib2fvNV 699
-#define _gloffset_VertexAttrib2sNV 700
-#define _gloffset_VertexAttrib2svNV 701
-#define _gloffset_VertexAttrib3dNV 702
-#define _gloffset_VertexAttrib3dvNV 703
-#define _gloffset_VertexAttrib3fNV 704
-#define _gloffset_VertexAttrib3fvNV 705
-#define _gloffset_VertexAttrib3sNV 706
-#define _gloffset_VertexAttrib3svNV 707
-#define _gloffset_VertexAttrib4dNV 708
-#define _gloffset_VertexAttrib4dvNV 709
-#define _gloffset_VertexAttrib4fNV 710
-#define _gloffset_VertexAttrib4fvNV 711
-#define _gloffset_VertexAttrib4sNV 712
-#define _gloffset_VertexAttrib4svNV 713
-#define _gloffset_VertexAttrib4ubNV 714
-#define _gloffset_VertexAttrib4ubvNV 715
-#define _gloffset_VertexAttribPointerNV 716
-#define _gloffset_VertexAttribs1dvNV 717
-#define _gloffset_VertexAttribs1fvNV 718
-#define _gloffset_VertexAttribs1svNV 719
-#define _gloffset_VertexAttribs2dvNV 720
-#define _gloffset_VertexAttribs2fvNV 721
-#define _gloffset_VertexAttribs2svNV 722
-#define _gloffset_VertexAttribs3dvNV 723
-#define _gloffset_VertexAttribs3fvNV 724
-#define _gloffset_VertexAttribs3svNV 725
-#define _gloffset_VertexAttribs4dvNV 726
-#define _gloffset_VertexAttribs4fvNV 727
-#define _gloffset_VertexAttribs4svNV 728
-#define _gloffset_VertexAttribs4ubvNV 729
-#define _gloffset_GetTexBumpParameterfvATI 730
-#define _gloffset_GetTexBumpParameterivATI 731
-#define _gloffset_TexBumpParameterfvATI 732
-#define _gloffset_TexBumpParameterivATI 733
-#define _gloffset_AlphaFragmentOp1ATI 734
-#define _gloffset_AlphaFragmentOp2ATI 735
-#define _gloffset_AlphaFragmentOp3ATI 736
-#define _gloffset_BeginFragmentShaderATI 737
-#define _gloffset_BindFragmentShaderATI 738
-#define _gloffset_ColorFragmentOp1ATI 739
-#define _gloffset_ColorFragmentOp2ATI 740
-#define _gloffset_ColorFragmentOp3ATI 741
-#define _gloffset_DeleteFragmentShaderATI 742
-#define _gloffset_EndFragmentShaderATI 743
-#define _gloffset_GenFragmentShadersATI 744
-#define _gloffset_PassTexCoordATI 745
-#define _gloffset_SampleMapATI 746
-#define _gloffset_SetFragmentShaderConstantATI 747
-#define _gloffset_PointParameteriNV 748
-#define _gloffset_PointParameterivNV 749
-#define _gloffset_ActiveStencilFaceEXT 750
-#define _gloffset_BindVertexArrayAPPLE 751
-#define _gloffset_DeleteVertexArraysAPPLE 752
-#define _gloffset_GenVertexArraysAPPLE 753
-#define _gloffset_IsVertexArrayAPPLE 754
-#define _gloffset_GetProgramNamedParameterdvNV 755
-#define _gloffset_GetProgramNamedParameterfvNV 756
-#define _gloffset_ProgramNamedParameter4dNV 757
-#define _gloffset_ProgramNamedParameter4dvNV 758
-#define _gloffset_ProgramNamedParameter4fNV 759
-#define _gloffset_ProgramNamedParameter4fvNV 760
-#define _gloffset_DepthBoundsEXT 761
-#define _gloffset_BlendEquationSeparateEXT 762
-#define _gloffset_BindFramebufferEXT 763
-#define _gloffset_BindRenderbufferEXT 764
-#define _gloffset_CheckFramebufferStatusEXT 765
-#define _gloffset_DeleteFramebuffersEXT 766
-#define _gloffset_DeleteRenderbuffersEXT 767
-#define _gloffset_FramebufferRenderbufferEXT 768
-#define _gloffset_FramebufferTexture1DEXT 769
-#define _gloffset_FramebufferTexture2DEXT 770
-#define _gloffset_FramebufferTexture3DEXT 771
-#define _gloffset_GenFramebuffersEXT 772
-#define _gloffset_GenRenderbuffersEXT 773
-#define _gloffset_GenerateMipmapEXT 774
-#define _gloffset_GetFramebufferAttachmentParameterivEXT 775
-#define _gloffset_GetRenderbufferParameterivEXT 776
-#define _gloffset_IsFramebufferEXT 777
-#define _gloffset_IsRenderbufferEXT 778
-#define _gloffset_RenderbufferStorageEXT 779
-#define _gloffset_BlitFramebufferEXT 780
-#define _gloffset_BufferParameteriAPPLE 781
-#define _gloffset_FlushMappedBufferRangeAPPLE 782
-#define _gloffset_FramebufferTextureLayerEXT 783
-#define _gloffset_ProvokingVertexEXT 784
-#define _gloffset_GetTexParameterPointervAPPLE 785
-#define _gloffset_TextureRangeAPPLE 786
-#define _gloffset_StencilFuncSeparateATI 787
-#define _gloffset_ProgramEnvParameters4fvEXT 788
-#define _gloffset_ProgramLocalParameters4fvEXT 789
-#define _gloffset_GetQueryObjecti64vEXT 790
-#define _gloffset_GetQueryObjectui64vEXT 791
-#define _gloffset_FIRST_DYNAMIC 792
+#define _gloffset_DrawElementsBaseVertex 574
+#define _gloffset_DrawRangeElementsBaseVertex 575
+#define _gloffset_MultiDrawElementsBaseVertex 576
+#define _gloffset_PolygonOffsetEXT 577
+#define _gloffset_GetPixelTexGenParameterfvSGIS 578
+#define _gloffset_GetPixelTexGenParameterivSGIS 579
+#define _gloffset_PixelTexGenParameterfSGIS 580
+#define _gloffset_PixelTexGenParameterfvSGIS 581
+#define _gloffset_PixelTexGenParameteriSGIS 582
+#define _gloffset_PixelTexGenParameterivSGIS 583
+#define _gloffset_SampleMaskSGIS 584
+#define _gloffset_SamplePatternSGIS 585
+#define _gloffset_ColorPointerEXT 586
+#define _gloffset_EdgeFlagPointerEXT 587
+#define _gloffset_IndexPointerEXT 588
+#define _gloffset_NormalPointerEXT 589
+#define _gloffset_TexCoordPointerEXT 590
+#define _gloffset_VertexPointerEXT 591
+#define _gloffset_PointParameterfEXT 592
+#define _gloffset_PointParameterfvEXT 593
+#define _gloffset_LockArraysEXT 594
+#define _gloffset_UnlockArraysEXT 595
+#define _gloffset_CullParameterdvEXT 596
+#define _gloffset_CullParameterfvEXT 597
+#define _gloffset_SecondaryColor3bEXT 598
+#define _gloffset_SecondaryColor3bvEXT 599
+#define _gloffset_SecondaryColor3dEXT 600
+#define _gloffset_SecondaryColor3dvEXT 601
+#define _gloffset_SecondaryColor3fEXT 602
+#define _gloffset_SecondaryColor3fvEXT 603
+#define _gloffset_SecondaryColor3iEXT 604
+#define _gloffset_SecondaryColor3ivEXT 605
+#define _gloffset_SecondaryColor3sEXT 606
+#define _gloffset_SecondaryColor3svEXT 607
+#define _gloffset_SecondaryColor3ubEXT 608
+#define _gloffset_SecondaryColor3ubvEXT 609
+#define _gloffset_SecondaryColor3uiEXT 610
+#define _gloffset_SecondaryColor3uivEXT 611
+#define _gloffset_SecondaryColor3usEXT 612
+#define _gloffset_SecondaryColor3usvEXT 613
+#define _gloffset_SecondaryColorPointerEXT 614
+#define _gloffset_MultiDrawArraysEXT 615
+#define _gloffset_MultiDrawElementsEXT 616
+#define _gloffset_FogCoordPointerEXT 617
+#define _gloffset_FogCoorddEXT 618
+#define _gloffset_FogCoorddvEXT 619
+#define _gloffset_FogCoordfEXT 620
+#define _gloffset_FogCoordfvEXT 621
+#define _gloffset_PixelTexGenSGIX 622
+#define _gloffset_BlendFuncSeparateEXT 623
+#define _gloffset_FlushVertexArrayRangeNV 624
+#define _gloffset_VertexArrayRangeNV 625
+#define _gloffset_CombinerInputNV 626
+#define _gloffset_CombinerOutputNV 627
+#define _gloffset_CombinerParameterfNV 628
+#define _gloffset_CombinerParameterfvNV 629
+#define _gloffset_CombinerParameteriNV 630
+#define _gloffset_CombinerParameterivNV 631
+#define _gloffset_FinalCombinerInputNV 632
+#define _gloffset_GetCombinerInputParameterfvNV 633
+#define _gloffset_GetCombinerInputParameterivNV 634
+#define _gloffset_GetCombinerOutputParameterfvNV 635
+#define _gloffset_GetCombinerOutputParameterivNV 636
+#define _gloffset_GetFinalCombinerInputParameterfvNV 637
+#define _gloffset_GetFinalCombinerInputParameterivNV 638
+#define _gloffset_ResizeBuffersMESA 639
+#define _gloffset_WindowPos2dMESA 640
+#define _gloffset_WindowPos2dvMESA 641
+#define _gloffset_WindowPos2fMESA 642
+#define _gloffset_WindowPos2fvMESA 643
+#define _gloffset_WindowPos2iMESA 644
+#define _gloffset_WindowPos2ivMESA 645
+#define _gloffset_WindowPos2sMESA 646
+#define _gloffset_WindowPos2svMESA 647
+#define _gloffset_WindowPos3dMESA 648
+#define _gloffset_WindowPos3dvMESA 649
+#define _gloffset_WindowPos3fMESA 650
+#define _gloffset_WindowPos3fvMESA 651
+#define _gloffset_WindowPos3iMESA 652
+#define _gloffset_WindowPos3ivMESA 653
+#define _gloffset_WindowPos3sMESA 654
+#define _gloffset_WindowPos3svMESA 655
+#define _gloffset_WindowPos4dMESA 656
+#define _gloffset_WindowPos4dvMESA 657
+#define _gloffset_WindowPos4fMESA 658
+#define _gloffset_WindowPos4fvMESA 659
+#define _gloffset_WindowPos4iMESA 660
+#define _gloffset_WindowPos4ivMESA 661
+#define _gloffset_WindowPos4sMESA 662
+#define _gloffset_WindowPos4svMESA 663
+#define _gloffset_MultiModeDrawArraysIBM 664
+#define _gloffset_MultiModeDrawElementsIBM 665
+#define _gloffset_DeleteFencesNV 666
+#define _gloffset_FinishFenceNV 667
+#define _gloffset_GenFencesNV 668
+#define _gloffset_GetFenceivNV 669
+#define _gloffset_IsFenceNV 670
+#define _gloffset_SetFenceNV 671
+#define _gloffset_TestFenceNV 672
+#define _gloffset_AreProgramsResidentNV 673
+#define _gloffset_BindProgramNV 674
+#define _gloffset_DeleteProgramsNV 675
+#define _gloffset_ExecuteProgramNV 676
+#define _gloffset_GenProgramsNV 677
+#define _gloffset_GetProgramParameterdvNV 678
+#define _gloffset_GetProgramParameterfvNV 679
+#define _gloffset_GetProgramStringNV 680
+#define _gloffset_GetProgramivNV 681
+#define _gloffset_GetTrackMatrixivNV 682
+#define _gloffset_GetVertexAttribPointervNV 683
+#define _gloffset_GetVertexAttribdvNV 684
+#define _gloffset_GetVertexAttribfvNV 685
+#define _gloffset_GetVertexAttribivNV 686
+#define _gloffset_IsProgramNV 687
+#define _gloffset_LoadProgramNV 688
+#define _gloffset_ProgramParameters4dvNV 689
+#define _gloffset_ProgramParameters4fvNV 690
+#define _gloffset_RequestResidentProgramsNV 691
+#define _gloffset_TrackMatrixNV 692
+#define _gloffset_VertexAttrib1dNV 693
+#define _gloffset_VertexAttrib1dvNV 694
+#define _gloffset_VertexAttrib1fNV 695
+#define _gloffset_VertexAttrib1fvNV 696
+#define _gloffset_VertexAttrib1sNV 697
+#define _gloffset_VertexAttrib1svNV 698
+#define _gloffset_VertexAttrib2dNV 699
+#define _gloffset_VertexAttrib2dvNV 700
+#define _gloffset_VertexAttrib2fNV 701
+#define _gloffset_VertexAttrib2fvNV 702
+#define _gloffset_VertexAttrib2sNV 703
+#define _gloffset_VertexAttrib2svNV 704
+#define _gloffset_VertexAttrib3dNV 705
+#define _gloffset_VertexAttrib3dvNV 706
+#define _gloffset_VertexAttrib3fNV 707
+#define _gloffset_VertexAttrib3fvNV 708
+#define _gloffset_VertexAttrib3sNV 709
+#define _gloffset_VertexAttrib3svNV 710
+#define _gloffset_VertexAttrib4dNV 711
+#define _gloffset_VertexAttrib4dvNV 712
+#define _gloffset_VertexAttrib4fNV 713
+#define _gloffset_VertexAttrib4fvNV 714
+#define _gloffset_VertexAttrib4sNV 715
+#define _gloffset_VertexAttrib4svNV 716
+#define _gloffset_VertexAttrib4ubNV 717
+#define _gloffset_VertexAttrib4ubvNV 718
+#define _gloffset_VertexAttribPointerNV 719
+#define _gloffset_VertexAttribs1dvNV 720
+#define _gloffset_VertexAttribs1fvNV 721
+#define _gloffset_VertexAttribs1svNV 722
+#define _gloffset_VertexAttribs2dvNV 723
+#define _gloffset_VertexAttribs2fvNV 724
+#define _gloffset_VertexAttribs2svNV 725
+#define _gloffset_VertexAttribs3dvNV 726
+#define _gloffset_VertexAttribs3fvNV 727
+#define _gloffset_VertexAttribs3svNV 728
+#define _gloffset_VertexAttribs4dvNV 729
+#define _gloffset_VertexAttribs4fvNV 730
+#define _gloffset_VertexAttribs4svNV 731
+#define _gloffset_VertexAttribs4ubvNV 732
+#define _gloffset_GetTexBumpParameterfvATI 733
+#define _gloffset_GetTexBumpParameterivATI 734
+#define _gloffset_TexBumpParameterfvATI 735
+#define _gloffset_TexBumpParameterivATI 736
+#define _gloffset_AlphaFragmentOp1ATI 737
+#define _gloffset_AlphaFragmentOp2ATI 738
+#define _gloffset_AlphaFragmentOp3ATI 739
+#define _gloffset_BeginFragmentShaderATI 740
+#define _gloffset_BindFragmentShaderATI 741
+#define _gloffset_ColorFragmentOp1ATI 742
+#define _gloffset_ColorFragmentOp2ATI 743
+#define _gloffset_ColorFragmentOp3ATI 744
+#define _gloffset_DeleteFragmentShaderATI 745
+#define _gloffset_EndFragmentShaderATI 746
+#define _gloffset_GenFragmentShadersATI 747
+#define _gloffset_PassTexCoordATI 748
+#define _gloffset_SampleMapATI 749
+#define _gloffset_SetFragmentShaderConstantATI 750
+#define _gloffset_PointParameteriNV 751
+#define _gloffset_PointParameterivNV 752
+#define _gloffset_ActiveStencilFaceEXT 753
+#define _gloffset_BindVertexArrayAPPLE 754
+#define _gloffset_DeleteVertexArraysAPPLE 755
+#define _gloffset_GenVertexArraysAPPLE 756
+#define _gloffset_IsVertexArrayAPPLE 757
+#define _gloffset_GetProgramNamedParameterdvNV 758
+#define _gloffset_GetProgramNamedParameterfvNV 759
+#define _gloffset_ProgramNamedParameter4dNV 760
+#define _gloffset_ProgramNamedParameter4dvNV 761
+#define _gloffset_ProgramNamedParameter4fNV 762
+#define _gloffset_ProgramNamedParameter4fvNV 763
+#define _gloffset_DepthBoundsEXT 764
+#define _gloffset_BlendEquationSeparateEXT 765
+#define _gloffset_BindFramebufferEXT 766
+#define _gloffset_BindRenderbufferEXT 767
+#define _gloffset_CheckFramebufferStatusEXT 768
+#define _gloffset_DeleteFramebuffersEXT 769
+#define _gloffset_DeleteRenderbuffersEXT 770
+#define _gloffset_FramebufferRenderbufferEXT 771
+#define _gloffset_FramebufferTexture1DEXT 772
+#define _gloffset_FramebufferTexture2DEXT 773
+#define _gloffset_FramebufferTexture3DEXT 774
+#define _gloffset_GenFramebuffersEXT 775
+#define _gloffset_GenRenderbuffersEXT 776
+#define _gloffset_GenerateMipmapEXT 777
+#define _gloffset_GetFramebufferAttachmentParameterivEXT 778
+#define _gloffset_GetRenderbufferParameterivEXT 779
+#define _gloffset_IsFramebufferEXT 780
+#define _gloffset_IsRenderbufferEXT 781
+#define _gloffset_RenderbufferStorageEXT 782
+#define _gloffset_BlitFramebufferEXT 783
+#define _gloffset_BufferParameteriAPPLE 784
+#define _gloffset_FlushMappedBufferRangeAPPLE 785
+#define _gloffset_FramebufferTextureLayerEXT 786
+#define _gloffset_ProvokingVertexEXT 787
+#define _gloffset_GetTexParameterPointervAPPLE 788
+#define _gloffset_TextureRangeAPPLE 789
+#define _gloffset_StencilFuncSeparateATI 790
+#define _gloffset_ProgramEnvParameters4fvEXT 791
+#define _gloffset_ProgramLocalParameters4fvEXT 792
+#define _gloffset_GetQueryObjecti64vEXT 793
+#define _gloffset_GetQueryObjectui64vEXT 794
+#define _gloffset_FIRST_DYNAMIC 795
 
 #else
 
@@ -994,6 +997,9 @@
 #define _gloffset_GetSynciv driDispatchRemapTable[GetSynciv_remap_index]
 #define _gloffset_IsSync driDispatchRemapTable[IsSync_remap_index]
 #define _gloffset_WaitSync driDispatchRemapTable[WaitSync_remap_index]
+#define _gloffset_DrawElementsBaseVertex driDispatchRemapTable[DrawElementsBaseVertex_remap_index]
+#define _gloffset_DrawRangeElementsBaseVertex driDispatchRemapTable[DrawRangeElementsBaseVertex_remap_index]
+#define _gloffset_MultiDrawElementsBaseVertex driDispatchRemapTable[MultiDrawElementsBaseVertex_remap_index]
 #define _gloffset_PolygonOffsetEXT driDispatchRemapTable[PolygonOffsetEXT_remap_index]
 #define _gloffset_GetPixelTexGenParameterfvSGIS driDispatchRemapTable[GetPixelTexGenParameterfvSGIS_remap_index]
 #define _gloffset_GetPixelTexGenParameterivSGIS driDispatchRemapTable[GetPixelTexGenParameterivSGIS_remap_index]
diff --git a/src/mesa/glapi/glapitable.h b/src/mesa/glapi/glapitable.h
index 855fcaa7fe..4f9e53b62d 100644
--- a/src/mesa/glapi/glapitable.h
+++ b/src/mesa/glapi/glapitable.h
@@ -614,224 +614,227 @@ struct _glapi_table
    void (GLAPIENTRYP GetSynciv)(GLsync sync, GLenum pname, GLsizei bufSize, GLsizei * length, GLint * values); /* 571 */
    GLboolean (GLAPIENTRYP IsSync)(GLsync sync); /* 572 */
    void (GLAPIENTRYP WaitSync)(GLsync sync, GLbitfield flags, GLuint64 timeout); /* 573 */
-   void (GLAPIENTRYP PolygonOffsetEXT)(GLfloat factor, GLfloat bias); /* 574 */
-   void (GLAPIENTRYP GetPixelTexGenParameterfvSGIS)(GLenum pname, GLfloat * params); /* 575 */
-   void (GLAPIENTRYP GetPixelTexGenParameterivSGIS)(GLenum pname, GLint * params); /* 576 */
-   void (GLAPIENTRYP PixelTexGenParameterfSGIS)(GLenum pname, GLfloat param); /* 577 */
-   void (GLAPIENTRYP PixelTexGenParameterfvSGIS)(GLenum pname, const GLfloat * params); /* 578 */
-   void (GLAPIENTRYP PixelTexGenParameteriSGIS)(GLenum pname, GLint param); /* 579 */
-   void (GLAPIENTRYP PixelTexGenParameterivSGIS)(GLenum pname, const GLint * params); /* 580 */
-   void (GLAPIENTRYP SampleMaskSGIS)(GLclampf value, GLboolean invert); /* 581 */
-   void (GLAPIENTRYP SamplePatternSGIS)(GLenum pattern); /* 582 */
-   void (GLAPIENTRYP ColorPointerEXT)(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 583 */
-   void (GLAPIENTRYP EdgeFlagPointerEXT)(GLsizei stride, GLsizei count, const GLboolean * pointer); /* 584 */
-   void (GLAPIENTRYP IndexPointerEXT)(GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 585 */
-   void (GLAPIENTRYP NormalPointerEXT)(GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 586 */
-   void (GLAPIENTRYP TexCoordPointerEXT)(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 587 */
-   void (GLAPIENTRYP VertexPointerEXT)(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 588 */
-   void (GLAPIENTRYP PointParameterfEXT)(GLenum pname, GLfloat param); /* 589 */
-   void (GLAPIENTRYP PointParameterfvEXT)(GLenum pname, const GLfloat * params); /* 590 */
-   void (GLAPIENTRYP LockArraysEXT)(GLint first, GLsizei count); /* 591 */
-   void (GLAPIENTRYP UnlockArraysEXT)(void); /* 592 */
-   void (GLAPIENTRYP CullParameterdvEXT)(GLenum pname, GLdouble * params); /* 593 */
-   void (GLAPIENTRYP CullParameterfvEXT)(GLenum pname, GLfloat * params); /* 594 */
-   void (GLAPIENTRYP SecondaryColor3bEXT)(GLbyte red, GLbyte green, GLbyte blue); /* 595 */
-   void (GLAPIENTRYP SecondaryColor3bvEXT)(const GLbyte * v); /* 596 */
-   void (GLAPIENTRYP SecondaryColor3dEXT)(GLdouble red, GLdouble green, GLdouble blue); /* 597 */
-   void (GLAPIENTRYP SecondaryColor3dvEXT)(const GLdouble * v); /* 598 */
-   void (GLAPIENTRYP SecondaryColor3fEXT)(GLfloat red, GLfloat green, GLfloat blue); /* 599 */
-   void (GLAPIENTRYP SecondaryColor3fvEXT)(const GLfloat * v); /* 600 */
-   void (GLAPIENTRYP SecondaryColor3iEXT)(GLint red, GLint green, GLint blue); /* 601 */
-   void (GLAPIENTRYP SecondaryColor3ivEXT)(const GLint * v); /* 602 */
-   void (GLAPIENTRYP SecondaryColor3sEXT)(GLshort red, GLshort green, GLshort blue); /* 603 */
-   void (GLAPIENTRYP SecondaryColor3svEXT)(const GLshort * v); /* 604 */
-   void (GLAPIENTRYP SecondaryColor3ubEXT)(GLubyte red, GLubyte green, GLubyte blue); /* 605 */
-   void (GLAPIENTRYP SecondaryColor3ubvEXT)(const GLubyte * v); /* 606 */
-   void (GLAPIENTRYP SecondaryColor3uiEXT)(GLuint red, GLuint green, GLuint blue); /* 607 */
-   void (GLAPIENTRYP SecondaryColor3uivEXT)(const GLuint * v); /* 608 */
-   void (GLAPIENTRYP SecondaryColor3usEXT)(GLushort red, GLushort green, GLushort blue); /* 609 */
-   void (GLAPIENTRYP SecondaryColor3usvEXT)(const GLushort * v); /* 610 */
-   void (GLAPIENTRYP SecondaryColorPointerEXT)(GLint size, GLenum type, GLsizei stride, const GLvoid * pointer); /* 611 */
-   void (GLAPIENTRYP MultiDrawArraysEXT)(GLenum mode, GLint * first, GLsizei * count, GLsizei primcount); /* 612 */
-   void (GLAPIENTRYP MultiDrawElementsEXT)(GLenum mode, const GLsizei * count, GLenum type, const GLvoid ** indices, GLsizei primcount); /* 613 */
-   void (GLAPIENTRYP FogCoordPointerEXT)(GLenum type, GLsizei stride, const GLvoid * pointer); /* 614 */
-   void (GLAPIENTRYP FogCoorddEXT)(GLdouble coord); /* 615 */
-   void (GLAPIENTRYP FogCoorddvEXT)(const GLdouble * coord); /* 616 */
-   void (GLAPIENTRYP FogCoordfEXT)(GLfloat coord); /* 617 */
-   void (GLAPIENTRYP FogCoordfvEXT)(const GLfloat * coord); /* 618 */
-   void (GLAPIENTRYP PixelTexGenSGIX)(GLenum mode); /* 619 */
-   void (GLAPIENTRYP BlendFuncSeparateEXT)(GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha); /* 620 */
-   void (GLAPIENTRYP FlushVertexArrayRangeNV)(void); /* 621 */
-   void (GLAPIENTRYP VertexArrayRangeNV)(GLsizei length, const GLvoid * pointer); /* 622 */
-   void (GLAPIENTRYP CombinerInputNV)(GLenum stage, GLenum portion, GLenum variable, GLenum input, GLenum mapping, GLenum componentUsage); /* 623 */
-   void (GLAPIENTRYP CombinerOutputNV)(GLenum stage, GLenum portion, GLenum abOutput, GLenum cdOutput, GLenum sumOutput, GLenum scale, GLenum bias, GLboolean abDotProduct, GLboolean cdDotProduct, GLboolean muxSum); /* 624 */
-   void (GLAPIENTRYP CombinerParameterfNV)(GLenum pname, GLfloat param); /* 625 */
-   void (GLAPIENTRYP CombinerParameterfvNV)(GLenum pname, const GLfloat * params); /* 626 */
-   void (GLAPIENTRYP CombinerParameteriNV)(GLenum pname, GLint param); /* 627 */
-   void (GLAPIENTRYP CombinerParameterivNV)(GLenum pname, const GLint * params); /* 628 */
-   void (GLAPIENTRYP FinalCombinerInputNV)(GLenum variable, GLenum input, GLenum mapping, GLenum componentUsage); /* 629 */
-   void (GLAPIENTRYP GetCombinerInputParameterfvNV)(GLenum stage, GLenum portion, GLenum variable, GLenum pname, GLfloat * params); /* 630 */
-   void (GLAPIENTRYP GetCombinerInputParameterivNV)(GLenum stage, GLenum portion, GLenum variable, GLenum pname, GLint * params); /* 631 */
-   void (GLAPIENTRYP GetCombinerOutputParameterfvNV)(GLenum stage, GLenum portion, GLenum pname, GLfloat * params); /* 632 */
-   void (GLAPIENTRYP GetCombinerOutputParameterivNV)(GLenum stage, GLenum portion, GLenum pname, GLint * params); /* 633 */
-   void (GLAPIENTRYP GetFinalCombinerInputParameterfvNV)(GLenum variable, GLenum pname, GLfloat * params); /* 634 */
-   void (GLAPIENTRYP GetFinalCombinerInputParameterivNV)(GLenum variable, GLenum pname, GLint * params); /* 635 */
-   void (GLAPIENTRYP ResizeBuffersMESA)(void); /* 636 */
-   void (GLAPIENTRYP WindowPos2dMESA)(GLdouble x, GLdouble y); /* 637 */
-   void (GLAPIENTRYP WindowPos2dvMESA)(const GLdouble * v); /* 638 */
-   void (GLAPIENTRYP WindowPos2fMESA)(GLfloat x, GLfloat y); /* 639 */
-   void (GLAPIENTRYP WindowPos2fvMESA)(const GLfloat * v); /* 640 */
-   void (GLAPIENTRYP WindowPos2iMESA)(GLint x, GLint y); /* 641 */
-   void (GLAPIENTRYP WindowPos2ivMESA)(const GLint * v); /* 642 */
-   void (GLAPIENTRYP WindowPos2sMESA)(GLshort x, GLshort y); /* 643 */
-   void (GLAPIENTRYP WindowPos2svMESA)(const GLshort * v); /* 644 */
-   void (GLAPIENTRYP WindowPos3dMESA)(GLdouble x, GLdouble y, GLdouble z); /* 645 */
-   void (GLAPIENTRYP WindowPos3dvMESA)(const GLdouble * v); /* 646 */
-   void (GLAPIENTRYP WindowPos3fMESA)(GLfloat x, GLfloat y, GLfloat z); /* 647 */
-   void (GLAPIENTRYP WindowPos3fvMESA)(const GLfloat * v); /* 648 */
-   void (GLAPIENTRYP WindowPos3iMESA)(GLint x, GLint y, GLint z); /* 649 */
-   void (GLAPIENTRYP WindowPos3ivMESA)(const GLint * v); /* 650 */
-   void (GLAPIENTRYP WindowPos3sMESA)(GLshort x, GLshort y, GLshort z); /* 651 */
-   void (GLAPIENTRYP WindowPos3svMESA)(const GLshort * v); /* 652 */
-   void (GLAPIENTRYP WindowPos4dMESA)(GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 653 */
-   void (GLAPIENTRYP WindowPos4dvMESA)(const GLdouble * v); /* 654 */
-   void (GLAPIENTRYP WindowPos4fMESA)(GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 655 */
-   void (GLAPIENTRYP WindowPos4fvMESA)(const GLfloat * v); /* 656 */
-   void (GLAPIENTRYP WindowPos4iMESA)(GLint x, GLint y, GLint z, GLint w); /* 657 */
-   void (GLAPIENTRYP WindowPos4ivMESA)(const GLint * v); /* 658 */
-   void (GLAPIENTRYP WindowPos4sMESA)(GLshort x, GLshort y, GLshort z, GLshort w); /* 659 */
-   void (GLAPIENTRYP WindowPos4svMESA)(const GLshort * v); /* 660 */
-   void (GLAPIENTRYP MultiModeDrawArraysIBM)(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride); /* 661 */
-   void (GLAPIENTRYP MultiModeDrawElementsIBM)(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride); /* 662 */
-   void (GLAPIENTRYP DeleteFencesNV)(GLsizei n, const GLuint * fences); /* 663 */
-   void (GLAPIENTRYP FinishFenceNV)(GLuint fence); /* 664 */
-   void (GLAPIENTRYP GenFencesNV)(GLsizei n, GLuint * fences); /* 665 */
-   void (GLAPIENTRYP GetFenceivNV)(GLuint fence, GLenum pname, GLint * params); /* 666 */
-   GLboolean (GLAPIENTRYP IsFenceNV)(GLuint fence); /* 667 */
-   void (GLAPIENTRYP SetFenceNV)(GLuint fence, GLenum condition); /* 668 */
-   GLboolean (GLAPIENTRYP TestFenceNV)(GLuint fence); /* 669 */
-   GLboolean (GLAPIENTRYP AreProgramsResidentNV)(GLsizei n, const GLuint * ids, GLboolean * residences); /* 670 */
-   void (GLAPIENTRYP BindProgramNV)(GLenum target, GLuint program); /* 671 */
-   void (GLAPIENTRYP DeleteProgramsNV)(GLsizei n, const GLuint * programs); /* 672 */
-   void (GLAPIENTRYP ExecuteProgramNV)(GLenum target, GLuint id, const GLfloat * params); /* 673 */
-   void (GLAPIENTRYP GenProgramsNV)(GLsizei n, GLuint * programs); /* 674 */
-   void (GLAPIENTRYP GetProgramParameterdvNV)(GLenum target, GLuint index, GLenum pname, GLdouble * params); /* 675 */
-   void (GLAPIENTRYP GetProgramParameterfvNV)(GLenum target, GLuint index, GLenum pname, GLfloat * params); /* 676 */
-   void (GLAPIENTRYP GetProgramStringNV)(GLuint id, GLenum pname, GLubyte * program); /* 677 */
-   void (GLAPIENTRYP GetProgramivNV)(GLuint id, GLenum pname, GLint * params); /* 678 */
-   void (GLAPIENTRYP GetTrackMatrixivNV)(GLenum target, GLuint address, GLenum pname, GLint * params); /* 679 */
-   void (GLAPIENTRYP GetVertexAttribPointervNV)(GLuint index, GLenum pname, GLvoid ** pointer); /* 680 */
-   void (GLAPIENTRYP GetVertexAttribdvNV)(GLuint index, GLenum pname, GLdouble * params); /* 681 */
-   void (GLAPIENTRYP GetVertexAttribfvNV)(GLuint index, GLenum pname, GLfloat * params); /* 682 */
-   void (GLAPIENTRYP GetVertexAttribivNV)(GLuint index, GLenum pname, GLint * params); /* 683 */
-   GLboolean (GLAPIENTRYP IsProgramNV)(GLuint program); /* 684 */
-   void (GLAPIENTRYP LoadProgramNV)(GLenum target, GLuint id, GLsizei len, const GLubyte * program); /* 685 */
-   void (GLAPIENTRYP ProgramParameters4dvNV)(GLenum target, GLuint index, GLuint num, const GLdouble * params); /* 686 */
-   void (GLAPIENTRYP ProgramParameters4fvNV)(GLenum target, GLuint index, GLuint num, const GLfloat * params); /* 687 */
-   void (GLAPIENTRYP RequestResidentProgramsNV)(GLsizei n, const GLuint * ids); /* 688 */
-   void (GLAPIENTRYP TrackMatrixNV)(GLenum target, GLuint address, GLenum matrix, GLenum transform); /* 689 */
-   void (GLAPIENTRYP VertexAttrib1dNV)(GLuint index, GLdouble x); /* 690 */
-   void (GLAPIENTRYP VertexAttrib1dvNV)(GLuint index, const GLdouble * v); /* 691 */
-   void (GLAPIENTRYP VertexAttrib1fNV)(GLuint index, GLfloat x); /* 692 */
-   void (GLAPIENTRYP VertexAttrib1fvNV)(GLuint index, const GLfloat * v); /* 693 */
-   void (GLAPIENTRYP VertexAttrib1sNV)(GLuint index, GLshort x); /* 694 */
-   void (GLAPIENTRYP VertexAttrib1svNV)(GLuint index, const GLshort * v); /* 695 */
-   void (GLAPIENTRYP VertexAttrib2dNV)(GLuint index, GLdouble x, GLdouble y); /* 696 */
-   void (GLAPIENTRYP VertexAttrib2dvNV)(GLuint index, const GLdouble * v); /* 697 */
-   void (GLAPIENTRYP VertexAttrib2fNV)(GLuint index, GLfloat x, GLfloat y); /* 698 */
-   void (GLAPIENTRYP VertexAttrib2fvNV)(GLuint index, const GLfloat * v); /* 699 */
-   void (GLAPIENTRYP VertexAttrib2sNV)(GLuint index, GLshort x, GLshort y); /* 700 */
-   void (GLAPIENTRYP VertexAttrib2svNV)(GLuint index, const GLshort * v); /* 701 */
-   void (GLAPIENTRYP VertexAttrib3dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z); /* 702 */
-   void (GLAPIENTRYP VertexAttrib3dvNV)(GLuint index, const GLdouble * v); /* 703 */
-   void (GLAPIENTRYP VertexAttrib3fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z); /* 704 */
-   void (GLAPIENTRYP VertexAttrib3fvNV)(GLuint index, const GLfloat * v); /* 705 */
-   void (GLAPIENTRYP VertexAttrib3sNV)(GLuint index, GLshort x, GLshort y, GLshort z); /* 706 */
-   void (GLAPIENTRYP VertexAttrib3svNV)(GLuint index, const GLshort * v); /* 707 */
-   void (GLAPIENTRYP VertexAttrib4dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 708 */
-   void (GLAPIENTRYP VertexAttrib4dvNV)(GLuint index, const GLdouble * v); /* 709 */
-   void (GLAPIENTRYP VertexAttrib4fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 710 */
-   void (GLAPIENTRYP VertexAttrib4fvNV)(GLuint index, const GLfloat * v); /* 711 */
-   void (GLAPIENTRYP VertexAttrib4sNV)(GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); /* 712 */
-   void (GLAPIENTRYP VertexAttrib4svNV)(GLuint index, const GLshort * v); /* 713 */
-   void (GLAPIENTRYP VertexAttrib4ubNV)(GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); /* 714 */
-   void (GLAPIENTRYP VertexAttrib4ubvNV)(GLuint index, const GLubyte * v); /* 715 */
-   void (GLAPIENTRYP VertexAttribPointerNV)(GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid * pointer); /* 716 */
-   void (GLAPIENTRYP VertexAttribs1dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 717 */
-   void (GLAPIENTRYP VertexAttribs1fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 718 */
-   void (GLAPIENTRYP VertexAttribs1svNV)(GLuint index, GLsizei n, const GLshort * v); /* 719 */
-   void (GLAPIENTRYP VertexAttribs2dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 720 */
-   void (GLAPIENTRYP VertexAttribs2fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 721 */
-   void (GLAPIENTRYP VertexAttribs2svNV)(GLuint index, GLsizei n, const GLshort * v); /* 722 */
-   void (GLAPIENTRYP VertexAttribs3dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 723 */
-   void (GLAPIENTRYP VertexAttribs3fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 724 */
-   void (GLAPIENTRYP VertexAttribs3svNV)(GLuint index, GLsizei n, const GLshort * v); /* 725 */
-   void (GLAPIENTRYP VertexAttribs4dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 726 */
-   void (GLAPIENTRYP VertexAttribs4fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 727 */
-   void (GLAPIENTRYP VertexAttribs4svNV)(GLuint index, GLsizei n, const GLshort * v); /* 728 */
-   void (GLAPIENTRYP VertexAttribs4ubvNV)(GLuint index, GLsizei n, const GLubyte * v); /* 729 */
-   void (GLAPIENTRYP GetTexBumpParameterfvATI)(GLenum pname, GLfloat * param); /* 730 */
-   void (GLAPIENTRYP GetTexBumpParameterivATI)(GLenum pname, GLint * param); /* 731 */
-   void (GLAPIENTRYP TexBumpParameterfvATI)(GLenum pname, const GLfloat * param); /* 732 */
-   void (GLAPIENTRYP TexBumpParameterivATI)(GLenum pname, const GLint * param); /* 733 */
-   void (GLAPIENTRYP AlphaFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 734 */
-   void (GLAPIENTRYP AlphaFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 735 */
-   void (GLAPIENTRYP AlphaFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 736 */
-   void (GLAPIENTRYP BeginFragmentShaderATI)(void); /* 737 */
-   void (GLAPIENTRYP BindFragmentShaderATI)(GLuint id); /* 738 */
-   void (GLAPIENTRYP ColorFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 739 */
-   void (GLAPIENTRYP ColorFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 740 */
-   void (GLAPIENTRYP ColorFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 741 */
-   void (GLAPIENTRYP DeleteFragmentShaderATI)(GLuint id); /* 742 */
-   void (GLAPIENTRYP EndFragmentShaderATI)(void); /* 743 */
-   GLuint (GLAPIENTRYP GenFragmentShadersATI)(GLuint range); /* 744 */
-   void (GLAPIENTRYP PassTexCoordATI)(GLuint dst, GLuint coord, GLenum swizzle); /* 745 */
-   void (GLAPIENTRYP SampleMapATI)(GLuint dst, GLuint interp, GLenum swizzle); /* 746 */
-   void (GLAPIENTRYP SetFragmentShaderConstantATI)(GLuint dst, const GLfloat * value); /* 747 */
-   void (GLAPIENTRYP PointParameteriNV)(GLenum pname, GLint param); /* 748 */
-   void (GLAPIENTRYP PointParameterivNV)(GLenum pname, const GLint * params); /* 749 */
-   void (GLAPIENTRYP ActiveStencilFaceEXT)(GLenum face); /* 750 */
-   void (GLAPIENTRYP BindVertexArrayAPPLE)(GLuint array); /* 751 */
-   void (GLAPIENTRYP DeleteVertexArraysAPPLE)(GLsizei n, const GLuint * arrays); /* 752 */
-   void (GLAPIENTRYP GenVertexArraysAPPLE)(GLsizei n, GLuint * arrays); /* 753 */
-   GLboolean (GLAPIENTRYP IsVertexArrayAPPLE)(GLuint array); /* 754 */
-   void (GLAPIENTRYP GetProgramNamedParameterdvNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble * params); /* 755 */
-   void (GLAPIENTRYP GetProgramNamedParameterfvNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat * params); /* 756 */
-   void (GLAPIENTRYP ProgramNamedParameter4dNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 757 */
-   void (GLAPIENTRYP ProgramNamedParameter4dvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLdouble * v); /* 758 */
-   void (GLAPIENTRYP ProgramNamedParameter4fNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 759 */
-   void (GLAPIENTRYP ProgramNamedParameter4fvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLfloat * v); /* 760 */
-   void (GLAPIENTRYP DepthBoundsEXT)(GLclampd zmin, GLclampd zmax); /* 761 */
-   void (GLAPIENTRYP BlendEquationSeparateEXT)(GLenum modeRGB, GLenum modeA); /* 762 */
-   void (GLAPIENTRYP BindFramebufferEXT)(GLenum target, GLuint framebuffer); /* 763 */
-   void (GLAPIENTRYP BindRenderbufferEXT)(GLenum target, GLuint renderbuffer); /* 764 */
-   GLenum (GLAPIENTRYP CheckFramebufferStatusEXT)(GLenum target); /* 765 */
-   void (GLAPIENTRYP DeleteFramebuffersEXT)(GLsizei n, const GLuint * framebuffers); /* 766 */
-   void (GLAPIENTRYP DeleteRenderbuffersEXT)(GLsizei n, const GLuint * renderbuffers); /* 767 */
-   void (GLAPIENTRYP FramebufferRenderbufferEXT)(GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); /* 768 */
-   void (GLAPIENTRYP FramebufferTexture1DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 769 */
-   void (GLAPIENTRYP FramebufferTexture2DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 770 */
-   void (GLAPIENTRYP FramebufferTexture3DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); /* 771 */
-   void (GLAPIENTRYP GenFramebuffersEXT)(GLsizei n, GLuint * framebuffers); /* 772 */
-   void (GLAPIENTRYP GenRenderbuffersEXT)(GLsizei n, GLuint * renderbuffers); /* 773 */
-   void (GLAPIENTRYP GenerateMipmapEXT)(GLenum target); /* 774 */
-   void (GLAPIENTRYP GetFramebufferAttachmentParameterivEXT)(GLenum target, GLenum attachment, GLenum pname, GLint * params); /* 775 */
-   void (GLAPIENTRYP GetRenderbufferParameterivEXT)(GLenum target, GLenum pname, GLint * params); /* 776 */
-   GLboolean (GLAPIENTRYP IsFramebufferEXT)(GLuint framebuffer); /* 777 */
-   GLboolean (GLAPIENTRYP IsRenderbufferEXT)(GLuint renderbuffer); /* 778 */
-   void (GLAPIENTRYP RenderbufferStorageEXT)(GLenum target, GLenum internalformat, GLsizei width, GLsizei height); /* 779 */
-   void (GLAPIENTRYP BlitFramebufferEXT)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); /* 780 */
-   void (GLAPIENTRYP BufferParameteriAPPLE)(GLenum target, GLenum pname, GLint param); /* 781 */
-   void (GLAPIENTRYP FlushMappedBufferRangeAPPLE)(GLenum target, GLintptr offset, GLsizeiptr size); /* 782 */
-   void (GLAPIENTRYP FramebufferTextureLayerEXT)(GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); /* 783 */
-   void (GLAPIENTRYP ProvokingVertexEXT)(GLenum mode); /* 784 */
-   void (GLAPIENTRYP GetTexParameterPointervAPPLE)(GLenum target, GLenum pname, GLvoid ** params); /* 785 */
-   void (GLAPIENTRYP TextureRangeAPPLE)(GLenum target, GLsizei length, GLvoid * pointer); /* 786 */
-   void (GLAPIENTRYP StencilFuncSeparateATI)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); /* 787 */
-   void (GLAPIENTRYP ProgramEnvParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 788 */
-   void (GLAPIENTRYP ProgramLocalParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 789 */
-   void (GLAPIENTRYP GetQueryObjecti64vEXT)(GLuint id, GLenum pname, GLint64EXT * params); /* 790 */
-   void (GLAPIENTRYP GetQueryObjectui64vEXT)(GLuint id, GLenum pname, GLuint64EXT * params); /* 791 */
+   void (GLAPIENTRYP DrawElementsBaseVertex)(GLenum mode, GLsizei count, GLenum type, const GLvoid * indices, GLint basevertex); /* 574 */
+   void (GLAPIENTRYP DrawRangeElementsBaseVertex)(GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid * indices, GLint basevertex); /* 575 */
+   void (GLAPIENTRYP MultiDrawElementsBaseVertex)(GLenum mode, const GLsizei * count, GLenum type, const GLvoid ** indices, GLsizei primcount, const GLint * basevertex); /* 576 */
+   void (GLAPIENTRYP PolygonOffsetEXT)(GLfloat factor, GLfloat bias); /* 577 */
+   void (GLAPIENTRYP GetPixelTexGenParameterfvSGIS)(GLenum pname, GLfloat * params); /* 578 */
+   void (GLAPIENTRYP GetPixelTexGenParameterivSGIS)(GLenum pname, GLint * params); /* 579 */
+   void (GLAPIENTRYP PixelTexGenParameterfSGIS)(GLenum pname, GLfloat param); /* 580 */
+   void (GLAPIENTRYP PixelTexGenParameterfvSGIS)(GLenum pname, const GLfloat * params); /* 581 */
+   void (GLAPIENTRYP PixelTexGenParameteriSGIS)(GLenum pname, GLint param); /* 582 */
+   void (GLAPIENTRYP PixelTexGenParameterivSGIS)(GLenum pname, const GLint * params); /* 583 */
+   void (GLAPIENTRYP SampleMaskSGIS)(GLclampf value, GLboolean invert); /* 584 */
+   void (GLAPIENTRYP SamplePatternSGIS)(GLenum pattern); /* 585 */
+   void (GLAPIENTRYP ColorPointerEXT)(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 586 */
+   void (GLAPIENTRYP EdgeFlagPointerEXT)(GLsizei stride, GLsizei count, const GLboolean * pointer); /* 587 */
+   void (GLAPIENTRYP IndexPointerEXT)(GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 588 */
+   void (GLAPIENTRYP NormalPointerEXT)(GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 589 */
+   void (GLAPIENTRYP TexCoordPointerEXT)(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 590 */
+   void (GLAPIENTRYP VertexPointerEXT)(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer); /* 591 */
+   void (GLAPIENTRYP PointParameterfEXT)(GLenum pname, GLfloat param); /* 592 */
+   void (GLAPIENTRYP PointParameterfvEXT)(GLenum pname, const GLfloat * params); /* 593 */
+   void (GLAPIENTRYP LockArraysEXT)(GLint first, GLsizei count); /* 594 */
+   void (GLAPIENTRYP UnlockArraysEXT)(void); /* 595 */
+   void (GLAPIENTRYP CullParameterdvEXT)(GLenum pname, GLdouble * params); /* 596 */
+   void (GLAPIENTRYP CullParameterfvEXT)(GLenum pname, GLfloat * params); /* 597 */
+   void (GLAPIENTRYP SecondaryColor3bEXT)(GLbyte red, GLbyte green, GLbyte blue); /* 598 */
+   void (GLAPIENTRYP SecondaryColor3bvEXT)(const GLbyte * v); /* 599 */
+   void (GLAPIENTRYP SecondaryColor3dEXT)(GLdouble red, GLdouble green, GLdouble blue); /* 600 */
+   void (GLAPIENTRYP SecondaryColor3dvEXT)(const GLdouble * v); /* 601 */
+   void (GLAPIENTRYP SecondaryColor3fEXT)(GLfloat red, GLfloat green, GLfloat blue); /* 602 */
+   void (GLAPIENTRYP SecondaryColor3fvEXT)(const GLfloat * v); /* 603 */
+   void (GLAPIENTRYP SecondaryColor3iEXT)(GLint red, GLint green, GLint blue); /* 604 */
+   void (GLAPIENTRYP SecondaryColor3ivEXT)(const GLint * v); /* 605 */
+   void (GLAPIENTRYP SecondaryColor3sEXT)(GLshort red, GLshort green, GLshort blue); /* 606 */
+   void (GLAPIENTRYP SecondaryColor3svEXT)(const GLshort * v); /* 607 */
+   void (GLAPIENTRYP SecondaryColor3ubEXT)(GLubyte red, GLubyte green, GLubyte blue); /* 608 */
+   void (GLAPIENTRYP SecondaryColor3ubvEXT)(const GLubyte * v); /* 609 */
+   void (GLAPIENTRYP SecondaryColor3uiEXT)(GLuint red, GLuint green, GLuint blue); /* 610 */
+   void (GLAPIENTRYP SecondaryColor3uivEXT)(const GLuint * v); /* 611 */
+   void (GLAPIENTRYP SecondaryColor3usEXT)(GLushort red, GLushort green, GLushort blue); /* 612 */
+   void (GLAPIENTRYP SecondaryColor3usvEXT)(const GLushort * v); /* 613 */
+   void (GLAPIENTRYP SecondaryColorPointerEXT)(GLint size, GLenum type, GLsizei stride, const GLvoid * pointer); /* 614 */
+   void (GLAPIENTRYP MultiDrawArraysEXT)(GLenum mode, GLint * first, GLsizei * count, GLsizei primcount); /* 615 */
+   void (GLAPIENTRYP MultiDrawElementsEXT)(GLenum mode, const GLsizei * count, GLenum type, const GLvoid ** indices, GLsizei primcount); /* 616 */
+   void (GLAPIENTRYP FogCoordPointerEXT)(GLenum type, GLsizei stride, const GLvoid * pointer); /* 617 */
+   void (GLAPIENTRYP FogCoorddEXT)(GLdouble coord); /* 618 */
+   void (GLAPIENTRYP FogCoorddvEXT)(const GLdouble * coord); /* 619 */
+   void (GLAPIENTRYP FogCoordfEXT)(GLfloat coord); /* 620 */
+   void (GLAPIENTRYP FogCoordfvEXT)(const GLfloat * coord); /* 621 */
+   void (GLAPIENTRYP PixelTexGenSGIX)(GLenum mode); /* 622 */
+   void (GLAPIENTRYP BlendFuncSeparateEXT)(GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha); /* 623 */
+   void (GLAPIENTRYP FlushVertexArrayRangeNV)(void); /* 624 */
+   void (GLAPIENTRYP VertexArrayRangeNV)(GLsizei length, const GLvoid * pointer); /* 625 */
+   void (GLAPIENTRYP CombinerInputNV)(GLenum stage, GLenum portion, GLenum variable, GLenum input, GLenum mapping, GLenum componentUsage); /* 626 */
+   void (GLAPIENTRYP CombinerOutputNV)(GLenum stage, GLenum portion, GLenum abOutput, GLenum cdOutput, GLenum sumOutput, GLenum scale, GLenum bias, GLboolean abDotProduct, GLboolean cdDotProduct, GLboolean muxSum); /* 627 */
+   void (GLAPIENTRYP CombinerParameterfNV)(GLenum pname, GLfloat param); /* 628 */
+   void (GLAPIENTRYP CombinerParameterfvNV)(GLenum pname, const GLfloat * params); /* 629 */
+   void (GLAPIENTRYP CombinerParameteriNV)(GLenum pname, GLint param); /* 630 */
+   void (GLAPIENTRYP CombinerParameterivNV)(GLenum pname, const GLint * params); /* 631 */
+   void (GLAPIENTRYP FinalCombinerInputNV)(GLenum variable, GLenum input, GLenum mapping, GLenum componentUsage); /* 632 */
+   void (GLAPIENTRYP GetCombinerInputParameterfvNV)(GLenum stage, GLenum portion, GLenum variable, GLenum pname, GLfloat * params); /* 633 */
+   void (GLAPIENTRYP GetCombinerInputParameterivNV)(GLenum stage, GLenum portion, GLenum variable, GLenum pname, GLint * params); /* 634 */
+   void (GLAPIENTRYP GetCombinerOutputParameterfvNV)(GLenum stage, GLenum portion, GLenum pname, GLfloat * params); /* 635 */
+   void (GLAPIENTRYP GetCombinerOutputParameterivNV)(GLenum stage, GLenum portion, GLenum pname, GLint * params); /* 636 */
+   void (GLAPIENTRYP GetFinalCombinerInputParameterfvNV)(GLenum variable, GLenum pname, GLfloat * params); /* 637 */
+   void (GLAPIENTRYP GetFinalCombinerInputParameterivNV)(GLenum variable, GLenum pname, GLint * params); /* 638 */
+   void (GLAPIENTRYP ResizeBuffersMESA)(void); /* 639 */
+   void (GLAPIENTRYP WindowPos2dMESA)(GLdouble x, GLdouble y); /* 640 */
+   void (GLAPIENTRYP WindowPos2dvMESA)(const GLdouble * v); /* 641 */
+   void (GLAPIENTRYP WindowPos2fMESA)(GLfloat x, GLfloat y); /* 642 */
+   void (GLAPIENTRYP WindowPos2fvMESA)(const GLfloat * v); /* 643 */
+   void (GLAPIENTRYP WindowPos2iMESA)(GLint x, GLint y); /* 644 */
+   void (GLAPIENTRYP WindowPos2ivMESA)(const GLint * v); /* 645 */
+   void (GLAPIENTRYP WindowPos2sMESA)(GLshort x, GLshort y); /* 646 */
+   void (GLAPIENTRYP WindowPos2svMESA)(const GLshort * v); /* 647 */
+   void (GLAPIENTRYP WindowPos3dMESA)(GLdouble x, GLdouble y, GLdouble z); /* 648 */
+   void (GLAPIENTRYP WindowPos3dvMESA)(const GLdouble * v); /* 649 */
+   void (GLAPIENTRYP WindowPos3fMESA)(GLfloat x, GLfloat y, GLfloat z); /* 650 */
+   void (GLAPIENTRYP WindowPos3fvMESA)(const GLfloat * v); /* 651 */
+   void (GLAPIENTRYP WindowPos3iMESA)(GLint x, GLint y, GLint z); /* 652 */
+   void (GLAPIENTRYP WindowPos3ivMESA)(const GLint * v); /* 653 */
+   void (GLAPIENTRYP WindowPos3sMESA)(GLshort x, GLshort y, GLshort z); /* 654 */
+   void (GLAPIENTRYP WindowPos3svMESA)(const GLshort * v); /* 655 */
+   void (GLAPIENTRYP WindowPos4dMESA)(GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 656 */
+   void (GLAPIENTRYP WindowPos4dvMESA)(const GLdouble * v); /* 657 */
+   void (GLAPIENTRYP WindowPos4fMESA)(GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 658 */
+   void (GLAPIENTRYP WindowPos4fvMESA)(const GLfloat * v); /* 659 */
+   void (GLAPIENTRYP WindowPos4iMESA)(GLint x, GLint y, GLint z, GLint w); /* 660 */
+   void (GLAPIENTRYP WindowPos4ivMESA)(const GLint * v); /* 661 */
+   void (GLAPIENTRYP WindowPos4sMESA)(GLshort x, GLshort y, GLshort z, GLshort w); /* 662 */
+   void (GLAPIENTRYP WindowPos4svMESA)(const GLshort * v); /* 663 */
+   void (GLAPIENTRYP MultiModeDrawArraysIBM)(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride); /* 664 */
+   void (GLAPIENTRYP MultiModeDrawElementsIBM)(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride); /* 665 */
+   void (GLAPIENTRYP DeleteFencesNV)(GLsizei n, const GLuint * fences); /* 666 */
+   void (GLAPIENTRYP FinishFenceNV)(GLuint fence); /* 667 */
+   void (GLAPIENTRYP GenFencesNV)(GLsizei n, GLuint * fences); /* 668 */
+   void (GLAPIENTRYP GetFenceivNV)(GLuint fence, GLenum pname, GLint * params); /* 669 */
+   GLboolean (GLAPIENTRYP IsFenceNV)(GLuint fence); /* 670 */
+   void (GLAPIENTRYP SetFenceNV)(GLuint fence, GLenum condition); /* 671 */
+   GLboolean (GLAPIENTRYP TestFenceNV)(GLuint fence); /* 672 */
+   GLboolean (GLAPIENTRYP AreProgramsResidentNV)(GLsizei n, const GLuint * ids, GLboolean * residences); /* 673 */
+   void (GLAPIENTRYP BindProgramNV)(GLenum target, GLuint program); /* 674 */
+   void (GLAPIENTRYP DeleteProgramsNV)(GLsizei n, const GLuint * programs); /* 675 */
+   void (GLAPIENTRYP ExecuteProgramNV)(GLenum target, GLuint id, const GLfloat * params); /* 676 */
+   void (GLAPIENTRYP GenProgramsNV)(GLsizei n, GLuint * programs); /* 677 */
+   void (GLAPIENTRYP GetProgramParameterdvNV)(GLenum target, GLuint index, GLenum pname, GLdouble * params); /* 678 */
+   void (GLAPIENTRYP GetProgramParameterfvNV)(GLenum target, GLuint index, GLenum pname, GLfloat * params); /* 679 */
+   void (GLAPIENTRYP GetProgramStringNV)(GLuint id, GLenum pname, GLubyte * program); /* 680 */
+   void (GLAPIENTRYP GetProgramivNV)(GLuint id, GLenum pname, GLint * params); /* 681 */
+   void (GLAPIENTRYP GetTrackMatrixivNV)(GLenum target, GLuint address, GLenum pname, GLint * params); /* 682 */
+   void (GLAPIENTRYP GetVertexAttribPointervNV)(GLuint index, GLenum pname, GLvoid ** pointer); /* 683 */
+   void (GLAPIENTRYP GetVertexAttribdvNV)(GLuint index, GLenum pname, GLdouble * params); /* 684 */
+   void (GLAPIENTRYP GetVertexAttribfvNV)(GLuint index, GLenum pname, GLfloat * params); /* 685 */
+   void (GLAPIENTRYP GetVertexAttribivNV)(GLuint index, GLenum pname, GLint * params); /* 686 */
+   GLboolean (GLAPIENTRYP IsProgramNV)(GLuint program); /* 687 */
+   void (GLAPIENTRYP LoadProgramNV)(GLenum target, GLuint id, GLsizei len, const GLubyte * program); /* 688 */
+   void (GLAPIENTRYP ProgramParameters4dvNV)(GLenum target, GLuint index, GLuint num, const GLdouble * params); /* 689 */
+   void (GLAPIENTRYP ProgramParameters4fvNV)(GLenum target, GLuint index, GLuint num, const GLfloat * params); /* 690 */
+   void (GLAPIENTRYP RequestResidentProgramsNV)(GLsizei n, const GLuint * ids); /* 691 */
+   void (GLAPIENTRYP TrackMatrixNV)(GLenum target, GLuint address, GLenum matrix, GLenum transform); /* 692 */
+   void (GLAPIENTRYP VertexAttrib1dNV)(GLuint index, GLdouble x); /* 693 */
+   void (GLAPIENTRYP VertexAttrib1dvNV)(GLuint index, const GLdouble * v); /* 694 */
+   void (GLAPIENTRYP VertexAttrib1fNV)(GLuint index, GLfloat x); /* 695 */
+   void (GLAPIENTRYP VertexAttrib1fvNV)(GLuint index, const GLfloat * v); /* 696 */
+   void (GLAPIENTRYP VertexAttrib1sNV)(GLuint index, GLshort x); /* 697 */
+   void (GLAPIENTRYP VertexAttrib1svNV)(GLuint index, const GLshort * v); /* 698 */
+   void (GLAPIENTRYP VertexAttrib2dNV)(GLuint index, GLdouble x, GLdouble y); /* 699 */
+   void (GLAPIENTRYP VertexAttrib2dvNV)(GLuint index, const GLdouble * v); /* 700 */
+   void (GLAPIENTRYP VertexAttrib2fNV)(GLuint index, GLfloat x, GLfloat y); /* 701 */
+   void (GLAPIENTRYP VertexAttrib2fvNV)(GLuint index, const GLfloat * v); /* 702 */
+   void (GLAPIENTRYP VertexAttrib2sNV)(GLuint index, GLshort x, GLshort y); /* 703 */
+   void (GLAPIENTRYP VertexAttrib2svNV)(GLuint index, const GLshort * v); /* 704 */
+   void (GLAPIENTRYP VertexAttrib3dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z); /* 705 */
+   void (GLAPIENTRYP VertexAttrib3dvNV)(GLuint index, const GLdouble * v); /* 706 */
+   void (GLAPIENTRYP VertexAttrib3fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z); /* 707 */
+   void (GLAPIENTRYP VertexAttrib3fvNV)(GLuint index, const GLfloat * v); /* 708 */
+   void (GLAPIENTRYP VertexAttrib3sNV)(GLuint index, GLshort x, GLshort y, GLshort z); /* 709 */
+   void (GLAPIENTRYP VertexAttrib3svNV)(GLuint index, const GLshort * v); /* 710 */
+   void (GLAPIENTRYP VertexAttrib4dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 711 */
+   void (GLAPIENTRYP VertexAttrib4dvNV)(GLuint index, const GLdouble * v); /* 712 */
+   void (GLAPIENTRYP VertexAttrib4fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 713 */
+   void (GLAPIENTRYP VertexAttrib4fvNV)(GLuint index, const GLfloat * v); /* 714 */
+   void (GLAPIENTRYP VertexAttrib4sNV)(GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); /* 715 */
+   void (GLAPIENTRYP VertexAttrib4svNV)(GLuint index, const GLshort * v); /* 716 */
+   void (GLAPIENTRYP VertexAttrib4ubNV)(GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); /* 717 */
+   void (GLAPIENTRYP VertexAttrib4ubvNV)(GLuint index, const GLubyte * v); /* 718 */
+   void (GLAPIENTRYP VertexAttribPointerNV)(GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid * pointer); /* 719 */
+   void (GLAPIENTRYP VertexAttribs1dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 720 */
+   void (GLAPIENTRYP VertexAttribs1fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 721 */
+   void (GLAPIENTRYP VertexAttribs1svNV)(GLuint index, GLsizei n, const GLshort * v); /* 722 */
+   void (GLAPIENTRYP VertexAttribs2dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 723 */
+   void (GLAPIENTRYP VertexAttribs2fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 724 */
+   void (GLAPIENTRYP VertexAttribs2svNV)(GLuint index, GLsizei n, const GLshort * v); /* 725 */
+   void (GLAPIENTRYP VertexAttribs3dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 726 */
+   void (GLAPIENTRYP VertexAttribs3fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 727 */
+   void (GLAPIENTRYP VertexAttribs3svNV)(GLuint index, GLsizei n, const GLshort * v); /* 728 */
+   void (GLAPIENTRYP VertexAttribs4dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 729 */
+   void (GLAPIENTRYP VertexAttribs4fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 730 */
+   void (GLAPIENTRYP VertexAttribs4svNV)(GLuint index, GLsizei n, const GLshort * v); /* 731 */
+   void (GLAPIENTRYP VertexAttribs4ubvNV)(GLuint index, GLsizei n, const GLubyte * v); /* 732 */
+   void (GLAPIENTRYP GetTexBumpParameterfvATI)(GLenum pname, GLfloat * param); /* 733 */
+   void (GLAPIENTRYP GetTexBumpParameterivATI)(GLenum pname, GLint * param); /* 734 */
+   void (GLAPIENTRYP TexBumpParameterfvATI)(GLenum pname, const GLfloat * param); /* 735 */
+   void (GLAPIENTRYP TexBumpParameterivATI)(GLenum pname, const GLint * param); /* 736 */
+   void (GLAPIENTRYP AlphaFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 737 */
+   void (GLAPIENTRYP AlphaFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 738 */
+   void (GLAPIENTRYP AlphaFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 739 */
+   void (GLAPIENTRYP BeginFragmentShaderATI)(void); /* 740 */
+   void (GLAPIENTRYP BindFragmentShaderATI)(GLuint id); /* 741 */
+   void (GLAPIENTRYP ColorFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 742 */
+   void (GLAPIENTRYP ColorFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 743 */
+   void (GLAPIENTRYP ColorFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 744 */
+   void (GLAPIENTRYP DeleteFragmentShaderATI)(GLuint id); /* 745 */
+   void (GLAPIENTRYP EndFragmentShaderATI)(void); /* 746 */
+   GLuint (GLAPIENTRYP GenFragmentShadersATI)(GLuint range); /* 747 */
+   void (GLAPIENTRYP PassTexCoordATI)(GLuint dst, GLuint coord, GLenum swizzle); /* 748 */
+   void (GLAPIENTRYP SampleMapATI)(GLuint dst, GLuint interp, GLenum swizzle); /* 749 */
+   void (GLAPIENTRYP SetFragmentShaderConstantATI)(GLuint dst, const GLfloat * value); /* 750 */
+   void (GLAPIENTRYP PointParameteriNV)(GLenum pname, GLint param); /* 751 */
+   void (GLAPIENTRYP PointParameterivNV)(GLenum pname, const GLint * params); /* 752 */
+   void (GLAPIENTRYP ActiveStencilFaceEXT)(GLenum face); /* 753 */
+   void (GLAPIENTRYP BindVertexArrayAPPLE)(GLuint array); /* 754 */
+   void (GLAPIENTRYP DeleteVertexArraysAPPLE)(GLsizei n, const GLuint * arrays); /* 755 */
+   void (GLAPIENTRYP GenVertexArraysAPPLE)(GLsizei n, GLuint * arrays); /* 756 */
+   GLboolean (GLAPIENTRYP IsVertexArrayAPPLE)(GLuint array); /* 757 */
+   void (GLAPIENTRYP GetProgramNamedParameterdvNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble * params); /* 758 */
+   void (GLAPIENTRYP GetProgramNamedParameterfvNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat * params); /* 759 */
+   void (GLAPIENTRYP ProgramNamedParameter4dNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 760 */
+   void (GLAPIENTRYP ProgramNamedParameter4dvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLdouble * v); /* 761 */
+   void (GLAPIENTRYP ProgramNamedParameter4fNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 762 */
+   void (GLAPIENTRYP ProgramNamedParameter4fvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLfloat * v); /* 763 */
+   void (GLAPIENTRYP DepthBoundsEXT)(GLclampd zmin, GLclampd zmax); /* 764 */
+   void (GLAPIENTRYP BlendEquationSeparateEXT)(GLenum modeRGB, GLenum modeA); /* 765 */
+   void (GLAPIENTRYP BindFramebufferEXT)(GLenum target, GLuint framebuffer); /* 766 */
+   void (GLAPIENTRYP BindRenderbufferEXT)(GLenum target, GLuint renderbuffer); /* 767 */
+   GLenum (GLAPIENTRYP CheckFramebufferStatusEXT)(GLenum target); /* 768 */
+   void (GLAPIENTRYP DeleteFramebuffersEXT)(GLsizei n, const GLuint * framebuffers); /* 769 */
+   void (GLAPIENTRYP DeleteRenderbuffersEXT)(GLsizei n, const GLuint * renderbuffers); /* 770 */
+   void (GLAPIENTRYP FramebufferRenderbufferEXT)(GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); /* 771 */
+   void (GLAPIENTRYP FramebufferTexture1DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 772 */
+   void (GLAPIENTRYP FramebufferTexture2DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 773 */
+   void (GLAPIENTRYP FramebufferTexture3DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); /* 774 */
+   void (GLAPIENTRYP GenFramebuffersEXT)(GLsizei n, GLuint * framebuffers); /* 775 */
+   void (GLAPIENTRYP GenRenderbuffersEXT)(GLsizei n, GLuint * renderbuffers); /* 776 */
+   void (GLAPIENTRYP GenerateMipmapEXT)(GLenum target); /* 777 */
+   void (GLAPIENTRYP GetFramebufferAttachmentParameterivEXT)(GLenum target, GLenum attachment, GLenum pname, GLint * params); /* 778 */
+   void (GLAPIENTRYP GetRenderbufferParameterivEXT)(GLenum target, GLenum pname, GLint * params); /* 779 */
+   GLboolean (GLAPIENTRYP IsFramebufferEXT)(GLuint framebuffer); /* 780 */
+   GLboolean (GLAPIENTRYP IsRenderbufferEXT)(GLuint renderbuffer); /* 781 */
+   void (GLAPIENTRYP RenderbufferStorageEXT)(GLenum target, GLenum internalformat, GLsizei width, GLsizei height); /* 782 */
+   void (GLAPIENTRYP BlitFramebufferEXT)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); /* 783 */
+   void (GLAPIENTRYP BufferParameteriAPPLE)(GLenum target, GLenum pname, GLint param); /* 784 */
+   void (GLAPIENTRYP FlushMappedBufferRangeAPPLE)(GLenum target, GLintptr offset, GLsizeiptr size); /* 785 */
+   void (GLAPIENTRYP FramebufferTextureLayerEXT)(GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); /* 786 */
+   void (GLAPIENTRYP ProvokingVertexEXT)(GLenum mode); /* 787 */
+   void (GLAPIENTRYP GetTexParameterPointervAPPLE)(GLenum target, GLenum pname, GLvoid ** params); /* 788 */
+   void (GLAPIENTRYP TextureRangeAPPLE)(GLenum target, GLsizei length, GLvoid * pointer); /* 789 */
+   void (GLAPIENTRYP StencilFuncSeparateATI)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); /* 790 */
+   void (GLAPIENTRYP ProgramEnvParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 791 */
+   void (GLAPIENTRYP ProgramLocalParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 792 */
+   void (GLAPIENTRYP GetQueryObjecti64vEXT)(GLuint id, GLenum pname, GLint64EXT * params); /* 793 */
+   void (GLAPIENTRYP GetQueryObjectui64vEXT)(GLuint id, GLenum pname, GLuint64EXT * params); /* 794 */
 };
 
 #endif /* !defined( _GLAPI_TABLE_H_ ) */
diff --git a/src/mesa/glapi/glapitemp.h b/src/mesa/glapi/glapitemp.h
index e08f9548f5..d9a3690f2a 100644
--- a/src/mesa/glapi/glapitemp.h
+++ b/src/mesa/glapi/glapitemp.h
@@ -4071,63 +4071,78 @@ KEYWORD1 void KEYWORD2 NAME(WaitSync)(GLsync sync, GLbitfield flags, GLuint64 ti
    DISPATCH(WaitSync, (sync, flags, timeout), (F, "glWaitSync(%d, %d, %d);\n", sync, flags, timeout));
 }
 
+KEYWORD1 void KEYWORD2 NAME(DrawElementsBaseVertex)(GLenum mode, GLsizei count, GLenum type, const GLvoid * indices, GLint basevertex)
+{
+   DISPATCH(DrawElementsBaseVertex, (mode, count, type, indices, basevertex), (F, "glDrawElementsBaseVertex(0x%x, %d, 0x%x, %p, %d);\n", mode, count, type, (const void *) indices, basevertex));
+}
+
+KEYWORD1 void KEYWORD2 NAME(DrawRangeElementsBaseVertex)(GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid * indices, GLint basevertex)
+{
+   DISPATCH(DrawRangeElementsBaseVertex, (mode, start, end, count, type, indices, basevertex), (F, "glDrawRangeElementsBaseVertex(0x%x, %d, %d, %d, 0x%x, %p, %d);\n", mode, start, end, count, type, (const void *) indices, basevertex));
+}
+
+KEYWORD1 void KEYWORD2 NAME(MultiDrawElementsBaseVertex)(GLenum mode, const GLsizei * count, GLenum type, const GLvoid ** indices, GLsizei primcount, const GLint * basevertex)
+{
+   DISPATCH(MultiDrawElementsBaseVertex, (mode, count, type, indices, primcount, basevertex), (F, "glMultiDrawElementsBaseVertex(0x%x, %p, 0x%x, %p, %d, %p);\n", mode, (const void *) count, type, (const void *) indices, primcount, (const void *) basevertex));
+}
+
 KEYWORD1 void KEYWORD2 NAME(PolygonOffsetEXT)(GLfloat factor, GLfloat bias)
 {
    DISPATCH(PolygonOffsetEXT, (factor, bias), (F, "glPolygonOffsetEXT(%f, %f);\n", factor, bias));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_575)(GLenum pname, GLfloat * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_578)(GLenum pname, GLfloat * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_575)(GLenum pname, GLfloat * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_578)(GLenum pname, GLfloat * params)
 {
    DISPATCH(GetPixelTexGenParameterfvSGIS, (pname, params), (F, "glGetPixelTexGenParameterfvSGIS(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_576)(GLenum pname, GLint * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_579)(GLenum pname, GLint * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_576)(GLenum pname, GLint * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_579)(GLenum pname, GLint * params)
 {
    DISPATCH(GetPixelTexGenParameterivSGIS, (pname, params), (F, "glGetPixelTexGenParameterivSGIS(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_577)(GLenum pname, GLfloat param);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_580)(GLenum pname, GLfloat param);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_577)(GLenum pname, GLfloat param)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_580)(GLenum pname, GLfloat param)
 {
    DISPATCH(PixelTexGenParameterfSGIS, (pname, param), (F, "glPixelTexGenParameterfSGIS(0x%x, %f);\n", pname, param));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_578)(GLenum pname, const GLfloat * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_581)(GLenum pname, const GLfloat * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_578)(GLenum pname, const GLfloat * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_581)(GLenum pname, const GLfloat * params)
 {
    DISPATCH(PixelTexGenParameterfvSGIS, (pname, params), (F, "glPixelTexGenParameterfvSGIS(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_579)(GLenum pname, GLint param);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_582)(GLenum pname, GLint param);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_579)(GLenum pname, GLint param)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_582)(GLenum pname, GLint param)
 {
    DISPATCH(PixelTexGenParameteriSGIS, (pname, param), (F, "glPixelTexGenParameteriSGIS(0x%x, %d);\n", pname, param));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_580)(GLenum pname, const GLint * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_583)(GLenum pname, const GLint * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_580)(GLenum pname, const GLint * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_583)(GLenum pname, const GLint * params)
 {
    DISPATCH(PixelTexGenParameterivSGIS, (pname, params), (F, "glPixelTexGenParameterivSGIS(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_581)(GLclampf value, GLboolean invert);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_584)(GLclampf value, GLboolean invert);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_581)(GLclampf value, GLboolean invert)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_584)(GLclampf value, GLboolean invert)
 {
    DISPATCH(SampleMaskSGIS, (value, invert), (F, "glSampleMaskSGIS(%f, %d);\n", value, invert));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_582)(GLenum pattern);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_585)(GLenum pattern);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_582)(GLenum pattern)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_585)(GLenum pattern)
 {
    DISPATCH(SamplePatternSGIS, (pattern), (F, "glSamplePatternSGIS(0x%x);\n", pattern));
 }
@@ -4177,9 +4192,9 @@ KEYWORD1 void KEYWORD2 NAME(PointParameterfEXT)(GLenum pname, GLfloat param)
    DISPATCH(PointParameterfEXT, (pname, param), (F, "glPointParameterfEXT(0x%x, %f);\n", pname, param));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_589)(GLenum pname, GLfloat param);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_592)(GLenum pname, GLfloat param);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_589)(GLenum pname, GLfloat param)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_592)(GLenum pname, GLfloat param)
 {
    DISPATCH(PointParameterfEXT, (pname, param), (F, "glPointParameterfSGIS(0x%x, %f);\n", pname, param));
 }
@@ -4199,9 +4214,9 @@ KEYWORD1 void KEYWORD2 NAME(PointParameterfvEXT)(GLenum pname, const GLfloat * p
    DISPATCH(PointParameterfvEXT, (pname, params), (F, "glPointParameterfvEXT(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_590)(GLenum pname, const GLfloat * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_593)(GLenum pname, const GLfloat * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_590)(GLenum pname, const GLfloat * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_593)(GLenum pname, const GLfloat * params)
 {
    DISPATCH(PointParameterfvEXT, (pname, params), (F, "glPointParameterfvSGIS(0x%x, %p);\n", pname, (const void *) params));
 }
@@ -4216,16 +4231,16 @@ KEYWORD1 void KEYWORD2 NAME(UnlockArraysEXT)(void)
    DISPATCH(UnlockArraysEXT, (), (F, "glUnlockArraysEXT();\n"));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_593)(GLenum pname, GLdouble * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_596)(GLenum pname, GLdouble * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_593)(GLenum pname, GLdouble * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_596)(GLenum pname, GLdouble * params)
 {
    DISPATCH(CullParameterdvEXT, (pname, params), (F, "glCullParameterdvEXT(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_594)(GLenum pname, GLfloat * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_597)(GLenum pname, GLfloat * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_594)(GLenum pname, GLfloat * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_597)(GLenum pname, GLfloat * params)
 {
    DISPATCH(CullParameterfvEXT, (pname, params), (F, "glCullParameterfvEXT(0x%x, %p);\n", pname, (const void *) params));
 }
@@ -4470,9 +4485,9 @@ KEYWORD1 void KEYWORD2 NAME(FogCoordfvEXT)(const GLfloat * coord)
    DISPATCH(FogCoordfvEXT, (coord), (F, "glFogCoordfvEXT(%p);\n", (const void *) coord));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_619)(GLenum mode);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_622)(GLenum mode);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_619)(GLenum mode)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_622)(GLenum mode)
 {
    DISPATCH(PixelTexGenSGIX, (mode), (F, "glPixelTexGenSGIX(0x%x);\n", mode));
 }
@@ -4487,9 +4502,9 @@ KEYWORD1 void KEYWORD2 NAME(BlendFuncSeparateEXT)(GLenum sfactorRGB, GLenum dfac
    DISPATCH(BlendFuncSeparateEXT, (sfactorRGB, dfactorRGB, sfactorAlpha, dfactorAlpha), (F, "glBlendFuncSeparateEXT(0x%x, 0x%x, 0x%x, 0x%x);\n", sfactorRGB, dfactorRGB, sfactorAlpha, dfactorAlpha));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_620)(GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_623)(GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_620)(GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_623)(GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha)
 {
    DISPATCH(BlendFuncSeparateEXT, (sfactorRGB, dfactorRGB, sfactorAlpha, dfactorAlpha), (F, "glBlendFuncSeparateINGR(0x%x, 0x%x, 0x%x, 0x%x);\n", sfactorRGB, dfactorRGB, sfactorAlpha, dfactorAlpha));
 }
@@ -4854,65 +4869,65 @@ KEYWORD1 void KEYWORD2 NAME(WindowPos4svMESA)(const GLshort * v)
    DISPATCH(WindowPos4svMESA, (v), (F, "glWindowPos4svMESA(%p);\n", (const void *) v));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_661)(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_664)(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_661)(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_664)(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride)
 {
    DISPATCH(MultiModeDrawArraysIBM, (mode, first, count, primcount, modestride), (F, "glMultiModeDrawArraysIBM(%p, %p, %p, %d, %d);\n", (const void *) mode, (const void *) first, (const void *) count, primcount, modestride));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_662)(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_665)(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_662)(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_665)(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride)
 {
    DISPATCH(MultiModeDrawElementsIBM, (mode, count, type, indices, primcount, modestride), (F, "glMultiModeDrawElementsIBM(%p, %p, 0x%x, %p, %d, %d);\n", (const void *) mode, (const void *) count, type, (const void *) indices, primcount, modestride));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_663)(GLsizei n, const GLuint * fences);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_666)(GLsizei n, const GLuint * fences);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_663)(GLsizei n, const GLuint * fences)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_666)(GLsizei n, const GLuint * fences)
 {
    DISPATCH(DeleteFencesNV, (n, fences), (F, "glDeleteFencesNV(%d, %p);\n", n, (const void *) fences));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_664)(GLuint fence);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_667)(GLuint fence);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_664)(GLuint fence)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_667)(GLuint fence)
 {
    DISPATCH(FinishFenceNV, (fence), (F, "glFinishFenceNV(%d);\n", fence));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_665)(GLsizei n, GLuint * fences);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_668)(GLsizei n, GLuint * fences);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_665)(GLsizei n, GLuint * fences)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_668)(GLsizei n, GLuint * fences)
 {
    DISPATCH(GenFencesNV, (n, fences), (F, "glGenFencesNV(%d, %p);\n", n, (const void *) fences));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_666)(GLuint fence, GLenum pname, GLint * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_669)(GLuint fence, GLenum pname, GLint * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_666)(GLuint fence, GLenum pname, GLint * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_669)(GLuint fence, GLenum pname, GLint * params)
 {
    DISPATCH(GetFenceivNV, (fence, pname, params), (F, "glGetFenceivNV(%d, 0x%x, %p);\n", fence, pname, (const void *) params));
 }
 
-KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_667)(GLuint fence);
+KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_670)(GLuint fence);
 
-KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_667)(GLuint fence)
+KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_670)(GLuint fence)
 {
    RETURN_DISPATCH(IsFenceNV, (fence), (F, "glIsFenceNV(%d);\n", fence));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_668)(GLuint fence, GLenum condition);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_671)(GLuint fence, GLenum condition);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_668)(GLuint fence, GLenum condition)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_671)(GLuint fence, GLenum condition)
 {
    DISPATCH(SetFenceNV, (fence, condition), (F, "glSetFenceNV(%d, 0x%x);\n", fence, condition));
 }
 
-KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_669)(GLuint fence);
+KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_672)(GLuint fence);
 
-KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_669)(GLuint fence)
+KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_672)(GLuint fence)
 {
    RETURN_DISPATCH(TestFenceNV, (fence), (F, "glTestFenceNV(%d);\n", fence));
 }
@@ -5357,16 +5372,16 @@ KEYWORD1 void KEYWORD2 NAME(PointParameterivNV)(GLenum pname, const GLint * para
    DISPATCH(PointParameterivNV, (pname, params), (F, "glPointParameterivNV(0x%x, %p);\n", pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_750)(GLenum face);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_753)(GLenum face);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_750)(GLenum face)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_753)(GLenum face)
 {
    DISPATCH(ActiveStencilFaceEXT, (face), (F, "glActiveStencilFaceEXT(0x%x);\n", face));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_751)(GLuint array);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_754)(GLuint array);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_751)(GLuint array)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_754)(GLuint array)
 {
    DISPATCH(BindVertexArrayAPPLE, (array), (F, "glBindVertexArrayAPPLE(%d);\n", array));
 }
@@ -5376,16 +5391,16 @@ KEYWORD1 void KEYWORD2 NAME(DeleteVertexArrays)(GLsizei n, const GLuint * arrays
    DISPATCH(DeleteVertexArraysAPPLE, (n, arrays), (F, "glDeleteVertexArrays(%d, %p);\n", n, (const void *) arrays));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_752)(GLsizei n, const GLuint * arrays);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_755)(GLsizei n, const GLuint * arrays);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_752)(GLsizei n, const GLuint * arrays)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_755)(GLsizei n, const GLuint * arrays)
 {
    DISPATCH(DeleteVertexArraysAPPLE, (n, arrays), (F, "glDeleteVertexArraysAPPLE(%d, %p);\n", n, (const void *) arrays));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_753)(GLsizei n, GLuint * arrays);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_756)(GLsizei n, GLuint * arrays);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_753)(GLsizei n, GLuint * arrays)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_756)(GLsizei n, GLuint * arrays)
 {
    DISPATCH(GenVertexArraysAPPLE, (n, arrays), (F, "glGenVertexArraysAPPLE(%d, %p);\n", n, (const void *) arrays));
 }
@@ -5395,9 +5410,9 @@ KEYWORD1 GLboolean KEYWORD2 NAME(IsVertexArray)(GLuint array)
    RETURN_DISPATCH(IsVertexArrayAPPLE, (array), (F, "glIsVertexArray(%d);\n", array));
 }
 
-KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_754)(GLuint array);
+KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_757)(GLuint array);
 
-KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_754)(GLuint array)
+KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_757)(GLuint array)
 {
    RETURN_DISPATCH(IsVertexArrayAPPLE, (array), (F, "glIsVertexArrayAPPLE(%d);\n", array));
 }
@@ -5432,9 +5447,9 @@ KEYWORD1 void KEYWORD2 NAME(ProgramNamedParameter4fvNV)(GLuint id, GLsizei len,
    DISPATCH(ProgramNamedParameter4fvNV, (id, len, name, v), (F, "glProgramNamedParameter4fvNV(%d, %d, %p, %p);\n", id, len, (const void *) name, (const void *) v));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_761)(GLclampd zmin, GLclampd zmax);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_764)(GLclampd zmin, GLclampd zmax);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_761)(GLclampd zmin, GLclampd zmax)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_764)(GLclampd zmin, GLclampd zmax)
 {
    DISPATCH(DepthBoundsEXT, (zmin, zmax), (F, "glDepthBoundsEXT(%f, %f);\n", zmin, zmax));
 }
@@ -5444,9 +5459,9 @@ KEYWORD1 void KEYWORD2 NAME(BlendEquationSeparate)(GLenum modeRGB, GLenum modeA)
    DISPATCH(BlendEquationSeparateEXT, (modeRGB, modeA), (F, "glBlendEquationSeparate(0x%x, 0x%x);\n", modeRGB, modeA));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_762)(GLenum modeRGB, GLenum modeA);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_765)(GLenum modeRGB, GLenum modeA);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_762)(GLenum modeRGB, GLenum modeA)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_765)(GLenum modeRGB, GLenum modeA)
 {
    DISPATCH(BlendEquationSeparateEXT, (modeRGB, modeA), (F, "glBlendEquationSeparateEXT(0x%x, 0x%x);\n", modeRGB, modeA));
 }
@@ -5626,23 +5641,23 @@ KEYWORD1 void KEYWORD2 NAME(BlitFramebuffer)(GLint srcX0, GLint srcY0, GLint src
    DISPATCH(BlitFramebufferEXT, (srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter), (F, "glBlitFramebuffer(%d, %d, %d, %d, %d, %d, %d, %d, %d, 0x%x);\n", srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_780)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_783)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_780)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_783)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter)
 {
    DISPATCH(BlitFramebufferEXT, (srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter), (F, "glBlitFramebufferEXT(%d, %d, %d, %d, %d, %d, %d, %d, %d, 0x%x);\n", srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_781)(GLenum target, GLenum pname, GLint param);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_784)(GLenum target, GLenum pname, GLint param);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_781)(GLenum target, GLenum pname, GLint param)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_784)(GLenum target, GLenum pname, GLint param)
 {
    DISPATCH(BufferParameteriAPPLE, (target, pname, param), (F, "glBufferParameteriAPPLE(0x%x, 0x%x, %d);\n", target, pname, param));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_782)(GLenum target, GLintptr offset, GLsizeiptr size);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_785)(GLenum target, GLintptr offset, GLsizeiptr size);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_782)(GLenum target, GLintptr offset, GLsizeiptr size)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_785)(GLenum target, GLintptr offset, GLsizeiptr size)
 {
    DISPATCH(FlushMappedBufferRangeAPPLE, (target, offset, size), (F, "glFlushMappedBufferRangeAPPLE(0x%x, %d, %d);\n", target, offset, size));
 }
@@ -5662,51 +5677,56 @@ KEYWORD1 void KEYWORD2 NAME(ProvokingVertexEXT)(GLenum mode)
    DISPATCH(ProvokingVertexEXT, (mode), (F, "glProvokingVertexEXT(0x%x);\n", mode));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_785)(GLenum target, GLenum pname, GLvoid ** params);
+KEYWORD1 void KEYWORD2 NAME(ProvokingVertex)(GLenum mode)
+{
+   DISPATCH(ProvokingVertexEXT, (mode), (F, "glProvokingVertex(0x%x);\n", mode));
+}
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_785)(GLenum target, GLenum pname, GLvoid ** params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_788)(GLenum target, GLenum pname, GLvoid ** params);
+
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_788)(GLenum target, GLenum pname, GLvoid ** params)
 {
    DISPATCH(GetTexParameterPointervAPPLE, (target, pname, params), (F, "glGetTexParameterPointervAPPLE(0x%x, 0x%x, %p);\n", target, pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_786)(GLenum target, GLsizei length, GLvoid * pointer);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_789)(GLenum target, GLsizei length, GLvoid * pointer);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_786)(GLenum target, GLsizei length, GLvoid * pointer)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_789)(GLenum target, GLsizei length, GLvoid * pointer)
 {
    DISPATCH(TextureRangeAPPLE, (target, length, pointer), (F, "glTextureRangeAPPLE(0x%x, %d, %p);\n", target, length, (const void *) pointer));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_787)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_790)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_787)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_790)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask)
 {
    DISPATCH(StencilFuncSeparateATI, (frontfunc, backfunc, ref, mask), (F, "glStencilFuncSeparateATI(0x%x, 0x%x, %d, %d);\n", frontfunc, backfunc, ref, mask));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_788)(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_791)(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_788)(GLenum target, GLuint index, GLsizei count, const GLfloat * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_791)(GLenum target, GLuint index, GLsizei count, const GLfloat * params)
 {
    DISPATCH(ProgramEnvParameters4fvEXT, (target, index, count, params), (F, "glProgramEnvParameters4fvEXT(0x%x, %d, %d, %p);\n", target, index, count, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_789)(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_792)(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_789)(GLenum target, GLuint index, GLsizei count, const GLfloat * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_792)(GLenum target, GLuint index, GLsizei count, const GLfloat * params)
 {
    DISPATCH(ProgramLocalParameters4fvEXT, (target, index, count, params), (F, "glProgramLocalParameters4fvEXT(0x%x, %d, %d, %p);\n", target, index, count, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_790)(GLuint id, GLenum pname, GLint64EXT * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_793)(GLuint id, GLenum pname, GLint64EXT * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_790)(GLuint id, GLenum pname, GLint64EXT * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_793)(GLuint id, GLenum pname, GLint64EXT * params)
 {
    DISPATCH(GetQueryObjecti64vEXT, (id, pname, params), (F, "glGetQueryObjecti64vEXT(%d, 0x%x, %p);\n", id, pname, (const void *) params));
 }
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_791)(GLuint id, GLenum pname, GLuint64EXT * params);
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_794)(GLuint id, GLenum pname, GLuint64EXT * params);
 
-KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_791)(GLuint id, GLenum pname, GLuint64EXT * params)
+KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_794)(GLuint id, GLenum pname, GLuint64EXT * params)
 {
    DISPATCH(GetQueryObjectui64vEXT, (id, pname, params), (F, "glGetQueryObjectui64vEXT(%d, 0x%x, %p);\n", id, pname, (const void *) params));
 }
@@ -6299,15 +6319,18 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = {
    TABLE_ENTRY(GetSynciv),
    TABLE_ENTRY(IsSync),
    TABLE_ENTRY(WaitSync),
+   TABLE_ENTRY(DrawElementsBaseVertex),
+   TABLE_ENTRY(DrawRangeElementsBaseVertex),
+   TABLE_ENTRY(MultiDrawElementsBaseVertex),
    TABLE_ENTRY(PolygonOffsetEXT),
-   TABLE_ENTRY(_dispatch_stub_575),
-   TABLE_ENTRY(_dispatch_stub_576),
-   TABLE_ENTRY(_dispatch_stub_577),
    TABLE_ENTRY(_dispatch_stub_578),
    TABLE_ENTRY(_dispatch_stub_579),
    TABLE_ENTRY(_dispatch_stub_580),
    TABLE_ENTRY(_dispatch_stub_581),
    TABLE_ENTRY(_dispatch_stub_582),
+   TABLE_ENTRY(_dispatch_stub_583),
+   TABLE_ENTRY(_dispatch_stub_584),
+   TABLE_ENTRY(_dispatch_stub_585),
    TABLE_ENTRY(ColorPointerEXT),
    TABLE_ENTRY(EdgeFlagPointerEXT),
    TABLE_ENTRY(IndexPointerEXT),
@@ -6318,8 +6341,8 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = {
    TABLE_ENTRY(PointParameterfvEXT),
    TABLE_ENTRY(LockArraysEXT),
    TABLE_ENTRY(UnlockArraysEXT),
-   TABLE_ENTRY(_dispatch_stub_593),
-   TABLE_ENTRY(_dispatch_stub_594),
+   TABLE_ENTRY(_dispatch_stub_596),
+   TABLE_ENTRY(_dispatch_stub_597),
    TABLE_ENTRY(SecondaryColor3bEXT),
    TABLE_ENTRY(SecondaryColor3bvEXT),
    TABLE_ENTRY(SecondaryColor3dEXT),
@@ -6344,7 +6367,7 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = {
    TABLE_ENTRY(FogCoorddvEXT),
    TABLE_ENTRY(FogCoordfEXT),
    TABLE_ENTRY(FogCoordfvEXT),
-   TABLE_ENTRY(_dispatch_stub_619),
+   TABLE_ENTRY(_dispatch_stub_622),
    TABLE_ENTRY(BlendFuncSeparateEXT),
    TABLE_ENTRY(FlushVertexArrayRangeNV),
    TABLE_ENTRY(VertexArrayRangeNV),
@@ -6386,15 +6409,15 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = {
    TABLE_ENTRY(WindowPos4ivMESA),
    TABLE_ENTRY(WindowPos4sMESA),
    TABLE_ENTRY(WindowPos4svMESA),
-   TABLE_ENTRY(_dispatch_stub_661),
-   TABLE_ENTRY(_dispatch_stub_662),
-   TABLE_ENTRY(_dispatch_stub_663),
    TABLE_ENTRY(_dispatch_stub_664),
    TABLE_ENTRY(_dispatch_stub_665),
    TABLE_ENTRY(_dispatch_stub_666),
    TABLE_ENTRY(_dispatch_stub_667),
    TABLE_ENTRY(_dispatch_stub_668),
    TABLE_ENTRY(_dispatch_stub_669),
+   TABLE_ENTRY(_dispatch_stub_670),
+   TABLE_ENTRY(_dispatch_stub_671),
+   TABLE_ENTRY(_dispatch_stub_672),
    TABLE_ENTRY(AreProgramsResidentNV),
    TABLE_ENTRY(BindProgramNV),
    TABLE_ENTRY(DeleteProgramsNV),
@@ -6475,19 +6498,19 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = {
    TABLE_ENTRY(SetFragmentShaderConstantATI),
    TABLE_ENTRY(PointParameteriNV),
    TABLE_ENTRY(PointParameterivNV),
-   TABLE_ENTRY(_dispatch_stub_750),
-   TABLE_ENTRY(_dispatch_stub_751),
-   TABLE_ENTRY(_dispatch_stub_752),
    TABLE_ENTRY(_dispatch_stub_753),
    TABLE_ENTRY(_dispatch_stub_754),
+   TABLE_ENTRY(_dispatch_stub_755),
+   TABLE_ENTRY(_dispatch_stub_756),
+   TABLE_ENTRY(_dispatch_stub_757),
    TABLE_ENTRY(GetProgramNamedParameterdvNV),
    TABLE_ENTRY(GetProgramNamedParameterfvNV),
    TABLE_ENTRY(ProgramNamedParameter4dNV),
    TABLE_ENTRY(ProgramNamedParameter4dvNV),
    TABLE_ENTRY(ProgramNamedParameter4fNV),
    TABLE_ENTRY(ProgramNamedParameter4fvNV),
-   TABLE_ENTRY(_dispatch_stub_761),
-   TABLE_ENTRY(_dispatch_stub_762),
+   TABLE_ENTRY(_dispatch_stub_764),
+   TABLE_ENTRY(_dispatch_stub_765),
    TABLE_ENTRY(BindFramebufferEXT),
    TABLE_ENTRY(BindRenderbufferEXT),
    TABLE_ENTRY(CheckFramebufferStatusEXT),
@@ -6505,18 +6528,18 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = {
    TABLE_ENTRY(IsFramebufferEXT),
    TABLE_ENTRY(IsRenderbufferEXT),
    TABLE_ENTRY(RenderbufferStorageEXT),
-   TABLE_ENTRY(_dispatch_stub_780),
-   TABLE_ENTRY(_dispatch_stub_781),
-   TABLE_ENTRY(_dispatch_stub_782),
+   TABLE_ENTRY(_dispatch_stub_783),
+   TABLE_ENTRY(_dispatch_stub_784),
+   TABLE_ENTRY(_dispatch_stub_785),
    TABLE_ENTRY(FramebufferTextureLayerEXT),
    TABLE_ENTRY(ProvokingVertexEXT),
-   TABLE_ENTRY(_dispatch_stub_785),
-   TABLE_ENTRY(_dispatch_stub_786),
-   TABLE_ENTRY(_dispatch_stub_787),
    TABLE_ENTRY(_dispatch_stub_788),
    TABLE_ENTRY(_dispatch_stub_789),
    TABLE_ENTRY(_dispatch_stub_790),
    TABLE_ENTRY(_dispatch_stub_791),
+   TABLE_ENTRY(_dispatch_stub_792),
+   TABLE_ENTRY(_dispatch_stub_793),
+   TABLE_ENTRY(_dispatch_stub_794),
    /* A whole bunch of no-op functions.  These might be called
     * when someone tries to call a dynamically-registered
     * extension function without a current rendering context.
@@ -6905,6 +6928,7 @@ static _glapi_proc UNUSED_TABLE_NAME[] = {
    TABLE_ENTRY(RenderbufferStorage),
    TABLE_ENTRY(BlitFramebuffer),
    TABLE_ENTRY(FramebufferTextureLayer),
+   TABLE_ENTRY(ProvokingVertex),
 };
 #endif /*UNUSED_TABLE_NAME*/
 
diff --git a/src/mesa/glapi/glprocs.h b/src/mesa/glapi/glprocs.h
index ae6603262e..c29f8b57be 100644
--- a/src/mesa/glapi/glprocs.h
+++ b/src/mesa/glapi/glprocs.h
@@ -626,6 +626,9 @@ static const char gl_string_table[] =
     "glGetSynciv\0"
     "glIsSync\0"
     "glWaitSync\0"
+    "glDrawElementsBaseVertex\0"
+    "glDrawRangeElementsBaseVertex\0"
+    "glMultiDrawElementsBaseVertex\0"
     "glPolygonOffsetEXT\0"
     "glGetPixelTexGenParameterfvSGIS\0"
     "glGetPixelTexGenParameterivSGIS\0"
@@ -1144,6 +1147,7 @@ static const char gl_string_table[] =
     "glRenderbufferStorage\0"
     "glBlitFramebuffer\0"
     "glFramebufferTextureLayer\0"
+    "glProvokingVertex\0"
     ;
 
 
@@ -1161,43 +1165,43 @@ static const char gl_string_table[] =
 #define gl_dispatch_stub_364 mgl_dispatch_stub_364
 #define gl_dispatch_stub_365 mgl_dispatch_stub_365
 #define gl_dispatch_stub_366 mgl_dispatch_stub_366
-#define gl_dispatch_stub_575 mgl_dispatch_stub_575
-#define gl_dispatch_stub_576 mgl_dispatch_stub_576
-#define gl_dispatch_stub_577 mgl_dispatch_stub_577
 #define gl_dispatch_stub_578 mgl_dispatch_stub_578
 #define gl_dispatch_stub_579 mgl_dispatch_stub_579
 #define gl_dispatch_stub_580 mgl_dispatch_stub_580
 #define gl_dispatch_stub_581 mgl_dispatch_stub_581
 #define gl_dispatch_stub_582 mgl_dispatch_stub_582
-#define gl_dispatch_stub_593 mgl_dispatch_stub_593
-#define gl_dispatch_stub_594 mgl_dispatch_stub_594
-#define gl_dispatch_stub_619 mgl_dispatch_stub_619
-#define gl_dispatch_stub_661 mgl_dispatch_stub_661
-#define gl_dispatch_stub_662 mgl_dispatch_stub_662
-#define gl_dispatch_stub_663 mgl_dispatch_stub_663
+#define gl_dispatch_stub_583 mgl_dispatch_stub_583
+#define gl_dispatch_stub_584 mgl_dispatch_stub_584
+#define gl_dispatch_stub_585 mgl_dispatch_stub_585
+#define gl_dispatch_stub_596 mgl_dispatch_stub_596
+#define gl_dispatch_stub_597 mgl_dispatch_stub_597
+#define gl_dispatch_stub_622 mgl_dispatch_stub_622
 #define gl_dispatch_stub_664 mgl_dispatch_stub_664
 #define gl_dispatch_stub_665 mgl_dispatch_stub_665
 #define gl_dispatch_stub_666 mgl_dispatch_stub_666
 #define gl_dispatch_stub_667 mgl_dispatch_stub_667
 #define gl_dispatch_stub_668 mgl_dispatch_stub_668
 #define gl_dispatch_stub_669 mgl_dispatch_stub_669
-#define gl_dispatch_stub_750 mgl_dispatch_stub_750
-#define gl_dispatch_stub_751 mgl_dispatch_stub_751
-#define gl_dispatch_stub_752 mgl_dispatch_stub_752
+#define gl_dispatch_stub_670 mgl_dispatch_stub_670
+#define gl_dispatch_stub_671 mgl_dispatch_stub_671
+#define gl_dispatch_stub_672 mgl_dispatch_stub_672
 #define gl_dispatch_stub_753 mgl_dispatch_stub_753
 #define gl_dispatch_stub_754 mgl_dispatch_stub_754
-#define gl_dispatch_stub_761 mgl_dispatch_stub_761
-#define gl_dispatch_stub_762 mgl_dispatch_stub_762
-#define gl_dispatch_stub_780 mgl_dispatch_stub_780
-#define gl_dispatch_stub_781 mgl_dispatch_stub_781
-#define gl_dispatch_stub_782 mgl_dispatch_stub_782
+#define gl_dispatch_stub_755 mgl_dispatch_stub_755
+#define gl_dispatch_stub_756 mgl_dispatch_stub_756
+#define gl_dispatch_stub_757 mgl_dispatch_stub_757
+#define gl_dispatch_stub_764 mgl_dispatch_stub_764
+#define gl_dispatch_stub_765 mgl_dispatch_stub_765
+#define gl_dispatch_stub_783 mgl_dispatch_stub_783
+#define gl_dispatch_stub_784 mgl_dispatch_stub_784
 #define gl_dispatch_stub_785 mgl_dispatch_stub_785
-#define gl_dispatch_stub_786 mgl_dispatch_stub_786
-#define gl_dispatch_stub_787 mgl_dispatch_stub_787
 #define gl_dispatch_stub_788 mgl_dispatch_stub_788
 #define gl_dispatch_stub_789 mgl_dispatch_stub_789
 #define gl_dispatch_stub_790 mgl_dispatch_stub_790
 #define gl_dispatch_stub_791 mgl_dispatch_stub_791
+#define gl_dispatch_stub_792 mgl_dispatch_stub_792
+#define gl_dispatch_stub_793 mgl_dispatch_stub_793
+#define gl_dispatch_stub_794 mgl_dispatch_stub_794
 #endif /* USE_MGL_NAMESPACE */
 
 
@@ -1215,43 +1219,43 @@ void GLAPIENTRY gl_dispatch_stub_363(GLenum target, GLenum pname, GLint * params
 void GLAPIENTRY gl_dispatch_stub_364(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid * values);
 void GLAPIENTRY gl_dispatch_stub_365(GLenum target, GLenum pname, GLfloat * params);
 void GLAPIENTRY gl_dispatch_stub_366(GLenum target, GLenum pname, GLint * params);
-void GLAPIENTRY gl_dispatch_stub_575(GLenum pname, GLfloat * params);
-void GLAPIENTRY gl_dispatch_stub_576(GLenum pname, GLint * params);
-void GLAPIENTRY gl_dispatch_stub_577(GLenum pname, GLfloat param);
-void GLAPIENTRY gl_dispatch_stub_578(GLenum pname, const GLfloat * params);
-void GLAPIENTRY gl_dispatch_stub_579(GLenum pname, GLint param);
-void GLAPIENTRY gl_dispatch_stub_580(GLenum pname, const GLint * params);
-void GLAPIENTRY gl_dispatch_stub_581(GLclampf value, GLboolean invert);
-void GLAPIENTRY gl_dispatch_stub_582(GLenum pattern);
-void GLAPIENTRY gl_dispatch_stub_593(GLenum pname, GLdouble * params);
-void GLAPIENTRY gl_dispatch_stub_594(GLenum pname, GLfloat * params);
-void GLAPIENTRY gl_dispatch_stub_619(GLenum mode);
-void GLAPIENTRY gl_dispatch_stub_661(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride);
-void GLAPIENTRY gl_dispatch_stub_662(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride);
-void GLAPIENTRY gl_dispatch_stub_663(GLsizei n, const GLuint * fences);
-void GLAPIENTRY gl_dispatch_stub_664(GLuint fence);
-void GLAPIENTRY gl_dispatch_stub_665(GLsizei n, GLuint * fences);
-void GLAPIENTRY gl_dispatch_stub_666(GLuint fence, GLenum pname, GLint * params);
-GLboolean GLAPIENTRY gl_dispatch_stub_667(GLuint fence);
-void GLAPIENTRY gl_dispatch_stub_668(GLuint fence, GLenum condition);
-GLboolean GLAPIENTRY gl_dispatch_stub_669(GLuint fence);
-void GLAPIENTRY gl_dispatch_stub_750(GLenum face);
-void GLAPIENTRY gl_dispatch_stub_751(GLuint array);
-void GLAPIENTRY gl_dispatch_stub_752(GLsizei n, const GLuint * arrays);
-void GLAPIENTRY gl_dispatch_stub_753(GLsizei n, GLuint * arrays);
-GLboolean GLAPIENTRY gl_dispatch_stub_754(GLuint array);
-void GLAPIENTRY gl_dispatch_stub_761(GLclampd zmin, GLclampd zmax);
-void GLAPIENTRY gl_dispatch_stub_762(GLenum modeRGB, GLenum modeA);
-void GLAPIENTRY gl_dispatch_stub_780(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
-void GLAPIENTRY gl_dispatch_stub_781(GLenum target, GLenum pname, GLint param);
-void GLAPIENTRY gl_dispatch_stub_782(GLenum target, GLintptr offset, GLsizeiptr size);
-void GLAPIENTRY gl_dispatch_stub_785(GLenum target, GLenum pname, GLvoid ** params);
-void GLAPIENTRY gl_dispatch_stub_786(GLenum target, GLsizei length, GLvoid * pointer);
-void GLAPIENTRY gl_dispatch_stub_787(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask);
-void GLAPIENTRY gl_dispatch_stub_788(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
-void GLAPIENTRY gl_dispatch_stub_789(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
-void GLAPIENTRY gl_dispatch_stub_790(GLuint id, GLenum pname, GLint64EXT * params);
-void GLAPIENTRY gl_dispatch_stub_791(GLuint id, GLenum pname, GLuint64EXT * params);
+void GLAPIENTRY gl_dispatch_stub_578(GLenum pname, GLfloat * params);
+void GLAPIENTRY gl_dispatch_stub_579(GLenum pname, GLint * params);
+void GLAPIENTRY gl_dispatch_stub_580(GLenum pname, GLfloat param);
+void GLAPIENTRY gl_dispatch_stub_581(GLenum pname, const GLfloat * params);
+void GLAPIENTRY gl_dispatch_stub_582(GLenum pname, GLint param);
+void GLAPIENTRY gl_dispatch_stub_583(GLenum pname, const GLint * params);
+void GLAPIENTRY gl_dispatch_stub_584(GLclampf value, GLboolean invert);
+void GLAPIENTRY gl_dispatch_stub_585(GLenum pattern);
+void GLAPIENTRY gl_dispatch_stub_596(GLenum pname, GLdouble * params);
+void GLAPIENTRY gl_dispatch_stub_597(GLenum pname, GLfloat * params);
+void GLAPIENTRY gl_dispatch_stub_622(GLenum mode);
+void GLAPIENTRY gl_dispatch_stub_664(const GLenum * mode, const GLint * first, const GLsizei * count, GLsizei primcount, GLint modestride);
+void GLAPIENTRY gl_dispatch_stub_665(const GLenum * mode, const GLsizei * count, GLenum type, const GLvoid * const * indices, GLsizei primcount, GLint modestride);
+void GLAPIENTRY gl_dispatch_stub_666(GLsizei n, const GLuint * fences);
+void GLAPIENTRY gl_dispatch_stub_667(GLuint fence);
+void GLAPIENTRY gl_dispatch_stub_668(GLsizei n, GLuint * fences);
+void GLAPIENTRY gl_dispatch_stub_669(GLuint fence, GLenum pname, GLint * params);
+GLboolean GLAPIENTRY gl_dispatch_stub_670(GLuint fence);
+void GLAPIENTRY gl_dispatch_stub_671(GLuint fence, GLenum condition);
+GLboolean GLAPIENTRY gl_dispatch_stub_672(GLuint fence);
+void GLAPIENTRY gl_dispatch_stub_753(GLenum face);
+void GLAPIENTRY gl_dispatch_stub_754(GLuint array);
+void GLAPIENTRY gl_dispatch_stub_755(GLsizei n, const GLuint * arrays);
+void GLAPIENTRY gl_dispatch_stub_756(GLsizei n, GLuint * arrays);
+GLboolean GLAPIENTRY gl_dispatch_stub_757(GLuint array);
+void GLAPIENTRY gl_dispatch_stub_764(GLclampd zmin, GLclampd zmax);
+void GLAPIENTRY gl_dispatch_stub_765(GLenum modeRGB, GLenum modeA);
+void GLAPIENTRY gl_dispatch_stub_783(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+void GLAPIENTRY gl_dispatch_stub_784(GLenum target, GLenum pname, GLint param);
+void GLAPIENTRY gl_dispatch_stub_785(GLenum target, GLintptr offset, GLsizeiptr size);
+void GLAPIENTRY gl_dispatch_stub_788(GLenum target, GLenum pname, GLvoid ** params);
+void GLAPIENTRY gl_dispatch_stub_789(GLenum target, GLsizei length, GLvoid * pointer);
+void GLAPIENTRY gl_dispatch_stub_790(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask);
+void GLAPIENTRY gl_dispatch_stub_791(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
+void GLAPIENTRY gl_dispatch_stub_792(GLenum target, GLuint index, GLsizei count, const GLfloat * params);
+void GLAPIENTRY gl_dispatch_stub_793(GLuint id, GLenum pname, GLint64EXT * params);
+void GLAPIENTRY gl_dispatch_stub_794(GLuint id, GLenum pname, GLuint64EXT * params);
 #endif /* defined(NEED_FUNCTION_POINTER) || defined(GLX_INDIRECT_RENDERING) */
 
 static const glprocs_table_t static_functions[] = {
@@ -1829,524 +1833,528 @@ static const glprocs_table_t static_functions[] = {
     NAME_FUNC_OFFSET( 9140, glGetSynciv, glGetSynciv, NULL, _gloffset_GetSynciv),
     NAME_FUNC_OFFSET( 9152, glIsSync, glIsSync, NULL, _gloffset_IsSync),
     NAME_FUNC_OFFSET( 9161, glWaitSync, glWaitSync, NULL, _gloffset_WaitSync),
-    NAME_FUNC_OFFSET( 9172, glPolygonOffsetEXT, glPolygonOffsetEXT, NULL, _gloffset_PolygonOffsetEXT),
-    NAME_FUNC_OFFSET( 9191, gl_dispatch_stub_575, gl_dispatch_stub_575, NULL, _gloffset_GetPixelTexGenParameterfvSGIS),
-    NAME_FUNC_OFFSET( 9223, gl_dispatch_stub_576, gl_dispatch_stub_576, NULL, _gloffset_GetPixelTexGenParameterivSGIS),
-    NAME_FUNC_OFFSET( 9255, gl_dispatch_stub_577, gl_dispatch_stub_577, NULL, _gloffset_PixelTexGenParameterfSGIS),
-    NAME_FUNC_OFFSET( 9283, gl_dispatch_stub_578, gl_dispatch_stub_578, NULL, _gloffset_PixelTexGenParameterfvSGIS),
-    NAME_FUNC_OFFSET( 9312, gl_dispatch_stub_579, gl_dispatch_stub_579, NULL, _gloffset_PixelTexGenParameteriSGIS),
-    NAME_FUNC_OFFSET( 9340, gl_dispatch_stub_580, gl_dispatch_stub_580, NULL, _gloffset_PixelTexGenParameterivSGIS),
-    NAME_FUNC_OFFSET( 9369, gl_dispatch_stub_581, gl_dispatch_stub_581, NULL, _gloffset_SampleMaskSGIS),
-    NAME_FUNC_OFFSET( 9386, gl_dispatch_stub_582, gl_dispatch_stub_582, NULL, _gloffset_SamplePatternSGIS),
-    NAME_FUNC_OFFSET( 9406, glColorPointerEXT, glColorPointerEXT, NULL, _gloffset_ColorPointerEXT),
-    NAME_FUNC_OFFSET( 9424, glEdgeFlagPointerEXT, glEdgeFlagPointerEXT, NULL, _gloffset_EdgeFlagPointerEXT),
-    NAME_FUNC_OFFSET( 9445, glIndexPointerEXT, glIndexPointerEXT, NULL, _gloffset_IndexPointerEXT),
-    NAME_FUNC_OFFSET( 9463, glNormalPointerEXT, glNormalPointerEXT, NULL, _gloffset_NormalPointerEXT),
-    NAME_FUNC_OFFSET( 9482, glTexCoordPointerEXT, glTexCoordPointerEXT, NULL, _gloffset_TexCoordPointerEXT),
-    NAME_FUNC_OFFSET( 9503, glVertexPointerEXT, glVertexPointerEXT, NULL, _gloffset_VertexPointerEXT),
-    NAME_FUNC_OFFSET( 9522, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET( 9543, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET( 9565, glLockArraysEXT, glLockArraysEXT, NULL, _gloffset_LockArraysEXT),
-    NAME_FUNC_OFFSET( 9581, glUnlockArraysEXT, glUnlockArraysEXT, NULL, _gloffset_UnlockArraysEXT),
-    NAME_FUNC_OFFSET( 9599, gl_dispatch_stub_593, gl_dispatch_stub_593, NULL, _gloffset_CullParameterdvEXT),
-    NAME_FUNC_OFFSET( 9620, gl_dispatch_stub_594, gl_dispatch_stub_594, NULL, _gloffset_CullParameterfvEXT),
-    NAME_FUNC_OFFSET( 9641, glSecondaryColor3bEXT, glSecondaryColor3bEXT, NULL, _gloffset_SecondaryColor3bEXT),
-    NAME_FUNC_OFFSET( 9663, glSecondaryColor3bvEXT, glSecondaryColor3bvEXT, NULL, _gloffset_SecondaryColor3bvEXT),
-    NAME_FUNC_OFFSET( 9686, glSecondaryColor3dEXT, glSecondaryColor3dEXT, NULL, _gloffset_SecondaryColor3dEXT),
-    NAME_FUNC_OFFSET( 9708, glSecondaryColor3dvEXT, glSecondaryColor3dvEXT, NULL, _gloffset_SecondaryColor3dvEXT),
-    NAME_FUNC_OFFSET( 9731, glSecondaryColor3fEXT, glSecondaryColor3fEXT, NULL, _gloffset_SecondaryColor3fEXT),
-    NAME_FUNC_OFFSET( 9753, glSecondaryColor3fvEXT, glSecondaryColor3fvEXT, NULL, _gloffset_SecondaryColor3fvEXT),
-    NAME_FUNC_OFFSET( 9776, glSecondaryColor3iEXT, glSecondaryColor3iEXT, NULL, _gloffset_SecondaryColor3iEXT),
-    NAME_FUNC_OFFSET( 9798, glSecondaryColor3ivEXT, glSecondaryColor3ivEXT, NULL, _gloffset_SecondaryColor3ivEXT),
-    NAME_FUNC_OFFSET( 9821, glSecondaryColor3sEXT, glSecondaryColor3sEXT, NULL, _gloffset_SecondaryColor3sEXT),
-    NAME_FUNC_OFFSET( 9843, glSecondaryColor3svEXT, glSecondaryColor3svEXT, NULL, _gloffset_SecondaryColor3svEXT),
-    NAME_FUNC_OFFSET( 9866, glSecondaryColor3ubEXT, glSecondaryColor3ubEXT, NULL, _gloffset_SecondaryColor3ubEXT),
-    NAME_FUNC_OFFSET( 9889, glSecondaryColor3ubvEXT, glSecondaryColor3ubvEXT, NULL, _gloffset_SecondaryColor3ubvEXT),
-    NAME_FUNC_OFFSET( 9913, glSecondaryColor3uiEXT, glSecondaryColor3uiEXT, NULL, _gloffset_SecondaryColor3uiEXT),
-    NAME_FUNC_OFFSET( 9936, glSecondaryColor3uivEXT, glSecondaryColor3uivEXT, NULL, _gloffset_SecondaryColor3uivEXT),
-    NAME_FUNC_OFFSET( 9960, glSecondaryColor3usEXT, glSecondaryColor3usEXT, NULL, _gloffset_SecondaryColor3usEXT),
-    NAME_FUNC_OFFSET( 9983, glSecondaryColor3usvEXT, glSecondaryColor3usvEXT, NULL, _gloffset_SecondaryColor3usvEXT),
-    NAME_FUNC_OFFSET(10007, glSecondaryColorPointerEXT, glSecondaryColorPointerEXT, NULL, _gloffset_SecondaryColorPointerEXT),
-    NAME_FUNC_OFFSET(10034, glMultiDrawArraysEXT, glMultiDrawArraysEXT, NULL, _gloffset_MultiDrawArraysEXT),
-    NAME_FUNC_OFFSET(10055, glMultiDrawElementsEXT, glMultiDrawElementsEXT, NULL, _gloffset_MultiDrawElementsEXT),
-    NAME_FUNC_OFFSET(10078, glFogCoordPointerEXT, glFogCoordPointerEXT, NULL, _gloffset_FogCoordPointerEXT),
-    NAME_FUNC_OFFSET(10099, glFogCoorddEXT, glFogCoorddEXT, NULL, _gloffset_FogCoorddEXT),
-    NAME_FUNC_OFFSET(10114, glFogCoorddvEXT, glFogCoorddvEXT, NULL, _gloffset_FogCoorddvEXT),
-    NAME_FUNC_OFFSET(10130, glFogCoordfEXT, glFogCoordfEXT, NULL, _gloffset_FogCoordfEXT),
-    NAME_FUNC_OFFSET(10145, glFogCoordfvEXT, glFogCoordfvEXT, NULL, _gloffset_FogCoordfvEXT),
-    NAME_FUNC_OFFSET(10161, gl_dispatch_stub_619, gl_dispatch_stub_619, NULL, _gloffset_PixelTexGenSGIX),
-    NAME_FUNC_OFFSET(10179, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
-    NAME_FUNC_OFFSET(10202, glFlushVertexArrayRangeNV, glFlushVertexArrayRangeNV, NULL, _gloffset_FlushVertexArrayRangeNV),
-    NAME_FUNC_OFFSET(10228, glVertexArrayRangeNV, glVertexArrayRangeNV, NULL, _gloffset_VertexArrayRangeNV),
-    NAME_FUNC_OFFSET(10249, glCombinerInputNV, glCombinerInputNV, NULL, _gloffset_CombinerInputNV),
-    NAME_FUNC_OFFSET(10267, glCombinerOutputNV, glCombinerOutputNV, NULL, _gloffset_CombinerOutputNV),
-    NAME_FUNC_OFFSET(10286, glCombinerParameterfNV, glCombinerParameterfNV, NULL, _gloffset_CombinerParameterfNV),
-    NAME_FUNC_OFFSET(10309, glCombinerParameterfvNV, glCombinerParameterfvNV, NULL, _gloffset_CombinerParameterfvNV),
-    NAME_FUNC_OFFSET(10333, glCombinerParameteriNV, glCombinerParameteriNV, NULL, _gloffset_CombinerParameteriNV),
-    NAME_FUNC_OFFSET(10356, glCombinerParameterivNV, glCombinerParameterivNV, NULL, _gloffset_CombinerParameterivNV),
-    NAME_FUNC_OFFSET(10380, glFinalCombinerInputNV, glFinalCombinerInputNV, NULL, _gloffset_FinalCombinerInputNV),
-    NAME_FUNC_OFFSET(10403, glGetCombinerInputParameterfvNV, glGetCombinerInputParameterfvNV, NULL, _gloffset_GetCombinerInputParameterfvNV),
-    NAME_FUNC_OFFSET(10435, glGetCombinerInputParameterivNV, glGetCombinerInputParameterivNV, NULL, _gloffset_GetCombinerInputParameterivNV),
-    NAME_FUNC_OFFSET(10467, glGetCombinerOutputParameterfvNV, glGetCombinerOutputParameterfvNV, NULL, _gloffset_GetCombinerOutputParameterfvNV),
-    NAME_FUNC_OFFSET(10500, glGetCombinerOutputParameterivNV, glGetCombinerOutputParameterivNV, NULL, _gloffset_GetCombinerOutputParameterivNV),
-    NAME_FUNC_OFFSET(10533, glGetFinalCombinerInputParameterfvNV, glGetFinalCombinerInputParameterfvNV, NULL, _gloffset_GetFinalCombinerInputParameterfvNV),
-    NAME_FUNC_OFFSET(10570, glGetFinalCombinerInputParameterivNV, glGetFinalCombinerInputParameterivNV, NULL, _gloffset_GetFinalCombinerInputParameterivNV),
-    NAME_FUNC_OFFSET(10607, glResizeBuffersMESA, glResizeBuffersMESA, NULL, _gloffset_ResizeBuffersMESA),
-    NAME_FUNC_OFFSET(10627, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
-    NAME_FUNC_OFFSET(10645, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
-    NAME_FUNC_OFFSET(10664, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
-    NAME_FUNC_OFFSET(10682, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
-    NAME_FUNC_OFFSET(10701, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
-    NAME_FUNC_OFFSET(10719, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
-    NAME_FUNC_OFFSET(10738, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
-    NAME_FUNC_OFFSET(10756, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
-    NAME_FUNC_OFFSET(10775, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
-    NAME_FUNC_OFFSET(10793, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
-    NAME_FUNC_OFFSET(10812, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
-    NAME_FUNC_OFFSET(10830, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
-    NAME_FUNC_OFFSET(10849, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
-    NAME_FUNC_OFFSET(10867, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
-    NAME_FUNC_OFFSET(10886, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
-    NAME_FUNC_OFFSET(10904, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
-    NAME_FUNC_OFFSET(10923, glWindowPos4dMESA, glWindowPos4dMESA, NULL, _gloffset_WindowPos4dMESA),
-    NAME_FUNC_OFFSET(10941, glWindowPos4dvMESA, glWindowPos4dvMESA, NULL, _gloffset_WindowPos4dvMESA),
-    NAME_FUNC_OFFSET(10960, glWindowPos4fMESA, glWindowPos4fMESA, NULL, _gloffset_WindowPos4fMESA),
-    NAME_FUNC_OFFSET(10978, glWindowPos4fvMESA, glWindowPos4fvMESA, NULL, _gloffset_WindowPos4fvMESA),
-    NAME_FUNC_OFFSET(10997, glWindowPos4iMESA, glWindowPos4iMESA, NULL, _gloffset_WindowPos4iMESA),
-    NAME_FUNC_OFFSET(11015, glWindowPos4ivMESA, glWindowPos4ivMESA, NULL, _gloffset_WindowPos4ivMESA),
-    NAME_FUNC_OFFSET(11034, glWindowPos4sMESA, glWindowPos4sMESA, NULL, _gloffset_WindowPos4sMESA),
-    NAME_FUNC_OFFSET(11052, glWindowPos4svMESA, glWindowPos4svMESA, NULL, _gloffset_WindowPos4svMESA),
-    NAME_FUNC_OFFSET(11071, gl_dispatch_stub_661, gl_dispatch_stub_661, NULL, _gloffset_MultiModeDrawArraysIBM),
-    NAME_FUNC_OFFSET(11096, gl_dispatch_stub_662, gl_dispatch_stub_662, NULL, _gloffset_MultiModeDrawElementsIBM),
-    NAME_FUNC_OFFSET(11123, gl_dispatch_stub_663, gl_dispatch_stub_663, NULL, _gloffset_DeleteFencesNV),
-    NAME_FUNC_OFFSET(11140, gl_dispatch_stub_664, gl_dispatch_stub_664, NULL, _gloffset_FinishFenceNV),
-    NAME_FUNC_OFFSET(11156, gl_dispatch_stub_665, gl_dispatch_stub_665, NULL, _gloffset_GenFencesNV),
-    NAME_FUNC_OFFSET(11170, gl_dispatch_stub_666, gl_dispatch_stub_666, NULL, _gloffset_GetFenceivNV),
-    NAME_FUNC_OFFSET(11185, gl_dispatch_stub_667, gl_dispatch_stub_667, NULL, _gloffset_IsFenceNV),
-    NAME_FUNC_OFFSET(11197, gl_dispatch_stub_668, gl_dispatch_stub_668, NULL, _gloffset_SetFenceNV),
-    NAME_FUNC_OFFSET(11210, gl_dispatch_stub_669, gl_dispatch_stub_669, NULL, _gloffset_TestFenceNV),
-    NAME_FUNC_OFFSET(11224, glAreProgramsResidentNV, glAreProgramsResidentNV, NULL, _gloffset_AreProgramsResidentNV),
-    NAME_FUNC_OFFSET(11248, glBindProgramNV, glBindProgramNV, NULL, _gloffset_BindProgramNV),
-    NAME_FUNC_OFFSET(11264, glDeleteProgramsNV, glDeleteProgramsNV, NULL, _gloffset_DeleteProgramsNV),
-    NAME_FUNC_OFFSET(11283, glExecuteProgramNV, glExecuteProgramNV, NULL, _gloffset_ExecuteProgramNV),
-    NAME_FUNC_OFFSET(11302, glGenProgramsNV, glGenProgramsNV, NULL, _gloffset_GenProgramsNV),
-    NAME_FUNC_OFFSET(11318, glGetProgramParameterdvNV, glGetProgramParameterdvNV, NULL, _gloffset_GetProgramParameterdvNV),
-    NAME_FUNC_OFFSET(11344, glGetProgramParameterfvNV, glGetProgramParameterfvNV, NULL, _gloffset_GetProgramParameterfvNV),
-    NAME_FUNC_OFFSET(11370, glGetProgramStringNV, glGetProgramStringNV, NULL, _gloffset_GetProgramStringNV),
-    NAME_FUNC_OFFSET(11391, glGetProgramivNV, glGetProgramivNV, NULL, _gloffset_GetProgramivNV),
-    NAME_FUNC_OFFSET(11408, glGetTrackMatrixivNV, glGetTrackMatrixivNV, NULL, _gloffset_GetTrackMatrixivNV),
-    NAME_FUNC_OFFSET(11429, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
-    NAME_FUNC_OFFSET(11457, glGetVertexAttribdvNV, glGetVertexAttribdvNV, NULL, _gloffset_GetVertexAttribdvNV),
-    NAME_FUNC_OFFSET(11479, glGetVertexAttribfvNV, glGetVertexAttribfvNV, NULL, _gloffset_GetVertexAttribfvNV),
-    NAME_FUNC_OFFSET(11501, glGetVertexAttribivNV, glGetVertexAttribivNV, NULL, _gloffset_GetVertexAttribivNV),
-    NAME_FUNC_OFFSET(11523, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV),
-    NAME_FUNC_OFFSET(11537, glLoadProgramNV, glLoadProgramNV, NULL, _gloffset_LoadProgramNV),
-    NAME_FUNC_OFFSET(11553, glProgramParameters4dvNV, glProgramParameters4dvNV, NULL, _gloffset_ProgramParameters4dvNV),
-    NAME_FUNC_OFFSET(11578, glProgramParameters4fvNV, glProgramParameters4fvNV, NULL, _gloffset_ProgramParameters4fvNV),
-    NAME_FUNC_OFFSET(11603, glRequestResidentProgramsNV, glRequestResidentProgramsNV, NULL, _gloffset_RequestResidentProgramsNV),
-    NAME_FUNC_OFFSET(11631, glTrackMatrixNV, glTrackMatrixNV, NULL, _gloffset_TrackMatrixNV),
-    NAME_FUNC_OFFSET(11647, glVertexAttrib1dNV, glVertexAttrib1dNV, NULL, _gloffset_VertexAttrib1dNV),
-    NAME_FUNC_OFFSET(11666, glVertexAttrib1dvNV, glVertexAttrib1dvNV, NULL, _gloffset_VertexAttrib1dvNV),
-    NAME_FUNC_OFFSET(11686, glVertexAttrib1fNV, glVertexAttrib1fNV, NULL, _gloffset_VertexAttrib1fNV),
-    NAME_FUNC_OFFSET(11705, glVertexAttrib1fvNV, glVertexAttrib1fvNV, NULL, _gloffset_VertexAttrib1fvNV),
-    NAME_FUNC_OFFSET(11725, glVertexAttrib1sNV, glVertexAttrib1sNV, NULL, _gloffset_VertexAttrib1sNV),
-    NAME_FUNC_OFFSET(11744, glVertexAttrib1svNV, glVertexAttrib1svNV, NULL, _gloffset_VertexAttrib1svNV),
-    NAME_FUNC_OFFSET(11764, glVertexAttrib2dNV, glVertexAttrib2dNV, NULL, _gloffset_VertexAttrib2dNV),
-    NAME_FUNC_OFFSET(11783, glVertexAttrib2dvNV, glVertexAttrib2dvNV, NULL, _gloffset_VertexAttrib2dvNV),
-    NAME_FUNC_OFFSET(11803, glVertexAttrib2fNV, glVertexAttrib2fNV, NULL, _gloffset_VertexAttrib2fNV),
-    NAME_FUNC_OFFSET(11822, glVertexAttrib2fvNV, glVertexAttrib2fvNV, NULL, _gloffset_VertexAttrib2fvNV),
-    NAME_FUNC_OFFSET(11842, glVertexAttrib2sNV, glVertexAttrib2sNV, NULL, _gloffset_VertexAttrib2sNV),
-    NAME_FUNC_OFFSET(11861, glVertexAttrib2svNV, glVertexAttrib2svNV, NULL, _gloffset_VertexAttrib2svNV),
-    NAME_FUNC_OFFSET(11881, glVertexAttrib3dNV, glVertexAttrib3dNV, NULL, _gloffset_VertexAttrib3dNV),
-    NAME_FUNC_OFFSET(11900, glVertexAttrib3dvNV, glVertexAttrib3dvNV, NULL, _gloffset_VertexAttrib3dvNV),
-    NAME_FUNC_OFFSET(11920, glVertexAttrib3fNV, glVertexAttrib3fNV, NULL, _gloffset_VertexAttrib3fNV),
-    NAME_FUNC_OFFSET(11939, glVertexAttrib3fvNV, glVertexAttrib3fvNV, NULL, _gloffset_VertexAttrib3fvNV),
-    NAME_FUNC_OFFSET(11959, glVertexAttrib3sNV, glVertexAttrib3sNV, NULL, _gloffset_VertexAttrib3sNV),
-    NAME_FUNC_OFFSET(11978, glVertexAttrib3svNV, glVertexAttrib3svNV, NULL, _gloffset_VertexAttrib3svNV),
-    NAME_FUNC_OFFSET(11998, glVertexAttrib4dNV, glVertexAttrib4dNV, NULL, _gloffset_VertexAttrib4dNV),
-    NAME_FUNC_OFFSET(12017, glVertexAttrib4dvNV, glVertexAttrib4dvNV, NULL, _gloffset_VertexAttrib4dvNV),
-    NAME_FUNC_OFFSET(12037, glVertexAttrib4fNV, glVertexAttrib4fNV, NULL, _gloffset_VertexAttrib4fNV),
-    NAME_FUNC_OFFSET(12056, glVertexAttrib4fvNV, glVertexAttrib4fvNV, NULL, _gloffset_VertexAttrib4fvNV),
-    NAME_FUNC_OFFSET(12076, glVertexAttrib4sNV, glVertexAttrib4sNV, NULL, _gloffset_VertexAttrib4sNV),
-    NAME_FUNC_OFFSET(12095, glVertexAttrib4svNV, glVertexAttrib4svNV, NULL, _gloffset_VertexAttrib4svNV),
-    NAME_FUNC_OFFSET(12115, glVertexAttrib4ubNV, glVertexAttrib4ubNV, NULL, _gloffset_VertexAttrib4ubNV),
-    NAME_FUNC_OFFSET(12135, glVertexAttrib4ubvNV, glVertexAttrib4ubvNV, NULL, _gloffset_VertexAttrib4ubvNV),
-    NAME_FUNC_OFFSET(12156, glVertexAttribPointerNV, glVertexAttribPointerNV, NULL, _gloffset_VertexAttribPointerNV),
-    NAME_FUNC_OFFSET(12180, glVertexAttribs1dvNV, glVertexAttribs1dvNV, NULL, _gloffset_VertexAttribs1dvNV),
-    NAME_FUNC_OFFSET(12201, glVertexAttribs1fvNV, glVertexAttribs1fvNV, NULL, _gloffset_VertexAttribs1fvNV),
-    NAME_FUNC_OFFSET(12222, glVertexAttribs1svNV, glVertexAttribs1svNV, NULL, _gloffset_VertexAttribs1svNV),
-    NAME_FUNC_OFFSET(12243, glVertexAttribs2dvNV, glVertexAttribs2dvNV, NULL, _gloffset_VertexAttribs2dvNV),
-    NAME_FUNC_OFFSET(12264, glVertexAttribs2fvNV, glVertexAttribs2fvNV, NULL, _gloffset_VertexAttribs2fvNV),
-    NAME_FUNC_OFFSET(12285, glVertexAttribs2svNV, glVertexAttribs2svNV, NULL, _gloffset_VertexAttribs2svNV),
-    NAME_FUNC_OFFSET(12306, glVertexAttribs3dvNV, glVertexAttribs3dvNV, NULL, _gloffset_VertexAttribs3dvNV),
-    NAME_FUNC_OFFSET(12327, glVertexAttribs3fvNV, glVertexAttribs3fvNV, NULL, _gloffset_VertexAttribs3fvNV),
-    NAME_FUNC_OFFSET(12348, glVertexAttribs3svNV, glVertexAttribs3svNV, NULL, _gloffset_VertexAttribs3svNV),
-    NAME_FUNC_OFFSET(12369, glVertexAttribs4dvNV, glVertexAttribs4dvNV, NULL, _gloffset_VertexAttribs4dvNV),
-    NAME_FUNC_OFFSET(12390, glVertexAttribs4fvNV, glVertexAttribs4fvNV, NULL, _gloffset_VertexAttribs4fvNV),
-    NAME_FUNC_OFFSET(12411, glVertexAttribs4svNV, glVertexAttribs4svNV, NULL, _gloffset_VertexAttribs4svNV),
-    NAME_FUNC_OFFSET(12432, glVertexAttribs4ubvNV, glVertexAttribs4ubvNV, NULL, _gloffset_VertexAttribs4ubvNV),
-    NAME_FUNC_OFFSET(12454, glGetTexBumpParameterfvATI, glGetTexBumpParameterfvATI, NULL, _gloffset_GetTexBumpParameterfvATI),
-    NAME_FUNC_OFFSET(12481, glGetTexBumpParameterivATI, glGetTexBumpParameterivATI, NULL, _gloffset_GetTexBumpParameterivATI),
-    NAME_FUNC_OFFSET(12508, glTexBumpParameterfvATI, glTexBumpParameterfvATI, NULL, _gloffset_TexBumpParameterfvATI),
-    NAME_FUNC_OFFSET(12532, glTexBumpParameterivATI, glTexBumpParameterivATI, NULL, _gloffset_TexBumpParameterivATI),
-    NAME_FUNC_OFFSET(12556, glAlphaFragmentOp1ATI, glAlphaFragmentOp1ATI, NULL, _gloffset_AlphaFragmentOp1ATI),
-    NAME_FUNC_OFFSET(12578, glAlphaFragmentOp2ATI, glAlphaFragmentOp2ATI, NULL, _gloffset_AlphaFragmentOp2ATI),
-    NAME_FUNC_OFFSET(12600, glAlphaFragmentOp3ATI, glAlphaFragmentOp3ATI, NULL, _gloffset_AlphaFragmentOp3ATI),
-    NAME_FUNC_OFFSET(12622, glBeginFragmentShaderATI, glBeginFragmentShaderATI, NULL, _gloffset_BeginFragmentShaderATI),
-    NAME_FUNC_OFFSET(12647, glBindFragmentShaderATI, glBindFragmentShaderATI, NULL, _gloffset_BindFragmentShaderATI),
-    NAME_FUNC_OFFSET(12671, glColorFragmentOp1ATI, glColorFragmentOp1ATI, NULL, _gloffset_ColorFragmentOp1ATI),
-    NAME_FUNC_OFFSET(12693, glColorFragmentOp2ATI, glColorFragmentOp2ATI, NULL, _gloffset_ColorFragmentOp2ATI),
-    NAME_FUNC_OFFSET(12715, glColorFragmentOp3ATI, glColorFragmentOp3ATI, NULL, _gloffset_ColorFragmentOp3ATI),
-    NAME_FUNC_OFFSET(12737, glDeleteFragmentShaderATI, glDeleteFragmentShaderATI, NULL, _gloffset_DeleteFragmentShaderATI),
-    NAME_FUNC_OFFSET(12763, glEndFragmentShaderATI, glEndFragmentShaderATI, NULL, _gloffset_EndFragmentShaderATI),
-    NAME_FUNC_OFFSET(12786, glGenFragmentShadersATI, glGenFragmentShadersATI, NULL, _gloffset_GenFragmentShadersATI),
-    NAME_FUNC_OFFSET(12810, glPassTexCoordATI, glPassTexCoordATI, NULL, _gloffset_PassTexCoordATI),
-    NAME_FUNC_OFFSET(12828, glSampleMapATI, glSampleMapATI, NULL, _gloffset_SampleMapATI),
-    NAME_FUNC_OFFSET(12843, glSetFragmentShaderConstantATI, glSetFragmentShaderConstantATI, NULL, _gloffset_SetFragmentShaderConstantATI),
-    NAME_FUNC_OFFSET(12874, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV),
-    NAME_FUNC_OFFSET(12894, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV),
-    NAME_FUNC_OFFSET(12915, gl_dispatch_stub_750, gl_dispatch_stub_750, NULL, _gloffset_ActiveStencilFaceEXT),
-    NAME_FUNC_OFFSET(12938, gl_dispatch_stub_751, gl_dispatch_stub_751, NULL, _gloffset_BindVertexArrayAPPLE),
-    NAME_FUNC_OFFSET(12961, gl_dispatch_stub_752, gl_dispatch_stub_752, NULL, _gloffset_DeleteVertexArraysAPPLE),
-    NAME_FUNC_OFFSET(12987, gl_dispatch_stub_753, gl_dispatch_stub_753, NULL, _gloffset_GenVertexArraysAPPLE),
-    NAME_FUNC_OFFSET(13010, gl_dispatch_stub_754, gl_dispatch_stub_754, NULL, _gloffset_IsVertexArrayAPPLE),
-    NAME_FUNC_OFFSET(13031, glGetProgramNamedParameterdvNV, glGetProgramNamedParameterdvNV, NULL, _gloffset_GetProgramNamedParameterdvNV),
-    NAME_FUNC_OFFSET(13062, glGetProgramNamedParameterfvNV, glGetProgramNamedParameterfvNV, NULL, _gloffset_GetProgramNamedParameterfvNV),
-    NAME_FUNC_OFFSET(13093, glProgramNamedParameter4dNV, glProgramNamedParameter4dNV, NULL, _gloffset_ProgramNamedParameter4dNV),
-    NAME_FUNC_OFFSET(13121, glProgramNamedParameter4dvNV, glProgramNamedParameter4dvNV, NULL, _gloffset_ProgramNamedParameter4dvNV),
-    NAME_FUNC_OFFSET(13150, glProgramNamedParameter4fNV, glProgramNamedParameter4fNV, NULL, _gloffset_ProgramNamedParameter4fNV),
-    NAME_FUNC_OFFSET(13178, glProgramNamedParameter4fvNV, glProgramNamedParameter4fvNV, NULL, _gloffset_ProgramNamedParameter4fvNV),
-    NAME_FUNC_OFFSET(13207, gl_dispatch_stub_761, gl_dispatch_stub_761, NULL, _gloffset_DepthBoundsEXT),
-    NAME_FUNC_OFFSET(13224, gl_dispatch_stub_762, gl_dispatch_stub_762, NULL, _gloffset_BlendEquationSeparateEXT),
-    NAME_FUNC_OFFSET(13251, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT),
-    NAME_FUNC_OFFSET(13272, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT),
-    NAME_FUNC_OFFSET(13294, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT),
-    NAME_FUNC_OFFSET(13322, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT),
-    NAME_FUNC_OFFSET(13346, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT),
-    NAME_FUNC_OFFSET(13371, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT),
-    NAME_FUNC_OFFSET(13400, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT),
-    NAME_FUNC_OFFSET(13426, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT),
-    NAME_FUNC_OFFSET(13452, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT),
-    NAME_FUNC_OFFSET(13478, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT),
-    NAME_FUNC_OFFSET(13499, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT),
-    NAME_FUNC_OFFSET(13521, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT),
-    NAME_FUNC_OFFSET(13541, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT),
-    NAME_FUNC_OFFSET(13582, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT),
-    NAME_FUNC_OFFSET(13614, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT),
-    NAME_FUNC_OFFSET(13633, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT),
-    NAME_FUNC_OFFSET(13653, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT),
-    NAME_FUNC_OFFSET(13678, gl_dispatch_stub_780, gl_dispatch_stub_780, NULL, _gloffset_BlitFramebufferEXT),
-    NAME_FUNC_OFFSET(13699, gl_dispatch_stub_781, gl_dispatch_stub_781, NULL, _gloffset_BufferParameteriAPPLE),
-    NAME_FUNC_OFFSET(13723, gl_dispatch_stub_782, gl_dispatch_stub_782, NULL, _gloffset_FlushMappedBufferRangeAPPLE),
-    NAME_FUNC_OFFSET(13753, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT),
-    NAME_FUNC_OFFSET(13782, glProvokingVertexEXT, glProvokingVertexEXT, NULL, _gloffset_ProvokingVertexEXT),
-    NAME_FUNC_OFFSET(13803, gl_dispatch_stub_785, gl_dispatch_stub_785, NULL, _gloffset_GetTexParameterPointervAPPLE),
-    NAME_FUNC_OFFSET(13834, gl_dispatch_stub_786, gl_dispatch_stub_786, NULL, _gloffset_TextureRangeAPPLE),
-    NAME_FUNC_OFFSET(13854, gl_dispatch_stub_787, gl_dispatch_stub_787, NULL, _gloffset_StencilFuncSeparateATI),
-    NAME_FUNC_OFFSET(13879, gl_dispatch_stub_788, gl_dispatch_stub_788, NULL, _gloffset_ProgramEnvParameters4fvEXT),
-    NAME_FUNC_OFFSET(13908, gl_dispatch_stub_789, gl_dispatch_stub_789, NULL, _gloffset_ProgramLocalParameters4fvEXT),
-    NAME_FUNC_OFFSET(13939, gl_dispatch_stub_790, gl_dispatch_stub_790, NULL, _gloffset_GetQueryObjecti64vEXT),
-    NAME_FUNC_OFFSET(13963, gl_dispatch_stub_791, gl_dispatch_stub_791, NULL, _gloffset_GetQueryObjectui64vEXT),
-    NAME_FUNC_OFFSET(13988, glArrayElement, glArrayElement, NULL, _gloffset_ArrayElement),
-    NAME_FUNC_OFFSET(14006, glBindTexture, glBindTexture, NULL, _gloffset_BindTexture),
-    NAME_FUNC_OFFSET(14023, glDrawArrays, glDrawArrays, NULL, _gloffset_DrawArrays),
-    NAME_FUNC_OFFSET(14039, glAreTexturesResident, glAreTexturesResidentEXT, glAreTexturesResidentEXT, _gloffset_AreTexturesResident),
-    NAME_FUNC_OFFSET(14064, glCopyTexImage1D, glCopyTexImage1D, NULL, _gloffset_CopyTexImage1D),
-    NAME_FUNC_OFFSET(14084, glCopyTexImage2D, glCopyTexImage2D, NULL, _gloffset_CopyTexImage2D),
-    NAME_FUNC_OFFSET(14104, glCopyTexSubImage1D, glCopyTexSubImage1D, NULL, _gloffset_CopyTexSubImage1D),
-    NAME_FUNC_OFFSET(14127, glCopyTexSubImage2D, glCopyTexSubImage2D, NULL, _gloffset_CopyTexSubImage2D),
-    NAME_FUNC_OFFSET(14150, glDeleteTextures, glDeleteTexturesEXT, glDeleteTexturesEXT, _gloffset_DeleteTextures),
-    NAME_FUNC_OFFSET(14170, glGenTextures, glGenTexturesEXT, glGenTexturesEXT, _gloffset_GenTextures),
-    NAME_FUNC_OFFSET(14187, glGetPointerv, glGetPointerv, NULL, _gloffset_GetPointerv),
-    NAME_FUNC_OFFSET(14204, glIsTexture, glIsTextureEXT, glIsTextureEXT, _gloffset_IsTexture),
-    NAME_FUNC_OFFSET(14219, glPrioritizeTextures, glPrioritizeTextures, NULL, _gloffset_PrioritizeTextures),
-    NAME_FUNC_OFFSET(14243, glTexSubImage1D, glTexSubImage1D, NULL, _gloffset_TexSubImage1D),
-    NAME_FUNC_OFFSET(14262, glTexSubImage2D, glTexSubImage2D, NULL, _gloffset_TexSubImage2D),
-    NAME_FUNC_OFFSET(14281, glBlendColor, glBlendColor, NULL, _gloffset_BlendColor),
-    NAME_FUNC_OFFSET(14297, glBlendEquation, glBlendEquation, NULL, _gloffset_BlendEquation),
-    NAME_FUNC_OFFSET(14316, glDrawRangeElements, glDrawRangeElements, NULL, _gloffset_DrawRangeElements),
-    NAME_FUNC_OFFSET(14339, glColorTable, glColorTable, NULL, _gloffset_ColorTable),
-    NAME_FUNC_OFFSET(14355, glColorTable, glColorTable, NULL, _gloffset_ColorTable),
-    NAME_FUNC_OFFSET(14371, glColorTableParameterfv, glColorTableParameterfv, NULL, _gloffset_ColorTableParameterfv),
-    NAME_FUNC_OFFSET(14398, glColorTableParameteriv, glColorTableParameteriv, NULL, _gloffset_ColorTableParameteriv),
-    NAME_FUNC_OFFSET(14425, glCopyColorTable, glCopyColorTable, NULL, _gloffset_CopyColorTable),
-    NAME_FUNC_OFFSET(14445, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable),
-    NAME_FUNC_OFFSET(14464, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable),
-    NAME_FUNC_OFFSET(14483, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv),
-    NAME_FUNC_OFFSET(14513, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv),
-    NAME_FUNC_OFFSET(14543, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv),
-    NAME_FUNC_OFFSET(14573, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv),
-    NAME_FUNC_OFFSET(14603, glColorSubTable, glColorSubTable, NULL, _gloffset_ColorSubTable),
-    NAME_FUNC_OFFSET(14622, glCopyColorSubTable, glCopyColorSubTable, NULL, _gloffset_CopyColorSubTable),
-    NAME_FUNC_OFFSET(14645, glConvolutionFilter1D, glConvolutionFilter1D, NULL, _gloffset_ConvolutionFilter1D),
-    NAME_FUNC_OFFSET(14670, glConvolutionFilter2D, glConvolutionFilter2D, NULL, _gloffset_ConvolutionFilter2D),
-    NAME_FUNC_OFFSET(14695, glConvolutionParameterf, glConvolutionParameterf, NULL, _gloffset_ConvolutionParameterf),
-    NAME_FUNC_OFFSET(14722, glConvolutionParameterfv, glConvolutionParameterfv, NULL, _gloffset_ConvolutionParameterfv),
-    NAME_FUNC_OFFSET(14750, glConvolutionParameteri, glConvolutionParameteri, NULL, _gloffset_ConvolutionParameteri),
-    NAME_FUNC_OFFSET(14777, glConvolutionParameteriv, glConvolutionParameteriv, NULL, _gloffset_ConvolutionParameteriv),
-    NAME_FUNC_OFFSET(14805, glCopyConvolutionFilter1D, glCopyConvolutionFilter1D, NULL, _gloffset_CopyConvolutionFilter1D),
-    NAME_FUNC_OFFSET(14834, glCopyConvolutionFilter2D, glCopyConvolutionFilter2D, NULL, _gloffset_CopyConvolutionFilter2D),
-    NAME_FUNC_OFFSET(14863, glGetConvolutionFilter, gl_dispatch_stub_356, gl_dispatch_stub_356, _gloffset_GetConvolutionFilter),
-    NAME_FUNC_OFFSET(14889, glGetConvolutionParameterfv, gl_dispatch_stub_357, gl_dispatch_stub_357, _gloffset_GetConvolutionParameterfv),
-    NAME_FUNC_OFFSET(14920, glGetConvolutionParameteriv, gl_dispatch_stub_358, gl_dispatch_stub_358, _gloffset_GetConvolutionParameteriv),
-    NAME_FUNC_OFFSET(14951, glGetSeparableFilter, gl_dispatch_stub_359, gl_dispatch_stub_359, _gloffset_GetSeparableFilter),
-    NAME_FUNC_OFFSET(14975, glSeparableFilter2D, glSeparableFilter2D, NULL, _gloffset_SeparableFilter2D),
-    NAME_FUNC_OFFSET(14998, glGetHistogram, gl_dispatch_stub_361, gl_dispatch_stub_361, _gloffset_GetHistogram),
-    NAME_FUNC_OFFSET(15016, glGetHistogramParameterfv, gl_dispatch_stub_362, gl_dispatch_stub_362, _gloffset_GetHistogramParameterfv),
-    NAME_FUNC_OFFSET(15045, glGetHistogramParameteriv, gl_dispatch_stub_363, gl_dispatch_stub_363, _gloffset_GetHistogramParameteriv),
-    NAME_FUNC_OFFSET(15074, glGetMinmax, gl_dispatch_stub_364, gl_dispatch_stub_364, _gloffset_GetMinmax),
-    NAME_FUNC_OFFSET(15089, glGetMinmaxParameterfv, gl_dispatch_stub_365, gl_dispatch_stub_365, _gloffset_GetMinmaxParameterfv),
-    NAME_FUNC_OFFSET(15115, glGetMinmaxParameteriv, gl_dispatch_stub_366, gl_dispatch_stub_366, _gloffset_GetMinmaxParameteriv),
-    NAME_FUNC_OFFSET(15141, glHistogram, glHistogram, NULL, _gloffset_Histogram),
-    NAME_FUNC_OFFSET(15156, glMinmax, glMinmax, NULL, _gloffset_Minmax),
-    NAME_FUNC_OFFSET(15168, glResetHistogram, glResetHistogram, NULL, _gloffset_ResetHistogram),
-    NAME_FUNC_OFFSET(15188, glResetMinmax, glResetMinmax, NULL, _gloffset_ResetMinmax),
-    NAME_FUNC_OFFSET(15205, glTexImage3D, glTexImage3D, NULL, _gloffset_TexImage3D),
-    NAME_FUNC_OFFSET(15221, glTexSubImage3D, glTexSubImage3D, NULL, _gloffset_TexSubImage3D),
-    NAME_FUNC_OFFSET(15240, glCopyTexSubImage3D, glCopyTexSubImage3D, NULL, _gloffset_CopyTexSubImage3D),
-    NAME_FUNC_OFFSET(15263, glActiveTextureARB, glActiveTextureARB, NULL, _gloffset_ActiveTextureARB),
-    NAME_FUNC_OFFSET(15279, glClientActiveTextureARB, glClientActiveTextureARB, NULL, _gloffset_ClientActiveTextureARB),
-    NAME_FUNC_OFFSET(15301, glMultiTexCoord1dARB, glMultiTexCoord1dARB, NULL, _gloffset_MultiTexCoord1dARB),
-    NAME_FUNC_OFFSET(15319, glMultiTexCoord1dvARB, glMultiTexCoord1dvARB, NULL, _gloffset_MultiTexCoord1dvARB),
-    NAME_FUNC_OFFSET(15338, glMultiTexCoord1fARB, glMultiTexCoord1fARB, NULL, _gloffset_MultiTexCoord1fARB),
-    NAME_FUNC_OFFSET(15356, glMultiTexCoord1fvARB, glMultiTexCoord1fvARB, NULL, _gloffset_MultiTexCoord1fvARB),
-    NAME_FUNC_OFFSET(15375, glMultiTexCoord1iARB, glMultiTexCoord1iARB, NULL, _gloffset_MultiTexCoord1iARB),
-    NAME_FUNC_OFFSET(15393, glMultiTexCoord1ivARB, glMultiTexCoord1ivARB, NULL, _gloffset_MultiTexCoord1ivARB),
-    NAME_FUNC_OFFSET(15412, glMultiTexCoord1sARB, glMultiTexCoord1sARB, NULL, _gloffset_MultiTexCoord1sARB),
-    NAME_FUNC_OFFSET(15430, glMultiTexCoord1svARB, glMultiTexCoord1svARB, NULL, _gloffset_MultiTexCoord1svARB),
-    NAME_FUNC_OFFSET(15449, glMultiTexCoord2dARB, glMultiTexCoord2dARB, NULL, _gloffset_MultiTexCoord2dARB),
-    NAME_FUNC_OFFSET(15467, glMultiTexCoord2dvARB, glMultiTexCoord2dvARB, NULL, _gloffset_MultiTexCoord2dvARB),
-    NAME_FUNC_OFFSET(15486, glMultiTexCoord2fARB, glMultiTexCoord2fARB, NULL, _gloffset_MultiTexCoord2fARB),
-    NAME_FUNC_OFFSET(15504, glMultiTexCoord2fvARB, glMultiTexCoord2fvARB, NULL, _gloffset_MultiTexCoord2fvARB),
-    NAME_FUNC_OFFSET(15523, glMultiTexCoord2iARB, glMultiTexCoord2iARB, NULL, _gloffset_MultiTexCoord2iARB),
-    NAME_FUNC_OFFSET(15541, glMultiTexCoord2ivARB, glMultiTexCoord2ivARB, NULL, _gloffset_MultiTexCoord2ivARB),
-    NAME_FUNC_OFFSET(15560, glMultiTexCoord2sARB, glMultiTexCoord2sARB, NULL, _gloffset_MultiTexCoord2sARB),
-    NAME_FUNC_OFFSET(15578, glMultiTexCoord2svARB, glMultiTexCoord2svARB, NULL, _gloffset_MultiTexCoord2svARB),
-    NAME_FUNC_OFFSET(15597, glMultiTexCoord3dARB, glMultiTexCoord3dARB, NULL, _gloffset_MultiTexCoord3dARB),
-    NAME_FUNC_OFFSET(15615, glMultiTexCoord3dvARB, glMultiTexCoord3dvARB, NULL, _gloffset_MultiTexCoord3dvARB),
-    NAME_FUNC_OFFSET(15634, glMultiTexCoord3fARB, glMultiTexCoord3fARB, NULL, _gloffset_MultiTexCoord3fARB),
-    NAME_FUNC_OFFSET(15652, glMultiTexCoord3fvARB, glMultiTexCoord3fvARB, NULL, _gloffset_MultiTexCoord3fvARB),
-    NAME_FUNC_OFFSET(15671, glMultiTexCoord3iARB, glMultiTexCoord3iARB, NULL, _gloffset_MultiTexCoord3iARB),
-    NAME_FUNC_OFFSET(15689, glMultiTexCoord3ivARB, glMultiTexCoord3ivARB, NULL, _gloffset_MultiTexCoord3ivARB),
-    NAME_FUNC_OFFSET(15708, glMultiTexCoord3sARB, glMultiTexCoord3sARB, NULL, _gloffset_MultiTexCoord3sARB),
-    NAME_FUNC_OFFSET(15726, glMultiTexCoord3svARB, glMultiTexCoord3svARB, NULL, _gloffset_MultiTexCoord3svARB),
-    NAME_FUNC_OFFSET(15745, glMultiTexCoord4dARB, glMultiTexCoord4dARB, NULL, _gloffset_MultiTexCoord4dARB),
-    NAME_FUNC_OFFSET(15763, glMultiTexCoord4dvARB, glMultiTexCoord4dvARB, NULL, _gloffset_MultiTexCoord4dvARB),
-    NAME_FUNC_OFFSET(15782, glMultiTexCoord4fARB, glMultiTexCoord4fARB, NULL, _gloffset_MultiTexCoord4fARB),
-    NAME_FUNC_OFFSET(15800, glMultiTexCoord4fvARB, glMultiTexCoord4fvARB, NULL, _gloffset_MultiTexCoord4fvARB),
-    NAME_FUNC_OFFSET(15819, glMultiTexCoord4iARB, glMultiTexCoord4iARB, NULL, _gloffset_MultiTexCoord4iARB),
-    NAME_FUNC_OFFSET(15837, glMultiTexCoord4ivARB, glMultiTexCoord4ivARB, NULL, _gloffset_MultiTexCoord4ivARB),
-    NAME_FUNC_OFFSET(15856, glMultiTexCoord4sARB, glMultiTexCoord4sARB, NULL, _gloffset_MultiTexCoord4sARB),
-    NAME_FUNC_OFFSET(15874, glMultiTexCoord4svARB, glMultiTexCoord4svARB, NULL, _gloffset_MultiTexCoord4svARB),
-    NAME_FUNC_OFFSET(15893, glStencilOpSeparate, glStencilOpSeparate, NULL, _gloffset_StencilOpSeparate),
-    NAME_FUNC_OFFSET(15916, glLoadTransposeMatrixdARB, glLoadTransposeMatrixdARB, NULL, _gloffset_LoadTransposeMatrixdARB),
-    NAME_FUNC_OFFSET(15939, glLoadTransposeMatrixfARB, glLoadTransposeMatrixfARB, NULL, _gloffset_LoadTransposeMatrixfARB),
-    NAME_FUNC_OFFSET(15962, glMultTransposeMatrixdARB, glMultTransposeMatrixdARB, NULL, _gloffset_MultTransposeMatrixdARB),
-    NAME_FUNC_OFFSET(15985, glMultTransposeMatrixfARB, glMultTransposeMatrixfARB, NULL, _gloffset_MultTransposeMatrixfARB),
-    NAME_FUNC_OFFSET(16008, glSampleCoverageARB, glSampleCoverageARB, NULL, _gloffset_SampleCoverageARB),
-    NAME_FUNC_OFFSET(16025, glCompressedTexImage1DARB, glCompressedTexImage1DARB, NULL, _gloffset_CompressedTexImage1DARB),
-    NAME_FUNC_OFFSET(16048, glCompressedTexImage2DARB, glCompressedTexImage2DARB, NULL, _gloffset_CompressedTexImage2DARB),
-    NAME_FUNC_OFFSET(16071, glCompressedTexImage3DARB, glCompressedTexImage3DARB, NULL, _gloffset_CompressedTexImage3DARB),
-    NAME_FUNC_OFFSET(16094, glCompressedTexSubImage1DARB, glCompressedTexSubImage1DARB, NULL, _gloffset_CompressedTexSubImage1DARB),
-    NAME_FUNC_OFFSET(16120, glCompressedTexSubImage2DARB, glCompressedTexSubImage2DARB, NULL, _gloffset_CompressedTexSubImage2DARB),
-    NAME_FUNC_OFFSET(16146, glCompressedTexSubImage3DARB, glCompressedTexSubImage3DARB, NULL, _gloffset_CompressedTexSubImage3DARB),
-    NAME_FUNC_OFFSET(16172, glGetCompressedTexImageARB, glGetCompressedTexImageARB, NULL, _gloffset_GetCompressedTexImageARB),
-    NAME_FUNC_OFFSET(16196, glDisableVertexAttribArrayARB, glDisableVertexAttribArrayARB, NULL, _gloffset_DisableVertexAttribArrayARB),
-    NAME_FUNC_OFFSET(16223, glEnableVertexAttribArrayARB, glEnableVertexAttribArrayARB, NULL, _gloffset_EnableVertexAttribArrayARB),
-    NAME_FUNC_OFFSET(16249, glGetVertexAttribdvARB, glGetVertexAttribdvARB, NULL, _gloffset_GetVertexAttribdvARB),
-    NAME_FUNC_OFFSET(16269, glGetVertexAttribfvARB, glGetVertexAttribfvARB, NULL, _gloffset_GetVertexAttribfvARB),
-    NAME_FUNC_OFFSET(16289, glGetVertexAttribivARB, glGetVertexAttribivARB, NULL, _gloffset_GetVertexAttribivARB),
-    NAME_FUNC_OFFSET(16309, glProgramEnvParameter4dARB, glProgramEnvParameter4dARB, NULL, _gloffset_ProgramEnvParameter4dARB),
-    NAME_FUNC_OFFSET(16332, glProgramEnvParameter4dvARB, glProgramEnvParameter4dvARB, NULL, _gloffset_ProgramEnvParameter4dvARB),
-    NAME_FUNC_OFFSET(16356, glProgramEnvParameter4fARB, glProgramEnvParameter4fARB, NULL, _gloffset_ProgramEnvParameter4fARB),
-    NAME_FUNC_OFFSET(16379, glProgramEnvParameter4fvARB, glProgramEnvParameter4fvARB, NULL, _gloffset_ProgramEnvParameter4fvARB),
-    NAME_FUNC_OFFSET(16403, glVertexAttrib1dARB, glVertexAttrib1dARB, NULL, _gloffset_VertexAttrib1dARB),
-    NAME_FUNC_OFFSET(16420, glVertexAttrib1dvARB, glVertexAttrib1dvARB, NULL, _gloffset_VertexAttrib1dvARB),
-    NAME_FUNC_OFFSET(16438, glVertexAttrib1fARB, glVertexAttrib1fARB, NULL, _gloffset_VertexAttrib1fARB),
-    NAME_FUNC_OFFSET(16455, glVertexAttrib1fvARB, glVertexAttrib1fvARB, NULL, _gloffset_VertexAttrib1fvARB),
-    NAME_FUNC_OFFSET(16473, glVertexAttrib1sARB, glVertexAttrib1sARB, NULL, _gloffset_VertexAttrib1sARB),
-    NAME_FUNC_OFFSET(16490, glVertexAttrib1svARB, glVertexAttrib1svARB, NULL, _gloffset_VertexAttrib1svARB),
-    NAME_FUNC_OFFSET(16508, glVertexAttrib2dARB, glVertexAttrib2dARB, NULL, _gloffset_VertexAttrib2dARB),
-    NAME_FUNC_OFFSET(16525, glVertexAttrib2dvARB, glVertexAttrib2dvARB, NULL, _gloffset_VertexAttrib2dvARB),
-    NAME_FUNC_OFFSET(16543, glVertexAttrib2fARB, glVertexAttrib2fARB, NULL, _gloffset_VertexAttrib2fARB),
-    NAME_FUNC_OFFSET(16560, glVertexAttrib2fvARB, glVertexAttrib2fvARB, NULL, _gloffset_VertexAttrib2fvARB),
-    NAME_FUNC_OFFSET(16578, glVertexAttrib2sARB, glVertexAttrib2sARB, NULL, _gloffset_VertexAttrib2sARB),
-    NAME_FUNC_OFFSET(16595, glVertexAttrib2svARB, glVertexAttrib2svARB, NULL, _gloffset_VertexAttrib2svARB),
-    NAME_FUNC_OFFSET(16613, glVertexAttrib3dARB, glVertexAttrib3dARB, NULL, _gloffset_VertexAttrib3dARB),
-    NAME_FUNC_OFFSET(16630, glVertexAttrib3dvARB, glVertexAttrib3dvARB, NULL, _gloffset_VertexAttrib3dvARB),
-    NAME_FUNC_OFFSET(16648, glVertexAttrib3fARB, glVertexAttrib3fARB, NULL, _gloffset_VertexAttrib3fARB),
-    NAME_FUNC_OFFSET(16665, glVertexAttrib3fvARB, glVertexAttrib3fvARB, NULL, _gloffset_VertexAttrib3fvARB),
-    NAME_FUNC_OFFSET(16683, glVertexAttrib3sARB, glVertexAttrib3sARB, NULL, _gloffset_VertexAttrib3sARB),
-    NAME_FUNC_OFFSET(16700, glVertexAttrib3svARB, glVertexAttrib3svARB, NULL, _gloffset_VertexAttrib3svARB),
-    NAME_FUNC_OFFSET(16718, glVertexAttrib4NbvARB, glVertexAttrib4NbvARB, NULL, _gloffset_VertexAttrib4NbvARB),
-    NAME_FUNC_OFFSET(16737, glVertexAttrib4NivARB, glVertexAttrib4NivARB, NULL, _gloffset_VertexAttrib4NivARB),
-    NAME_FUNC_OFFSET(16756, glVertexAttrib4NsvARB, glVertexAttrib4NsvARB, NULL, _gloffset_VertexAttrib4NsvARB),
-    NAME_FUNC_OFFSET(16775, glVertexAttrib4NubARB, glVertexAttrib4NubARB, NULL, _gloffset_VertexAttrib4NubARB),
-    NAME_FUNC_OFFSET(16794, glVertexAttrib4NubvARB, glVertexAttrib4NubvARB, NULL, _gloffset_VertexAttrib4NubvARB),
-    NAME_FUNC_OFFSET(16814, glVertexAttrib4NuivARB, glVertexAttrib4NuivARB, NULL, _gloffset_VertexAttrib4NuivARB),
-    NAME_FUNC_OFFSET(16834, glVertexAttrib4NusvARB, glVertexAttrib4NusvARB, NULL, _gloffset_VertexAttrib4NusvARB),
-    NAME_FUNC_OFFSET(16854, glVertexAttrib4bvARB, glVertexAttrib4bvARB, NULL, _gloffset_VertexAttrib4bvARB),
-    NAME_FUNC_OFFSET(16872, glVertexAttrib4dARB, glVertexAttrib4dARB, NULL, _gloffset_VertexAttrib4dARB),
-    NAME_FUNC_OFFSET(16889, glVertexAttrib4dvARB, glVertexAttrib4dvARB, NULL, _gloffset_VertexAttrib4dvARB),
-    NAME_FUNC_OFFSET(16907, glVertexAttrib4fARB, glVertexAttrib4fARB, NULL, _gloffset_VertexAttrib4fARB),
-    NAME_FUNC_OFFSET(16924, glVertexAttrib4fvARB, glVertexAttrib4fvARB, NULL, _gloffset_VertexAttrib4fvARB),
-    NAME_FUNC_OFFSET(16942, glVertexAttrib4ivARB, glVertexAttrib4ivARB, NULL, _gloffset_VertexAttrib4ivARB),
-    NAME_FUNC_OFFSET(16960, glVertexAttrib4sARB, glVertexAttrib4sARB, NULL, _gloffset_VertexAttrib4sARB),
-    NAME_FUNC_OFFSET(16977, glVertexAttrib4svARB, glVertexAttrib4svARB, NULL, _gloffset_VertexAttrib4svARB),
-    NAME_FUNC_OFFSET(16995, glVertexAttrib4ubvARB, glVertexAttrib4ubvARB, NULL, _gloffset_VertexAttrib4ubvARB),
-    NAME_FUNC_OFFSET(17014, glVertexAttrib4uivARB, glVertexAttrib4uivARB, NULL, _gloffset_VertexAttrib4uivARB),
-    NAME_FUNC_OFFSET(17033, glVertexAttrib4usvARB, glVertexAttrib4usvARB, NULL, _gloffset_VertexAttrib4usvARB),
-    NAME_FUNC_OFFSET(17052, glVertexAttribPointerARB, glVertexAttribPointerARB, NULL, _gloffset_VertexAttribPointerARB),
-    NAME_FUNC_OFFSET(17074, glBindBufferARB, glBindBufferARB, NULL, _gloffset_BindBufferARB),
-    NAME_FUNC_OFFSET(17087, glBufferDataARB, glBufferDataARB, NULL, _gloffset_BufferDataARB),
-    NAME_FUNC_OFFSET(17100, glBufferSubDataARB, glBufferSubDataARB, NULL, _gloffset_BufferSubDataARB),
-    NAME_FUNC_OFFSET(17116, glDeleteBuffersARB, glDeleteBuffersARB, NULL, _gloffset_DeleteBuffersARB),
-    NAME_FUNC_OFFSET(17132, glGenBuffersARB, glGenBuffersARB, NULL, _gloffset_GenBuffersARB),
-    NAME_FUNC_OFFSET(17145, glGetBufferParameterivARB, glGetBufferParameterivARB, NULL, _gloffset_GetBufferParameterivARB),
-    NAME_FUNC_OFFSET(17168, glGetBufferPointervARB, glGetBufferPointervARB, NULL, _gloffset_GetBufferPointervARB),
-    NAME_FUNC_OFFSET(17188, glGetBufferSubDataARB, glGetBufferSubDataARB, NULL, _gloffset_GetBufferSubDataARB),
-    NAME_FUNC_OFFSET(17207, glIsBufferARB, glIsBufferARB, NULL, _gloffset_IsBufferARB),
-    NAME_FUNC_OFFSET(17218, glMapBufferARB, glMapBufferARB, NULL, _gloffset_MapBufferARB),
-    NAME_FUNC_OFFSET(17230, glUnmapBufferARB, glUnmapBufferARB, NULL, _gloffset_UnmapBufferARB),
-    NAME_FUNC_OFFSET(17244, glBeginQueryARB, glBeginQueryARB, NULL, _gloffset_BeginQueryARB),
-    NAME_FUNC_OFFSET(17257, glDeleteQueriesARB, glDeleteQueriesARB, NULL, _gloffset_DeleteQueriesARB),
-    NAME_FUNC_OFFSET(17273, glEndQueryARB, glEndQueryARB, NULL, _gloffset_EndQueryARB),
-    NAME_FUNC_OFFSET(17284, glGenQueriesARB, glGenQueriesARB, NULL, _gloffset_GenQueriesARB),
-    NAME_FUNC_OFFSET(17297, glGetQueryObjectivARB, glGetQueryObjectivARB, NULL, _gloffset_GetQueryObjectivARB),
-    NAME_FUNC_OFFSET(17316, glGetQueryObjectuivARB, glGetQueryObjectuivARB, NULL, _gloffset_GetQueryObjectuivARB),
-    NAME_FUNC_OFFSET(17336, glGetQueryivARB, glGetQueryivARB, NULL, _gloffset_GetQueryivARB),
-    NAME_FUNC_OFFSET(17349, glIsQueryARB, glIsQueryARB, NULL, _gloffset_IsQueryARB),
-    NAME_FUNC_OFFSET(17359, glCompileShaderARB, glCompileShaderARB, NULL, _gloffset_CompileShaderARB),
-    NAME_FUNC_OFFSET(17375, glGetActiveUniformARB, glGetActiveUniformARB, NULL, _gloffset_GetActiveUniformARB),
-    NAME_FUNC_OFFSET(17394, glGetShaderSourceARB, glGetShaderSourceARB, NULL, _gloffset_GetShaderSourceARB),
-    NAME_FUNC_OFFSET(17412, glGetUniformLocationARB, glGetUniformLocationARB, NULL, _gloffset_GetUniformLocationARB),
-    NAME_FUNC_OFFSET(17433, glGetUniformfvARB, glGetUniformfvARB, NULL, _gloffset_GetUniformfvARB),
-    NAME_FUNC_OFFSET(17448, glGetUniformivARB, glGetUniformivARB, NULL, _gloffset_GetUniformivARB),
-    NAME_FUNC_OFFSET(17463, glLinkProgramARB, glLinkProgramARB, NULL, _gloffset_LinkProgramARB),
-    NAME_FUNC_OFFSET(17477, glShaderSourceARB, glShaderSourceARB, NULL, _gloffset_ShaderSourceARB),
-    NAME_FUNC_OFFSET(17492, glUniform1fARB, glUniform1fARB, NULL, _gloffset_Uniform1fARB),
-    NAME_FUNC_OFFSET(17504, glUniform1fvARB, glUniform1fvARB, NULL, _gloffset_Uniform1fvARB),
-    NAME_FUNC_OFFSET(17517, glUniform1iARB, glUniform1iARB, NULL, _gloffset_Uniform1iARB),
-    NAME_FUNC_OFFSET(17529, glUniform1ivARB, glUniform1ivARB, NULL, _gloffset_Uniform1ivARB),
-    NAME_FUNC_OFFSET(17542, glUniform2fARB, glUniform2fARB, NULL, _gloffset_Uniform2fARB),
-    NAME_FUNC_OFFSET(17554, glUniform2fvARB, glUniform2fvARB, NULL, _gloffset_Uniform2fvARB),
-    NAME_FUNC_OFFSET(17567, glUniform2iARB, glUniform2iARB, NULL, _gloffset_Uniform2iARB),
-    NAME_FUNC_OFFSET(17579, glUniform2ivARB, glUniform2ivARB, NULL, _gloffset_Uniform2ivARB),
-    NAME_FUNC_OFFSET(17592, glUniform3fARB, glUniform3fARB, NULL, _gloffset_Uniform3fARB),
-    NAME_FUNC_OFFSET(17604, glUniform3fvARB, glUniform3fvARB, NULL, _gloffset_Uniform3fvARB),
-    NAME_FUNC_OFFSET(17617, glUniform3iARB, glUniform3iARB, NULL, _gloffset_Uniform3iARB),
-    NAME_FUNC_OFFSET(17629, glUniform3ivARB, glUniform3ivARB, NULL, _gloffset_Uniform3ivARB),
-    NAME_FUNC_OFFSET(17642, glUniform4fARB, glUniform4fARB, NULL, _gloffset_Uniform4fARB),
-    NAME_FUNC_OFFSET(17654, glUniform4fvARB, glUniform4fvARB, NULL, _gloffset_Uniform4fvARB),
-    NAME_FUNC_OFFSET(17667, glUniform4iARB, glUniform4iARB, NULL, _gloffset_Uniform4iARB),
-    NAME_FUNC_OFFSET(17679, glUniform4ivARB, glUniform4ivARB, NULL, _gloffset_Uniform4ivARB),
-    NAME_FUNC_OFFSET(17692, glUniformMatrix2fvARB, glUniformMatrix2fvARB, NULL, _gloffset_UniformMatrix2fvARB),
-    NAME_FUNC_OFFSET(17711, glUniformMatrix3fvARB, glUniformMatrix3fvARB, NULL, _gloffset_UniformMatrix3fvARB),
-    NAME_FUNC_OFFSET(17730, glUniformMatrix4fvARB, glUniformMatrix4fvARB, NULL, _gloffset_UniformMatrix4fvARB),
-    NAME_FUNC_OFFSET(17749, glUseProgramObjectARB, glUseProgramObjectARB, NULL, _gloffset_UseProgramObjectARB),
-    NAME_FUNC_OFFSET(17762, glValidateProgramARB, glValidateProgramARB, NULL, _gloffset_ValidateProgramARB),
-    NAME_FUNC_OFFSET(17780, glBindAttribLocationARB, glBindAttribLocationARB, NULL, _gloffset_BindAttribLocationARB),
-    NAME_FUNC_OFFSET(17801, glGetActiveAttribARB, glGetActiveAttribARB, NULL, _gloffset_GetActiveAttribARB),
-    NAME_FUNC_OFFSET(17819, glGetAttribLocationARB, glGetAttribLocationARB, NULL, _gloffset_GetAttribLocationARB),
-    NAME_FUNC_OFFSET(17839, glDrawBuffersARB, glDrawBuffersARB, NULL, _gloffset_DrawBuffersARB),
-    NAME_FUNC_OFFSET(17853, glDrawBuffersARB, glDrawBuffersARB, NULL, _gloffset_DrawBuffersARB),
-    NAME_FUNC_OFFSET(17870, gl_dispatch_stub_581, gl_dispatch_stub_581, NULL, _gloffset_SampleMaskSGIS),
-    NAME_FUNC_OFFSET(17886, gl_dispatch_stub_582, gl_dispatch_stub_582, NULL, _gloffset_SamplePatternSGIS),
-    NAME_FUNC_OFFSET(17905, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET(17923, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET(17944, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET(17966, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET(17985, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET(18007, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET(18030, glSecondaryColor3bEXT, glSecondaryColor3bEXT, NULL, _gloffset_SecondaryColor3bEXT),
-    NAME_FUNC_OFFSET(18049, glSecondaryColor3bvEXT, glSecondaryColor3bvEXT, NULL, _gloffset_SecondaryColor3bvEXT),
-    NAME_FUNC_OFFSET(18069, glSecondaryColor3dEXT, glSecondaryColor3dEXT, NULL, _gloffset_SecondaryColor3dEXT),
-    NAME_FUNC_OFFSET(18088, glSecondaryColor3dvEXT, glSecondaryColor3dvEXT, NULL, _gloffset_SecondaryColor3dvEXT),
-    NAME_FUNC_OFFSET(18108, glSecondaryColor3fEXT, glSecondaryColor3fEXT, NULL, _gloffset_SecondaryColor3fEXT),
-    NAME_FUNC_OFFSET(18127, glSecondaryColor3fvEXT, glSecondaryColor3fvEXT, NULL, _gloffset_SecondaryColor3fvEXT),
-    NAME_FUNC_OFFSET(18147, glSecondaryColor3iEXT, glSecondaryColor3iEXT, NULL, _gloffset_SecondaryColor3iEXT),
-    NAME_FUNC_OFFSET(18166, glSecondaryColor3ivEXT, glSecondaryColor3ivEXT, NULL, _gloffset_SecondaryColor3ivEXT),
-    NAME_FUNC_OFFSET(18186, glSecondaryColor3sEXT, glSecondaryColor3sEXT, NULL, _gloffset_SecondaryColor3sEXT),
-    NAME_FUNC_OFFSET(18205, glSecondaryColor3svEXT, glSecondaryColor3svEXT, NULL, _gloffset_SecondaryColor3svEXT),
-    NAME_FUNC_OFFSET(18225, glSecondaryColor3ubEXT, glSecondaryColor3ubEXT, NULL, _gloffset_SecondaryColor3ubEXT),
-    NAME_FUNC_OFFSET(18245, glSecondaryColor3ubvEXT, glSecondaryColor3ubvEXT, NULL, _gloffset_SecondaryColor3ubvEXT),
-    NAME_FUNC_OFFSET(18266, glSecondaryColor3uiEXT, glSecondaryColor3uiEXT, NULL, _gloffset_SecondaryColor3uiEXT),
-    NAME_FUNC_OFFSET(18286, glSecondaryColor3uivEXT, glSecondaryColor3uivEXT, NULL, _gloffset_SecondaryColor3uivEXT),
-    NAME_FUNC_OFFSET(18307, glSecondaryColor3usEXT, glSecondaryColor3usEXT, NULL, _gloffset_SecondaryColor3usEXT),
-    NAME_FUNC_OFFSET(18327, glSecondaryColor3usvEXT, glSecondaryColor3usvEXT, NULL, _gloffset_SecondaryColor3usvEXT),
-    NAME_FUNC_OFFSET(18348, glSecondaryColorPointerEXT, glSecondaryColorPointerEXT, NULL, _gloffset_SecondaryColorPointerEXT),
-    NAME_FUNC_OFFSET(18372, glMultiDrawArraysEXT, glMultiDrawArraysEXT, NULL, _gloffset_MultiDrawArraysEXT),
-    NAME_FUNC_OFFSET(18390, glMultiDrawElementsEXT, glMultiDrawElementsEXT, NULL, _gloffset_MultiDrawElementsEXT),
-    NAME_FUNC_OFFSET(18410, glFogCoordPointerEXT, glFogCoordPointerEXT, NULL, _gloffset_FogCoordPointerEXT),
-    NAME_FUNC_OFFSET(18428, glFogCoorddEXT, glFogCoorddEXT, NULL, _gloffset_FogCoorddEXT),
-    NAME_FUNC_OFFSET(18440, glFogCoorddvEXT, glFogCoorddvEXT, NULL, _gloffset_FogCoorddvEXT),
-    NAME_FUNC_OFFSET(18453, glFogCoordfEXT, glFogCoordfEXT, NULL, _gloffset_FogCoordfEXT),
-    NAME_FUNC_OFFSET(18465, glFogCoordfvEXT, glFogCoordfvEXT, NULL, _gloffset_FogCoordfvEXT),
-    NAME_FUNC_OFFSET(18478, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
-    NAME_FUNC_OFFSET(18498, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
-    NAME_FUNC_OFFSET(18522, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
-    NAME_FUNC_OFFSET(18536, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
-    NAME_FUNC_OFFSET(18553, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
-    NAME_FUNC_OFFSET(18568, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
-    NAME_FUNC_OFFSET(18586, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
-    NAME_FUNC_OFFSET(18600, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
-    NAME_FUNC_OFFSET(18617, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
-    NAME_FUNC_OFFSET(18632, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
-    NAME_FUNC_OFFSET(18650, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
-    NAME_FUNC_OFFSET(18664, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
-    NAME_FUNC_OFFSET(18681, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
-    NAME_FUNC_OFFSET(18696, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
-    NAME_FUNC_OFFSET(18714, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
-    NAME_FUNC_OFFSET(18728, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
-    NAME_FUNC_OFFSET(18745, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
-    NAME_FUNC_OFFSET(18760, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
-    NAME_FUNC_OFFSET(18778, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
-    NAME_FUNC_OFFSET(18792, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
-    NAME_FUNC_OFFSET(18809, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
-    NAME_FUNC_OFFSET(18824, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
-    NAME_FUNC_OFFSET(18842, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
-    NAME_FUNC_OFFSET(18856, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
-    NAME_FUNC_OFFSET(18873, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
-    NAME_FUNC_OFFSET(18888, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
-    NAME_FUNC_OFFSET(18906, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
-    NAME_FUNC_OFFSET(18920, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
-    NAME_FUNC_OFFSET(18937, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
-    NAME_FUNC_OFFSET(18952, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
-    NAME_FUNC_OFFSET(18970, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
-    NAME_FUNC_OFFSET(18984, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
-    NAME_FUNC_OFFSET(19001, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
-    NAME_FUNC_OFFSET(19016, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
-    NAME_FUNC_OFFSET(19034, glBindProgramNV, glBindProgramNV, NULL, _gloffset_BindProgramNV),
-    NAME_FUNC_OFFSET(19051, glDeleteProgramsNV, glDeleteProgramsNV, NULL, _gloffset_DeleteProgramsNV),
-    NAME_FUNC_OFFSET(19071, glGenProgramsNV, glGenProgramsNV, NULL, _gloffset_GenProgramsNV),
-    NAME_FUNC_OFFSET(19088, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
-    NAME_FUNC_OFFSET(19114, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
-    NAME_FUNC_OFFSET(19143, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV),
-    NAME_FUNC_OFFSET(19158, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV),
-    NAME_FUNC_OFFSET(19176, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV),
-    NAME_FUNC_OFFSET(19195, gl_dispatch_stub_752, gl_dispatch_stub_752, NULL, _gloffset_DeleteVertexArraysAPPLE),
-    NAME_FUNC_OFFSET(19216, gl_dispatch_stub_754, gl_dispatch_stub_754, NULL, _gloffset_IsVertexArrayAPPLE),
-    NAME_FUNC_OFFSET(19232, gl_dispatch_stub_762, gl_dispatch_stub_762, NULL, _gloffset_BlendEquationSeparateEXT),
-    NAME_FUNC_OFFSET(19256, gl_dispatch_stub_762, gl_dispatch_stub_762, NULL, _gloffset_BlendEquationSeparateEXT),
-    NAME_FUNC_OFFSET(19283, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT),
-    NAME_FUNC_OFFSET(19301, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT),
-    NAME_FUNC_OFFSET(19320, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT),
-    NAME_FUNC_OFFSET(19345, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT),
-    NAME_FUNC_OFFSET(19366, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT),
-    NAME_FUNC_OFFSET(19388, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT),
-    NAME_FUNC_OFFSET(19414, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT),
-    NAME_FUNC_OFFSET(19437, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT),
-    NAME_FUNC_OFFSET(19460, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT),
-    NAME_FUNC_OFFSET(19483, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT),
-    NAME_FUNC_OFFSET(19501, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT),
-    NAME_FUNC_OFFSET(19520, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT),
-    NAME_FUNC_OFFSET(19537, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT),
-    NAME_FUNC_OFFSET(19575, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT),
-    NAME_FUNC_OFFSET(19604, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT),
-    NAME_FUNC_OFFSET(19620, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT),
-    NAME_FUNC_OFFSET(19637, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT),
-    NAME_FUNC_OFFSET(19659, gl_dispatch_stub_780, gl_dispatch_stub_780, NULL, _gloffset_BlitFramebufferEXT),
-    NAME_FUNC_OFFSET(19677, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT),
+    NAME_FUNC_OFFSET( 9172, glDrawElementsBaseVertex, glDrawElementsBaseVertex, NULL, _gloffset_DrawElementsBaseVertex),
+    NAME_FUNC_OFFSET( 9197, glDrawRangeElementsBaseVertex, glDrawRangeElementsBaseVertex, NULL, _gloffset_DrawRangeElementsBaseVertex),
+    NAME_FUNC_OFFSET( 9227, glMultiDrawElementsBaseVertex, glMultiDrawElementsBaseVertex, NULL, _gloffset_MultiDrawElementsBaseVertex),
+    NAME_FUNC_OFFSET( 9257, glPolygonOffsetEXT, glPolygonOffsetEXT, NULL, _gloffset_PolygonOffsetEXT),
+    NAME_FUNC_OFFSET( 9276, gl_dispatch_stub_578, gl_dispatch_stub_578, NULL, _gloffset_GetPixelTexGenParameterfvSGIS),
+    NAME_FUNC_OFFSET( 9308, gl_dispatch_stub_579, gl_dispatch_stub_579, NULL, _gloffset_GetPixelTexGenParameterivSGIS),
+    NAME_FUNC_OFFSET( 9340, gl_dispatch_stub_580, gl_dispatch_stub_580, NULL, _gloffset_PixelTexGenParameterfSGIS),
+    NAME_FUNC_OFFSET( 9368, gl_dispatch_stub_581, gl_dispatch_stub_581, NULL, _gloffset_PixelTexGenParameterfvSGIS),
+    NAME_FUNC_OFFSET( 9397, gl_dispatch_stub_582, gl_dispatch_stub_582, NULL, _gloffset_PixelTexGenParameteriSGIS),
+    NAME_FUNC_OFFSET( 9425, gl_dispatch_stub_583, gl_dispatch_stub_583, NULL, _gloffset_PixelTexGenParameterivSGIS),
+    NAME_FUNC_OFFSET( 9454, gl_dispatch_stub_584, gl_dispatch_stub_584, NULL, _gloffset_SampleMaskSGIS),
+    NAME_FUNC_OFFSET( 9471, gl_dispatch_stub_585, gl_dispatch_stub_585, NULL, _gloffset_SamplePatternSGIS),
+    NAME_FUNC_OFFSET( 9491, glColorPointerEXT, glColorPointerEXT, NULL, _gloffset_ColorPointerEXT),
+    NAME_FUNC_OFFSET( 9509, glEdgeFlagPointerEXT, glEdgeFlagPointerEXT, NULL, _gloffset_EdgeFlagPointerEXT),
+    NAME_FUNC_OFFSET( 9530, glIndexPointerEXT, glIndexPointerEXT, NULL, _gloffset_IndexPointerEXT),
+    NAME_FUNC_OFFSET( 9548, glNormalPointerEXT, glNormalPointerEXT, NULL, _gloffset_NormalPointerEXT),
+    NAME_FUNC_OFFSET( 9567, glTexCoordPointerEXT, glTexCoordPointerEXT, NULL, _gloffset_TexCoordPointerEXT),
+    NAME_FUNC_OFFSET( 9588, glVertexPointerEXT, glVertexPointerEXT, NULL, _gloffset_VertexPointerEXT),
+    NAME_FUNC_OFFSET( 9607, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET( 9628, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET( 9650, glLockArraysEXT, glLockArraysEXT, NULL, _gloffset_LockArraysEXT),
+    NAME_FUNC_OFFSET( 9666, glUnlockArraysEXT, glUnlockArraysEXT, NULL, _gloffset_UnlockArraysEXT),
+    NAME_FUNC_OFFSET( 9684, gl_dispatch_stub_596, gl_dispatch_stub_596, NULL, _gloffset_CullParameterdvEXT),
+    NAME_FUNC_OFFSET( 9705, gl_dispatch_stub_597, gl_dispatch_stub_597, NULL, _gloffset_CullParameterfvEXT),
+    NAME_FUNC_OFFSET( 9726, glSecondaryColor3bEXT, glSecondaryColor3bEXT, NULL, _gloffset_SecondaryColor3bEXT),
+    NAME_FUNC_OFFSET( 9748, glSecondaryColor3bvEXT, glSecondaryColor3bvEXT, NULL, _gloffset_SecondaryColor3bvEXT),
+    NAME_FUNC_OFFSET( 9771, glSecondaryColor3dEXT, glSecondaryColor3dEXT, NULL, _gloffset_SecondaryColor3dEXT),
+    NAME_FUNC_OFFSET( 9793, glSecondaryColor3dvEXT, glSecondaryColor3dvEXT, NULL, _gloffset_SecondaryColor3dvEXT),
+    NAME_FUNC_OFFSET( 9816, glSecondaryColor3fEXT, glSecondaryColor3fEXT, NULL, _gloffset_SecondaryColor3fEXT),
+    NAME_FUNC_OFFSET( 9838, glSecondaryColor3fvEXT, glSecondaryColor3fvEXT, NULL, _gloffset_SecondaryColor3fvEXT),
+    NAME_FUNC_OFFSET( 9861, glSecondaryColor3iEXT, glSecondaryColor3iEXT, NULL, _gloffset_SecondaryColor3iEXT),
+    NAME_FUNC_OFFSET( 9883, glSecondaryColor3ivEXT, glSecondaryColor3ivEXT, NULL, _gloffset_SecondaryColor3ivEXT),
+    NAME_FUNC_OFFSET( 9906, glSecondaryColor3sEXT, glSecondaryColor3sEXT, NULL, _gloffset_SecondaryColor3sEXT),
+    NAME_FUNC_OFFSET( 9928, glSecondaryColor3svEXT, glSecondaryColor3svEXT, NULL, _gloffset_SecondaryColor3svEXT),
+    NAME_FUNC_OFFSET( 9951, glSecondaryColor3ubEXT, glSecondaryColor3ubEXT, NULL, _gloffset_SecondaryColor3ubEXT),
+    NAME_FUNC_OFFSET( 9974, glSecondaryColor3ubvEXT, glSecondaryColor3ubvEXT, NULL, _gloffset_SecondaryColor3ubvEXT),
+    NAME_FUNC_OFFSET( 9998, glSecondaryColor3uiEXT, glSecondaryColor3uiEXT, NULL, _gloffset_SecondaryColor3uiEXT),
+    NAME_FUNC_OFFSET(10021, glSecondaryColor3uivEXT, glSecondaryColor3uivEXT, NULL, _gloffset_SecondaryColor3uivEXT),
+    NAME_FUNC_OFFSET(10045, glSecondaryColor3usEXT, glSecondaryColor3usEXT, NULL, _gloffset_SecondaryColor3usEXT),
+    NAME_FUNC_OFFSET(10068, glSecondaryColor3usvEXT, glSecondaryColor3usvEXT, NULL, _gloffset_SecondaryColor3usvEXT),
+    NAME_FUNC_OFFSET(10092, glSecondaryColorPointerEXT, glSecondaryColorPointerEXT, NULL, _gloffset_SecondaryColorPointerEXT),
+    NAME_FUNC_OFFSET(10119, glMultiDrawArraysEXT, glMultiDrawArraysEXT, NULL, _gloffset_MultiDrawArraysEXT),
+    NAME_FUNC_OFFSET(10140, glMultiDrawElementsEXT, glMultiDrawElementsEXT, NULL, _gloffset_MultiDrawElementsEXT),
+    NAME_FUNC_OFFSET(10163, glFogCoordPointerEXT, glFogCoordPointerEXT, NULL, _gloffset_FogCoordPointerEXT),
+    NAME_FUNC_OFFSET(10184, glFogCoorddEXT, glFogCoorddEXT, NULL, _gloffset_FogCoorddEXT),
+    NAME_FUNC_OFFSET(10199, glFogCoorddvEXT, glFogCoorddvEXT, NULL, _gloffset_FogCoorddvEXT),
+    NAME_FUNC_OFFSET(10215, glFogCoordfEXT, glFogCoordfEXT, NULL, _gloffset_FogCoordfEXT),
+    NAME_FUNC_OFFSET(10230, glFogCoordfvEXT, glFogCoordfvEXT, NULL, _gloffset_FogCoordfvEXT),
+    NAME_FUNC_OFFSET(10246, gl_dispatch_stub_622, gl_dispatch_stub_622, NULL, _gloffset_PixelTexGenSGIX),
+    NAME_FUNC_OFFSET(10264, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
+    NAME_FUNC_OFFSET(10287, glFlushVertexArrayRangeNV, glFlushVertexArrayRangeNV, NULL, _gloffset_FlushVertexArrayRangeNV),
+    NAME_FUNC_OFFSET(10313, glVertexArrayRangeNV, glVertexArrayRangeNV, NULL, _gloffset_VertexArrayRangeNV),
+    NAME_FUNC_OFFSET(10334, glCombinerInputNV, glCombinerInputNV, NULL, _gloffset_CombinerInputNV),
+    NAME_FUNC_OFFSET(10352, glCombinerOutputNV, glCombinerOutputNV, NULL, _gloffset_CombinerOutputNV),
+    NAME_FUNC_OFFSET(10371, glCombinerParameterfNV, glCombinerParameterfNV, NULL, _gloffset_CombinerParameterfNV),
+    NAME_FUNC_OFFSET(10394, glCombinerParameterfvNV, glCombinerParameterfvNV, NULL, _gloffset_CombinerParameterfvNV),
+    NAME_FUNC_OFFSET(10418, glCombinerParameteriNV, glCombinerParameteriNV, NULL, _gloffset_CombinerParameteriNV),
+    NAME_FUNC_OFFSET(10441, glCombinerParameterivNV, glCombinerParameterivNV, NULL, _gloffset_CombinerParameterivNV),
+    NAME_FUNC_OFFSET(10465, glFinalCombinerInputNV, glFinalCombinerInputNV, NULL, _gloffset_FinalCombinerInputNV),
+    NAME_FUNC_OFFSET(10488, glGetCombinerInputParameterfvNV, glGetCombinerInputParameterfvNV, NULL, _gloffset_GetCombinerInputParameterfvNV),
+    NAME_FUNC_OFFSET(10520, glGetCombinerInputParameterivNV, glGetCombinerInputParameterivNV, NULL, _gloffset_GetCombinerInputParameterivNV),
+    NAME_FUNC_OFFSET(10552, glGetCombinerOutputParameterfvNV, glGetCombinerOutputParameterfvNV, NULL, _gloffset_GetCombinerOutputParameterfvNV),
+    NAME_FUNC_OFFSET(10585, glGetCombinerOutputParameterivNV, glGetCombinerOutputParameterivNV, NULL, _gloffset_GetCombinerOutputParameterivNV),
+    NAME_FUNC_OFFSET(10618, glGetFinalCombinerInputParameterfvNV, glGetFinalCombinerInputParameterfvNV, NULL, _gloffset_GetFinalCombinerInputParameterfvNV),
+    NAME_FUNC_OFFSET(10655, glGetFinalCombinerInputParameterivNV, glGetFinalCombinerInputParameterivNV, NULL, _gloffset_GetFinalCombinerInputParameterivNV),
+    NAME_FUNC_OFFSET(10692, glResizeBuffersMESA, glResizeBuffersMESA, NULL, _gloffset_ResizeBuffersMESA),
+    NAME_FUNC_OFFSET(10712, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
+    NAME_FUNC_OFFSET(10730, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
+    NAME_FUNC_OFFSET(10749, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
+    NAME_FUNC_OFFSET(10767, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
+    NAME_FUNC_OFFSET(10786, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
+    NAME_FUNC_OFFSET(10804, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
+    NAME_FUNC_OFFSET(10823, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
+    NAME_FUNC_OFFSET(10841, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
+    NAME_FUNC_OFFSET(10860, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
+    NAME_FUNC_OFFSET(10878, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
+    NAME_FUNC_OFFSET(10897, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
+    NAME_FUNC_OFFSET(10915, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
+    NAME_FUNC_OFFSET(10934, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
+    NAME_FUNC_OFFSET(10952, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
+    NAME_FUNC_OFFSET(10971, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
+    NAME_FUNC_OFFSET(10989, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
+    NAME_FUNC_OFFSET(11008, glWindowPos4dMESA, glWindowPos4dMESA, NULL, _gloffset_WindowPos4dMESA),
+    NAME_FUNC_OFFSET(11026, glWindowPos4dvMESA, glWindowPos4dvMESA, NULL, _gloffset_WindowPos4dvMESA),
+    NAME_FUNC_OFFSET(11045, glWindowPos4fMESA, glWindowPos4fMESA, NULL, _gloffset_WindowPos4fMESA),
+    NAME_FUNC_OFFSET(11063, glWindowPos4fvMESA, glWindowPos4fvMESA, NULL, _gloffset_WindowPos4fvMESA),
+    NAME_FUNC_OFFSET(11082, glWindowPos4iMESA, glWindowPos4iMESA, NULL, _gloffset_WindowPos4iMESA),
+    NAME_FUNC_OFFSET(11100, glWindowPos4ivMESA, glWindowPos4ivMESA, NULL, _gloffset_WindowPos4ivMESA),
+    NAME_FUNC_OFFSET(11119, glWindowPos4sMESA, glWindowPos4sMESA, NULL, _gloffset_WindowPos4sMESA),
+    NAME_FUNC_OFFSET(11137, glWindowPos4svMESA, glWindowPos4svMESA, NULL, _gloffset_WindowPos4svMESA),
+    NAME_FUNC_OFFSET(11156, gl_dispatch_stub_664, gl_dispatch_stub_664, NULL, _gloffset_MultiModeDrawArraysIBM),
+    NAME_FUNC_OFFSET(11181, gl_dispatch_stub_665, gl_dispatch_stub_665, NULL, _gloffset_MultiModeDrawElementsIBM),
+    NAME_FUNC_OFFSET(11208, gl_dispatch_stub_666, gl_dispatch_stub_666, NULL, _gloffset_DeleteFencesNV),
+    NAME_FUNC_OFFSET(11225, gl_dispatch_stub_667, gl_dispatch_stub_667, NULL, _gloffset_FinishFenceNV),
+    NAME_FUNC_OFFSET(11241, gl_dispatch_stub_668, gl_dispatch_stub_668, NULL, _gloffset_GenFencesNV),
+    NAME_FUNC_OFFSET(11255, gl_dispatch_stub_669, gl_dispatch_stub_669, NULL, _gloffset_GetFenceivNV),
+    NAME_FUNC_OFFSET(11270, gl_dispatch_stub_670, gl_dispatch_stub_670, NULL, _gloffset_IsFenceNV),
+    NAME_FUNC_OFFSET(11282, gl_dispatch_stub_671, gl_dispatch_stub_671, NULL, _gloffset_SetFenceNV),
+    NAME_FUNC_OFFSET(11295, gl_dispatch_stub_672, gl_dispatch_stub_672, NULL, _gloffset_TestFenceNV),
+    NAME_FUNC_OFFSET(11309, glAreProgramsResidentNV, glAreProgramsResidentNV, NULL, _gloffset_AreProgramsResidentNV),
+    NAME_FUNC_OFFSET(11333, glBindProgramNV, glBindProgramNV, NULL, _gloffset_BindProgramNV),
+    NAME_FUNC_OFFSET(11349, glDeleteProgramsNV, glDeleteProgramsNV, NULL, _gloffset_DeleteProgramsNV),
+    NAME_FUNC_OFFSET(11368, glExecuteProgramNV, glExecuteProgramNV, NULL, _gloffset_ExecuteProgramNV),
+    NAME_FUNC_OFFSET(11387, glGenProgramsNV, glGenProgramsNV, NULL, _gloffset_GenProgramsNV),
+    NAME_FUNC_OFFSET(11403, glGetProgramParameterdvNV, glGetProgramParameterdvNV, NULL, _gloffset_GetProgramParameterdvNV),
+    NAME_FUNC_OFFSET(11429, glGetProgramParameterfvNV, glGetProgramParameterfvNV, NULL, _gloffset_GetProgramParameterfvNV),
+    NAME_FUNC_OFFSET(11455, glGetProgramStringNV, glGetProgramStringNV, NULL, _gloffset_GetProgramStringNV),
+    NAME_FUNC_OFFSET(11476, glGetProgramivNV, glGetProgramivNV, NULL, _gloffset_GetProgramivNV),
+    NAME_FUNC_OFFSET(11493, glGetTrackMatrixivNV, glGetTrackMatrixivNV, NULL, _gloffset_GetTrackMatrixivNV),
+    NAME_FUNC_OFFSET(11514, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
+    NAME_FUNC_OFFSET(11542, glGetVertexAttribdvNV, glGetVertexAttribdvNV, NULL, _gloffset_GetVertexAttribdvNV),
+    NAME_FUNC_OFFSET(11564, glGetVertexAttribfvNV, glGetVertexAttribfvNV, NULL, _gloffset_GetVertexAttribfvNV),
+    NAME_FUNC_OFFSET(11586, glGetVertexAttribivNV, glGetVertexAttribivNV, NULL, _gloffset_GetVertexAttribivNV),
+    NAME_FUNC_OFFSET(11608, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV),
+    NAME_FUNC_OFFSET(11622, glLoadProgramNV, glLoadProgramNV, NULL, _gloffset_LoadProgramNV),
+    NAME_FUNC_OFFSET(11638, glProgramParameters4dvNV, glProgramParameters4dvNV, NULL, _gloffset_ProgramParameters4dvNV),
+    NAME_FUNC_OFFSET(11663, glProgramParameters4fvNV, glProgramParameters4fvNV, NULL, _gloffset_ProgramParameters4fvNV),
+    NAME_FUNC_OFFSET(11688, glRequestResidentProgramsNV, glRequestResidentProgramsNV, NULL, _gloffset_RequestResidentProgramsNV),
+    NAME_FUNC_OFFSET(11716, glTrackMatrixNV, glTrackMatrixNV, NULL, _gloffset_TrackMatrixNV),
+    NAME_FUNC_OFFSET(11732, glVertexAttrib1dNV, glVertexAttrib1dNV, NULL, _gloffset_VertexAttrib1dNV),
+    NAME_FUNC_OFFSET(11751, glVertexAttrib1dvNV, glVertexAttrib1dvNV, NULL, _gloffset_VertexAttrib1dvNV),
+    NAME_FUNC_OFFSET(11771, glVertexAttrib1fNV, glVertexAttrib1fNV, NULL, _gloffset_VertexAttrib1fNV),
+    NAME_FUNC_OFFSET(11790, glVertexAttrib1fvNV, glVertexAttrib1fvNV, NULL, _gloffset_VertexAttrib1fvNV),
+    NAME_FUNC_OFFSET(11810, glVertexAttrib1sNV, glVertexAttrib1sNV, NULL, _gloffset_VertexAttrib1sNV),
+    NAME_FUNC_OFFSET(11829, glVertexAttrib1svNV, glVertexAttrib1svNV, NULL, _gloffset_VertexAttrib1svNV),
+    NAME_FUNC_OFFSET(11849, glVertexAttrib2dNV, glVertexAttrib2dNV, NULL, _gloffset_VertexAttrib2dNV),
+    NAME_FUNC_OFFSET(11868, glVertexAttrib2dvNV, glVertexAttrib2dvNV, NULL, _gloffset_VertexAttrib2dvNV),
+    NAME_FUNC_OFFSET(11888, glVertexAttrib2fNV, glVertexAttrib2fNV, NULL, _gloffset_VertexAttrib2fNV),
+    NAME_FUNC_OFFSET(11907, glVertexAttrib2fvNV, glVertexAttrib2fvNV, NULL, _gloffset_VertexAttrib2fvNV),
+    NAME_FUNC_OFFSET(11927, glVertexAttrib2sNV, glVertexAttrib2sNV, NULL, _gloffset_VertexAttrib2sNV),
+    NAME_FUNC_OFFSET(11946, glVertexAttrib2svNV, glVertexAttrib2svNV, NULL, _gloffset_VertexAttrib2svNV),
+    NAME_FUNC_OFFSET(11966, glVertexAttrib3dNV, glVertexAttrib3dNV, NULL, _gloffset_VertexAttrib3dNV),
+    NAME_FUNC_OFFSET(11985, glVertexAttrib3dvNV, glVertexAttrib3dvNV, NULL, _gloffset_VertexAttrib3dvNV),
+    NAME_FUNC_OFFSET(12005, glVertexAttrib3fNV, glVertexAttrib3fNV, NULL, _gloffset_VertexAttrib3fNV),
+    NAME_FUNC_OFFSET(12024, glVertexAttrib3fvNV, glVertexAttrib3fvNV, NULL, _gloffset_VertexAttrib3fvNV),
+    NAME_FUNC_OFFSET(12044, glVertexAttrib3sNV, glVertexAttrib3sNV, NULL, _gloffset_VertexAttrib3sNV),
+    NAME_FUNC_OFFSET(12063, glVertexAttrib3svNV, glVertexAttrib3svNV, NULL, _gloffset_VertexAttrib3svNV),
+    NAME_FUNC_OFFSET(12083, glVertexAttrib4dNV, glVertexAttrib4dNV, NULL, _gloffset_VertexAttrib4dNV),
+    NAME_FUNC_OFFSET(12102, glVertexAttrib4dvNV, glVertexAttrib4dvNV, NULL, _gloffset_VertexAttrib4dvNV),
+    NAME_FUNC_OFFSET(12122, glVertexAttrib4fNV, glVertexAttrib4fNV, NULL, _gloffset_VertexAttrib4fNV),
+    NAME_FUNC_OFFSET(12141, glVertexAttrib4fvNV, glVertexAttrib4fvNV, NULL, _gloffset_VertexAttrib4fvNV),
+    NAME_FUNC_OFFSET(12161, glVertexAttrib4sNV, glVertexAttrib4sNV, NULL, _gloffset_VertexAttrib4sNV),
+    NAME_FUNC_OFFSET(12180, glVertexAttrib4svNV, glVertexAttrib4svNV, NULL, _gloffset_VertexAttrib4svNV),
+    NAME_FUNC_OFFSET(12200, glVertexAttrib4ubNV, glVertexAttrib4ubNV, NULL, _gloffset_VertexAttrib4ubNV),
+    NAME_FUNC_OFFSET(12220, glVertexAttrib4ubvNV, glVertexAttrib4ubvNV, NULL, _gloffset_VertexAttrib4ubvNV),
+    NAME_FUNC_OFFSET(12241, glVertexAttribPointerNV, glVertexAttribPointerNV, NULL, _gloffset_VertexAttribPointerNV),
+    NAME_FUNC_OFFSET(12265, glVertexAttribs1dvNV, glVertexAttribs1dvNV, NULL, _gloffset_VertexAttribs1dvNV),
+    NAME_FUNC_OFFSET(12286, glVertexAttribs1fvNV, glVertexAttribs1fvNV, NULL, _gloffset_VertexAttribs1fvNV),
+    NAME_FUNC_OFFSET(12307, glVertexAttribs1svNV, glVertexAttribs1svNV, NULL, _gloffset_VertexAttribs1svNV),
+    NAME_FUNC_OFFSET(12328, glVertexAttribs2dvNV, glVertexAttribs2dvNV, NULL, _gloffset_VertexAttribs2dvNV),
+    NAME_FUNC_OFFSET(12349, glVertexAttribs2fvNV, glVertexAttribs2fvNV, NULL, _gloffset_VertexAttribs2fvNV),
+    NAME_FUNC_OFFSET(12370, glVertexAttribs2svNV, glVertexAttribs2svNV, NULL, _gloffset_VertexAttribs2svNV),
+    NAME_FUNC_OFFSET(12391, glVertexAttribs3dvNV, glVertexAttribs3dvNV, NULL, _gloffset_VertexAttribs3dvNV),
+    NAME_FUNC_OFFSET(12412, glVertexAttribs3fvNV, glVertexAttribs3fvNV, NULL, _gloffset_VertexAttribs3fvNV),
+    NAME_FUNC_OFFSET(12433, glVertexAttribs3svNV, glVertexAttribs3svNV, NULL, _gloffset_VertexAttribs3svNV),
+    NAME_FUNC_OFFSET(12454, glVertexAttribs4dvNV, glVertexAttribs4dvNV, NULL, _gloffset_VertexAttribs4dvNV),
+    NAME_FUNC_OFFSET(12475, glVertexAttribs4fvNV, glVertexAttribs4fvNV, NULL, _gloffset_VertexAttribs4fvNV),
+    NAME_FUNC_OFFSET(12496, glVertexAttribs4svNV, glVertexAttribs4svNV, NULL, _gloffset_VertexAttribs4svNV),
+    NAME_FUNC_OFFSET(12517, glVertexAttribs4ubvNV, glVertexAttribs4ubvNV, NULL, _gloffset_VertexAttribs4ubvNV),
+    NAME_FUNC_OFFSET(12539, glGetTexBumpParameterfvATI, glGetTexBumpParameterfvATI, NULL, _gloffset_GetTexBumpParameterfvATI),
+    NAME_FUNC_OFFSET(12566, glGetTexBumpParameterivATI, glGetTexBumpParameterivATI, NULL, _gloffset_GetTexBumpParameterivATI),
+    NAME_FUNC_OFFSET(12593, glTexBumpParameterfvATI, glTexBumpParameterfvATI, NULL, _gloffset_TexBumpParameterfvATI),
+    NAME_FUNC_OFFSET(12617, glTexBumpParameterivATI, glTexBumpParameterivATI, NULL, _gloffset_TexBumpParameterivATI),
+    NAME_FUNC_OFFSET(12641, glAlphaFragmentOp1ATI, glAlphaFragmentOp1ATI, NULL, _gloffset_AlphaFragmentOp1ATI),
+    NAME_FUNC_OFFSET(12663, glAlphaFragmentOp2ATI, glAlphaFragmentOp2ATI, NULL, _gloffset_AlphaFragmentOp2ATI),
+    NAME_FUNC_OFFSET(12685, glAlphaFragmentOp3ATI, glAlphaFragmentOp3ATI, NULL, _gloffset_AlphaFragmentOp3ATI),
+    NAME_FUNC_OFFSET(12707, glBeginFragmentShaderATI, glBeginFragmentShaderATI, NULL, _gloffset_BeginFragmentShaderATI),
+    NAME_FUNC_OFFSET(12732, glBindFragmentShaderATI, glBindFragmentShaderATI, NULL, _gloffset_BindFragmentShaderATI),
+    NAME_FUNC_OFFSET(12756, glColorFragmentOp1ATI, glColorFragmentOp1ATI, NULL, _gloffset_ColorFragmentOp1ATI),
+    NAME_FUNC_OFFSET(12778, glColorFragmentOp2ATI, glColorFragmentOp2ATI, NULL, _gloffset_ColorFragmentOp2ATI),
+    NAME_FUNC_OFFSET(12800, glColorFragmentOp3ATI, glColorFragmentOp3ATI, NULL, _gloffset_ColorFragmentOp3ATI),
+    NAME_FUNC_OFFSET(12822, glDeleteFragmentShaderATI, glDeleteFragmentShaderATI, NULL, _gloffset_DeleteFragmentShaderATI),
+    NAME_FUNC_OFFSET(12848, glEndFragmentShaderATI, glEndFragmentShaderATI, NULL, _gloffset_EndFragmentShaderATI),
+    NAME_FUNC_OFFSET(12871, glGenFragmentShadersATI, glGenFragmentShadersATI, NULL, _gloffset_GenFragmentShadersATI),
+    NAME_FUNC_OFFSET(12895, glPassTexCoordATI, glPassTexCoordATI, NULL, _gloffset_PassTexCoordATI),
+    NAME_FUNC_OFFSET(12913, glSampleMapATI, glSampleMapATI, NULL, _gloffset_SampleMapATI),
+    NAME_FUNC_OFFSET(12928, glSetFragmentShaderConstantATI, glSetFragmentShaderConstantATI, NULL, _gloffset_SetFragmentShaderConstantATI),
+    NAME_FUNC_OFFSET(12959, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV),
+    NAME_FUNC_OFFSET(12979, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV),
+    NAME_FUNC_OFFSET(13000, gl_dispatch_stub_753, gl_dispatch_stub_753, NULL, _gloffset_ActiveStencilFaceEXT),
+    NAME_FUNC_OFFSET(13023, gl_dispatch_stub_754, gl_dispatch_stub_754, NULL, _gloffset_BindVertexArrayAPPLE),
+    NAME_FUNC_OFFSET(13046, gl_dispatch_stub_755, gl_dispatch_stub_755, NULL, _gloffset_DeleteVertexArraysAPPLE),
+    NAME_FUNC_OFFSET(13072, gl_dispatch_stub_756, gl_dispatch_stub_756, NULL, _gloffset_GenVertexArraysAPPLE),
+    NAME_FUNC_OFFSET(13095, gl_dispatch_stub_757, gl_dispatch_stub_757, NULL, _gloffset_IsVertexArrayAPPLE),
+    NAME_FUNC_OFFSET(13116, glGetProgramNamedParameterdvNV, glGetProgramNamedParameterdvNV, NULL, _gloffset_GetProgramNamedParameterdvNV),
+    NAME_FUNC_OFFSET(13147, glGetProgramNamedParameterfvNV, glGetProgramNamedParameterfvNV, NULL, _gloffset_GetProgramNamedParameterfvNV),
+    NAME_FUNC_OFFSET(13178, glProgramNamedParameter4dNV, glProgramNamedParameter4dNV, NULL, _gloffset_ProgramNamedParameter4dNV),
+    NAME_FUNC_OFFSET(13206, glProgramNamedParameter4dvNV, glProgramNamedParameter4dvNV, NULL, _gloffset_ProgramNamedParameter4dvNV),
+    NAME_FUNC_OFFSET(13235, glProgramNamedParameter4fNV, glProgramNamedParameter4fNV, NULL, _gloffset_ProgramNamedParameter4fNV),
+    NAME_FUNC_OFFSET(13263, glProgramNamedParameter4fvNV, glProgramNamedParameter4fvNV, NULL, _gloffset_ProgramNamedParameter4fvNV),
+    NAME_FUNC_OFFSET(13292, gl_dispatch_stub_764, gl_dispatch_stub_764, NULL, _gloffset_DepthBoundsEXT),
+    NAME_FUNC_OFFSET(13309, gl_dispatch_stub_765, gl_dispatch_stub_765, NULL, _gloffset_BlendEquationSeparateEXT),
+    NAME_FUNC_OFFSET(13336, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT),
+    NAME_FUNC_OFFSET(13357, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT),
+    NAME_FUNC_OFFSET(13379, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT),
+    NAME_FUNC_OFFSET(13407, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT),
+    NAME_FUNC_OFFSET(13431, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT),
+    NAME_FUNC_OFFSET(13456, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT),
+    NAME_FUNC_OFFSET(13485, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT),
+    NAME_FUNC_OFFSET(13511, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT),
+    NAME_FUNC_OFFSET(13537, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT),
+    NAME_FUNC_OFFSET(13563, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT),
+    NAME_FUNC_OFFSET(13584, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT),
+    NAME_FUNC_OFFSET(13606, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT),
+    NAME_FUNC_OFFSET(13626, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT),
+    NAME_FUNC_OFFSET(13667, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT),
+    NAME_FUNC_OFFSET(13699, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT),
+    NAME_FUNC_OFFSET(13718, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT),
+    NAME_FUNC_OFFSET(13738, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT),
+    NAME_FUNC_OFFSET(13763, gl_dispatch_stub_783, gl_dispatch_stub_783, NULL, _gloffset_BlitFramebufferEXT),
+    NAME_FUNC_OFFSET(13784, gl_dispatch_stub_784, gl_dispatch_stub_784, NULL, _gloffset_BufferParameteriAPPLE),
+    NAME_FUNC_OFFSET(13808, gl_dispatch_stub_785, gl_dispatch_stub_785, NULL, _gloffset_FlushMappedBufferRangeAPPLE),
+    NAME_FUNC_OFFSET(13838, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT),
+    NAME_FUNC_OFFSET(13867, glProvokingVertexEXT, glProvokingVertexEXT, NULL, _gloffset_ProvokingVertexEXT),
+    NAME_FUNC_OFFSET(13888, gl_dispatch_stub_788, gl_dispatch_stub_788, NULL, _gloffset_GetTexParameterPointervAPPLE),
+    NAME_FUNC_OFFSET(13919, gl_dispatch_stub_789, gl_dispatch_stub_789, NULL, _gloffset_TextureRangeAPPLE),
+    NAME_FUNC_OFFSET(13939, gl_dispatch_stub_790, gl_dispatch_stub_790, NULL, _gloffset_StencilFuncSeparateATI),
+    NAME_FUNC_OFFSET(13964, gl_dispatch_stub_791, gl_dispatch_stub_791, NULL, _gloffset_ProgramEnvParameters4fvEXT),
+    NAME_FUNC_OFFSET(13993, gl_dispatch_stub_792, gl_dispatch_stub_792, NULL, _gloffset_ProgramLocalParameters4fvEXT),
+    NAME_FUNC_OFFSET(14024, gl_dispatch_stub_793, gl_dispatch_stub_793, NULL, _gloffset_GetQueryObjecti64vEXT),
+    NAME_FUNC_OFFSET(14048, gl_dispatch_stub_794, gl_dispatch_stub_794, NULL, _gloffset_GetQueryObjectui64vEXT),
+    NAME_FUNC_OFFSET(14073, glArrayElement, glArrayElement, NULL, _gloffset_ArrayElement),
+    NAME_FUNC_OFFSET(14091, glBindTexture, glBindTexture, NULL, _gloffset_BindTexture),
+    NAME_FUNC_OFFSET(14108, glDrawArrays, glDrawArrays, NULL, _gloffset_DrawArrays),
+    NAME_FUNC_OFFSET(14124, glAreTexturesResident, glAreTexturesResidentEXT, glAreTexturesResidentEXT, _gloffset_AreTexturesResident),
+    NAME_FUNC_OFFSET(14149, glCopyTexImage1D, glCopyTexImage1D, NULL, _gloffset_CopyTexImage1D),
+    NAME_FUNC_OFFSET(14169, glCopyTexImage2D, glCopyTexImage2D, NULL, _gloffset_CopyTexImage2D),
+    NAME_FUNC_OFFSET(14189, glCopyTexSubImage1D, glCopyTexSubImage1D, NULL, _gloffset_CopyTexSubImage1D),
+    NAME_FUNC_OFFSET(14212, glCopyTexSubImage2D, glCopyTexSubImage2D, NULL, _gloffset_CopyTexSubImage2D),
+    NAME_FUNC_OFFSET(14235, glDeleteTextures, glDeleteTexturesEXT, glDeleteTexturesEXT, _gloffset_DeleteTextures),
+    NAME_FUNC_OFFSET(14255, glGenTextures, glGenTexturesEXT, glGenTexturesEXT, _gloffset_GenTextures),
+    NAME_FUNC_OFFSET(14272, glGetPointerv, glGetPointerv, NULL, _gloffset_GetPointerv),
+    NAME_FUNC_OFFSET(14289, glIsTexture, glIsTextureEXT, glIsTextureEXT, _gloffset_IsTexture),
+    NAME_FUNC_OFFSET(14304, glPrioritizeTextures, glPrioritizeTextures, NULL, _gloffset_PrioritizeTextures),
+    NAME_FUNC_OFFSET(14328, glTexSubImage1D, glTexSubImage1D, NULL, _gloffset_TexSubImage1D),
+    NAME_FUNC_OFFSET(14347, glTexSubImage2D, glTexSubImage2D, NULL, _gloffset_TexSubImage2D),
+    NAME_FUNC_OFFSET(14366, glBlendColor, glBlendColor, NULL, _gloffset_BlendColor),
+    NAME_FUNC_OFFSET(14382, glBlendEquation, glBlendEquation, NULL, _gloffset_BlendEquation),
+    NAME_FUNC_OFFSET(14401, glDrawRangeElements, glDrawRangeElements, NULL, _gloffset_DrawRangeElements),
+    NAME_FUNC_OFFSET(14424, glColorTable, glColorTable, NULL, _gloffset_ColorTable),
+    NAME_FUNC_OFFSET(14440, glColorTable, glColorTable, NULL, _gloffset_ColorTable),
+    NAME_FUNC_OFFSET(14456, glColorTableParameterfv, glColorTableParameterfv, NULL, _gloffset_ColorTableParameterfv),
+    NAME_FUNC_OFFSET(14483, glColorTableParameteriv, glColorTableParameteriv, NULL, _gloffset_ColorTableParameteriv),
+    NAME_FUNC_OFFSET(14510, glCopyColorTable, glCopyColorTable, NULL, _gloffset_CopyColorTable),
+    NAME_FUNC_OFFSET(14530, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable),
+    NAME_FUNC_OFFSET(14549, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable),
+    NAME_FUNC_OFFSET(14568, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv),
+    NAME_FUNC_OFFSET(14598, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv),
+    NAME_FUNC_OFFSET(14628, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv),
+    NAME_FUNC_OFFSET(14658, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv),
+    NAME_FUNC_OFFSET(14688, glColorSubTable, glColorSubTable, NULL, _gloffset_ColorSubTable),
+    NAME_FUNC_OFFSET(14707, glCopyColorSubTable, glCopyColorSubTable, NULL, _gloffset_CopyColorSubTable),
+    NAME_FUNC_OFFSET(14730, glConvolutionFilter1D, glConvolutionFilter1D, NULL, _gloffset_ConvolutionFilter1D),
+    NAME_FUNC_OFFSET(14755, glConvolutionFilter2D, glConvolutionFilter2D, NULL, _gloffset_ConvolutionFilter2D),
+    NAME_FUNC_OFFSET(14780, glConvolutionParameterf, glConvolutionParameterf, NULL, _gloffset_ConvolutionParameterf),
+    NAME_FUNC_OFFSET(14807, glConvolutionParameterfv, glConvolutionParameterfv, NULL, _gloffset_ConvolutionParameterfv),
+    NAME_FUNC_OFFSET(14835, glConvolutionParameteri, glConvolutionParameteri, NULL, _gloffset_ConvolutionParameteri),
+    NAME_FUNC_OFFSET(14862, glConvolutionParameteriv, glConvolutionParameteriv, NULL, _gloffset_ConvolutionParameteriv),
+    NAME_FUNC_OFFSET(14890, glCopyConvolutionFilter1D, glCopyConvolutionFilter1D, NULL, _gloffset_CopyConvolutionFilter1D),
+    NAME_FUNC_OFFSET(14919, glCopyConvolutionFilter2D, glCopyConvolutionFilter2D, NULL, _gloffset_CopyConvolutionFilter2D),
+    NAME_FUNC_OFFSET(14948, glGetConvolutionFilter, gl_dispatch_stub_356, gl_dispatch_stub_356, _gloffset_GetConvolutionFilter),
+    NAME_FUNC_OFFSET(14974, glGetConvolutionParameterfv, gl_dispatch_stub_357, gl_dispatch_stub_357, _gloffset_GetConvolutionParameterfv),
+    NAME_FUNC_OFFSET(15005, glGetConvolutionParameteriv, gl_dispatch_stub_358, gl_dispatch_stub_358, _gloffset_GetConvolutionParameteriv),
+    NAME_FUNC_OFFSET(15036, glGetSeparableFilter, gl_dispatch_stub_359, gl_dispatch_stub_359, _gloffset_GetSeparableFilter),
+    NAME_FUNC_OFFSET(15060, glSeparableFilter2D, glSeparableFilter2D, NULL, _gloffset_SeparableFilter2D),
+    NAME_FUNC_OFFSET(15083, glGetHistogram, gl_dispatch_stub_361, gl_dispatch_stub_361, _gloffset_GetHistogram),
+    NAME_FUNC_OFFSET(15101, glGetHistogramParameterfv, gl_dispatch_stub_362, gl_dispatch_stub_362, _gloffset_GetHistogramParameterfv),
+    NAME_FUNC_OFFSET(15130, glGetHistogramParameteriv, gl_dispatch_stub_363, gl_dispatch_stub_363, _gloffset_GetHistogramParameteriv),
+    NAME_FUNC_OFFSET(15159, glGetMinmax, gl_dispatch_stub_364, gl_dispatch_stub_364, _gloffset_GetMinmax),
+    NAME_FUNC_OFFSET(15174, glGetMinmaxParameterfv, gl_dispatch_stub_365, gl_dispatch_stub_365, _gloffset_GetMinmaxParameterfv),
+    NAME_FUNC_OFFSET(15200, glGetMinmaxParameteriv, gl_dispatch_stub_366, gl_dispatch_stub_366, _gloffset_GetMinmaxParameteriv),
+    NAME_FUNC_OFFSET(15226, glHistogram, glHistogram, NULL, _gloffset_Histogram),
+    NAME_FUNC_OFFSET(15241, glMinmax, glMinmax, NULL, _gloffset_Minmax),
+    NAME_FUNC_OFFSET(15253, glResetHistogram, glResetHistogram, NULL, _gloffset_ResetHistogram),
+    NAME_FUNC_OFFSET(15273, glResetMinmax, glResetMinmax, NULL, _gloffset_ResetMinmax),
+    NAME_FUNC_OFFSET(15290, glTexImage3D, glTexImage3D, NULL, _gloffset_TexImage3D),
+    NAME_FUNC_OFFSET(15306, glTexSubImage3D, glTexSubImage3D, NULL, _gloffset_TexSubImage3D),
+    NAME_FUNC_OFFSET(15325, glCopyTexSubImage3D, glCopyTexSubImage3D, NULL, _gloffset_CopyTexSubImage3D),
+    NAME_FUNC_OFFSET(15348, glActiveTextureARB, glActiveTextureARB, NULL, _gloffset_ActiveTextureARB),
+    NAME_FUNC_OFFSET(15364, glClientActiveTextureARB, glClientActiveTextureARB, NULL, _gloffset_ClientActiveTextureARB),
+    NAME_FUNC_OFFSET(15386, glMultiTexCoord1dARB, glMultiTexCoord1dARB, NULL, _gloffset_MultiTexCoord1dARB),
+    NAME_FUNC_OFFSET(15404, glMultiTexCoord1dvARB, glMultiTexCoord1dvARB, NULL, _gloffset_MultiTexCoord1dvARB),
+    NAME_FUNC_OFFSET(15423, glMultiTexCoord1fARB, glMultiTexCoord1fARB, NULL, _gloffset_MultiTexCoord1fARB),
+    NAME_FUNC_OFFSET(15441, glMultiTexCoord1fvARB, glMultiTexCoord1fvARB, NULL, _gloffset_MultiTexCoord1fvARB),
+    NAME_FUNC_OFFSET(15460, glMultiTexCoord1iARB, glMultiTexCoord1iARB, NULL, _gloffset_MultiTexCoord1iARB),
+    NAME_FUNC_OFFSET(15478, glMultiTexCoord1ivARB, glMultiTexCoord1ivARB, NULL, _gloffset_MultiTexCoord1ivARB),
+    NAME_FUNC_OFFSET(15497, glMultiTexCoord1sARB, glMultiTexCoord1sARB, NULL, _gloffset_MultiTexCoord1sARB),
+    NAME_FUNC_OFFSET(15515, glMultiTexCoord1svARB, glMultiTexCoord1svARB, NULL, _gloffset_MultiTexCoord1svARB),
+    NAME_FUNC_OFFSET(15534, glMultiTexCoord2dARB, glMultiTexCoord2dARB, NULL, _gloffset_MultiTexCoord2dARB),
+    NAME_FUNC_OFFSET(15552, glMultiTexCoord2dvARB, glMultiTexCoord2dvARB, NULL, _gloffset_MultiTexCoord2dvARB),
+    NAME_FUNC_OFFSET(15571, glMultiTexCoord2fARB, glMultiTexCoord2fARB, NULL, _gloffset_MultiTexCoord2fARB),
+    NAME_FUNC_OFFSET(15589, glMultiTexCoord2fvARB, glMultiTexCoord2fvARB, NULL, _gloffset_MultiTexCoord2fvARB),
+    NAME_FUNC_OFFSET(15608, glMultiTexCoord2iARB, glMultiTexCoord2iARB, NULL, _gloffset_MultiTexCoord2iARB),
+    NAME_FUNC_OFFSET(15626, glMultiTexCoord2ivARB, glMultiTexCoord2ivARB, NULL, _gloffset_MultiTexCoord2ivARB),
+    NAME_FUNC_OFFSET(15645, glMultiTexCoord2sARB, glMultiTexCoord2sARB, NULL, _gloffset_MultiTexCoord2sARB),
+    NAME_FUNC_OFFSET(15663, glMultiTexCoord2svARB, glMultiTexCoord2svARB, NULL, _gloffset_MultiTexCoord2svARB),
+    NAME_FUNC_OFFSET(15682, glMultiTexCoord3dARB, glMultiTexCoord3dARB, NULL, _gloffset_MultiTexCoord3dARB),
+    NAME_FUNC_OFFSET(15700, glMultiTexCoord3dvARB, glMultiTexCoord3dvARB, NULL, _gloffset_MultiTexCoord3dvARB),
+    NAME_FUNC_OFFSET(15719, glMultiTexCoord3fARB, glMultiTexCoord3fARB, NULL, _gloffset_MultiTexCoord3fARB),
+    NAME_FUNC_OFFSET(15737, glMultiTexCoord3fvARB, glMultiTexCoord3fvARB, NULL, _gloffset_MultiTexCoord3fvARB),
+    NAME_FUNC_OFFSET(15756, glMultiTexCoord3iARB, glMultiTexCoord3iARB, NULL, _gloffset_MultiTexCoord3iARB),
+    NAME_FUNC_OFFSET(15774, glMultiTexCoord3ivARB, glMultiTexCoord3ivARB, NULL, _gloffset_MultiTexCoord3ivARB),
+    NAME_FUNC_OFFSET(15793, glMultiTexCoord3sARB, glMultiTexCoord3sARB, NULL, _gloffset_MultiTexCoord3sARB),
+    NAME_FUNC_OFFSET(15811, glMultiTexCoord3svARB, glMultiTexCoord3svARB, NULL, _gloffset_MultiTexCoord3svARB),
+    NAME_FUNC_OFFSET(15830, glMultiTexCoord4dARB, glMultiTexCoord4dARB, NULL, _gloffset_MultiTexCoord4dARB),
+    NAME_FUNC_OFFSET(15848, glMultiTexCoord4dvARB, glMultiTexCoord4dvARB, NULL, _gloffset_MultiTexCoord4dvARB),
+    NAME_FUNC_OFFSET(15867, glMultiTexCoord4fARB, glMultiTexCoord4fARB, NULL, _gloffset_MultiTexCoord4fARB),
+    NAME_FUNC_OFFSET(15885, glMultiTexCoord4fvARB, glMultiTexCoord4fvARB, NULL, _gloffset_MultiTexCoord4fvARB),
+    NAME_FUNC_OFFSET(15904, glMultiTexCoord4iARB, glMultiTexCoord4iARB, NULL, _gloffset_MultiTexCoord4iARB),
+    NAME_FUNC_OFFSET(15922, glMultiTexCoord4ivARB, glMultiTexCoord4ivARB, NULL, _gloffset_MultiTexCoord4ivARB),
+    NAME_FUNC_OFFSET(15941, glMultiTexCoord4sARB, glMultiTexCoord4sARB, NULL, _gloffset_MultiTexCoord4sARB),
+    NAME_FUNC_OFFSET(15959, glMultiTexCoord4svARB, glMultiTexCoord4svARB, NULL, _gloffset_MultiTexCoord4svARB),
+    NAME_FUNC_OFFSET(15978, glStencilOpSeparate, glStencilOpSeparate, NULL, _gloffset_StencilOpSeparate),
+    NAME_FUNC_OFFSET(16001, glLoadTransposeMatrixdARB, glLoadTransposeMatrixdARB, NULL, _gloffset_LoadTransposeMatrixdARB),
+    NAME_FUNC_OFFSET(16024, glLoadTransposeMatrixfARB, glLoadTransposeMatrixfARB, NULL, _gloffset_LoadTransposeMatrixfARB),
+    NAME_FUNC_OFFSET(16047, glMultTransposeMatrixdARB, glMultTransposeMatrixdARB, NULL, _gloffset_MultTransposeMatrixdARB),
+    NAME_FUNC_OFFSET(16070, glMultTransposeMatrixfARB, glMultTransposeMatrixfARB, NULL, _gloffset_MultTransposeMatrixfARB),
+    NAME_FUNC_OFFSET(16093, glSampleCoverageARB, glSampleCoverageARB, NULL, _gloffset_SampleCoverageARB),
+    NAME_FUNC_OFFSET(16110, glCompressedTexImage1DARB, glCompressedTexImage1DARB, NULL, _gloffset_CompressedTexImage1DARB),
+    NAME_FUNC_OFFSET(16133, glCompressedTexImage2DARB, glCompressedTexImage2DARB, NULL, _gloffset_CompressedTexImage2DARB),
+    NAME_FUNC_OFFSET(16156, glCompressedTexImage3DARB, glCompressedTexImage3DARB, NULL, _gloffset_CompressedTexImage3DARB),
+    NAME_FUNC_OFFSET(16179, glCompressedTexSubImage1DARB, glCompressedTexSubImage1DARB, NULL, _gloffset_CompressedTexSubImage1DARB),
+    NAME_FUNC_OFFSET(16205, glCompressedTexSubImage2DARB, glCompressedTexSubImage2DARB, NULL, _gloffset_CompressedTexSubImage2DARB),
+    NAME_FUNC_OFFSET(16231, glCompressedTexSubImage3DARB, glCompressedTexSubImage3DARB, NULL, _gloffset_CompressedTexSubImage3DARB),
+    NAME_FUNC_OFFSET(16257, glGetCompressedTexImageARB, glGetCompressedTexImageARB, NULL, _gloffset_GetCompressedTexImageARB),
+    NAME_FUNC_OFFSET(16281, glDisableVertexAttribArrayARB, glDisableVertexAttribArrayARB, NULL, _gloffset_DisableVertexAttribArrayARB),
+    NAME_FUNC_OFFSET(16308, glEnableVertexAttribArrayARB, glEnableVertexAttribArrayARB, NULL, _gloffset_EnableVertexAttribArrayARB),
+    NAME_FUNC_OFFSET(16334, glGetVertexAttribdvARB, glGetVertexAttribdvARB, NULL, _gloffset_GetVertexAttribdvARB),
+    NAME_FUNC_OFFSET(16354, glGetVertexAttribfvARB, glGetVertexAttribfvARB, NULL, _gloffset_GetVertexAttribfvARB),
+    NAME_FUNC_OFFSET(16374, glGetVertexAttribivARB, glGetVertexAttribivARB, NULL, _gloffset_GetVertexAttribivARB),
+    NAME_FUNC_OFFSET(16394, glProgramEnvParameter4dARB, glProgramEnvParameter4dARB, NULL, _gloffset_ProgramEnvParameter4dARB),
+    NAME_FUNC_OFFSET(16417, glProgramEnvParameter4dvARB, glProgramEnvParameter4dvARB, NULL, _gloffset_ProgramEnvParameter4dvARB),
+    NAME_FUNC_OFFSET(16441, glProgramEnvParameter4fARB, glProgramEnvParameter4fARB, NULL, _gloffset_ProgramEnvParameter4fARB),
+    NAME_FUNC_OFFSET(16464, glProgramEnvParameter4fvARB, glProgramEnvParameter4fvARB, NULL, _gloffset_ProgramEnvParameter4fvARB),
+    NAME_FUNC_OFFSET(16488, glVertexAttrib1dARB, glVertexAttrib1dARB, NULL, _gloffset_VertexAttrib1dARB),
+    NAME_FUNC_OFFSET(16505, glVertexAttrib1dvARB, glVertexAttrib1dvARB, NULL, _gloffset_VertexAttrib1dvARB),
+    NAME_FUNC_OFFSET(16523, glVertexAttrib1fARB, glVertexAttrib1fARB, NULL, _gloffset_VertexAttrib1fARB),
+    NAME_FUNC_OFFSET(16540, glVertexAttrib1fvARB, glVertexAttrib1fvARB, NULL, _gloffset_VertexAttrib1fvARB),
+    NAME_FUNC_OFFSET(16558, glVertexAttrib1sARB, glVertexAttrib1sARB, NULL, _gloffset_VertexAttrib1sARB),
+    NAME_FUNC_OFFSET(16575, glVertexAttrib1svARB, glVertexAttrib1svARB, NULL, _gloffset_VertexAttrib1svARB),
+    NAME_FUNC_OFFSET(16593, glVertexAttrib2dARB, glVertexAttrib2dARB, NULL, _gloffset_VertexAttrib2dARB),
+    NAME_FUNC_OFFSET(16610, glVertexAttrib2dvARB, glVertexAttrib2dvARB, NULL, _gloffset_VertexAttrib2dvARB),
+    NAME_FUNC_OFFSET(16628, glVertexAttrib2fARB, glVertexAttrib2fARB, NULL, _gloffset_VertexAttrib2fARB),
+    NAME_FUNC_OFFSET(16645, glVertexAttrib2fvARB, glVertexAttrib2fvARB, NULL, _gloffset_VertexAttrib2fvARB),
+    NAME_FUNC_OFFSET(16663, glVertexAttrib2sARB, glVertexAttrib2sARB, NULL, _gloffset_VertexAttrib2sARB),
+    NAME_FUNC_OFFSET(16680, glVertexAttrib2svARB, glVertexAttrib2svARB, NULL, _gloffset_VertexAttrib2svARB),
+    NAME_FUNC_OFFSET(16698, glVertexAttrib3dARB, glVertexAttrib3dARB, NULL, _gloffset_VertexAttrib3dARB),
+    NAME_FUNC_OFFSET(16715, glVertexAttrib3dvARB, glVertexAttrib3dvARB, NULL, _gloffset_VertexAttrib3dvARB),
+    NAME_FUNC_OFFSET(16733, glVertexAttrib3fARB, glVertexAttrib3fARB, NULL, _gloffset_VertexAttrib3fARB),
+    NAME_FUNC_OFFSET(16750, glVertexAttrib3fvARB, glVertexAttrib3fvARB, NULL, _gloffset_VertexAttrib3fvARB),
+    NAME_FUNC_OFFSET(16768, glVertexAttrib3sARB, glVertexAttrib3sARB, NULL, _gloffset_VertexAttrib3sARB),
+    NAME_FUNC_OFFSET(16785, glVertexAttrib3svARB, glVertexAttrib3svARB, NULL, _gloffset_VertexAttrib3svARB),
+    NAME_FUNC_OFFSET(16803, glVertexAttrib4NbvARB, glVertexAttrib4NbvARB, NULL, _gloffset_VertexAttrib4NbvARB),
+    NAME_FUNC_OFFSET(16822, glVertexAttrib4NivARB, glVertexAttrib4NivARB, NULL, _gloffset_VertexAttrib4NivARB),
+    NAME_FUNC_OFFSET(16841, glVertexAttrib4NsvARB, glVertexAttrib4NsvARB, NULL, _gloffset_VertexAttrib4NsvARB),
+    NAME_FUNC_OFFSET(16860, glVertexAttrib4NubARB, glVertexAttrib4NubARB, NULL, _gloffset_VertexAttrib4NubARB),
+    NAME_FUNC_OFFSET(16879, glVertexAttrib4NubvARB, glVertexAttrib4NubvARB, NULL, _gloffset_VertexAttrib4NubvARB),
+    NAME_FUNC_OFFSET(16899, glVertexAttrib4NuivARB, glVertexAttrib4NuivARB, NULL, _gloffset_VertexAttrib4NuivARB),
+    NAME_FUNC_OFFSET(16919, glVertexAttrib4NusvARB, glVertexAttrib4NusvARB, NULL, _gloffset_VertexAttrib4NusvARB),
+    NAME_FUNC_OFFSET(16939, glVertexAttrib4bvARB, glVertexAttrib4bvARB, NULL, _gloffset_VertexAttrib4bvARB),
+    NAME_FUNC_OFFSET(16957, glVertexAttrib4dARB, glVertexAttrib4dARB, NULL, _gloffset_VertexAttrib4dARB),
+    NAME_FUNC_OFFSET(16974, glVertexAttrib4dvARB, glVertexAttrib4dvARB, NULL, _gloffset_VertexAttrib4dvARB),
+    NAME_FUNC_OFFSET(16992, glVertexAttrib4fARB, glVertexAttrib4fARB, NULL, _gloffset_VertexAttrib4fARB),
+    NAME_FUNC_OFFSET(17009, glVertexAttrib4fvARB, glVertexAttrib4fvARB, NULL, _gloffset_VertexAttrib4fvARB),
+    NAME_FUNC_OFFSET(17027, glVertexAttrib4ivARB, glVertexAttrib4ivARB, NULL, _gloffset_VertexAttrib4ivARB),
+    NAME_FUNC_OFFSET(17045, glVertexAttrib4sARB, glVertexAttrib4sARB, NULL, _gloffset_VertexAttrib4sARB),
+    NAME_FUNC_OFFSET(17062, glVertexAttrib4svARB, glVertexAttrib4svARB, NULL, _gloffset_VertexAttrib4svARB),
+    NAME_FUNC_OFFSET(17080, glVertexAttrib4ubvARB, glVertexAttrib4ubvARB, NULL, _gloffset_VertexAttrib4ubvARB),
+    NAME_FUNC_OFFSET(17099, glVertexAttrib4uivARB, glVertexAttrib4uivARB, NULL, _gloffset_VertexAttrib4uivARB),
+    NAME_FUNC_OFFSET(17118, glVertexAttrib4usvARB, glVertexAttrib4usvARB, NULL, _gloffset_VertexAttrib4usvARB),
+    NAME_FUNC_OFFSET(17137, glVertexAttribPointerARB, glVertexAttribPointerARB, NULL, _gloffset_VertexAttribPointerARB),
+    NAME_FUNC_OFFSET(17159, glBindBufferARB, glBindBufferARB, NULL, _gloffset_BindBufferARB),
+    NAME_FUNC_OFFSET(17172, glBufferDataARB, glBufferDataARB, NULL, _gloffset_BufferDataARB),
+    NAME_FUNC_OFFSET(17185, glBufferSubDataARB, glBufferSubDataARB, NULL, _gloffset_BufferSubDataARB),
+    NAME_FUNC_OFFSET(17201, glDeleteBuffersARB, glDeleteBuffersARB, NULL, _gloffset_DeleteBuffersARB),
+    NAME_FUNC_OFFSET(17217, glGenBuffersARB, glGenBuffersARB, NULL, _gloffset_GenBuffersARB),
+    NAME_FUNC_OFFSET(17230, glGetBufferParameterivARB, glGetBufferParameterivARB, NULL, _gloffset_GetBufferParameterivARB),
+    NAME_FUNC_OFFSET(17253, glGetBufferPointervARB, glGetBufferPointervARB, NULL, _gloffset_GetBufferPointervARB),
+    NAME_FUNC_OFFSET(17273, glGetBufferSubDataARB, glGetBufferSubDataARB, NULL, _gloffset_GetBufferSubDataARB),
+    NAME_FUNC_OFFSET(17292, glIsBufferARB, glIsBufferARB, NULL, _gloffset_IsBufferARB),
+    NAME_FUNC_OFFSET(17303, glMapBufferARB, glMapBufferARB, NULL, _gloffset_MapBufferARB),
+    NAME_FUNC_OFFSET(17315, glUnmapBufferARB, glUnmapBufferARB, NULL, _gloffset_UnmapBufferARB),
+    NAME_FUNC_OFFSET(17329, glBeginQueryARB, glBeginQueryARB, NULL, _gloffset_BeginQueryARB),
+    NAME_FUNC_OFFSET(17342, glDeleteQueriesARB, glDeleteQueriesARB, NULL, _gloffset_DeleteQueriesARB),
+    NAME_FUNC_OFFSET(17358, glEndQueryARB, glEndQueryARB, NULL, _gloffset_EndQueryARB),
+    NAME_FUNC_OFFSET(17369, glGenQueriesARB, glGenQueriesARB, NULL, _gloffset_GenQueriesARB),
+    NAME_FUNC_OFFSET(17382, glGetQueryObjectivARB, glGetQueryObjectivARB, NULL, _gloffset_GetQueryObjectivARB),
+    NAME_FUNC_OFFSET(17401, glGetQueryObjectuivARB, glGetQueryObjectuivARB, NULL, _gloffset_GetQueryObjectuivARB),
+    NAME_FUNC_OFFSET(17421, glGetQueryivARB, glGetQueryivARB, NULL, _gloffset_GetQueryivARB),
+    NAME_FUNC_OFFSET(17434, glIsQueryARB, glIsQueryARB, NULL, _gloffset_IsQueryARB),
+    NAME_FUNC_OFFSET(17444, glCompileShaderARB, glCompileShaderARB, NULL, _gloffset_CompileShaderARB),
+    NAME_FUNC_OFFSET(17460, glGetActiveUniformARB, glGetActiveUniformARB, NULL, _gloffset_GetActiveUniformARB),
+    NAME_FUNC_OFFSET(17479, glGetShaderSourceARB, glGetShaderSourceARB, NULL, _gloffset_GetShaderSourceARB),
+    NAME_FUNC_OFFSET(17497, glGetUniformLocationARB, glGetUniformLocationARB, NULL, _gloffset_GetUniformLocationARB),
+    NAME_FUNC_OFFSET(17518, glGetUniformfvARB, glGetUniformfvARB, NULL, _gloffset_GetUniformfvARB),
+    NAME_FUNC_OFFSET(17533, glGetUniformivARB, glGetUniformivARB, NULL, _gloffset_GetUniformivARB),
+    NAME_FUNC_OFFSET(17548, glLinkProgramARB, glLinkProgramARB, NULL, _gloffset_LinkProgramARB),
+    NAME_FUNC_OFFSET(17562, glShaderSourceARB, glShaderSourceARB, NULL, _gloffset_ShaderSourceARB),
+    NAME_FUNC_OFFSET(17577, glUniform1fARB, glUniform1fARB, NULL, _gloffset_Uniform1fARB),
+    NAME_FUNC_OFFSET(17589, glUniform1fvARB, glUniform1fvARB, NULL, _gloffset_Uniform1fvARB),
+    NAME_FUNC_OFFSET(17602, glUniform1iARB, glUniform1iARB, NULL, _gloffset_Uniform1iARB),
+    NAME_FUNC_OFFSET(17614, glUniform1ivARB, glUniform1ivARB, NULL, _gloffset_Uniform1ivARB),
+    NAME_FUNC_OFFSET(17627, glUniform2fARB, glUniform2fARB, NULL, _gloffset_Uniform2fARB),
+    NAME_FUNC_OFFSET(17639, glUniform2fvARB, glUniform2fvARB, NULL, _gloffset_Uniform2fvARB),
+    NAME_FUNC_OFFSET(17652, glUniform2iARB, glUniform2iARB, NULL, _gloffset_Uniform2iARB),
+    NAME_FUNC_OFFSET(17664, glUniform2ivARB, glUniform2ivARB, NULL, _gloffset_Uniform2ivARB),
+    NAME_FUNC_OFFSET(17677, glUniform3fARB, glUniform3fARB, NULL, _gloffset_Uniform3fARB),
+    NAME_FUNC_OFFSET(17689, glUniform3fvARB, glUniform3fvARB, NULL, _gloffset_Uniform3fvARB),
+    NAME_FUNC_OFFSET(17702, glUniform3iARB, glUniform3iARB, NULL, _gloffset_Uniform3iARB),
+    NAME_FUNC_OFFSET(17714, glUniform3ivARB, glUniform3ivARB, NULL, _gloffset_Uniform3ivARB),
+    NAME_FUNC_OFFSET(17727, glUniform4fARB, glUniform4fARB, NULL, _gloffset_Uniform4fARB),
+    NAME_FUNC_OFFSET(17739, glUniform4fvARB, glUniform4fvARB, NULL, _gloffset_Uniform4fvARB),
+    NAME_FUNC_OFFSET(17752, glUniform4iARB, glUniform4iARB, NULL, _gloffset_Uniform4iARB),
+    NAME_FUNC_OFFSET(17764, glUniform4ivARB, glUniform4ivARB, NULL, _gloffset_Uniform4ivARB),
+    NAME_FUNC_OFFSET(17777, glUniformMatrix2fvARB, glUniformMatrix2fvARB, NULL, _gloffset_UniformMatrix2fvARB),
+    NAME_FUNC_OFFSET(17796, glUniformMatrix3fvARB, glUniformMatrix3fvARB, NULL, _gloffset_UniformMatrix3fvARB),
+    NAME_FUNC_OFFSET(17815, glUniformMatrix4fvARB, glUniformMatrix4fvARB, NULL, _gloffset_UniformMatrix4fvARB),
+    NAME_FUNC_OFFSET(17834, glUseProgramObjectARB, glUseProgramObjectARB, NULL, _gloffset_UseProgramObjectARB),
+    NAME_FUNC_OFFSET(17847, glValidateProgramARB, glValidateProgramARB, NULL, _gloffset_ValidateProgramARB),
+    NAME_FUNC_OFFSET(17865, glBindAttribLocationARB, glBindAttribLocationARB, NULL, _gloffset_BindAttribLocationARB),
+    NAME_FUNC_OFFSET(17886, glGetActiveAttribARB, glGetActiveAttribARB, NULL, _gloffset_GetActiveAttribARB),
+    NAME_FUNC_OFFSET(17904, glGetAttribLocationARB, glGetAttribLocationARB, NULL, _gloffset_GetAttribLocationARB),
+    NAME_FUNC_OFFSET(17924, glDrawBuffersARB, glDrawBuffersARB, NULL, _gloffset_DrawBuffersARB),
+    NAME_FUNC_OFFSET(17938, glDrawBuffersARB, glDrawBuffersARB, NULL, _gloffset_DrawBuffersARB),
+    NAME_FUNC_OFFSET(17955, gl_dispatch_stub_584, gl_dispatch_stub_584, NULL, _gloffset_SampleMaskSGIS),
+    NAME_FUNC_OFFSET(17971, gl_dispatch_stub_585, gl_dispatch_stub_585, NULL, _gloffset_SamplePatternSGIS),
+    NAME_FUNC_OFFSET(17990, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET(18008, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET(18029, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET(18051, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET(18070, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET(18092, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET(18115, glSecondaryColor3bEXT, glSecondaryColor3bEXT, NULL, _gloffset_SecondaryColor3bEXT),
+    NAME_FUNC_OFFSET(18134, glSecondaryColor3bvEXT, glSecondaryColor3bvEXT, NULL, _gloffset_SecondaryColor3bvEXT),
+    NAME_FUNC_OFFSET(18154, glSecondaryColor3dEXT, glSecondaryColor3dEXT, NULL, _gloffset_SecondaryColor3dEXT),
+    NAME_FUNC_OFFSET(18173, glSecondaryColor3dvEXT, glSecondaryColor3dvEXT, NULL, _gloffset_SecondaryColor3dvEXT),
+    NAME_FUNC_OFFSET(18193, glSecondaryColor3fEXT, glSecondaryColor3fEXT, NULL, _gloffset_SecondaryColor3fEXT),
+    NAME_FUNC_OFFSET(18212, glSecondaryColor3fvEXT, glSecondaryColor3fvEXT, NULL, _gloffset_SecondaryColor3fvEXT),
+    NAME_FUNC_OFFSET(18232, glSecondaryColor3iEXT, glSecondaryColor3iEXT, NULL, _gloffset_SecondaryColor3iEXT),
+    NAME_FUNC_OFFSET(18251, glSecondaryColor3ivEXT, glSecondaryColor3ivEXT, NULL, _gloffset_SecondaryColor3ivEXT),
+    NAME_FUNC_OFFSET(18271, glSecondaryColor3sEXT, glSecondaryColor3sEXT, NULL, _gloffset_SecondaryColor3sEXT),
+    NAME_FUNC_OFFSET(18290, glSecondaryColor3svEXT, glSecondaryColor3svEXT, NULL, _gloffset_SecondaryColor3svEXT),
+    NAME_FUNC_OFFSET(18310, glSecondaryColor3ubEXT, glSecondaryColor3ubEXT, NULL, _gloffset_SecondaryColor3ubEXT),
+    NAME_FUNC_OFFSET(18330, glSecondaryColor3ubvEXT, glSecondaryColor3ubvEXT, NULL, _gloffset_SecondaryColor3ubvEXT),
+    NAME_FUNC_OFFSET(18351, glSecondaryColor3uiEXT, glSecondaryColor3uiEXT, NULL, _gloffset_SecondaryColor3uiEXT),
+    NAME_FUNC_OFFSET(18371, glSecondaryColor3uivEXT, glSecondaryColor3uivEXT, NULL, _gloffset_SecondaryColor3uivEXT),
+    NAME_FUNC_OFFSET(18392, glSecondaryColor3usEXT, glSecondaryColor3usEXT, NULL, _gloffset_SecondaryColor3usEXT),
+    NAME_FUNC_OFFSET(18412, glSecondaryColor3usvEXT, glSecondaryColor3usvEXT, NULL, _gloffset_SecondaryColor3usvEXT),
+    NAME_FUNC_OFFSET(18433, glSecondaryColorPointerEXT, glSecondaryColorPointerEXT, NULL, _gloffset_SecondaryColorPointerEXT),
+    NAME_FUNC_OFFSET(18457, glMultiDrawArraysEXT, glMultiDrawArraysEXT, NULL, _gloffset_MultiDrawArraysEXT),
+    NAME_FUNC_OFFSET(18475, glMultiDrawElementsEXT, glMultiDrawElementsEXT, NULL, _gloffset_MultiDrawElementsEXT),
+    NAME_FUNC_OFFSET(18495, glFogCoordPointerEXT, glFogCoordPointerEXT, NULL, _gloffset_FogCoordPointerEXT),
+    NAME_FUNC_OFFSET(18513, glFogCoorddEXT, glFogCoorddEXT, NULL, _gloffset_FogCoorddEXT),
+    NAME_FUNC_OFFSET(18525, glFogCoorddvEXT, glFogCoorddvEXT, NULL, _gloffset_FogCoorddvEXT),
+    NAME_FUNC_OFFSET(18538, glFogCoordfEXT, glFogCoordfEXT, NULL, _gloffset_FogCoordfEXT),
+    NAME_FUNC_OFFSET(18550, glFogCoordfvEXT, glFogCoordfvEXT, NULL, _gloffset_FogCoordfvEXT),
+    NAME_FUNC_OFFSET(18563, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
+    NAME_FUNC_OFFSET(18583, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
+    NAME_FUNC_OFFSET(18607, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
+    NAME_FUNC_OFFSET(18621, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
+    NAME_FUNC_OFFSET(18638, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
+    NAME_FUNC_OFFSET(18653, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
+    NAME_FUNC_OFFSET(18671, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
+    NAME_FUNC_OFFSET(18685, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
+    NAME_FUNC_OFFSET(18702, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
+    NAME_FUNC_OFFSET(18717, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
+    NAME_FUNC_OFFSET(18735, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
+    NAME_FUNC_OFFSET(18749, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
+    NAME_FUNC_OFFSET(18766, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
+    NAME_FUNC_OFFSET(18781, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
+    NAME_FUNC_OFFSET(18799, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
+    NAME_FUNC_OFFSET(18813, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
+    NAME_FUNC_OFFSET(18830, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
+    NAME_FUNC_OFFSET(18845, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
+    NAME_FUNC_OFFSET(18863, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
+    NAME_FUNC_OFFSET(18877, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
+    NAME_FUNC_OFFSET(18894, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
+    NAME_FUNC_OFFSET(18909, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
+    NAME_FUNC_OFFSET(18927, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
+    NAME_FUNC_OFFSET(18941, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
+    NAME_FUNC_OFFSET(18958, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
+    NAME_FUNC_OFFSET(18973, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
+    NAME_FUNC_OFFSET(18991, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
+    NAME_FUNC_OFFSET(19005, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
+    NAME_FUNC_OFFSET(19022, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
+    NAME_FUNC_OFFSET(19037, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
+    NAME_FUNC_OFFSET(19055, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
+    NAME_FUNC_OFFSET(19069, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
+    NAME_FUNC_OFFSET(19086, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
+    NAME_FUNC_OFFSET(19101, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
+    NAME_FUNC_OFFSET(19119, glBindProgramNV, glBindProgramNV, NULL, _gloffset_BindProgramNV),
+    NAME_FUNC_OFFSET(19136, glDeleteProgramsNV, glDeleteProgramsNV, NULL, _gloffset_DeleteProgramsNV),
+    NAME_FUNC_OFFSET(19156, glGenProgramsNV, glGenProgramsNV, NULL, _gloffset_GenProgramsNV),
+    NAME_FUNC_OFFSET(19173, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
+    NAME_FUNC_OFFSET(19199, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
+    NAME_FUNC_OFFSET(19228, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV),
+    NAME_FUNC_OFFSET(19243, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV),
+    NAME_FUNC_OFFSET(19261, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV),
+    NAME_FUNC_OFFSET(19280, gl_dispatch_stub_755, gl_dispatch_stub_755, NULL, _gloffset_DeleteVertexArraysAPPLE),
+    NAME_FUNC_OFFSET(19301, gl_dispatch_stub_757, gl_dispatch_stub_757, NULL, _gloffset_IsVertexArrayAPPLE),
+    NAME_FUNC_OFFSET(19317, gl_dispatch_stub_765, gl_dispatch_stub_765, NULL, _gloffset_BlendEquationSeparateEXT),
+    NAME_FUNC_OFFSET(19341, gl_dispatch_stub_765, gl_dispatch_stub_765, NULL, _gloffset_BlendEquationSeparateEXT),
+    NAME_FUNC_OFFSET(19368, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT),
+    NAME_FUNC_OFFSET(19386, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT),
+    NAME_FUNC_OFFSET(19405, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT),
+    NAME_FUNC_OFFSET(19430, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT),
+    NAME_FUNC_OFFSET(19451, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT),
+    NAME_FUNC_OFFSET(19473, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT),
+    NAME_FUNC_OFFSET(19499, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT),
+    NAME_FUNC_OFFSET(19522, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT),
+    NAME_FUNC_OFFSET(19545, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT),
+    NAME_FUNC_OFFSET(19568, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT),
+    NAME_FUNC_OFFSET(19586, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT),
+    NAME_FUNC_OFFSET(19605, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT),
+    NAME_FUNC_OFFSET(19622, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT),
+    NAME_FUNC_OFFSET(19660, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT),
+    NAME_FUNC_OFFSET(19689, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT),
+    NAME_FUNC_OFFSET(19705, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT),
+    NAME_FUNC_OFFSET(19722, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT),
+    NAME_FUNC_OFFSET(19744, gl_dispatch_stub_783, gl_dispatch_stub_783, NULL, _gloffset_BlitFramebufferEXT),
+    NAME_FUNC_OFFSET(19762, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT),
+    NAME_FUNC_OFFSET(19788, glProvokingVertexEXT, glProvokingVertexEXT, NULL, _gloffset_ProvokingVertexEXT),
     NAME_FUNC_OFFSET(-1, NULL, NULL, NULL, 0)
 };
 
diff --git a/src/mesa/glapi/mesadef.py b/src/mesa/glapi/mesadef.py
index 0f410fc482..342c9cde46 100644
--- a/src/mesa/glapi/mesadef.py
+++ b/src/mesa/glapi/mesadef.py
@@ -155,6 +155,15 @@ def PrintTail():
 	print '\t_mesa_strcmp'
 	print '\t_mesa_test_proxy_teximage'
 	print '\t_mesa_Viewport'
+	print '\t_mesa_meta_CopyColorSubTable'
+	print '\t_mesa_meta_CopyColorTable'
+	print '\t_mesa_meta_CopyConvolutionFilter1D'
+	print '\t_mesa_meta_CopyConvolutionFilter2D'
+	print '\t_mesa_meta_CopyTexImage1D'
+	print '\t_mesa_meta_CopyTexImage2D'
+	print '\t_mesa_meta_CopyTexSubImage1D'
+	print '\t_mesa_meta_CopyTexSubImage2D'
+	print '\t_mesa_meta_CopyTexSubImage3D'
 	print '\t_swrast_Accum'
 	print '\t_swrast_alloc_buffers'
 	print '\t_swrast_Bitmap'
@@ -164,15 +173,6 @@ def PrintTail():
 	print '\t_swrast_Clear'
 	print '\t_swrast_choose_line'
 	print '\t_swrast_choose_triangle'
-	print '\t_swrast_CopyColorSubTable'
-	print '\t_swrast_CopyColorTable'
-	print '\t_swrast_CopyConvolutionFilter1D'
-	print '\t_swrast_CopyConvolutionFilter2D'
-	print '\t_swrast_copy_teximage1d'
-	print '\t_swrast_copy_teximage2d'
-	print '\t_swrast_copy_texsubimage1d'
-	print '\t_swrast_copy_texsubimage2d'
-	print '\t_swrast_copy_texsubimage3d'
 	print '\t_swrast_CreateContext'
 	print '\t_swrast_DestroyContext'
 	print '\t_swrast_InvalidateState'
diff --git a/src/mesa/main/accum.c b/src/mesa/main/accum.c
index 2345695f3c..032e13b96e 100644
--- a/src/mesa/main/accum.c
+++ b/src/mesa/main/accum.c
@@ -29,6 +29,10 @@
 #include "macros.h"
 #include "state.h"
 #include "mtypes.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_accum
 
 
 void GLAPIENTRY
@@ -51,7 +55,7 @@ _mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Accum( GLenum op, GLfloat value )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -99,6 +103,16 @@ _mesa_Accum( GLenum op, GLfloat value )
 }
 
 
+void
+_mesa_init_accum_dispatch(struct _glapi_table *disp)
+{
+   SET_Accum(disp, _mesa_Accum);
+   SET_ClearAccum(disp, _mesa_ClearAccum);
+}
+
+
+#endif /* FEATURE_accum */
+
 
 void 
 _mesa_init_accum( GLcontext *ctx )
diff --git a/src/mesa/main/accum.h b/src/mesa/main/accum.h
index ce92688a5b..63740f07ed 100644
--- a/src/mesa/main/accum.h
+++ b/src/mesa/main/accum.h
@@ -38,25 +38,40 @@
 #define ACCUM_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
-#if _HAVE_FULL_GL
-
-extern void GLAPIENTRY
-_mesa_Accum( GLenum op, GLfloat value );
+#if FEATURE_accum
 
+#define _MESA_INIT_ACCUM_FUNCTIONS(driver, impl) \
+   do {                                          \
+      (driver)->Accum = impl ## Accum;           \
+   } while (0)
 
 extern void GLAPIENTRY
 _mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha );
 
-extern void 
-_mesa_init_accum( GLcontext *ctx );
+extern void
+_mesa_init_accum_dispatch(struct _glapi_table *disp);
+
+#else /* FEATURE_accum */
 
-#else
+#define _MESA_INIT_ACCUM_FUNCTIONS(driver, impl) do { } while (0)
 
-/** No-op */
-#define _mesa_init_accum( c ) ((void)0)
+static INLINE void
+_mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha )
+{
+   /* this is used in _mesa_PopAttrib */
+   ASSERT_NO_FEATURE();
+}
 
-#endif
+static INLINE void
+_mesa_init_accum_dispatch(struct _glapi_table *disp)
+{
+}
+
+#endif /* FEATURE_accum */
+
+extern void
+_mesa_init_accum( GLcontext *ctx );
 
-#endif
+#endif /* ACCUM_H */
diff --git a/src/mesa/main/api_arrayelt.c b/src/mesa/main/api_arrayelt.c
index 2462a1b003..a058227110 100644
--- a/src/mesa/main/api_arrayelt.c
+++ b/src/mesa/main/api_arrayelt.c
@@ -71,6 +71,10 @@ typedef struct {
  */
 #define TYPE_IDX(t) ( (t) == GL_DOUBLE ? 7 : (t) & 7 )
 
+
+#if FEATURE_arrayelt
+
+
 static const int ColorFuncs[2][8] = {
    {
       _gloffset_Color3bv,
@@ -1160,7 +1164,7 @@ static void _ae_update_state( GLcontext *ctx )
          at->array = attribArray;
          /* Note: we can't grab the _glapi_Dispatch->VertexAttrib1fvNV
           * function pointer here (for float arrays) since the pointer may
-          * change from one execution of _ae_loopback_array_elt() to
+          * change from one execution of _ae_ArrayElement() to
           * the next.  Doing so caused UT to break.
           */
          if (ctx->VertexProgram._Enabled
@@ -1254,7 +1258,7 @@ void _ae_unmap_vbos( GLcontext *ctx )
  * for all enabled vertex arrays (for elt-th element).
  * Note: this may be called during display list construction.
  */
-void GLAPIENTRY _ae_loopback_array_elt( GLint elt )
+void GLAPIENTRY _ae_ArrayElement( GLint elt )
 {
    GET_CURRENT_CONTEXT(ctx);
    const AEcontext *actx = AE_CONTEXT(ctx);
@@ -1317,3 +1321,13 @@ void _ae_invalidate_state( GLcontext *ctx, GLuint new_state )
       actx->NewState |= new_state;
    }
 }
+
+
+void _mesa_install_arrayelt_vtxfmt(struct _glapi_table *disp,
+                                   const GLvertexformat *vfmt)
+{
+   SET_ArrayElement(disp, vfmt->ArrayElement);
+}
+
+
+#endif /* FEATURE_arrayelt */
diff --git a/src/mesa/main/api_arrayelt.h b/src/mesa/main/api_arrayelt.h
index e621724fb2..d18c0792c3 100644
--- a/src/mesa/main/api_arrayelt.h
+++ b/src/mesa/main/api_arrayelt.h
@@ -27,16 +27,57 @@
 #ifndef API_ARRAYELT_H
 #define API_ARRAYELT_H
 
-#include "mtypes.h"
+
+#include "main/mtypes.h"
+
+#if FEATURE_arrayelt
+
+#define _MESA_INIT_ARRAYELT_VTXFMT(vfmt, impl)     \
+   do {                                            \
+      (vfmt)->ArrayElement = impl ## ArrayElement; \
+   } while (0)
 
 extern GLboolean _ae_create_context( GLcontext *ctx );
 extern void _ae_destroy_context( GLcontext *ctx );
 extern void _ae_invalidate_state( GLcontext *ctx, GLuint new_state );
-extern void GLAPIENTRY _ae_loopback_array_elt( GLint elt );
+extern void GLAPIENTRY _ae_ArrayElement( GLint elt );
 
 /* May optionally be called before a batch of element calls:
  */
 extern void _ae_map_vbos( GLcontext *ctx );
 extern void _ae_unmap_vbos( GLcontext *ctx );
 
-#endif
+extern void
+_mesa_install_arrayelt_vtxfmt(struct _glapi_table *disp,
+                              const GLvertexformat *vfmt);
+
+#else /* FEATURE_arrayelt */
+
+#define _MESA_INIT_ARRAYELT_VTXFMT(vfmt, impl) do { } while (0)
+
+static INLINE GLboolean
+_ae_create_context( GLcontext *ctx )
+{
+   return GL_TRUE;
+}
+
+static INLINE void
+_ae_destroy_context( GLcontext *ctx )
+{
+}
+
+static INLINE void
+_ae_invalidate_state( GLcontext *ctx, GLuint new_state )
+{
+}
+
+static INLINE void
+_mesa_install_arrayelt_vtxfmt(struct _glapi_table *disp,
+                              const GLvertexformat *vfmt)
+{
+}
+
+#endif /* FEATURE_arrayelt */
+
+
+#endif /* API_ARRAYELT_H */
diff --git a/src/mesa/main/api_exec.c b/src/mesa/main/api_exec.c
index 02550ae108..1559984f43 100644
--- a/src/mesa/main/api_exec.c
+++ b/src/mesa/main/api_exec.c
@@ -30,9 +30,7 @@
 
 
 #include "mfeatures.h"
-#if FEATURE_accum
 #include "accum.h"
-#endif
 #include "api_loopback.h"
 #include "api_exec.h"
 #if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program
@@ -41,9 +39,7 @@
 #if FEATURE_ATI_fragment_shader
 #include "shader/atifragshader.h"
 #endif
-#if FEATURE_attrib_stack
 #include "attrib.h"
-#endif
 #include "blend.h"
 #if FEATURE_ARB_vertex_buffer_object
 #include "bufferobj.h"
@@ -54,29 +50,17 @@
 #endif
 #include "clear.h"
 #include "clip.h"
-#if FEATURE_colortable
 #include "colortab.h"
-#endif
 #include "context.h"
-#if FEATURE_convolve
 #include "convolve.h"
-#endif
 #include "depth.h"
-#if FEATURE_dlist
 #include "dlist.h"
-#endif
-#if FEATURE_drawpix
 #include "drawpix.h"
 #include "rastpos.h"
-#endif
 #include "enable.h"
-#if FEATURE_evaluators
 #include "eval.h"
-#endif
 #include "get.h"
-#if FEATURE_feedback
 #include "feedback.h"
-#endif
 #include "fog.h"
 #if FEATURE_EXT_framebuffer_object
 #include "fbobject.h"
@@ -84,24 +68,18 @@
 #include "ffvertex_prog.h"
 #include "framebuffer.h"
 #include "hint.h"
-#if FEATURE_histogram
 #include "histogram.h"
-#endif
 #include "imports.h"
 #include "light.h"
 #include "lines.h"
 #include "macros.h"
 #include "matrix.h"
 #include "multisample.h"
-#if FEATURE_pixel_transfer
 #include "pixel.h"
-#endif
 #include "pixelstore.h"
 #include "points.h"
 #include "polygon.h"
-#if FEATURE_ARB_occlusion_query || FEATURE_EXT_timer_query
 #include "queryobj.h"
-#endif
 #include "readpix.h"
 #include "scissor.h"
 #include "state.h"
@@ -109,9 +87,7 @@
 #include "texenv.h"
 #include "texgetimage.h"
 #include "teximage.h"
-#if FEATURE_texgen
 #include "texgen.h"
-#endif
 #include "texobj.h"
 #include "texparam.h"
 #include "texstate.h"
@@ -199,20 +175,10 @@ _mesa_init_exec_table(struct _glapi_table *exec)
    SET_TexParameteri(exec, _mesa_TexParameteri);
    SET_Translatef(exec, _mesa_Translatef);
    SET_Viewport(exec, _mesa_Viewport);
-#if FEATURE_accum
-   SET_Accum(exec, _mesa_Accum);
-   SET_ClearAccum(exec, _mesa_ClearAccum);
-#endif
-#if FEATURE_dlist
-   SET_CallList(exec, _mesa_CallList);
-   SET_CallLists(exec, _mesa_CallLists);
-   SET_DeleteLists(exec, _mesa_DeleteLists);
-   SET_EndList(exec, _mesa_EndList);
-   SET_GenLists(exec, _mesa_GenLists);
-   SET_IsList(exec, _mesa_IsList);
-   SET_ListBase(exec, _mesa_ListBase);
-   SET_NewList(exec, _mesa_NewList);
-#endif
+
+   _mesa_init_accum_dispatch(exec);
+   _mesa_init_dlist_dispatch(exec);
+
    SET_ClearDepth(exec, _mesa_ClearDepth);
    SET_ClearIndex(exec, _mesa_ClearIndex);
    SET_ClipPlane(exec, _mesa_ClipPlane);
@@ -222,21 +188,10 @@ _mesa_init_exec_table(struct _glapi_table *exec)
    SET_DepthFunc(exec, _mesa_DepthFunc);
    SET_DepthMask(exec, _mesa_DepthMask);
    SET_DepthRange(exec, _mesa_DepthRange);
-#if FEATURE_drawpix
-   SET_Bitmap(exec, _mesa_Bitmap);
-   SET_CopyPixels(exec, _mesa_CopyPixels);
-   SET_DrawPixels(exec, _mesa_DrawPixels);
-#endif
-#if FEATURE_feedback
-   SET_InitNames(exec, _mesa_InitNames);
-   SET_FeedbackBuffer(exec, _mesa_FeedbackBuffer);
-   SET_LoadName(exec, _mesa_LoadName);
-   SET_PassThrough(exec, _mesa_PassThrough);
-   SET_PopName(exec, _mesa_PopName);
-   SET_PushName(exec, _mesa_PushName);
-   SET_SelectBuffer(exec, _mesa_SelectBuffer);
-   SET_RenderMode(exec, _mesa_RenderMode);
-#endif
+
+   _mesa_init_drawpix_dispatch(exec);
+   _mesa_init_feedback_dispatch(exec);
+
    SET_FogCoordPointerEXT(exec, _mesa_FogCoordPointerEXT);
    SET_Fogf(exec, _mesa_Fogf);
    SET_Fogfv(exec, _mesa_Fogfv);
@@ -270,68 +225,22 @@ _mesa_init_exec_table(struct _glapi_table *exec)
    SET_Lighti(exec, _mesa_Lighti);
    SET_Lightiv(exec, _mesa_Lightiv);
    SET_LoadMatrixd(exec, _mesa_LoadMatrixd);
-#if FEATURE_evaluators
-   SET_GetMapdv(exec, _mesa_GetMapdv);
-   SET_GetMapfv(exec, _mesa_GetMapfv);
-   SET_GetMapiv(exec, _mesa_GetMapiv);
-   SET_Map1d(exec, _mesa_Map1d);
-   SET_Map1f(exec, _mesa_Map1f);
-   SET_Map2d(exec, _mesa_Map2d);
-   SET_Map2f(exec, _mesa_Map2f);
-   SET_MapGrid1d(exec, _mesa_MapGrid1d);
-   SET_MapGrid1f(exec, _mesa_MapGrid1f);
-   SET_MapGrid2d(exec, _mesa_MapGrid2d);
-   SET_MapGrid2f(exec, _mesa_MapGrid2f);
-#endif
+
+   _mesa_init_eval_dispatch(exec);
+
    SET_MultMatrixd(exec, _mesa_MultMatrixd);
-#if FEATURE_pixel_transfer
-   SET_GetPixelMapfv(exec, _mesa_GetPixelMapfv);
-   SET_GetPixelMapuiv(exec, _mesa_GetPixelMapuiv);
-   SET_GetPixelMapusv(exec, _mesa_GetPixelMapusv);
-   SET_PixelMapfv(exec, _mesa_PixelMapfv);
-   SET_PixelMapuiv(exec, _mesa_PixelMapuiv);
-   SET_PixelMapusv(exec, _mesa_PixelMapusv);
-   SET_PixelTransferf(exec, _mesa_PixelTransferf);
-   SET_PixelTransferi(exec, _mesa_PixelTransferi);
-   SET_PixelZoom(exec, _mesa_PixelZoom);
-#endif
+
+   _mesa_init_pixel_dispatch(exec);
+
    SET_PixelStoref(exec, _mesa_PixelStoref);
    SET_PointSize(exec, _mesa_PointSize);
    SET_PolygonMode(exec, _mesa_PolygonMode);
    SET_PolygonOffset(exec, _mesa_PolygonOffset);
    SET_PolygonStipple(exec, _mesa_PolygonStipple);
-#if FEATURE_attrib_stack
-   SET_PopAttrib(exec, _mesa_PopAttrib);
-   SET_PushAttrib(exec, _mesa_PushAttrib);
-   SET_PopClientAttrib(exec, _mesa_PopClientAttrib);
-   SET_PushClientAttrib(exec, _mesa_PushClientAttrib);
-#endif
-#if FEATURE_drawpix
-   SET_RasterPos2f(exec, _mesa_RasterPos2f);
-   SET_RasterPos2fv(exec, _mesa_RasterPos2fv);
-   SET_RasterPos2i(exec, _mesa_RasterPos2i);
-   SET_RasterPos2iv(exec, _mesa_RasterPos2iv);
-   SET_RasterPos2d(exec, _mesa_RasterPos2d);
-   SET_RasterPos2dv(exec, _mesa_RasterPos2dv);
-   SET_RasterPos2s(exec, _mesa_RasterPos2s);
-   SET_RasterPos2sv(exec, _mesa_RasterPos2sv);
-   SET_RasterPos3d(exec, _mesa_RasterPos3d);
-   SET_RasterPos3dv(exec, _mesa_RasterPos3dv);
-   SET_RasterPos3f(exec, _mesa_RasterPos3f);
-   SET_RasterPos3fv(exec, _mesa_RasterPos3fv);
-   SET_RasterPos3i(exec, _mesa_RasterPos3i);
-   SET_RasterPos3iv(exec, _mesa_RasterPos3iv);
-   SET_RasterPos3s(exec, _mesa_RasterPos3s);
-   SET_RasterPos3sv(exec, _mesa_RasterPos3sv);
-   SET_RasterPos4d(exec, _mesa_RasterPos4d);
-   SET_RasterPos4dv(exec, _mesa_RasterPos4dv);
-   SET_RasterPos4f(exec, _mesa_RasterPos4f);
-   SET_RasterPos4fv(exec, _mesa_RasterPos4fv);
-   SET_RasterPos4i(exec, _mesa_RasterPos4i);
-   SET_RasterPos4iv(exec, _mesa_RasterPos4iv);
-   SET_RasterPos4s(exec, _mesa_RasterPos4s);
-   SET_RasterPos4sv(exec, _mesa_RasterPos4sv);
-#endif
+
+   _mesa_init_attrib_dispatch(exec);
+   _mesa_init_rastpos_dispatch(exec);
+
    SET_ReadPixels(exec, _mesa_ReadPixels);
    SET_Rotated(exec, _mesa_Rotated);
    SET_Scaled(exec, _mesa_Scaled);
@@ -339,17 +248,7 @@ _mesa_init_exec_table(struct _glapi_table *exec)
    SET_TexEnvf(exec, _mesa_TexEnvf);
    SET_TexEnviv(exec, _mesa_TexEnviv);
 
-#if FEATURE_texgen
-   SET_GetTexGendv(exec, _mesa_GetTexGendv);
-   SET_GetTexGenfv(exec, _mesa_GetTexGenfv);
-   SET_GetTexGeniv(exec, _mesa_GetTexGeniv);
-   SET_TexGend(exec, _mesa_TexGend);
-   SET_TexGendv(exec, _mesa_TexGendv);
-   SET_TexGenf(exec, _mesa_TexGenf);
-   SET_TexGenfv(exec, _mesa_TexGenfv);
-   SET_TexGeni(exec, _mesa_TexGeni);
-   SET_TexGeniv(exec, _mesa_TexGeniv);
-#endif
+   _mesa_init_texgen_dispatch(exec);
 
    SET_TexImage1D(exec, _mesa_TexImage1D);
    SET_TexParameterf(exec, _mesa_TexParameterf);
@@ -395,45 +294,9 @@ _mesa_init_exec_table(struct _glapi_table *exec)
    SET_BlendEquation(exec, _mesa_BlendEquation);
    SET_BlendEquationSeparateEXT(exec, _mesa_BlendEquationSeparateEXT);
 
-#if FEATURE_colortable
-   SET_ColorSubTable(exec, _mesa_ColorSubTable);
-   SET_ColorTable(exec, _mesa_ColorTable);
-   SET_ColorTableParameterfv(exec, _mesa_ColorTableParameterfv);
-   SET_ColorTableParameteriv(exec, _mesa_ColorTableParameteriv);
-   SET_CopyColorSubTable(exec, _mesa_CopyColorSubTable);
-   SET_CopyColorTable(exec, _mesa_CopyColorTable);
-   SET_GetColorTable(exec, _mesa_GetColorTable);
-   SET_GetColorTableParameterfv(exec, _mesa_GetColorTableParameterfv);
-   SET_GetColorTableParameteriv(exec, _mesa_GetColorTableParameteriv);
-#endif
-
-#if FEATURE_convolve
-   SET_ConvolutionFilter1D(exec, _mesa_ConvolutionFilter1D);
-   SET_ConvolutionFilter2D(exec, _mesa_ConvolutionFilter2D);
-   SET_ConvolutionParameterf(exec, _mesa_ConvolutionParameterf);
-   SET_ConvolutionParameterfv(exec, _mesa_ConvolutionParameterfv);
-   SET_ConvolutionParameteri(exec, _mesa_ConvolutionParameteri);
-   SET_ConvolutionParameteriv(exec, _mesa_ConvolutionParameteriv);
-   SET_CopyConvolutionFilter1D(exec, _mesa_CopyConvolutionFilter1D);
-   SET_CopyConvolutionFilter2D(exec, _mesa_CopyConvolutionFilter2D);
-   SET_GetConvolutionFilter(exec, _mesa_GetConvolutionFilter);
-   SET_GetConvolutionParameterfv(exec, _mesa_GetConvolutionParameterfv);
-   SET_GetConvolutionParameteriv(exec, _mesa_GetConvolutionParameteriv);
-   SET_SeparableFilter2D(exec, _mesa_SeparableFilter2D);
-#endif
-#if FEATURE_histogram
-   SET_GetHistogram(exec, _mesa_GetHistogram);
-   SET_GetHistogramParameterfv(exec, _mesa_GetHistogramParameterfv);
-   SET_GetHistogramParameteriv(exec, _mesa_GetHistogramParameteriv);
-   SET_GetMinmax(exec, _mesa_GetMinmax);
-   SET_GetMinmaxParameterfv(exec, _mesa_GetMinmaxParameterfv);
-   SET_GetMinmaxParameteriv(exec, _mesa_GetMinmaxParameteriv);
-   SET_GetSeparableFilter(exec, _mesa_GetSeparableFilter);
-   SET_Histogram(exec, _mesa_Histogram);
-   SET_Minmax(exec, _mesa_Minmax);
-   SET_ResetHistogram(exec, _mesa_ResetHistogram);
-   SET_ResetMinmax(exec, _mesa_ResetMinmax);
-#endif
+   _mesa_init_colortable_dispatch(exec);
+   _mesa_init_convolve_dispatch(exec);
+   _mesa_init_histogram_dispatch(exec);
 
    /* OpenGL 2.0 */
    SET_StencilFuncSeparate(exec, _mesa_StencilFuncSeparate);
@@ -545,32 +408,7 @@ _mesa_init_exec_table(struct _glapi_table *exec)
 #endif
 
    /* 197. GL_MESA_window_pos */
-#if FEATURE_drawpix
-   SET_WindowPos2dMESA(exec, _mesa_WindowPos2dMESA);
-   SET_WindowPos2dvMESA(exec, _mesa_WindowPos2dvMESA);
-   SET_WindowPos2fMESA(exec, _mesa_WindowPos2fMESA);
-   SET_WindowPos2fvMESA(exec, _mesa_WindowPos2fvMESA);
-   SET_WindowPos2iMESA(exec, _mesa_WindowPos2iMESA);
-   SET_WindowPos2ivMESA(exec, _mesa_WindowPos2ivMESA);
-   SET_WindowPos2sMESA(exec, _mesa_WindowPos2sMESA);
-   SET_WindowPos2svMESA(exec, _mesa_WindowPos2svMESA);
-   SET_WindowPos3dMESA(exec, _mesa_WindowPos3dMESA);
-   SET_WindowPos3dvMESA(exec, _mesa_WindowPos3dvMESA);
-   SET_WindowPos3fMESA(exec, _mesa_WindowPos3fMESA);
-   SET_WindowPos3fvMESA(exec, _mesa_WindowPos3fvMESA);
-   SET_WindowPos3iMESA(exec, _mesa_WindowPos3iMESA);
-   SET_WindowPos3ivMESA(exec, _mesa_WindowPos3ivMESA);
-   SET_WindowPos3sMESA(exec, _mesa_WindowPos3sMESA);
-   SET_WindowPos3svMESA(exec, _mesa_WindowPos3svMESA);
-   SET_WindowPos4dMESA(exec, _mesa_WindowPos4dMESA);
-   SET_WindowPos4dvMESA(exec, _mesa_WindowPos4dvMESA);
-   SET_WindowPos4fMESA(exec, _mesa_WindowPos4fMESA);
-   SET_WindowPos4fvMESA(exec, _mesa_WindowPos4fvMESA);
-   SET_WindowPos4iMESA(exec, _mesa_WindowPos4iMESA);
-   SET_WindowPos4ivMESA(exec, _mesa_WindowPos4ivMESA);
-   SET_WindowPos4sMESA(exec, _mesa_WindowPos4sMESA);
-   SET_WindowPos4svMESA(exec, _mesa_WindowPos4svMESA);
-#endif
+   /* part of _mesa_init_rastpos_dispatch(exec); */
 
    /* 200. GL_IBM_multimode_draw_arrays */
 #if _HAVE_FULL_GL
@@ -762,16 +600,7 @@ _mesa_init_exec_table(struct _glapi_table *exec)
 #endif
 
    /* ARB 29. GL_ARB_occlusion_query */
-#if FEATURE_ARB_occlusion_query
-   SET_GenQueriesARB(exec, _mesa_GenQueriesARB);
-   SET_DeleteQueriesARB(exec, _mesa_DeleteQueriesARB);
-   SET_IsQueryARB(exec, _mesa_IsQueryARB);
-   SET_BeginQueryARB(exec, _mesa_BeginQueryARB);
-   SET_EndQueryARB(exec, _mesa_EndQueryARB);
-   SET_GetQueryivARB(exec, _mesa_GetQueryivARB);
-   SET_GetQueryObjectivARB(exec, _mesa_GetQueryObjectivARB);
-   SET_GetQueryObjectuivARB(exec, _mesa_GetQueryObjectuivARB);
-#endif
+   _mesa_init_queryobj_dispatch(exec);
 
    /* ARB 37. GL_ARB_draw_buffers */
 #if FEATURE_draw_read_buffer
@@ -881,11 +710,6 @@ _mesa_init_exec_table(struct _glapi_table *exec)
    SET_GenerateMipmapEXT(exec, _mesa_GenerateMipmapEXT);
 #endif
 
-#if FEATURE_EXT_timer_query
-   SET_GetQueryObjecti64vEXT(exec, _mesa_GetQueryObjecti64vEXT);
-   SET_GetQueryObjectui64vEXT(exec, _mesa_GetQueryObjectui64vEXT);
-#endif
-
 #if FEATURE_EXT_framebuffer_blit
    SET_BlitFramebufferEXT(exec, _mesa_BlitFramebufferEXT);
 #endif
diff --git a/src/mesa/main/api_loopback.c b/src/mesa/main/api_loopback.c
index 0e3f5ff957..3d466ac44a 100644
--- a/src/mesa/main/api_loopback.c
+++ b/src/mesa/main/api_loopback.c
@@ -77,6 +77,10 @@
 #define FOGCOORDF(x)                CALL_FogCoordfEXT(GET_DISPATCH(), (x))
 #define SECONDARYCOLORF(a,b,c)      CALL_SecondaryColor3fEXT(GET_DISPATCH(), (a,b,c))
 
+
+#if FEATURE_beginend
+
+
 static void GLAPIENTRY
 loopback_Color3b_f( GLbyte red, GLbyte green, GLbyte blue )
 {
@@ -1655,3 +1659,6 @@ _mesa_loopback_init_api_table( struct _glapi_table *dest )
    SET_VertexAttrib4NusvARB(dest, loopback_VertexAttrib4NusvARB);
    SET_VertexAttrib4NuivARB(dest, loopback_VertexAttrib4NuivARB);
 }
+
+
+#endif /* FEATURE_beginend */
diff --git a/src/mesa/main/api_loopback.h b/src/mesa/main/api_loopback.h
index 6f85bbc1d9..3140eb515e 100644
--- a/src/mesa/main/api_loopback.h
+++ b/src/mesa/main/api_loopback.h
@@ -27,11 +27,19 @@
 #ifndef API_LOOPBACK_H
 #define API_LOOPBACK_H
 
-#include "glheader.h"
+#include "main/mtypes.h"
 
-
-struct _glapi_table;
+#if FEATURE_beginend
 
 extern void _mesa_loopback_init_api_table( struct _glapi_table *dest );
 
-#endif
+#else /* FEATURE_beginend */
+
+static INLINE void
+_mesa_loopback_init_api_table( struct _glapi_table *dest )
+{
+}
+
+#endif /* FEATURE_beginend */
+
+#endif /* API_LOOPBACK_H */
diff --git a/src/mesa/main/api_noop.c b/src/mesa/main/api_noop.c
index 09ba7e5062..f72f957300 100644
--- a/src/mesa/main/api_noop.c
+++ b/src/mesa/main/api_noop.c
@@ -30,9 +30,8 @@
 #include "context.h"
 #include "light.h"
 #include "macros.h"
-#if FEATURE_dlist
 #include "dlist.h"
-#endif
+#include "eval.h"
 #include "glapi/dispatch.h"
 
 
@@ -44,6 +43,9 @@
  */
 
 
+#if FEATURE_beginend
+
+
 static void GLAPIENTRY _mesa_noop_EdgeFlag( GLboolean b )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -731,7 +733,7 @@ _mesa_noop_DrawElements(GLenum mode, GLsizei count, GLenum type,
    GET_CURRENT_CONTEXT(ctx);
    GLint i;
 
-   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
+   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices, 0 ))
       return;
 
    CALL_Begin(GET_DISPATCH(), (mode));
@@ -757,6 +759,43 @@ _mesa_noop_DrawElements(GLenum mode, GLsizei count, GLenum type,
    CALL_End(GET_DISPATCH(), ());
 }
 
+static void GLAPIENTRY
+_mesa_noop_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+				  const GLvoid *indices, GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices,
+				     basevertex ))
+      return;
+
+   CALL_Begin(GET_DISPATCH(), (mode));
+
+   switch (type) {
+   case GL_UNSIGNED_BYTE:
+      for (i = 0 ; i < count ; i++)
+	  CALL_ArrayElement(GET_DISPATCH(), ( ((GLubyte *)indices)[i] +
+					      basevertex));
+      break;
+   case GL_UNSIGNED_SHORT:
+      for (i = 0 ; i < count ; i++)
+	  CALL_ArrayElement(GET_DISPATCH(), ( ((GLushort *)indices)[i] +
+					      basevertex ));
+      break;
+   case GL_UNSIGNED_INT:
+      for (i = 0 ; i < count ; i++)
+	  CALL_ArrayElement(GET_DISPATCH(), ( ((GLuint *)indices)[i] +
+					      basevertex ));
+      break;
+   default:
+      _mesa_error( ctx, GL_INVALID_ENUM, "glDrawElementsBaseVertex(type)" );
+      break;
+   }
+
+   CALL_End(GET_DISPATCH(), ());
+}
+
 
 static void GLAPIENTRY
 _mesa_noop_DrawRangeElements(GLenum mode,
@@ -768,7 +807,7 @@ _mesa_noop_DrawRangeElements(GLenum mode,
 
    if (_mesa_validate_DrawRangeElements( ctx, mode,
 					 start, end,
-					 count, type, indices ))
+					 count, type, indices, 0 ))
        CALL_DrawElements(GET_DISPATCH(), (mode, count, type, indices));
 }
 
@@ -786,6 +825,40 @@ _mesa_noop_MultiDrawElements(GLenum mode, const GLsizei *count, GLenum type,
    }
 }
 
+static void GLAPIENTRY
+_mesa_noop_DrawRangeElementsBaseVertex(GLenum mode,
+				       GLuint start, GLuint end,
+				       GLsizei count, GLenum type,
+				       const GLvoid *indices, GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (_mesa_validate_DrawRangeElements( ctx, mode,
+					 start, end,
+					 count, type, indices, basevertex ))
+      CALL_DrawElementsBaseVertex(GET_DISPATCH(), (mode, count, type, indices,
+						   basevertex));
+}
+
+/* GL_EXT_multi_draw_arrays */
+void GLAPIENTRY
+_mesa_noop_MultiDrawElementsBaseVertex(GLenum mode, const GLsizei *count,
+				       GLenum type,
+				       const GLvoid **indices,
+				       GLsizei primcount,
+				       const GLint *basevertex)
+{
+   GLsizei i;
+
+   for (i = 0; i < primcount; i++) {
+      if (count[i] > 0) {
+	 CALL_DrawElementsBaseVertex(GET_DISPATCH(), (mode, count[i], type,
+						      indices[i],
+						      basevertex[i]));
+      }
+   }
+}
+
 /*
  * Eval Mesh
  */
@@ -921,26 +994,21 @@ _mesa_noop_EvalMesh2( GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2 )
 void
 _mesa_noop_vtxfmt_init( GLvertexformat *vfmt )
 {
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
+   _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_);
+
    vfmt->Begin = _mesa_noop_Begin;
-#if FEATURE_dlist
-   vfmt->CallList = _mesa_CallList;
-   vfmt->CallLists = _mesa_CallLists;
-#endif
+
+   _MESA_INIT_DLIST_VTXFMT(vfmt, _mesa_);
+
    vfmt->Color3f = _mesa_noop_Color3f;
    vfmt->Color3fv = _mesa_noop_Color3fv;
    vfmt->Color4f = _mesa_noop_Color4f;
    vfmt->Color4fv = _mesa_noop_Color4fv;
    vfmt->EdgeFlag = _mesa_noop_EdgeFlag;
    vfmt->End = _mesa_noop_End;
-#if FEATURE_evaluators
-   vfmt->EvalCoord1f = _mesa_noop_EvalCoord1f;
-   vfmt->EvalCoord1fv = _mesa_noop_EvalCoord1fv;
-   vfmt->EvalCoord2f = _mesa_noop_EvalCoord2f;
-   vfmt->EvalCoord2fv = _mesa_noop_EvalCoord2fv;
-   vfmt->EvalPoint1 = _mesa_noop_EvalPoint1;
-   vfmt->EvalPoint2 = _mesa_noop_EvalPoint2;
-#endif
+
+   _MESA_INIT_EVAL_VTXFMT(vfmt, _mesa_noop_);
+
    vfmt->FogCoordfEXT = _mesa_noop_FogCoordfEXT;
    vfmt->FogCoordfvEXT = _mesa_noop_FogCoordfvEXT;
    vfmt->Indexf = _mesa_noop_Indexf;
@@ -995,6 +1063,10 @@ _mesa_noop_vtxfmt_init( GLvertexformat *vfmt )
    vfmt->DrawElements = _mesa_noop_DrawElements;
    vfmt->DrawRangeElements = _mesa_noop_DrawRangeElements;
    vfmt->MultiDrawElementsEXT = _mesa_noop_MultiDrawElements;
-   vfmt->EvalMesh1 = _mesa_noop_EvalMesh1;
-   vfmt->EvalMesh2 = _mesa_noop_EvalMesh2;
+   vfmt->DrawElementsBaseVertex = _mesa_noop_DrawElementsBaseVertex;
+   vfmt->DrawRangeElementsBaseVertex = _mesa_noop_DrawRangeElementsBaseVertex;
+   vfmt->MultiDrawElementsBaseVertex = _mesa_noop_MultiDrawElementsBaseVertex;
 }
+
+
+#endif /* FEATURE_beginend */
diff --git a/src/mesa/main/api_noop.h b/src/mesa/main/api_noop.h
index a7956d00b3..e7fd49bafb 100644
--- a/src/mesa/main/api_noop.h
+++ b/src/mesa/main/api_noop.h
@@ -25,8 +25,9 @@
 #ifndef _API_NOOP_H
 #define _API_NOOP_H
 
-#include "mtypes.h"
-#include "context.h"
+#include "main/mtypes.h"
+
+#if FEATURE_beginend
 
 extern void GLAPIENTRY
 _mesa_noop_Rectf(GLfloat x1, GLfloat y1, GLfloat x2, GLfloat y2);
@@ -44,7 +45,16 @@ extern void GLAPIENTRY
 _mesa_noop_MultiDrawElements(GLenum mode, const GLsizei *count, GLenum type,
 			     const GLvoid **indices, GLsizei primcount);
 
+extern void GLAPIENTRY
+_mesa_noop_MultiDrawElementsBaseVertex(GLenum mode, const GLsizei *count,
+				       GLenum type,
+				       const GLvoid **indices,
+				       GLsizei primcount,
+				       const GLint *basevertex);
+
 extern void
 _mesa_noop_vtxfmt_init(GLvertexformat *vfmt);
 
-#endif
+#endif /* FEATURE_beginend */
+
+#endif /* _API_NOOP_H */
diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c
index 507e21fe81..e71e5a6ce8 100644
--- a/src/mesa/main/api_validate.c
+++ b/src/mesa/main/api_validate.c
@@ -29,7 +29,7 @@
 #include "imports.h"
 #include "mtypes.h"
 #include "state.h"
-
+#include "vbo/vbo.h"
 
 
 /**
@@ -124,6 +124,40 @@ check_valid_to_render(GLcontext *ctx, const char *function)
    return GL_TRUE;
 }
 
+static GLboolean
+check_index_bounds(GLcontext *ctx, GLsizei count, GLenum type,
+		   const GLvoid *indices, GLint basevertex)
+{
+   struct _mesa_prim prim;
+   struct _mesa_index_buffer ib;
+   GLuint min, max;
+
+   /* Only the X Server needs to do this -- otherwise, accessing outside
+    * array/BO bounds allows application termination.
+    */
+   if (!ctx->Const.CheckArrayBounds)
+      return GL_TRUE;
+
+   memset(&prim, 0, sizeof(prim));
+   prim.count = count;
+
+   memset(&ib, 0, sizeof(ib));
+   ib.type = type;
+   ib.ptr = indices;
+   ib.obj = ctx->Array.ElementArrayBufferObj;
+
+   vbo_get_minmax_index(ctx, &prim, &ib, &min, &max);
+
+   if (min + basevertex < 0 ||
+       max + basevertex > ctx->Array.ArrayObj->_MaxElement) {
+      /* the max element is out of bounds of one or more enabled arrays */
+      _mesa_warning(ctx, "glDrawElements() index=%u is "
+		    "out of bounds (max=%u)", max, ctx->Array.ArrayObj->_MaxElement);
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
 
 /**
  * Error checking for glDrawElements().  Includes parameter checking
@@ -133,7 +167,7 @@ check_valid_to_render(GLcontext *ctx, const char *function)
 GLboolean
 _mesa_validate_DrawElements(GLcontext *ctx,
 			    GLenum mode, GLsizei count, GLenum type,
-			    const GLvoid *indices)
+			    const GLvoid *indices, GLint basevertex)
 {
    ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx,  GL_FALSE);
 
@@ -177,17 +211,8 @@ _mesa_validate_DrawElements(GLcontext *ctx,
          return GL_FALSE;
    }
 
-   if (ctx->Const.CheckArrayBounds) {
-      /* find max array index */
-      GLuint max = _mesa_max_buffer_index(ctx, count, type, indices,
-                                          ctx->Array.ElementArrayBufferObj);
-      if (max >= ctx->Array.ArrayObj->_MaxElement) {
-         /* the max element is out of bounds of one or more enabled arrays */
-         _mesa_warning(ctx, "glDrawElements() index=%u is "
-                       "out of bounds (max=%u)", max, ctx->Array.ArrayObj->_MaxElement);
-         return GL_FALSE;
-      }
-   }
+   if (!check_index_bounds(ctx, count, type, indices, basevertex))
+      return GL_FALSE;
 
    return GL_TRUE;
 }
@@ -202,7 +227,7 @@ GLboolean
 _mesa_validate_DrawRangeElements(GLcontext *ctx, GLenum mode,
 				 GLuint start, GLuint end,
 				 GLsizei count, GLenum type,
-				 const GLvoid *indices)
+				 const GLvoid *indices, GLint basevertex)
 {
    ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx, GL_FALSE);
 
@@ -250,14 +275,8 @@ _mesa_validate_DrawRangeElements(GLcontext *ctx, GLenum mode,
          return GL_FALSE;
    }
 
-   if (ctx->Const.CheckArrayBounds) {
-      GLuint max = _mesa_max_buffer_index(ctx, count, type, indices,
-                                          ctx->Array.ElementArrayBufferObj);
-      if (max >= ctx->Array.ArrayObj->_MaxElement) {
-         /* the max element is out of bounds of one or more enabled arrays */
-         return GL_FALSE;
-      }
-   }
+   if (!check_index_bounds(ctx, count, type, indices, basevertex))
+      return GL_FALSE;
 
    return GL_TRUE;
 }
diff --git a/src/mesa/main/api_validate.h b/src/mesa/main/api_validate.h
index ff82a2966c..6064d15fe6 100644
--- a/src/mesa/main/api_validate.h
+++ b/src/mesa/main/api_validate.h
@@ -43,13 +43,13 @@ _mesa_validate_DrawArrays(GLcontext *ctx,
 extern GLboolean
 _mesa_validate_DrawElements(GLcontext *ctx,
 			    GLenum mode, GLsizei count, GLenum type,
-			    const GLvoid *indices);
+			    const GLvoid *indices, GLint basevertex);
 
 extern GLboolean
 _mesa_validate_DrawRangeElements(GLcontext *ctx, GLenum mode,
 				 GLuint start, GLuint end,
 				 GLsizei count, GLenum type,
-				 const GLvoid *indices);
+				 const GLvoid *indices, GLint basevertex);
 
 
 #endif
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index ab99ca1c64..246c5521b7 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -57,6 +57,7 @@
 #include "varray.h"
 #include "viewport.h"
 #include "mtypes.h"
+#include "glapi/dispatch.h"
 
 
 /**
@@ -74,6 +75,7 @@ struct gl_enable_attrib
    GLboolean Convolution2D;
    GLboolean Separable2D;
    GLboolean CullFace;
+   GLboolean DepthClamp;
    GLboolean DepthTest;
    GLboolean Dither;
    GLboolean Fog;
@@ -173,6 +175,9 @@ struct texture_state
 };
 
 
+#if FEATURE_attrib_stack
+
+
 /**
  * Allocate new attribute node of given type/kind.  Attach payload data.
  * Insert it into the linked list named by 'head'.
@@ -265,6 +270,7 @@ _mesa_PushAttrib(GLbitfield mask)
       attr->Convolution2D = ctx->Pixel.Convolution2DEnabled;
       attr->Separable2D = ctx->Pixel.Separable2DEnabled;
       attr->CullFace = ctx->Polygon.CullFlag;
+      attr->DepthClamp = ctx->Transform.DepthClamp;
       attr->DepthTest = ctx->Depth.Test;
       attr->Dither = ctx->Color.DitherFlag;
       attr->Fog = ctx->Fog.Enabled;
@@ -514,6 +520,8 @@ pop_enable_group(GLcontext *ctx, const struct gl_enable_attrib *enable)
                    enable->ColorTable[COLORTABLE_POSTCOLORMATRIX],
                    GL_POST_COLOR_MATRIX_COLOR_TABLE);
    TEST_AND_UPDATE(ctx->Polygon.CullFlag, enable->CullFace, GL_CULL_FACE);
+   TEST_AND_UPDATE(ctx->Transform.DepthClamp, enable->DepthClamp,
+		   GL_DEPTH_CLAMP);
    TEST_AND_UPDATE(ctx->Depth.Test, enable->DepthTest, GL_DEPTH_TEST);
    TEST_AND_UPDATE(ctx->Color.DitherFlag, enable->Dither, GL_DITHER);
    TEST_AND_UPDATE(ctx->Pixel.Convolution1DEnabled, enable->Convolution1D,
@@ -1221,6 +1229,9 @@ _mesa_PopAttrib(void)
                if (xform->RescaleNormals != ctx->Transform.RescaleNormals)
                   _mesa_set_enable(ctx, GL_RESCALE_NORMAL_EXT,
                                    ctx->Transform.RescaleNormals);
+               if (xform->DepthClamp != ctx->Transform.DepthClamp)
+                  _mesa_set_enable(ctx, GL_DEPTH_CLAMP,
+                                   ctx->Transform.DepthClamp);
             }
             break;
          case GL_TEXTURE_BIT:
@@ -1457,6 +1468,19 @@ _mesa_PopClientAttrib(void)
 }
 
 
+void
+_mesa_init_attrib_dispatch(struct _glapi_table *disp)
+{
+   SET_PopAttrib(disp, _mesa_PopAttrib);
+   SET_PushAttrib(disp, _mesa_PushAttrib);
+   SET_PopClientAttrib(disp, _mesa_PopClientAttrib);
+   SET_PushClientAttrib(disp, _mesa_PushClientAttrib);
+}
+
+
+#endif /* FEATURE_attrib_stack */
+
+
 /**
  * Free any attribute state data that might be attached to the context.
  */
diff --git a/src/mesa/main/attrib.h b/src/mesa/main/attrib.h
index 2cf8fe6934..6b48a17663 100644
--- a/src/mesa/main/attrib.h
+++ b/src/mesa/main/attrib.h
@@ -26,10 +26,10 @@
 #define ATTRIB_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 
-#if _HAVE_FULL_GL
+#if FEATURE_attrib_stack
 
 extern void GLAPIENTRY
 _mesa_PushAttrib( GLbitfield mask );
@@ -43,18 +43,34 @@ _mesa_PushClientAttrib( GLbitfield mask );
 extern void GLAPIENTRY
 _mesa_PopClientAttrib( void );
 
+extern void
+_mesa_init_attrib_dispatch(struct _glapi_table *disp);
+
+#else /* FEATURE_attrib_stack */
+
+static INLINE void
+_mesa_PushClientAttrib( GLbitfield mask )
+{
+   ASSERT_NO_FEATURE();
+}
+
+static INLINE void
+_mesa_PopClientAttrib( void )
+{
+   ASSERT_NO_FEATURE();
+}
+
+static INLINE void
+_mesa_init_attrib_dispatch(struct _glapi_table *disp)
+{
+}
+
+#endif /* FEATURE_attrib_stack */
+
 extern void 
 _mesa_init_attrib( GLcontext *ctx );
 
 extern void 
 _mesa_free_attrib_data( GLcontext *ctx );
 
-#else
-
-/** No-op */
-#define _mesa_init_attrib( c ) ((void)0)
-#define _mesa_free_attrib_data( c ) ((void)0)
-
-#endif
-
-#endif
+#endif /* ATTRIB_H */
diff --git a/src/mesa/main/colortab.c b/src/mesa/main/colortab.c
index 5a7de5f209..5ede76c1fb 100644
--- a/src/mesa/main/colortab.c
+++ b/src/mesa/main/colortab.c
@@ -31,6 +31,11 @@
 #include "macros.h"
 #include "state.h"
 #include "teximage.h"
+#include "texstate.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_colortable
 
 
 /**
@@ -278,7 +283,7 @@ _mesa_ColorTable( GLenum target, GLenum internalFormat,
    static const GLfloat one[4] = { 1.0, 1.0, 1.0, 1.0 };
    static const GLfloat zero[4] = { 0.0, 0.0, 0.0, 0.0 };
    GET_CURRENT_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
    struct gl_texture_object *texObj = NULL;
    struct gl_color_table *table = NULL;
    GLboolean proxy = GL_FALSE;
@@ -443,7 +448,7 @@ _mesa_ColorSubTable( GLenum target, GLsizei start,
    static const GLfloat one[4] = { 1.0, 1.0, 1.0, 1.0 };
    static const GLfloat zero[4] = { 0.0, 0.0, 0.0, 0.0 };
    GET_CURRENT_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
    struct gl_texture_object *texObj = NULL;
    struct gl_color_table *table = NULL;
    const GLfloat *scale = one, *bias = zero;
@@ -535,37 +540,44 @@ _mesa_ColorSubTable( GLenum target, GLsizei start,
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_CopyColorTable(GLenum target, GLenum internalformat,
                      GLint x, GLint y, GLsizei width)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
-   /* Select buffer to read from */
+   if (!ctx->ReadBuffer->_ColorReadBuffer) {
+      return;      /* no readbuffer - OK */
+   }
+
    ctx->Driver.CopyColorTable( ctx, target, internalformat, x, y, width );
 }
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_CopyColorSubTable(GLenum target, GLsizei start,
                         GLint x, GLint y, GLsizei width)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
+   if (!ctx->ReadBuffer->_ColorReadBuffer) {
+      return;      /* no readbuffer - OK */
+   }
+
    ctx->Driver.CopyColorSubTable( ctx, target, start, x, y, width );
 }
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetColorTable( GLenum target, GLenum format,
                      GLenum type, GLvoid *data )
 {
    GET_CURRENT_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
    struct gl_color_table *table = NULL;
    GLfloat rgba[MAX_COLOR_TABLE_SIZE][4];
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
@@ -694,7 +706,7 @@ _mesa_GetColorTable( GLenum target, GLenum format,
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 {
    GLfloat *scale, *bias;
@@ -739,7 +751,7 @@ _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params)
 {
    GLfloat fparams[4];
@@ -762,11 +774,11 @@ _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params)
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params )
 {
    GET_CURRENT_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
    struct gl_color_table *table = NULL;
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -889,11 +901,11 @@ _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params )
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params )
 {
    GET_CURRENT_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
    struct gl_color_table *table = NULL;
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -1044,6 +1056,25 @@ _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params )
    }
 }
 
+
+void
+_mesa_init_colortable_dispatch(struct _glapi_table *disp)
+{
+   SET_ColorSubTable(disp, _mesa_ColorSubTable);
+   SET_ColorTable(disp, _mesa_ColorTable);
+   SET_ColorTableParameterfv(disp, _mesa_ColorTableParameterfv);
+   SET_ColorTableParameteriv(disp, _mesa_ColorTableParameteriv);
+   SET_CopyColorSubTable(disp, _mesa_CopyColorSubTable);
+   SET_CopyColorTable(disp, _mesa_CopyColorTable);
+   SET_GetColorTable(disp, _mesa_GetColorTable);
+   SET_GetColorTableParameterfv(disp, _mesa_GetColorTableParameterfv);
+   SET_GetColorTableParameteriv(disp, _mesa_GetColorTableParameteriv);
+}
+
+
+#endif /* FEATURE_colortable */
+
+
 /**********************************************************************/
 /*****                      Initialization                        *****/
 /**********************************************************************/
diff --git a/src/mesa/main/colortab.h b/src/mesa/main/colortab.h
index b6ff737a65..652fb58246 100644
--- a/src/mesa/main/colortab.h
+++ b/src/mesa/main/colortab.h
@@ -27,9 +27,16 @@
 #define COLORTAB_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
-#if _HAVE_FULL_GL
+#if FEATURE_colortable
+
+#define _MESA_INIT_COLORTABLE_FUNCTIONS(driver, impl)                \
+   do {                                                              \
+      (driver)->CopyColorTable       = impl ## CopyColorTable;       \
+      (driver)->CopyColorSubTable    = impl ## CopyColorSubTable;    \
+      (driver)->UpdateTexturePalette = impl ## UpdateTexturePalette; \
+   } while (0)
 
 extern void GLAPIENTRY
 _mesa_ColorTable( GLenum target, GLenum internalformat,
@@ -41,32 +48,35 @@ _mesa_ColorSubTable( GLenum target, GLsizei start,
                      GLsizei count, GLenum format, GLenum type,
                      const GLvoid *table );
 
-extern void GLAPIENTRY
-_mesa_CopyColorSubTable(GLenum target, GLsizei start,
-                        GLint x, GLint y, GLsizei width);
-
-extern void GLAPIENTRY
-_mesa_CopyColorTable(GLenum target, GLenum internalformat,
-                     GLint x, GLint y, GLsizei width);
+extern void
+_mesa_init_colortable_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY
-_mesa_GetColorTable( GLenum target, GLenum format,
-                     GLenum type, GLvoid *table );
+#else /* FEATURE_colortable */
 
-extern void GLAPIENTRY
-_mesa_ColorTableParameterfv(GLenum target, GLenum pname,
-                            const GLfloat *params);
+#define _MESA_INIT_COLORTABLE_FUNCTIONS(driver, impl) do { } while (0)
 
-extern void GLAPIENTRY
-_mesa_ColorTableParameteriv(GLenum target, GLenum pname,
-                            const GLint *params);
+static INLINE void GLAPIENTRY
+_mesa_ColorTable( GLenum target, GLenum internalformat,
+                  GLsizei width, GLenum format, GLenum type,
+                  const GLvoid *table )
+{
+   ASSERT_NO_FEATURE();
+}
 
-extern void GLAPIENTRY
-_mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params );
+static INLINE void GLAPIENTRY
+_mesa_ColorSubTable( GLenum target, GLsizei start,
+                     GLsizei count, GLenum format, GLenum type,
+                     const GLvoid *table )
+{
+   ASSERT_NO_FEATURE();
+}
 
-extern void GLAPIENTRY
-_mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params );
+static INLINE void
+_mesa_init_colortable_dispatch(struct _glapi_table *disp)
+{
+}
 
+#endif /* FEATURE_colortable */
 
 
 extern void
@@ -81,20 +91,5 @@ _mesa_init_colortables( GLcontext *ctx );
 extern void 
 _mesa_free_colortables_data( GLcontext *ctx );
 
-#else
-
-/** No-op */
-#define _mesa_init_colortable( p ) ((void) 0)
-
-/** No-op */
-#define _mesa_free_colortable_data( p ) ((void) 0)
-
-/** No-op */
-#define _mesa_init_colortables( p ) ((void)0)
-
-/** No-op */
-#define _mesa_free_colortables_data( p ) ((void)0)
-
-#endif
 
-#endif
+#endif /* COLORTAB_H */
diff --git a/src/mesa/main/compiler.h b/src/mesa/main/compiler.h
index 9319505a75..380663ec97 100644
--- a/src/mesa/main/compiler.h
+++ b/src/mesa/main/compiler.h
@@ -107,8 +107,7 @@ extern "C" {
 /**
  * finite macro.
  */
-#if defined(_WIN32) && !defined(__WIN32__) && !defined(__CYGWIN__) && !defined(BUILD_FOR_SNAP)
-#  define __WIN32__
+#if defined(_MSC_VER)
 #  define finite _finite
 #elif defined(__WATCOMC__)
 #  define finite _finite
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index e844a7432d..490b8f0f33 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -79,42 +79,28 @@
 #include "glheader.h"
 #include "mfeatures.h"
 #include "imports.h"
-#if FEATURE_accum
 #include "accum.h"
-#endif
 #include "api_exec.h"
 #include "arrayobj.h"
-#if FEATURE_attrib_stack
 #include "attrib.h"
-#endif
 #include "blend.h"
 #include "buffers.h"
 #include "bufferobj.h"
-#if FEATURE_colortable
 #include "colortab.h"
-#endif
 #include "context.h"
 #include "cpuinfo.h"
 #include "debug.h"
 #include "depth.h"
-#if FEATURE_dlist
 #include "dlist.h"
-#endif
-#if FEATURE_evaluators
 #include "eval.h"
-#endif
 #include "enums.h"
 #include "extensions.h"
 #include "fbobject.h"
-#if FEATURE_feedback
 #include "feedback.h"
-#endif
 #include "fog.h"
 #include "framebuffer.h"
 #include "get.h"
-#if FEATURE_histogram
 #include "histogram.h"
-#endif
 #include "hint.h"
 #include "hash.h"
 #include "light.h"
@@ -126,15 +112,11 @@
 #include "pixelstore.h"
 #include "points.h"
 #include "polygon.h"
-#if FEATURE_ARB_occlusion_query
 #include "queryobj.h"
-#endif
 #if FEATURE_ARB_sync
 #include "syncobj.h"
 #endif
-#if FEATURE_drawpix
 #include "rastpos.h"
-#endif
 #include "scissor.h"
 #include "shared.h"
 #include "simple_list.h"
@@ -191,6 +173,8 @@ GLfloat _mesa_ubyte_to_float_color_tab[256];
 void
 _mesa_notifySwapBuffers(__GLcontext *ctx)
 {
+   if (MESA_VERBOSE & VERBOSE_SWAPBUFFERS)
+      _mesa_debug(ctx, "SwapBuffers\n");
    FLUSH_CURRENT( ctx, 0 );
    if (ctx->Driver.Flush) {
       ctx->Driver.Flush(ctx);
@@ -678,36 +662,20 @@ init_attrib_groups(GLcontext *ctx)
    _mesa_init_extensions( ctx );
 
    /* Attribute Groups */
-#if FEATURE_accum
    _mesa_init_accum( ctx );
-#endif
-#if FEATURE_attrib_stack
    _mesa_init_attrib( ctx );
-#endif
    _mesa_init_buffer_objects( ctx );
    _mesa_init_color( ctx );
-#if FEATURE_colortable
    _mesa_init_colortables( ctx );
-#endif
    _mesa_init_current( ctx );
    _mesa_init_depth( ctx );
    _mesa_init_debug( ctx );
-#if FEATURE_dlist
    _mesa_init_display_list( ctx );
-#endif
-#if FEATURE_evaluators
    _mesa_init_eval( ctx );
-#endif
    _mesa_init_fbobjects( ctx );
-#if FEATURE_feedback
    _mesa_init_feedback( ctx );
-#else
-   ctx->RenderMode = GL_RENDER;
-#endif
    _mesa_init_fog( ctx );
-#if FEATURE_histogram
    _mesa_init_histogram( ctx );
-#endif
    _mesa_init_hint( ctx );
    _mesa_init_line( ctx );
    _mesa_init_lighting( ctx );
@@ -718,15 +686,11 @@ init_attrib_groups(GLcontext *ctx)
    _mesa_init_point( ctx );
    _mesa_init_polygon( ctx );
    _mesa_init_program( ctx );
-#if FEATURE_ARB_occlusion_query
-   _mesa_init_query( ctx );
-#endif
+   _mesa_init_queryobj( ctx );
 #if FEATURE_ARB_sync
    _mesa_init_sync( ctx );
 #endif
-#if FEATURE_drawpix
    _mesa_init_rastpos( ctx );
-#endif
    _mesa_init_scissor( ctx );
    _mesa_init_shader_state( ctx );
    _mesa_init_stencil( ctx );
@@ -904,10 +868,12 @@ _mesa_initialize_context(GLcontext *ctx,
    _mesa_init_exec_table(ctx->Exec);
 #endif
    ctx->CurrentDispatch = ctx->Exec;
+
 #if FEATURE_dlist
-   _mesa_init_dlist_table(ctx->Save);
+   _mesa_init_save_table(ctx->Save);
    _mesa_install_save_vtxfmt( ctx, &ctx->ListState.ListVtxfmt );
 #endif
+
    /* Neutral tnl module stuff */
    _mesa_init_exec_vtxfmt( ctx ); 
    ctx->TnlModule.Current = NULL;
@@ -1006,24 +972,16 @@ _mesa_free_context_data( GLcontext *ctx )
    _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL);
    _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, NULL);
 
-#if FEATURE_attrib_stack
    _mesa_free_attrib_data(ctx);
-#endif
    _mesa_free_lighting_data( ctx );
-#if FEATURE_evaluators
    _mesa_free_eval_data( ctx );
-#endif
    _mesa_free_texture_data( ctx );
    _mesa_free_matrix_data( ctx );
    _mesa_free_viewport_data( ctx );
-#if FEATURE_colortable
    _mesa_free_colortables_data( ctx );
-#endif
    _mesa_free_program_data(ctx);
    _mesa_free_shader_state(ctx);
-#if FEATURE_ARB_occlusion_query
-   _mesa_free_query_data(ctx);
-#endif
+   _mesa_free_queryobj_data(ctx);
 #if FEATURE_ARB_sync
    _mesa_free_sync_data(ctx);
 #endif
@@ -1055,6 +1013,9 @@ _mesa_free_context_data( GLcontext *ctx )
       _mesa_free_shared_state( ctx, ctx->Shared );
    }
 
+   /* needs to be after freeing shared state */
+   _mesa_free_display_list_data(ctx);
+
    if (ctx->Extensions.String)
       _mesa_free((void *) ctx->Extensions.String);
 
diff --git a/src/mesa/main/convolve.c b/src/mesa/main/convolve.c
index 70951112a1..8db3e79d38 100644
--- a/src/mesa/main/convolve.c
+++ b/src/mesa/main/convolve.c
@@ -40,6 +40,10 @@
 #include "mtypes.h"
 #include "pixel.h"
 #include "state.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_convolve
 
 
 /*
@@ -256,7 +260,7 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalFormat, GLsizei width, G
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -299,7 +303,7 @@ _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -351,7 +355,7 @@ _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -394,7 +398,7 @@ _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -459,7 +463,7 @@ _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLint y, GLsizei width)
 {
    GLint baseFormat;
@@ -482,12 +486,16 @@ _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLi
       return;
    }
 
+   if (!ctx->ReadBuffer->_ColorReadBuffer) {
+      return;      /* no readbuffer - OK */
+   }
+
    ctx->Driver.CopyConvolutionFilter1D( ctx, target, 
 					internalFormat, x, y, width);
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLint y, GLsizei width, GLsizei height)
 {
    GLint baseFormat;
@@ -514,12 +522,16 @@ _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLi
       return;
    }
 
+   if (!ctx->ReadBuffer->_ColorReadBuffer) {
+      return;      /* no readbuffer - OK */
+   }
+
    ctx->Driver.CopyConvolutionFilter2D( ctx, target, internalFormat, x, y, 
 					width, height );
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type,
                            GLvoid *image)
 {
@@ -578,7 +590,7 @@ _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -639,7 +651,7 @@ _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -709,7 +721,7 @@ _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type,
                          GLvoid *row, GLvoid *column, GLvoid *span)
 {
@@ -776,7 +788,7 @@ _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_SeparableFilter2D(GLenum target, GLenum internalFormat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *row, const GLvoid *column)
 {
    const GLint colStart = MAX_CONVOLUTION_WIDTH * 4;
@@ -1425,3 +1437,25 @@ _mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions,
       *height = *height - (MAX2(ctx->Separable2D.Height, 1) - 1);
    }
 }
+
+
+void
+_mesa_init_convolve_dispatch(struct _glapi_table *disp)
+{
+   SET_ConvolutionFilter1D(disp, _mesa_ConvolutionFilter1D);
+   SET_ConvolutionFilter2D(disp, _mesa_ConvolutionFilter2D);
+   SET_ConvolutionParameterf(disp, _mesa_ConvolutionParameterf);
+   SET_ConvolutionParameterfv(disp, _mesa_ConvolutionParameterfv);
+   SET_ConvolutionParameteri(disp, _mesa_ConvolutionParameteri);
+   SET_ConvolutionParameteriv(disp, _mesa_ConvolutionParameteriv);
+   SET_CopyConvolutionFilter1D(disp, _mesa_CopyConvolutionFilter1D);
+   SET_CopyConvolutionFilter2D(disp, _mesa_CopyConvolutionFilter2D);
+   SET_GetConvolutionFilter(disp, _mesa_GetConvolutionFilter);
+   SET_GetConvolutionParameterfv(disp, _mesa_GetConvolutionParameterfv);
+   SET_GetConvolutionParameteriv(disp, _mesa_GetConvolutionParameteriv);
+   SET_SeparableFilter2D(disp, _mesa_SeparableFilter2D);
+   SET_GetSeparableFilter(disp, _mesa_GetSeparableFilter);
+}
+
+
+#endif /* FEATURE_convolve */
diff --git a/src/mesa/main/convolve.h b/src/mesa/main/convolve.h
index 4505cdae01..59492bc7c5 100644
--- a/src/mesa/main/convolve.h
+++ b/src/mesa/main/convolve.h
@@ -28,10 +28,17 @@
 #define CONVOLVE_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 
-#if _HAVE_FULL_GL
+#if FEATURE_convolve
+
+#define _MESA_INIT_CONVOLVE_FUNCTIONS(driver, impl)                        \
+   do {                                                                    \
+      (driver)->CopyConvolutionFilter1D = impl ## CopyConvolutionFilter1D; \
+      (driver)->CopyConvolutionFilter2D = impl ## CopyConvolutionFilter2D; \
+   } while (0)
+
 extern void GLAPIENTRY
 _mesa_ConvolutionFilter1D(GLenum target, GLenum internalformat, GLsizei width,
                           GLenum format, GLenum type, const GLvoid *image);
@@ -41,74 +48,79 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalformat, GLsizei width,
                           GLsizei height, GLenum format, GLenum type,
                           const GLvoid *image);
 
-extern void GLAPIENTRY
-_mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat params);
-
-extern void GLAPIENTRY
-_mesa_ConvolutionParameterfv(GLenum target, GLenum pname,
-                             const GLfloat *params);
-
-extern void GLAPIENTRY
-_mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint params);
-
-extern void GLAPIENTRY
-_mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params);
-
-extern void GLAPIENTRY
-_mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalformat,
-                              GLint x, GLint y, GLsizei width);
+extern void
+_mesa_convolve_1d_image(const GLcontext *ctx, GLsizei *width,
+                        const GLfloat *srcImage, GLfloat *dstImage);
 
-extern void GLAPIENTRY
-_mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalformat,
-                              GLint x, GLint y, GLsizei width, GLsizei height);
+extern void
+_mesa_convolve_2d_image(const GLcontext *ctx, GLsizei *width, GLsizei *height,
+                        const GLfloat *srcImage, GLfloat *dstImage);
 
-extern void GLAPIENTRY
-_mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type,
-                           GLvoid *image);
+extern void
+_mesa_convolve_sep_image(const GLcontext *ctx,
+                         GLsizei *width, GLsizei *height,
+                         const GLfloat *srcImage, GLfloat *dstImage);
 
-extern void GLAPIENTRY
-_mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params);
+extern void
+_mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions,
+                                   GLsizei *width, GLsizei *height);
 
-extern void GLAPIENTRY
-_mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params);
+extern void
+_mesa_init_convolve_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY
-_mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type,
-                         GLvoid *row, GLvoid *column, GLvoid *span);
+#else /* FEATURE_convolve */
 
-extern void GLAPIENTRY
-_mesa_SeparableFilter2D(GLenum target, GLenum internalformat,
-                        GLsizei width, GLsizei height,
-                        GLenum format, GLenum type,
-                        const GLvoid *row, const GLvoid *column);
+#define _MESA_INIT_CONVOLVE_FUNCTIONS(driver, impl) do { } while (0)
 
+static INLINE void GLAPIENTRY
+_mesa_ConvolutionFilter1D(GLenum target, GLenum internalformat, GLsizei width,
+                          GLenum format, GLenum type, const GLvoid *image)
+{
+   ASSERT_NO_FEATURE();
+}
 
+static INLINE void GLAPIENTRY
+_mesa_ConvolutionFilter2D(GLenum target, GLenum internalformat, GLsizei width,
+                          GLsizei height, GLenum format, GLenum type,
+                          const GLvoid *image)
+{
+   ASSERT_NO_FEATURE();
+}
 
-extern void
+static INLINE void
 _mesa_convolve_1d_image(const GLcontext *ctx, GLsizei *width,
-                        const GLfloat *srcImage, GLfloat *dstImage);
-
+                        const GLfloat *srcImage, GLfloat *dstImage)
+{
+   ASSERT_NO_FEATURE();
+}
 
-extern void
+static INLINE void
 _mesa_convolve_2d_image(const GLcontext *ctx, GLsizei *width, GLsizei *height,
-                        const GLfloat *srcImage, GLfloat *dstImage);
+                        const GLfloat *srcImage, GLfloat *dstImage)
+{
+   ASSERT_NO_FEATURE();
+}
 
 
-extern void
+static INLINE void
 _mesa_convolve_sep_image(const GLcontext *ctx,
                          GLsizei *width, GLsizei *height,
-                         const GLfloat *srcImage, GLfloat *dstImage);
-
+                         const GLfloat *srcImage, GLfloat *dstImage)
+{
+   ASSERT_NO_FEATURE();
+}
 
-extern void
+static INLINE void
 _mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions,
-                                   GLsizei *width, GLsizei *height);
+                                   GLsizei *width, GLsizei *height)
+{
+}
+
+static INLINE void
+_mesa_init_convolve_dispatch(struct _glapi_table *disp)
+{
+}
 
-#else
-#define _mesa_adjust_image_for_convolution(c, d, w, h) ((void)0)
-#define _mesa_convolve_1d_image(c,w,s,d) ((void)0)
-#define _mesa_convolve_2d_image(c,w,h,s,d) ((void)0)
-#define _mesa_convolve_sep_image(c,w,h,s,d) ((void)0)
-#endif
+#endif /* FEATURE_convolve */
 
-#endif
+#endif /* CONVOLVE_H */
diff --git a/src/mesa/main/dd.h b/src/mesa/main/dd.h
index 4a700b5cb4..25aaddea81 100644
--- a/src/mesa/main/dd.h
+++ b/src/mesa/main/dd.h
@@ -545,12 +545,6 @@ struct dd_function_table {
                                    struct gl_texture_object *t );
 
    /**
-    * Called by glPrioritizeTextures().
-    */
-   void (*PrioritizeTexture)( GLcontext *ctx,  struct gl_texture_object *t,
-                              GLclampf priority );
-
-   /**
     * Called by glActiveTextureARB() to set current texture unit.
     */
    void (*ActiveTexture)( GLcontext *ctx, GLuint texUnitNumber );
@@ -717,7 +711,6 @@ struct dd_function_table {
    void (*TexParameter)(GLcontext *ctx, GLenum target,
                         struct gl_texture_object *texObj,
                         GLenum pname, const GLfloat *params);
-   void (*TextureMatrix)(GLcontext *ctx, GLuint unit, const GLmatrix *mat);
    /** Set the viewport */
    void (*Viewport)(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h);
    /*@}*/
@@ -1172,7 +1165,22 @@ typedef struct {
 					    GLenum type,
 					    const GLvoid **indices,
 					    GLsizei primcount);
-  /*@}*/
+   void (GLAPIENTRYP DrawElementsBaseVertex)( GLenum mode, GLsizei count,
+					      GLenum type,
+					      const GLvoid *indices,
+					      GLint basevertex );
+   void (GLAPIENTRYP DrawRangeElementsBaseVertex)( GLenum mode, GLuint start,
+						   GLuint end, GLsizei count,
+						   GLenum type,
+						   const GLvoid *indices,
+						   GLint basevertex);
+   void (GLAPIENTRYP MultiDrawElementsBaseVertex)( GLenum mode,
+						   const GLsizei *count,
+						   GLenum type,
+						   const GLvoid **indices,
+						   GLsizei primcount,
+						   const GLint *basevertex);
+   /*@}*/
 
    /**
     * \name Eval
diff --git a/src/mesa/main/debug.c b/src/mesa/main/debug.c
index 8492c8561d..0e35617575 100644
--- a/src/mesa/main/debug.c
+++ b/src/mesa/main/debug.c
@@ -27,6 +27,7 @@
 #include "attrib.h"
 #include "colormac.h"
 #include "context.h"
+#include "enums.h"
 #include "hash.h"
 #include "imports.h"
 #include "debug.h"
@@ -167,14 +168,16 @@ static void add_debug_flags( const char *debug )
    static const struct debug_option debug_opt[] = {
       { "varray",    VERBOSE_VARRAY },
       { "tex",       VERBOSE_TEXTURE },
-      { "imm",       VERBOSE_IMMEDIATE },
+      { "mat",       VERBOSE_MATERIAL },
       { "pipe",      VERBOSE_PIPELINE },
       { "driver",    VERBOSE_DRIVER },
       { "state",     VERBOSE_STATE },
       { "api",       VERBOSE_API },
       { "list",      VERBOSE_DISPLAY_LIST },
       { "lighting",  VERBOSE_LIGHTING },
-      { "disassem",  VERBOSE_DISASSEM }
+      { "disassem",  VERBOSE_DISASSEM },
+      { "draw",      VERBOSE_DRAW },
+      { "swap",      VERBOSE_SWAPBUFFERS }
    };
    GLuint i;
 
@@ -262,10 +265,13 @@ write_ppm(const char *filename, const GLubyte *buffer, int width, int height,
 
 
 /**
- * Write level[0] image to a ppm file.
+ * Write a texture image to a ppm file.
+ * \param face  cube face in [0,5]
+ * \param level  mipmap level
  */
 static void
-write_texture_image(struct gl_texture_object *texObj, GLuint face, GLuint level)
+write_texture_image(struct gl_texture_object *texObj,
+                    GLuint face, GLuint level)
 {
    struct gl_texture_image *img = texObj->Image[face][level];
    if (img) {
@@ -297,16 +303,53 @@ write_texture_image(struct gl_texture_object *texObj, GLuint face, GLuint level)
 }
 
 
+/**
+ * Write renderbuffer image to a ppm file.
+ */
+static void
+write_renderbuffer_image(const struct gl_renderbuffer *rb)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLubyte *buffer;
+   char s[100];
+   GLenum format, type;
+
+   if (rb->_BaseFormat == GL_RGB || 
+       rb->_BaseFormat == GL_RGBA) {
+      format = GL_RGBA;
+      type = GL_UNSIGNED_BYTE;
+   }
+   else if (rb->_BaseFormat == GL_DEPTH_STENCIL) {
+      format = GL_DEPTH_STENCIL;
+      type = GL_UNSIGNED_INT_24_8;
+   }
+   else {
+      return;
+   }
+
+   buffer = (GLubyte *) _mesa_malloc(rb->Width * rb->Height * 4);
+
+   ctx->Driver.ReadPixels(ctx, 0, 0, rb->Width, rb->Height,
+                          format, type, &ctx->DefaultPacking, buffer);
+
+   /* make filename */
+   _mesa_sprintf(s, "/tmp/renderbuffer%u.ppm", rb->Name);
+
+   _mesa_printf("  Writing renderbuffer image to %s\n", s);
+   write_ppm(s, buffer, rb->Width, rb->Height, 4, 0, 1, 2, GL_TRUE);
+
+   _mesa_free(buffer);
+}
+
+
 static GLboolean DumpImages;
 
 
 static void
-dump_texture_cb(GLuint id, void *data, void *userData)
+dump_texture(struct gl_texture_object *texObj)
 {
-   struct gl_texture_object *texObj = (struct gl_texture_object *) data;
    int i;
    GLboolean written = GL_FALSE;
-   (void) userData;
 
    _mesa_printf("Texture %u\n", texObj->Name);
    _mesa_printf("  Target 0x%x\n", texObj->Target);
@@ -327,6 +370,30 @@ dump_texture_cb(GLuint id, void *data, void *userData)
 
 
 /**
+ * Dump a single texture.
+ */
+void
+_mesa_dump_texture(GLuint texture, GLboolean dumpImages)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_texture_object *texObj = _mesa_lookup_texture(ctx, texture);
+   if (texObj) {
+      DumpImages = dumpImages;
+      dump_texture(texObj);
+   }
+}
+
+
+static void
+dump_texture_cb(GLuint id, void *data, void *userData)
+{
+   struct gl_texture_object *texObj = (struct gl_texture_object *) data;
+   (void) userData;
+   dump_texture(texObj);
+}
+
+
+/**
  * Print basic info about all texture objext to stdout.
  * If dumpImages is true, write PPM of level[0] image to a file.
  */
@@ -339,6 +406,35 @@ _mesa_dump_textures(GLboolean dumpImages)
 }
 
 
+static void
+dump_renderbuffer_cb(GLuint id, void *data, void *userData)
+{
+   const struct gl_renderbuffer *rb = (const struct gl_renderbuffer *) data;
+   (void) userData;
+
+   _mesa_printf("Renderbuffer %u: %u x %u  IntFormat = %s\n",
+                rb->Name, rb->Width, rb->Height,
+                _mesa_lookup_enum_by_nr(rb->InternalFormat));
+   if (DumpImages) {
+      write_renderbuffer_image(rb);
+   }
+}
+
+
+/**
+ * Print basic info about all renderbuffers to stdout.
+ * If dumpImages is true, write PPM of level[0] image to a file.
+ */
+void
+_mesa_dump_renderbuffers(GLboolean dumpImages)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   DumpImages = dumpImages;
+   _mesa_HashWalk(ctx->Shared->RenderBuffers, dump_renderbuffer_cb, ctx);
+}
+
+
+
 void
 _mesa_dump_color_buffer(const char *filename)
 {
diff --git a/src/mesa/main/debug.h b/src/mesa/main/debug.h
index bb384c4324..f66f774a45 100644
--- a/src/mesa/main/debug.h
+++ b/src/mesa/main/debug.h
@@ -58,9 +58,15 @@ extern void _mesa_init_debug( GLcontext *ctx );
 #endif
 
 extern void
+_mesa_dump_texture(GLuint texture, GLboolean dumpImages);
+
+extern void
 _mesa_dump_textures(GLboolean dumpImages);
 
 extern void
+_mesa_dump_renderbuffers(GLboolean dumpImages);
+
+extern void
 _mesa_dump_color_buffer(const char *filename);
 
 extern void
diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c
index 41a5b61406..b692c335a7 100644
--- a/src/mesa/main/dlist.c
+++ b/src/mesa/main/dlist.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.7
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -73,6 +74,7 @@
 #include "texstate.h"
 #include "mtypes.h"
 #include "varray.h"
+#include "vtxfmt.h"
 #if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program
 #include "shader/arbprogram.h"
 #include "shader/program.h"
@@ -90,6 +92,33 @@
 #include "glapi/dispatch.h"
 
 
+
+/**
+ * Other parts of Mesa (such as the VBO module) can plug into the display
+ * list system.  This structure describes new display list instructions.
+ */
+struct gl_list_instruction
+{
+   GLuint Size;
+   void (*Execute)( GLcontext *ctx, void *data );
+   void (*Destroy)( GLcontext *ctx, void *data );
+   void (*Print)( GLcontext *ctx, void *data );
+};
+
+
+#define MAX_DLIST_EXT_OPCODES 16
+
+/**
+ * Used by device drivers to hook new commands into display lists.
+ */
+struct gl_list_extensions
+{
+   struct gl_list_instruction Opcode[MAX_DLIST_EXT_OPCODES];
+   GLuint NumOpcodes;
+};
+
+
+
 /**
  * Flush vertices.
  *
@@ -438,6 +467,10 @@ typedef union gl_dlist_node Node;
  */
 static GLuint InstSize[OPCODE_END_OF_LIST + 1];
 
+
+#if FEATURE_dlist
+
+
 void mesa_print_display_list(GLuint list);
 
 
@@ -472,6 +505,49 @@ lookup_list(GLcontext *ctx, GLuint list)
 }
 
 
+/** Is the given opcode an extension code? */
+static INLINE GLboolean
+is_ext_opcode(OpCode opcode)
+{
+   return (opcode >= OPCODE_EXT_0);
+}
+
+
+/** Destroy an extended opcode instruction */
+static GLint
+ext_opcode_destroy(GLcontext *ctx, Node *node)
+{
+   const GLint i = node[0].opcode - OPCODE_EXT_0;
+   GLint step;
+   ctx->ListExt->Opcode[i].Destroy(ctx, &node[1]);
+   step = ctx->ListExt->Opcode[i].Size;
+   return step;
+}
+
+
+/** Execute an extended opcode instruction */
+static GLint
+ext_opcode_execute(GLcontext *ctx, Node *node)
+{
+   const GLint i = node[0].opcode - OPCODE_EXT_0;
+   GLint step;
+   ctx->ListExt->Opcode[i].Execute(ctx, &node[1]);
+   step = ctx->ListExt->Opcode[i].Size;
+   return step;
+}
+
+
+/** Print an extended opcode instruction */
+static GLint
+ext_opcode_print(GLcontext *ctx, Node *node)
+{
+   const GLint i = node[0].opcode - OPCODE_EXT_0;
+   GLint step;
+   ctx->ListExt->Opcode[i].Print(ctx, &node[1]);
+   step = ctx->ListExt->Opcode[i].Size;
+   return step;
+}
+
 
 /**
  * Delete the named display list, but don't remove from hash table.
@@ -487,16 +563,14 @@ _mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist)
 
    done = block ? GL_FALSE : GL_TRUE;
    while (!done) {
+      const OpCode opcode = n[0].opcode;
 
       /* check for extension opcodes first */
-
-      GLint i = (GLint) n[0].opcode - (GLint) OPCODE_EXT_0;
-      if (i >= 0 && i < (GLint) ctx->ListExt.NumOpcodes) {
-         ctx->ListExt.Opcode[i].Destroy(ctx, &n[1]);
-         n += ctx->ListExt.Opcode[i].Size;
+      if (is_ext_opcode(opcode)) {
+         n += ext_opcode_destroy(ctx, n);
       }
       else {
-         switch (n[0].opcode) {
+         switch (opcode) {
             /* for some commands, we need to free malloc'd memory */
          case OPCODE_MAP1:
             _mesa_free(n[6].data);
@@ -784,14 +858,13 @@ unpack_image(GLcontext *ctx, GLuint dimensions,
 
 
 /**
- * Allocate space for a display list instruction.
+ * Allocate space for a display list instruction (opcode + payload space).
  * \param opcode  the instruction opcode (OPCODE_* value)
- * \param bytes   instruction size in bytes, not counting opcode.
- * \return pointer to the usable data area (not including the internal
- *         opcode).
+ * \param bytes   instruction payload size (not counting opcode)
+ * \return pointer to allocated memory (the opcode space)
  */
-void *
-_mesa_alloc_instruction(GLcontext *ctx, GLuint opcode, GLuint bytes)
+static Node *
+dlist_alloc(GLcontext *ctx, OpCode opcode, GLuint bytes)
 {
    const GLuint numNodes = 1 + (bytes + sizeof(Node) - 1) / sizeof(Node);
    Node *n;
@@ -825,9 +898,30 @@ _mesa_alloc_instruction(GLcontext *ctx, GLuint opcode, GLuint bytes)
    n = ctx->ListState.CurrentBlock + ctx->ListState.CurrentPos;
    ctx->ListState.CurrentPos += numNodes;
 
-   n[0].opcode = (OpCode) opcode;
+   n[0].opcode = opcode;
 
-   return (void *) (n + 1);     /* return ptr to node following opcode */
+   return n;
+}
+
+
+
+/**
+ * Allocate space for a display list instruction.  Used by callers outside
+ * this file for things like VBO vertex data.
+ *
+ * \param opcode  the instruction opcode (OPCODE_* value)
+ * \param bytes   instruction size in bytes, not counting opcode.
+ * \return pointer to the usable data area (not including the internal
+ *         opcode).
+ */
+void *
+_mesa_dlist_alloc(GLcontext *ctx, GLuint opcode, GLuint bytes)
+{
+   Node *n = dlist_alloc(ctx, (OpCode) opcode, bytes);
+   if (n)
+      return n + 1;  /* return pointer to payload area, after opcode */
+   else
+      return NULL;
 }
 
 
@@ -842,35 +936,39 @@ _mesa_alloc_instruction(GLcontext *ctx, GLuint opcode, GLuint bytes)
  * \return  the new opcode number or -1 if error
  */
 GLint
-_mesa_alloc_opcode(GLcontext *ctx,
-                   GLuint size,
-                   void (*execute) (GLcontext *, void *),
-                   void (*destroy) (GLcontext *, void *),
-                   void (*print) (GLcontext *, void *))
-{
-   if (ctx->ListExt.NumOpcodes < MAX_DLIST_EXT_OPCODES) {
-      const GLuint i = ctx->ListExt.NumOpcodes++;
-      ctx->ListExt.Opcode[i].Size =
+_mesa_dlist_alloc_opcode(GLcontext *ctx,
+                         GLuint size,
+                         void (*execute) (GLcontext *, void *),
+                         void (*destroy) (GLcontext *, void *),
+                         void (*print) (GLcontext *, void *))
+{
+   if (ctx->ListExt->NumOpcodes < MAX_DLIST_EXT_OPCODES) {
+      const GLuint i = ctx->ListExt->NumOpcodes++;
+      ctx->ListExt->Opcode[i].Size =
          1 + (size + sizeof(Node) - 1) / sizeof(Node);
-      ctx->ListExt.Opcode[i].Execute = execute;
-      ctx->ListExt.Opcode[i].Destroy = destroy;
-      ctx->ListExt.Opcode[i].Print = print;
+      ctx->ListExt->Opcode[i].Execute = execute;
+      ctx->ListExt->Opcode[i].Destroy = destroy;
+      ctx->ListExt->Opcode[i].Print = print;
       return i + OPCODE_EXT_0;
    }
    return -1;
 }
 
 
-
 /**
- * Allocate display list instruction.  Returns Node ptr to where the opcode
- * is stored.
- *   - nParams is the number of function parameters
- *   - return value a pointer to sizeof(Node) before the actual
- *     usable data area.
+ * Allocate space for a display list instruction.  The space is basically
+ * an array of Nodes where node[0] holds the opcode, node[1] is the first
+ * function parameter, node[2] is the second parameter, etc.
+ *
+ * \param opcode  one of OPCODE_x
+ * \param nparams  number of function parameters
+ * \return  pointer to start of instruction space
  */
-#define ALLOC_INSTRUCTION(CTX, OPCODE, NPARAMS) \
-    ((Node *)_mesa_alloc_instruction(CTX, OPCODE, (NPARAMS)*sizeof(Node)) - 1)
+static INLINE Node *
+alloc_instruction(GLcontext *ctx, OpCode opcode, GLuint nparams)
+{
+   return dlist_alloc(ctx, opcode, nparams * sizeof(Node));
+}
 
 
 
@@ -883,7 +981,7 @@ save_Accum(GLenum op, GLfloat value)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ACCUM, 2);
+   n = alloc_instruction(ctx, OPCODE_ACCUM, 2);
    if (n) {
       n[1].e = op;
       n[2].f = value;
@@ -900,7 +998,7 @@ save_AlphaFunc(GLenum func, GLclampf ref)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ALPHA_FUNC, 2);
+   n = alloc_instruction(ctx, OPCODE_ALPHA_FUNC, 2);
    if (n) {
       n[1].e = func;
       n[2].f = (GLfloat) ref;
@@ -917,7 +1015,7 @@ save_BindTexture(GLenum target, GLuint texture)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BIND_TEXTURE, 2);
+   n = alloc_instruction(ctx, OPCODE_BIND_TEXTURE, 2);
    if (n) {
       n[1].e = target;
       n[2].ui = texture;
@@ -936,7 +1034,7 @@ save_Bitmap(GLsizei width, GLsizei height,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BITMAP, 7);
+   n = alloc_instruction(ctx, OPCODE_BITMAP, 7);
    if (n) {
       n[1].i = (GLint) width;
       n[2].i = (GLint) height;
@@ -959,7 +1057,7 @@ save_BlendEquation(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BLEND_EQUATION, 1);
+   n = alloc_instruction(ctx, OPCODE_BLEND_EQUATION, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -975,7 +1073,7 @@ save_BlendEquationSeparateEXT(GLenum modeRGB, GLenum modeA)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BLEND_EQUATION_SEPARATE, 2);
+   n = alloc_instruction(ctx, OPCODE_BLEND_EQUATION_SEPARATE, 2);
    if (n) {
       n[1].e = modeRGB;
       n[2].e = modeA;
@@ -993,7 +1091,7 @@ save_BlendFuncSeparateEXT(GLenum sfactorRGB, GLenum dfactorRGB,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BLEND_FUNC_SEPARATE, 4);
+   n = alloc_instruction(ctx, OPCODE_BLEND_FUNC_SEPARATE, 4);
    if (n) {
       n[1].e = sfactorRGB;
       n[2].e = dfactorRGB;
@@ -1020,7 +1118,7 @@ save_BlendColor(GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BLEND_COLOR, 4);
+   n = alloc_instruction(ctx, OPCODE_BLEND_COLOR, 4);
    if (n) {
       n[1].f = red;
       n[2].f = green;
@@ -1047,14 +1145,14 @@ static void invalidate_saved_current_state( GLcontext *ctx )
    ctx->Driver.CurrentSavePrimitive = PRIM_UNKNOWN;
 }
 
-void GLAPIENTRY
-_mesa_save_CallList(GLuint list)
+static void GLAPIENTRY
+save_CallList(GLuint list)
 {
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CALL_LIST, 1);
+   n = alloc_instruction(ctx, OPCODE_CALL_LIST, 1);
    if (n) {
       n[1].ui = list;
    }
@@ -1070,8 +1168,8 @@ _mesa_save_CallList(GLuint list)
 }
 
 
-void GLAPIENTRY
-_mesa_save_CallLists(GLsizei num, GLenum type, const GLvoid * lists)
+static void GLAPIENTRY
+save_CallLists(GLsizei num, GLenum type, const GLvoid * lists)
 {
    GET_CURRENT_CONTEXT(ctx);
    GLint i;
@@ -1098,7 +1196,7 @@ _mesa_save_CallLists(GLsizei num, GLenum type, const GLvoid * lists)
 
    for (i = 0; i < num; i++) {
       GLint list = translate_id(i, type, lists);
-      Node *n = ALLOC_INSTRUCTION(ctx, OPCODE_CALL_LIST_OFFSET, 2);
+      Node *n = alloc_instruction(ctx, OPCODE_CALL_LIST_OFFSET, 2);
       if (n) {
          n[1].i = list;
          n[2].b = typeErrorFlag;
@@ -1122,7 +1220,7 @@ save_Clear(GLbitfield mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLEAR, 1);
+   n = alloc_instruction(ctx, OPCODE_CLEAR, 1);
    if (n) {
       n[1].bf = mask;
    }
@@ -1138,7 +1236,7 @@ save_ClearAccum(GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLEAR_ACCUM, 4);
+   n = alloc_instruction(ctx, OPCODE_CLEAR_ACCUM, 4);
    if (n) {
       n[1].f = red;
       n[2].f = green;
@@ -1157,7 +1255,7 @@ save_ClearColor(GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLEAR_COLOR, 4);
+   n = alloc_instruction(ctx, OPCODE_CLEAR_COLOR, 4);
    if (n) {
       n[1].f = red;
       n[2].f = green;
@@ -1176,7 +1274,7 @@ save_ClearDepth(GLclampd depth)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLEAR_DEPTH, 1);
+   n = alloc_instruction(ctx, OPCODE_CLEAR_DEPTH, 1);
    if (n) {
       n[1].f = (GLfloat) depth;
    }
@@ -1192,7 +1290,7 @@ save_ClearIndex(GLfloat c)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLEAR_INDEX, 1);
+   n = alloc_instruction(ctx, OPCODE_CLEAR_INDEX, 1);
    if (n) {
       n[1].f = c;
    }
@@ -1208,7 +1306,7 @@ save_ClearStencil(GLint s)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLEAR_STENCIL, 1);
+   n = alloc_instruction(ctx, OPCODE_CLEAR_STENCIL, 1);
    if (n) {
       n[1].i = s;
    }
@@ -1224,7 +1322,7 @@ save_ClipPlane(GLenum plane, const GLdouble * equ)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CLIP_PLANE, 5);
+   n = alloc_instruction(ctx, OPCODE_CLIP_PLANE, 5);
    if (n) {
       n[1].e = plane;
       n[2].f = (GLfloat) equ[0];
@@ -1246,7 +1344,7 @@ save_ColorMask(GLboolean red, GLboolean green,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COLOR_MASK, 4);
+   n = alloc_instruction(ctx, OPCODE_COLOR_MASK, 4);
    if (n) {
       n[1].b = red;
       n[2].b = green;
@@ -1266,7 +1364,7 @@ save_ColorMaterial(GLenum face, GLenum mode)
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COLOR_MATERIAL, 2);
+   n = alloc_instruction(ctx, OPCODE_COLOR_MATERIAL, 2);
    if (n) {
       n[1].e = face;
       n[2].e = mode;
@@ -1291,7 +1389,7 @@ save_ColorTable(GLenum target, GLenum internalFormat,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_COLOR_TABLE, 6);
+      n = alloc_instruction(ctx, OPCODE_COLOR_TABLE, 6);
       if (n) {
          n[1].e = target;
          n[2].e = internalFormat;
@@ -1319,7 +1417,7 @@ save_ColorTableParameterfv(GLenum target, GLenum pname,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COLOR_TABLE_PARAMETER_FV, 6);
+   n = alloc_instruction(ctx, OPCODE_COLOR_TABLE_PARAMETER_FV, 6);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -1348,7 +1446,7 @@ save_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params)
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COLOR_TABLE_PARAMETER_IV, 6);
+   n = alloc_instruction(ctx, OPCODE_COLOR_TABLE_PARAMETER_IV, 6);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -1377,7 +1475,7 @@ save_ColorSubTable(GLenum target, GLsizei start, GLsizei count,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COLOR_SUB_TABLE, 6);
+   n = alloc_instruction(ctx, OPCODE_COLOR_SUB_TABLE, 6);
    if (n) {
       n[1].e = target;
       n[2].i = start;
@@ -1402,7 +1500,7 @@ save_CopyColorSubTable(GLenum target, GLsizei start,
    Node *n;
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_COLOR_SUB_TABLE, 5);
+   n = alloc_instruction(ctx, OPCODE_COPY_COLOR_SUB_TABLE, 5);
    if (n) {
       n[1].e = target;
       n[2].i = start;
@@ -1424,7 +1522,7 @@ save_CopyColorTable(GLenum target, GLenum internalformat,
    Node *n;
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_COLOR_TABLE, 5);
+   n = alloc_instruction(ctx, OPCODE_COPY_COLOR_TABLE, 5);
    if (n) {
       n[1].e = target;
       n[2].e = internalformat;
@@ -1447,7 +1545,7 @@ save_ConvolutionFilter1D(GLenum target, GLenum internalFormat, GLsizei width,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CONVOLUTION_FILTER_1D, 6);
+   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_FILTER_1D, 6);
    if (n) {
       n[1].e = target;
       n[2].e = internalFormat;
@@ -1474,7 +1572,7 @@ save_ConvolutionFilter2D(GLenum target, GLenum internalFormat,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CONVOLUTION_FILTER_2D, 7);
+   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_FILTER_2D, 7);
    if (n) {
       n[1].e = target;
       n[2].e = internalFormat;
@@ -1499,7 +1597,7 @@ save_ConvolutionParameteri(GLenum target, GLenum pname, GLint param)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CONVOLUTION_PARAMETER_I, 3);
+   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_I, 3);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -1517,7 +1615,7 @@ save_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CONVOLUTION_PARAMETER_IV, 6);
+   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_IV, 6);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -1545,7 +1643,7 @@ save_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CONVOLUTION_PARAMETER_F, 3);
+   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_F, 3);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -1564,7 +1662,7 @@ save_ConvolutionParameterfv(GLenum target, GLenum pname,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CONVOLUTION_PARAMETER_FV, 6);
+   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_FV, 6);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -1592,7 +1690,7 @@ save_CopyPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum type)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_PIXELS, 5);
+   n = alloc_instruction(ctx, OPCODE_COPY_PIXELS, 5);
    if (n) {
       n[1].i = x;
       n[2].i = y;
@@ -1614,7 +1712,7 @@ save_CopyTexImage1D(GLenum target, GLint level, GLenum internalformat,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_TEX_IMAGE1D, 7);
+   n = alloc_instruction(ctx, OPCODE_COPY_TEX_IMAGE1D, 7);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -1640,7 +1738,7 @@ save_CopyTexImage2D(GLenum target, GLint level,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_TEX_IMAGE2D, 8);
+   n = alloc_instruction(ctx, OPCODE_COPY_TEX_IMAGE2D, 8);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -1666,7 +1764,7 @@ save_CopyTexSubImage1D(GLenum target, GLint level,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_TEX_SUB_IMAGE1D, 6);
+   n = alloc_instruction(ctx, OPCODE_COPY_TEX_SUB_IMAGE1D, 6);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -1690,7 +1788,7 @@ save_CopyTexSubImage2D(GLenum target, GLint level,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_TEX_SUB_IMAGE2D, 8);
+   n = alloc_instruction(ctx, OPCODE_COPY_TEX_SUB_IMAGE2D, 8);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -1716,7 +1814,7 @@ save_CopyTexSubImage3D(GLenum target, GLint level,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COPY_TEX_SUB_IMAGE3D, 9);
+   n = alloc_instruction(ctx, OPCODE_COPY_TEX_SUB_IMAGE3D, 9);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -1742,7 +1840,7 @@ save_CullFace(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_CULL_FACE, 1);
+   n = alloc_instruction(ctx, OPCODE_CULL_FACE, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -1758,7 +1856,7 @@ save_DepthFunc(GLenum func)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DEPTH_FUNC, 1);
+   n = alloc_instruction(ctx, OPCODE_DEPTH_FUNC, 1);
    if (n) {
       n[1].e = func;
    }
@@ -1774,7 +1872,7 @@ save_DepthMask(GLboolean mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DEPTH_MASK, 1);
+   n = alloc_instruction(ctx, OPCODE_DEPTH_MASK, 1);
    if (n) {
       n[1].b = mask;
    }
@@ -1790,7 +1888,7 @@ save_DepthRange(GLclampd nearval, GLclampd farval)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DEPTH_RANGE, 2);
+   n = alloc_instruction(ctx, OPCODE_DEPTH_RANGE, 2);
    if (n) {
       n[1].f = (GLfloat) nearval;
       n[2].f = (GLfloat) farval;
@@ -1807,7 +1905,7 @@ save_Disable(GLenum cap)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DISABLE, 1);
+   n = alloc_instruction(ctx, OPCODE_DISABLE, 1);
    if (n) {
       n[1].e = cap;
    }
@@ -1823,7 +1921,7 @@ save_DrawBuffer(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DRAW_BUFFER, 1);
+   n = alloc_instruction(ctx, OPCODE_DRAW_BUFFER, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -1842,7 +1940,7 @@ save_DrawPixels(GLsizei width, GLsizei height,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DRAW_PIXELS, 5);
+   n = alloc_instruction(ctx, OPCODE_DRAW_PIXELS, 5);
    if (n) {
       n[1].i = width;
       n[2].i = height;
@@ -1864,7 +1962,7 @@ save_Enable(GLenum cap)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ENABLE, 1);
+   n = alloc_instruction(ctx, OPCODE_ENABLE, 1);
    if (n) {
       n[1].e = cap;
    }
@@ -1876,12 +1974,12 @@ save_Enable(GLenum cap)
 
 
 static void GLAPIENTRY
-_mesa_save_EvalMesh1(GLenum mode, GLint i1, GLint i2)
+save_EvalMesh1(GLenum mode, GLint i1, GLint i2)
 {
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EVALMESH1, 3);
+   n = alloc_instruction(ctx, OPCODE_EVALMESH1, 3);
    if (n) {
       n[1].e = mode;
       n[2].i = i1;
@@ -1894,12 +1992,12 @@ _mesa_save_EvalMesh1(GLenum mode, GLint i1, GLint i2)
 
 
 static void GLAPIENTRY
-_mesa_save_EvalMesh2(GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2)
+save_EvalMesh2(GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2)
 {
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EVALMESH2, 5);
+   n = alloc_instruction(ctx, OPCODE_EVALMESH2, 5);
    if (n) {
       n[1].e = mode;
       n[2].i = i1;
@@ -1921,7 +2019,7 @@ save_Fogfv(GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_FOG, 5);
+   n = alloc_instruction(ctx, OPCODE_FOG, 5);
    if (n) {
       n[1].e = pname;
       n[2].f = params[0];
@@ -1990,7 +2088,7 @@ save_FrontFace(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_FRONT_FACE, 1);
+   n = alloc_instruction(ctx, OPCODE_FRONT_FACE, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -2007,7 +2105,7 @@ save_Frustum(GLdouble left, GLdouble right,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_FRUSTUM, 6);
+   n = alloc_instruction(ctx, OPCODE_FRUSTUM, 6);
    if (n) {
       n[1].f = (GLfloat) left;
       n[2].f = (GLfloat) right;
@@ -2028,7 +2126,7 @@ save_Hint(GLenum target, GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_HINT, 2);
+   n = alloc_instruction(ctx, OPCODE_HINT, 2);
    if (n) {
       n[1].e = target;
       n[2].e = mode;
@@ -2047,7 +2145,7 @@ save_Histogram(GLenum target, GLsizei width, GLenum internalFormat,
    Node *n;
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_HISTOGRAM, 4);
+   n = alloc_instruction(ctx, OPCODE_HISTOGRAM, 4);
    if (n) {
       n[1].e = target;
       n[2].i = width;
@@ -2066,7 +2164,7 @@ save_IndexMask(GLuint mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_INDEX_MASK, 1);
+   n = alloc_instruction(ctx, OPCODE_INDEX_MASK, 1);
    if (n) {
       n[1].ui = mask;
    }
@@ -2081,7 +2179,7 @@ save_InitNames(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_INIT_NAMES, 0);
+   (void) alloc_instruction(ctx, OPCODE_INIT_NAMES, 0);
    if (ctx->ExecuteFlag) {
       CALL_InitNames(ctx->Exec, ());
    }
@@ -2094,7 +2192,7 @@ save_Lightfv(GLenum light, GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LIGHT, 6);
+   n = alloc_instruction(ctx, OPCODE_LIGHT, 6);
    if (n) {
       GLint i, nParams;
       n[1].e = light;
@@ -2208,7 +2306,7 @@ save_LightModelfv(GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LIGHT_MODEL, 5);
+   n = alloc_instruction(ctx, OPCODE_LIGHT_MODEL, 5);
    if (n) {
       n[1].e = pname;
       n[2].f = params[0];
@@ -2275,7 +2373,7 @@ save_LineStipple(GLint factor, GLushort pattern)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LINE_STIPPLE, 2);
+   n = alloc_instruction(ctx, OPCODE_LINE_STIPPLE, 2);
    if (n) {
       n[1].i = factor;
       n[2].us = pattern;
@@ -2292,7 +2390,7 @@ save_LineWidth(GLfloat width)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LINE_WIDTH, 1);
+   n = alloc_instruction(ctx, OPCODE_LINE_WIDTH, 1);
    if (n) {
       n[1].f = width;
    }
@@ -2308,7 +2406,7 @@ save_ListBase(GLuint base)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LIST_BASE, 1);
+   n = alloc_instruction(ctx, OPCODE_LIST_BASE, 1);
    if (n) {
       n[1].ui = base;
    }
@@ -2323,7 +2421,7 @@ save_LoadIdentity(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_LOAD_IDENTITY, 0);
+   (void) alloc_instruction(ctx, OPCODE_LOAD_IDENTITY, 0);
    if (ctx->ExecuteFlag) {
       CALL_LoadIdentity(ctx->Exec, ());
    }
@@ -2336,7 +2434,7 @@ save_LoadMatrixf(const GLfloat * m)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LOAD_MATRIX, 16);
+   n = alloc_instruction(ctx, OPCODE_LOAD_MATRIX, 16);
    if (n) {
       GLuint i;
       for (i = 0; i < 16; i++) {
@@ -2367,7 +2465,7 @@ save_LoadName(GLuint name)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LOAD_NAME, 1);
+   n = alloc_instruction(ctx, OPCODE_LOAD_NAME, 1);
    if (n) {
       n[1].ui = name;
    }
@@ -2383,7 +2481,7 @@ save_LogicOp(GLenum opcode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LOGIC_OP, 1);
+   n = alloc_instruction(ctx, OPCODE_LOGIC_OP, 1);
    if (n) {
       n[1].e = opcode;
    }
@@ -2400,7 +2498,7 @@ save_Map1d(GLenum target, GLdouble u1, GLdouble u2, GLint stride,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MAP1, 6);
+   n = alloc_instruction(ctx, OPCODE_MAP1, 6);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points1d(target, stride, order, points);
       n[1].e = target;
@@ -2422,7 +2520,7 @@ save_Map1f(GLenum target, GLfloat u1, GLfloat u2, GLint stride,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MAP1, 6);
+   n = alloc_instruction(ctx, OPCODE_MAP1, 6);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points1f(target, stride, order, points);
       n[1].e = target;
@@ -2447,7 +2545,7 @@ save_Map2d(GLenum target,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MAP2, 10);
+   n = alloc_instruction(ctx, OPCODE_MAP2, 10);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points2d(target, ustride, uorder,
                                               vstride, vorder, points);
@@ -2480,7 +2578,7 @@ save_Map2f(GLenum target,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MAP2, 10);
+   n = alloc_instruction(ctx, OPCODE_MAP2, 10);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points2f(target, ustride, uorder,
                                               vstride, vorder, points);
@@ -2509,7 +2607,7 @@ save_MapGrid1f(GLint un, GLfloat u1, GLfloat u2)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MAPGRID1, 3);
+   n = alloc_instruction(ctx, OPCODE_MAPGRID1, 3);
    if (n) {
       n[1].i = un;
       n[2].f = u1;
@@ -2535,7 +2633,7 @@ save_MapGrid2f(GLint un, GLfloat u1, GLfloat u2,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MAPGRID2, 6);
+   n = alloc_instruction(ctx, OPCODE_MAPGRID2, 6);
    if (n) {
       n[1].i = un;
       n[2].f = u1;
@@ -2566,7 +2664,7 @@ save_MatrixMode(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MATRIX_MODE, 1);
+   n = alloc_instruction(ctx, OPCODE_MATRIX_MODE, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -2583,7 +2681,7 @@ save_Minmax(GLenum target, GLenum internalFormat, GLboolean sink)
    Node *n;
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MIN_MAX, 3);
+   n = alloc_instruction(ctx, OPCODE_MIN_MAX, 3);
    if (n) {
       n[1].e = target;
       n[2].e = internalFormat;
@@ -2601,7 +2699,7 @@ save_MultMatrixf(const GLfloat * m)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MULT_MATRIX, 16);
+   n = alloc_instruction(ctx, OPCODE_MULT_MATRIX, 16);
    if (n) {
       GLuint i;
       for (i = 0; i < 16; i++) {
@@ -2645,7 +2743,7 @@ save_Ortho(GLdouble left, GLdouble right,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ORTHO, 6);
+   n = alloc_instruction(ctx, OPCODE_ORTHO, 6);
    if (n) {
       n[1].f = (GLfloat) left;
       n[2].f = (GLfloat) right;
@@ -2666,7 +2764,7 @@ save_PixelMapfv(GLenum map, GLint mapsize, const GLfloat *values)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PIXEL_MAP, 3);
+   n = alloc_instruction(ctx, OPCODE_PIXEL_MAP, 3);
    if (n) {
       n[1].e = map;
       n[2].i = mapsize;
@@ -2723,7 +2821,7 @@ save_PixelTransferf(GLenum pname, GLfloat param)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PIXEL_TRANSFER, 2);
+   n = alloc_instruction(ctx, OPCODE_PIXEL_TRANSFER, 2);
    if (n) {
       n[1].e = pname;
       n[2].f = param;
@@ -2747,7 +2845,7 @@ save_PixelZoom(GLfloat xfactor, GLfloat yfactor)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PIXEL_ZOOM, 2);
+   n = alloc_instruction(ctx, OPCODE_PIXEL_ZOOM, 2);
    if (n) {
       n[1].f = xfactor;
       n[2].f = yfactor;
@@ -2764,7 +2862,7 @@ save_PointParameterfvEXT(GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_POINT_PARAMETERS, 4);
+   n = alloc_instruction(ctx, OPCODE_POINT_PARAMETERS, 4);
    if (n) {
       n[1].e = pname;
       n[2].f = params[0];
@@ -2811,7 +2909,7 @@ save_PointSize(GLfloat size)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_POINT_SIZE, 1);
+   n = alloc_instruction(ctx, OPCODE_POINT_SIZE, 1);
    if (n) {
       n[1].f = size;
    }
@@ -2827,7 +2925,7 @@ save_PolygonMode(GLenum face, GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_POLYGON_MODE, 2);
+   n = alloc_instruction(ctx, OPCODE_POLYGON_MODE, 2);
    if (n) {
       n[1].e = face;
       n[2].e = mode;
@@ -2846,7 +2944,7 @@ save_PolygonStipple(const GLubyte * pattern)
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_POLYGON_STIPPLE, 1);
+   n = alloc_instruction(ctx, OPCODE_POLYGON_STIPPLE, 1);
    if (n) {
       n[1].data = unpack_image(ctx, 2, 32, 32, 1, GL_COLOR_INDEX, GL_BITMAP,
                                pattern, &ctx->Unpack);
@@ -2863,7 +2961,7 @@ save_PolygonOffset(GLfloat factor, GLfloat units)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_POLYGON_OFFSET, 2);
+   n = alloc_instruction(ctx, OPCODE_POLYGON_OFFSET, 2);
    if (n) {
       n[1].f = factor;
       n[2].f = units;
@@ -2888,7 +2986,7 @@ save_PopAttrib(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_POP_ATTRIB, 0);
+   (void) alloc_instruction(ctx, OPCODE_POP_ATTRIB, 0);
    if (ctx->ExecuteFlag) {
       CALL_PopAttrib(ctx->Exec, ());
    }
@@ -2900,7 +2998,7 @@ save_PopMatrix(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_POP_MATRIX, 0);
+   (void) alloc_instruction(ctx, OPCODE_POP_MATRIX, 0);
    if (ctx->ExecuteFlag) {
       CALL_PopMatrix(ctx->Exec, ());
    }
@@ -2912,7 +3010,7 @@ save_PopName(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_POP_NAME, 0);
+   (void) alloc_instruction(ctx, OPCODE_POP_NAME, 0);
    if (ctx->ExecuteFlag) {
       CALL_PopName(ctx->Exec, ());
    }
@@ -2929,7 +3027,7 @@ save_PrioritizeTextures(GLsizei num, const GLuint * textures,
 
    for (i = 0; i < num; i++) {
       Node *n;
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_PRIORITIZE_TEXTURE, 2);
+      n = alloc_instruction(ctx, OPCODE_PRIORITIZE_TEXTURE, 2);
       if (n) {
          n[1].ui = textures[i];
          n[2].f = priorities[i];
@@ -2947,7 +3045,7 @@ save_PushAttrib(GLbitfield mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PUSH_ATTRIB, 1);
+   n = alloc_instruction(ctx, OPCODE_PUSH_ATTRIB, 1);
    if (n) {
       n[1].bf = mask;
    }
@@ -2962,7 +3060,7 @@ save_PushMatrix(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_PUSH_MATRIX, 0);
+   (void) alloc_instruction(ctx, OPCODE_PUSH_MATRIX, 0);
    if (ctx->ExecuteFlag) {
       CALL_PushMatrix(ctx->Exec, ());
    }
@@ -2975,7 +3073,7 @@ save_PushName(GLuint name)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PUSH_NAME, 1);
+   n = alloc_instruction(ctx, OPCODE_PUSH_NAME, 1);
    if (n) {
       n[1].ui = name;
    }
@@ -2991,7 +3089,7 @@ save_RasterPos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_RASTER_POS, 4);
+   n = alloc_instruction(ctx, OPCODE_RASTER_POS, 4);
    if (n) {
       n[1].f = x;
       n[2].f = y;
@@ -3150,7 +3248,7 @@ save_PassThrough(GLfloat token)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PASSTHROUGH, 1);
+   n = alloc_instruction(ctx, OPCODE_PASSTHROUGH, 1);
    if (n) {
       n[1].f = token;
    }
@@ -3166,7 +3264,7 @@ save_ReadBuffer(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_READ_BUFFER, 1);
+   n = alloc_instruction(ctx, OPCODE_READ_BUFFER, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -3182,7 +3280,7 @@ save_ResetHistogram(GLenum target)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_RESET_HISTOGRAM, 1);
+   n = alloc_instruction(ctx, OPCODE_RESET_HISTOGRAM, 1);
    if (n) {
       n[1].e = target;
    }
@@ -3198,7 +3296,7 @@ save_ResetMinmax(GLenum target)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_RESET_MIN_MAX, 1);
+   n = alloc_instruction(ctx, OPCODE_RESET_MIN_MAX, 1);
    if (n) {
       n[1].e = target;
    }
@@ -3214,7 +3312,7 @@ save_Rotatef(GLfloat angle, GLfloat x, GLfloat y, GLfloat z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ROTATE, 4);
+   n = alloc_instruction(ctx, OPCODE_ROTATE, 4);
    if (n) {
       n[1].f = angle;
       n[2].f = x;
@@ -3240,7 +3338,7 @@ save_Scalef(GLfloat x, GLfloat y, GLfloat z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_SCALE, 3);
+   n = alloc_instruction(ctx, OPCODE_SCALE, 3);
    if (n) {
       n[1].f = x;
       n[2].f = y;
@@ -3265,7 +3363,7 @@ save_Scissor(GLint x, GLint y, GLsizei width, GLsizei height)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_SCISSOR, 4);
+   n = alloc_instruction(ctx, OPCODE_SCISSOR, 4);
    if (n) {
       n[1].i = x;
       n[2].i = y;
@@ -3299,7 +3397,7 @@ save_ShadeModel(GLenum mode)
    if (ctx->Driver.CurrentSavePrimitive == PRIM_OUTSIDE_BEGIN_END)
       ctx->ListState.Current.ShadeModel = mode;
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_SHADE_MODEL, 1);
+   n = alloc_instruction(ctx, OPCODE_SHADE_MODEL, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -3312,7 +3410,7 @@ save_StencilFunc(GLenum func, GLint ref, GLuint mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_FUNC, 3);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_FUNC, 3);
    if (n) {
       n[1].e = func;
       n[2].i = ref;
@@ -3330,7 +3428,7 @@ save_StencilMask(GLuint mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_MASK, 1);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_MASK, 1);
    if (n) {
       n[1].ui = mask;
    }
@@ -3346,7 +3444,7 @@ save_StencilOp(GLenum fail, GLenum zfail, GLenum zpass)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_OP, 3);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_OP, 3);
    if (n) {
       n[1].e = fail;
       n[2].e = zfail;
@@ -3364,7 +3462,7 @@ save_StencilFuncSeparate(GLenum face, GLenum func, GLint ref, GLuint mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
    if (n) {
       n[1].e = face;
       n[2].e = func;
@@ -3385,7 +3483,7 @@ save_StencilFuncSeparateATI(GLenum frontfunc, GLenum backfunc, GLint ref,
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
    /* GL_FRONT */
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
    if (n) {
       n[1].e = GL_FRONT;
       n[2].e = frontfunc;
@@ -3393,7 +3491,7 @@ save_StencilFuncSeparateATI(GLenum frontfunc, GLenum backfunc, GLint ref,
       n[4].ui = mask;
    }
    /* GL_BACK */
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
    if (n) {
       n[1].e = GL_BACK;
       n[2].e = backfunc;
@@ -3413,7 +3511,7 @@ save_StencilMaskSeparate(GLenum face, GLuint mask)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_MASK_SEPARATE, 2);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_MASK_SEPARATE, 2);
    if (n) {
       n[1].e = face;
       n[2].ui = mask;
@@ -3430,7 +3528,7 @@ save_StencilOpSeparate(GLenum face, GLenum fail, GLenum zfail, GLenum zpass)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_OP_SEPARATE, 4);
+   n = alloc_instruction(ctx, OPCODE_STENCIL_OP_SEPARATE, 4);
    if (n) {
       n[1].e = face;
       n[2].e = fail;
@@ -3449,7 +3547,7 @@ save_TexEnvfv(GLenum target, GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEXENV, 6);
+   n = alloc_instruction(ctx, OPCODE_TEXENV, 6);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -3514,7 +3612,7 @@ save_TexGenfv(GLenum coord, GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEXGEN, 6);
+   n = alloc_instruction(ctx, OPCODE_TEXGEN, 6);
    if (n) {
       n[1].e = coord;
       n[2].e = pname;
@@ -3589,7 +3687,7 @@ save_TexParameterfv(GLenum target, GLenum pname, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEXPARAMETER, 6);
+   n = alloc_instruction(ctx, OPCODE_TEXPARAMETER, 6);
    if (n) {
       n[1].e = target;
       n[2].e = pname;
@@ -3649,7 +3747,7 @@ save_TexImage1D(GLenum target,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_IMAGE1D, 8);
+      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE1D, 8);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -3684,7 +3782,7 @@ save_TexImage2D(GLenum target,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_IMAGE2D, 9);
+      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE2D, 9);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -3722,7 +3820,7 @@ save_TexImage3D(GLenum target,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_IMAGE3D, 10);
+      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE3D, 10);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -3755,7 +3853,7 @@ save_TexSubImage1D(GLenum target, GLint level, GLint xoffset,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_SUB_IMAGE1D, 7);
+   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE1D, 7);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -3784,7 +3882,7 @@ save_TexSubImage2D(GLenum target, GLint level,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_SUB_IMAGE2D, 9);
+   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE2D, 9);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -3815,7 +3913,7 @@ save_TexSubImage3D(GLenum target, GLint level,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_SUB_IMAGE3D, 11);
+   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE3D, 11);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -3845,7 +3943,7 @@ save_Translatef(GLfloat x, GLfloat y, GLfloat z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TRANSLATE, 3);
+   n = alloc_instruction(ctx, OPCODE_TRANSLATE, 3);
    if (n) {
       n[1].f = x;
       n[2].f = y;
@@ -3871,7 +3969,7 @@ save_Viewport(GLint x, GLint y, GLsizei width, GLsizei height)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_VIEWPORT, 4);
+   n = alloc_instruction(ctx, OPCODE_VIEWPORT, 4);
    if (n) {
       n[1].i = x;
       n[2].i = y;
@@ -3890,7 +3988,7 @@ save_WindowPos4fMESA(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_WINDOW_POS, 4);
+   n = alloc_instruction(ctx, OPCODE_WINDOW_POS, 4);
    if (n) {
       n[1].f = x;
       n[2].f = y;
@@ -4051,7 +4149,7 @@ save_ActiveTextureARB(GLenum target)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ACTIVE_TEXTURE, 1);
+   n = alloc_instruction(ctx, OPCODE_ACTIVE_TEXTURE, 1);
    if (n) {
       n[1].e = target;
    }
@@ -4124,7 +4222,7 @@ save_CompressedTexImage1DARB(GLenum target, GLint level,
          return;
       }
       MEMCPY(image, data, imageSize);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_COMPRESSED_TEX_IMAGE_1D, 7);
+      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_1D, 7);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4170,7 +4268,7 @@ save_CompressedTexImage2DARB(GLenum target, GLint level,
          return;
       }
       MEMCPY(image, data, imageSize);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_COMPRESSED_TEX_IMAGE_2D, 8);
+      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_2D, 8);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4217,7 +4315,7 @@ save_CompressedTexImage3DARB(GLenum target, GLint level,
          return;
       }
       MEMCPY(image, data, imageSize);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_COMPRESSED_TEX_IMAGE_3D, 9);
+      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_3D, 9);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4260,7 +4358,7 @@ save_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset,
       return;
    }
    MEMCPY(image, data, imageSize);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_1D, 7);
+   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_1D, 7);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4300,7 +4398,7 @@ save_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset,
       return;
    }
    MEMCPY(image, data, imageSize);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_2D, 9);
+   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_2D, 9);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4342,7 +4440,7 @@ save_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset,
       return;
    }
    MEMCPY(image, data, imageSize);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_3D, 11);
+   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_3D, 11);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4375,7 +4473,7 @@ save_SampleCoverageARB(GLclampf value, GLboolean invert)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_SAMPLE_COVERAGE, 2);
+   n = alloc_instruction(ctx, OPCODE_SAMPLE_COVERAGE, 2);
    if (n) {
       n[1].f = value;
       n[2].b = invert;
@@ -4396,7 +4494,7 @@ save_BindProgramNV(GLenum target, GLuint id)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BIND_PROGRAM_NV, 2);
+   n = alloc_instruction(ctx, OPCODE_BIND_PROGRAM_NV, 2);
    if (n) {
       n[1].e = target;
       n[2].ui = id;
@@ -4413,7 +4511,7 @@ save_ProgramEnvParameter4fARB(GLenum target, GLuint index,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6);
    if (n) {
       n[1].e = target;
       n[2].ui = index;
@@ -4450,7 +4548,7 @@ save_ProgramEnvParameters4fvEXT(GLenum target, GLuint index, GLsizei count,
       const GLfloat * p = params;
 
       for (i = 0 ; i < count ; i++) {
-	 n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6);
+	 n = alloc_instruction(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6);
 	 if (n) {
 	    n[1].e = target;
 	    n[2].ui = index;
@@ -4498,7 +4596,7 @@ save_ExecuteProgramNV(GLenum target, GLuint id, const GLfloat *params)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EXECUTE_PROGRAM_NV, 6);
+   n = alloc_instruction(ctx, OPCODE_EXECUTE_PROGRAM_NV, 6);
    if (n) {
       n[1].e = target;
       n[2].ui = id;
@@ -4544,7 +4642,7 @@ save_LoadProgramNV(GLenum target, GLuint id, GLsizei len,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_LOAD_PROGRAM_NV, 4);
+   n = alloc_instruction(ctx, OPCODE_LOAD_PROGRAM_NV, 4);
    if (n) {
       GLubyte *programCopy = (GLubyte *) _mesa_malloc(len);
       if (!programCopy) {
@@ -4571,7 +4669,7 @@ save_RequestResidentProgramsNV(GLsizei num, const GLuint * ids)
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TRACK_MATRIX_NV, 2);
+   n = alloc_instruction(ctx, OPCODE_TRACK_MATRIX_NV, 2);
    if (n) {
       GLuint *idCopy = (GLuint *) _mesa_malloc(num * sizeof(GLuint));
       if (!idCopy) {
@@ -4595,7 +4693,7 @@ save_TrackMatrixNV(GLenum target, GLuint address,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TRACK_MATRIX_NV, 4);
+   n = alloc_instruction(ctx, OPCODE_TRACK_MATRIX_NV, 4);
    if (n) {
       n[1].e = target;
       n[2].ui = address;
@@ -4620,7 +4718,7 @@ save_ProgramLocalParameter4fARB(GLenum target, GLuint index,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
    if (n) {
       n[1].e = target;
       n[2].ui = index;
@@ -4642,7 +4740,7 @@ save_ProgramLocalParameter4fvARB(GLenum target, GLuint index,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
    if (n) {
       n[1].e = target;
       n[2].ui = index;
@@ -4670,7 +4768,7 @@ save_ProgramLocalParameters4fvEXT(GLenum target, GLuint index, GLsizei count,
       const GLfloat * p = params;
 
       for (i = 0 ; i < count ; i++) {
-	 n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
+	 n = alloc_instruction(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
 	 if (n) {
 	    n[1].e = target;
 	    n[2].ui = index;
@@ -4697,7 +4795,7 @@ save_ProgramLocalParameter4dARB(GLenum target, GLuint index,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
    if (n) {
       n[1].e = target;
       n[2].ui = index;
@@ -4719,7 +4817,7 @@ save_ProgramLocalParameter4dvARB(GLenum target, GLuint index,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, 6);
    if (n) {
       n[1].e = target;
       n[2].ui = index;
@@ -4742,7 +4840,7 @@ save_ProgramNamedParameter4fNV(GLuint id, GLsizei len, const GLubyte * name,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_NAMED_PARAMETER_NV, 6);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_NAMED_PARAMETER_NV, 6);
    if (n) {
       GLubyte *nameCopy = (GLubyte *) _mesa_malloc(len);
       if (!nameCopy) {
@@ -4801,7 +4899,7 @@ save_ActiveStencilFaceEXT(GLenum face)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ACTIVE_STENCIL_FACE_EXT, 1);
+   n = alloc_instruction(ctx, OPCODE_ACTIVE_STENCIL_FACE_EXT, 1);
    if (n) {
       n[1].e = face;
    }
@@ -4818,7 +4916,7 @@ save_DepthBoundsEXT(GLclampd zmin, GLclampd zmax)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DEPTH_BOUNDS_EXT, 2);
+   n = alloc_instruction(ctx, OPCODE_DEPTH_BOUNDS_EXT, 2);
    if (n) {
       n[1].f = (GLfloat) zmin;
       n[2].f = (GLfloat) zmax;
@@ -4841,7 +4939,7 @@ save_ProgramStringARB(GLenum target, GLenum format, GLsizei len,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_STRING_ARB, 4);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_STRING_ARB, 4);
    if (n) {
       GLubyte *programCopy = (GLubyte *) _mesa_malloc(len);
       if (!programCopy) {
@@ -4862,7 +4960,7 @@ save_ProgramStringARB(GLenum target, GLenum format, GLsizei len,
 #endif /* FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program */
 
 
-#if FEATURE_ARB_occlusion_query
+#if FEATURE_queryobj
 
 static void GLAPIENTRY
 save_BeginQueryARB(GLenum target, GLuint id)
@@ -4870,7 +4968,7 @@ save_BeginQueryARB(GLenum target, GLuint id)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BEGIN_QUERY_ARB, 2);
+   n = alloc_instruction(ctx, OPCODE_BEGIN_QUERY_ARB, 2);
    if (n) {
       n[1].e = target;
       n[2].ui = id;
@@ -4887,7 +4985,7 @@ save_EndQueryARB(GLenum target)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_END_QUERY_ARB, 1);
+   n = alloc_instruction(ctx, OPCODE_END_QUERY_ARB, 1);
    if (n) {
       n[1].e = target;
    }
@@ -4896,7 +4994,7 @@ save_EndQueryARB(GLenum target)
    }
 }
 
-#endif /* FEATURE_ARB_occlusion_query */
+#endif /* FEATURE_queryobj */
 
 
 static void GLAPIENTRY
@@ -4905,7 +5003,7 @@ save_DrawBuffersARB(GLsizei count, const GLenum * buffers)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_DRAW_BUFFERS_ARB, 1 + MAX_DRAW_BUFFERS);
+   n = alloc_instruction(ctx, OPCODE_DRAW_BUFFERS_ARB, 1 + MAX_DRAW_BUFFERS);
    if (n) {
       GLint i;
       n[1].i = count;
@@ -4926,7 +5024,7 @@ save_TexBumpParameterfvATI(GLenum pname, const GLfloat *param)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_TEX_BUMP_PARAMETER_ATI, 5);
+   n = alloc_instruction(ctx, OPCODE_TEX_BUMP_PARAMETER_ATI, 5);
    if (n) {
       n[1].ui = pname;
       n[2].f = param[0];
@@ -4957,7 +5055,7 @@ save_BindFragmentShaderATI(GLuint id)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BIND_FRAGMENT_SHADER_ATI, 1);
+   n = alloc_instruction(ctx, OPCODE_BIND_FRAGMENT_SHADER_ATI, 1);
    if (n) {
       n[1].ui = id;
    }
@@ -4972,7 +5070,7 @@ save_SetFragmentShaderConstantATI(GLuint dst, const GLfloat *value)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_SET_FRAGMENT_SHADER_CONSTANTS_ATI, 5);
+   n = alloc_instruction(ctx, OPCODE_SET_FRAGMENT_SHADER_CONSTANTS_ATI, 5);
    if (n) {
       n[1].ui = dst;
       n[2].f = value[0];
@@ -4992,7 +5090,7 @@ save_Attr1fNV(GLenum attr, GLfloat x)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_1F_NV, 2);
+   n = alloc_instruction(ctx, OPCODE_ATTR_1F_NV, 2);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5013,7 +5111,7 @@ save_Attr2fNV(GLenum attr, GLfloat x, GLfloat y)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_2F_NV, 3);
+   n = alloc_instruction(ctx, OPCODE_ATTR_2F_NV, 3);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5035,7 +5133,7 @@ save_Attr3fNV(GLenum attr, GLfloat x, GLfloat y, GLfloat z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_3F_NV, 4);
+   n = alloc_instruction(ctx, OPCODE_ATTR_3F_NV, 4);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5058,7 +5156,7 @@ save_Attr4fNV(GLenum attr, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_4F_NV, 5);
+   n = alloc_instruction(ctx, OPCODE_ATTR_4F_NV, 5);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5083,7 +5181,7 @@ save_Attr1fARB(GLenum attr, GLfloat x)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_1F_ARB, 2);
+   n = alloc_instruction(ctx, OPCODE_ATTR_1F_ARB, 2);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5104,7 +5202,7 @@ save_Attr2fARB(GLenum attr, GLfloat x, GLfloat y)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_2F_ARB, 3);
+   n = alloc_instruction(ctx, OPCODE_ATTR_2F_ARB, 3);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5126,7 +5224,7 @@ save_Attr3fARB(GLenum attr, GLfloat x, GLfloat y, GLfloat z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_3F_ARB, 4);
+   n = alloc_instruction(ctx, OPCODE_ATTR_3F_ARB, 4);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5149,7 +5247,7 @@ save_Attr4fARB(GLenum attr, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ATTR_4F_ARB, 5);
+   n = alloc_instruction(ctx, OPCODE_ATTR_4F_ARB, 5);
    if (n) {
       n[1].e = attr;
       n[2].f = x;
@@ -5174,7 +5272,7 @@ save_EvalCoord1f(GLfloat x)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EVAL_C1, 1);
+   n = alloc_instruction(ctx, OPCODE_EVAL_C1, 1);
    if (n) {
       n[1].f = x;
    }
@@ -5195,7 +5293,7 @@ save_EvalCoord2f(GLfloat x, GLfloat y)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EVAL_C2, 2);
+   n = alloc_instruction(ctx, OPCODE_EVAL_C2, 2);
    if (n) {
       n[1].f = x;
       n[2].f = y;
@@ -5218,7 +5316,7 @@ save_EvalPoint1(GLint x)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EVAL_P1, 1);
+   n = alloc_instruction(ctx, OPCODE_EVAL_P1, 1);
    if (n) {
       n[1].i = x;
    }
@@ -5233,7 +5331,7 @@ save_EvalPoint2(GLint x, GLint y)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_EVAL_P2, 2);
+   n = alloc_instruction(ctx, OPCODE_EVAL_P2, 2);
    if (n) {
       n[1].i = x;
       n[2].i = y;
@@ -5336,7 +5434,7 @@ save_Materialfv(GLenum face, GLenum pname, const GLfloat * param)
 
    SAVE_FLUSH_VERTICES(ctx);
 
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_MATERIAL, 6);
+   n = alloc_instruction(ctx, OPCODE_MATERIAL, 6);
    if (n) {
       n[1].e = face;
       n[2].e = pname;
@@ -5379,7 +5477,7 @@ save_Begin(GLenum mode)
          return;
 
       SAVE_FLUSH_VERTICES(ctx);
-      n = ALLOC_INSTRUCTION(ctx, OPCODE_BEGIN, 1);
+      n = alloc_instruction(ctx, OPCODE_BEGIN, 1);
       if (n) {
          n[1].e = mode;
       }
@@ -5395,7 +5493,7 @@ save_End(void)
 {
    GET_CURRENT_CONTEXT(ctx);
    SAVE_FLUSH_VERTICES(ctx);
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_END, 0);
+   (void) alloc_instruction(ctx, OPCODE_END, 0);
    ctx->Driver.CurrentSavePrimitive = PRIM_OUTSIDE_BEGIN_END;
    if (ctx->ExecuteFlag) {
       CALL_End(ctx->Exec, ());
@@ -5408,7 +5506,7 @@ save_Rectf(GLfloat a, GLfloat b, GLfloat c, GLfloat d)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    SAVE_FLUSH_VERTICES(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_RECTF, 4);
+   n = alloc_instruction(ctx, OPCODE_RECTF, 4);
    if (n) {
       n[1].f = a;
       n[2].f = b;
@@ -5822,7 +5920,7 @@ save_BlitFramebufferEXT(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_BLIT_FRAMEBUFFER, 10);
+   n = alloc_instruction(ctx, OPCODE_BLIT_FRAMEBUFFER, 10);
    if (n) {
       n[1].i = srcX0;
       n[2].i = srcY0;
@@ -5851,7 +5949,7 @@ save_ProvokingVertexEXT(GLenum mode)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_PROVOKING_VERTEX, 1);
+   n = alloc_instruction(ctx, OPCODE_PROVOKING_VERTEX, 1);
    if (n) {
       n[1].e = mode;
    }
@@ -5869,7 +5967,7 @@ save_UseProgramObjectARB(GLhandleARB program)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_USE_PROGRAM, 1);
+   n = alloc_instruction(ctx, OPCODE_USE_PROGRAM, 1);
    if (n) {
       n[1].ui = program;
    }
@@ -5885,7 +5983,7 @@ save_Uniform1fARB(GLint location, GLfloat x)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1F, 2);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1F, 2);
    if (n) {
       n[1].i = location;
       n[2].f = x;
@@ -5902,7 +6000,7 @@ save_Uniform2fARB(GLint location, GLfloat x, GLfloat y)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2F, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2F, 3);
    if (n) {
       n[1].i = location;
       n[2].f = x;
@@ -5920,7 +6018,7 @@ save_Uniform3fARB(GLint location, GLfloat x, GLfloat y, GLfloat z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3F, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3F, 4);
    if (n) {
       n[1].i = location;
       n[2].f = x;
@@ -5939,7 +6037,7 @@ save_Uniform4fARB(GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4F, 5);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4F, 5);
    if (n) {
       n[1].i = location;
       n[2].f = x;
@@ -5970,7 +6068,7 @@ save_Uniform1fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1FV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -5987,7 +6085,7 @@ save_Uniform2fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2FV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6004,7 +6102,7 @@ save_Uniform3fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3FV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6021,7 +6119,7 @@ save_Uniform4fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4FV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6039,7 +6137,7 @@ save_Uniform1iARB(GLint location, GLint x)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1I, 2);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1I, 2);
    if (n) {
       n[1].i = location;
       n[2].i = x;
@@ -6055,7 +6153,7 @@ save_Uniform2iARB(GLint location, GLint x, GLint y)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2I, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2I, 3);
    if (n) {
       n[1].i = location;
       n[2].i = x;
@@ -6072,7 +6170,7 @@ save_Uniform3iARB(GLint location, GLint x, GLint y, GLint z)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3I, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3I, 4);
    if (n) {
       n[1].i = location;
       n[2].i = x;
@@ -6090,7 +6188,7 @@ save_Uniform4iARB(GLint location, GLint x, GLint y, GLint z, GLint w)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4I, 5);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4I, 5);
    if (n) {
       n[1].i = location;
       n[2].i = x;
@@ -6111,7 +6209,7 @@ save_Uniform1ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_1IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1IV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6128,7 +6226,7 @@ save_Uniform2ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_2IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2IV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6145,7 +6243,7 @@ save_Uniform3ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_3IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3IV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6162,7 +6260,7 @@ save_Uniform4ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_4IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4IV, 3);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6181,7 +6279,7 @@ save_UniformMatrix2fvARB(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX22, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX22, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6200,7 +6298,7 @@ save_UniformMatrix3fvARB(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX33, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX33, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6219,7 +6317,7 @@ save_UniformMatrix4fvARB(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX44, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX44, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6239,7 +6337,7 @@ save_UniformMatrix2x3fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX23, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX23, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6258,7 +6356,7 @@ save_UniformMatrix3x2fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX32, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX32, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6278,7 +6376,7 @@ save_UniformMatrix2x4fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX24, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX24, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6297,7 +6395,7 @@ save_UniformMatrix4x2fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX42, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX42, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6317,7 +6415,7 @@ save_UniformMatrix3x4fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX34, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX34, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6336,7 +6434,7 @@ save_UniformMatrix4x3fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_UNIFORM_MATRIX43, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX43, 4);
    if (n) {
       n[1].i = location;
       n[2].i = count;
@@ -6360,7 +6458,7 @@ static void
 save_error(GLcontext *ctx, GLenum error, const char *s)
 {
    Node *n;
-   n = ALLOC_INSTRUCTION(ctx, OPCODE_ERROR, 2);
+   n = alloc_instruction(ctx, OPCODE_ERROR, 2);
    if (n) {
       n[1].e = error;
       n[2].data = (void *) s;
@@ -6436,13 +6534,10 @@ execute_list(GLcontext *ctx, GLuint list)
 
    done = GL_FALSE;
    while (!done) {
-      OpCode opcode = n[0].opcode;
-      int i = (int) n[0].opcode - (int) OPCODE_EXT_0;
+      const OpCode opcode = n[0].opcode;
 
-      if (i >= 0 && i < (GLint) ctx->ListExt.NumOpcodes) {
-         /* this is a driver-extended opcode */
-         ctx->ListExt.Opcode[i].Execute(ctx, &n[1]);
-         n += ctx->ListExt.Opcode[i].Size;
+      if (is_ext_opcode(opcode)) {
+         n += ext_opcode_execute(ctx, n);
       }
       else {
          switch (opcode) {
@@ -7161,7 +7256,7 @@ execute_list(GLcontext *ctx, GLuint list)
                                                       n[6].f));
             break;
 #endif
-#if FEATURE_ARB_occlusion_query
+#if FEATURE_queryobj
          case OPCODE_BEGIN_QUERY_ARB:
             CALL_BeginQueryARB(ctx->Exec, (n[1].e, n[2].ui));
             break;
@@ -7432,7 +7527,7 @@ execute_list(GLcontext *ctx, GLuint list)
 /**
  * Test if a display list number is valid.
  */
-GLboolean GLAPIENTRY
+static GLboolean GLAPIENTRY
 _mesa_IsList(GLuint list)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -7445,7 +7540,7 @@ _mesa_IsList(GLuint list)
 /**
  * Delete a sequence of consecutive display lists.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_DeleteLists(GLuint list, GLsizei range)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -7467,7 +7562,7 @@ _mesa_DeleteLists(GLuint list, GLsizei range)
  * Return a display list number, n, such that lists n through n+range-1
  * are free.
  */
-GLuint GLAPIENTRY
+static GLuint GLAPIENTRY
 _mesa_GenLists(GLsizei range)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -7507,7 +7602,7 @@ _mesa_GenLists(GLsizei range)
 /**
  * Begin a new display list.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_NewList(GLuint name, GLenum mode)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -7557,7 +7652,7 @@ _mesa_NewList(GLuint name, GLenum mode)
 /**
  * End definition of current display list. 
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_EndList(void)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -7578,7 +7673,7 @@ _mesa_EndList(void)
     */
    ctx->Driver.EndList(ctx);
 
-   (void) ALLOC_INSTRUCTION(ctx, OPCODE_END_OF_LIST, 0);
+   (void) alloc_instruction(ctx, OPCODE_END_OF_LIST, 0);
 
    /* Destroy old list, if any */
    destroy_list(ctx, ctx->ListState.CurrentList->Name);
@@ -7691,7 +7786,7 @@ _mesa_CallLists(GLsizei n, GLenum type, const GLvoid * lists)
 /**
  * Set the offset added to list numbers in glCallLists.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ListBase(GLuint base)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -8426,7 +8521,7 @@ exec_MultiModeDrawElementsIBM(const GLenum * mode,
  * struct.
  */
 void
-_mesa_init_dlist_table(struct _glapi_table *table)
+_mesa_init_save_table(struct _glapi_table *table)
 {
    _mesa_loopback_init_api_table(table);
 
@@ -8435,8 +8530,8 @@ _mesa_init_dlist_table(struct _glapi_table *table)
    SET_AlphaFunc(table, save_AlphaFunc);
    SET_Bitmap(table, save_Bitmap);
    SET_BlendFunc(table, save_BlendFunc);
-   SET_CallList(table, _mesa_save_CallList);
-   SET_CallLists(table, _mesa_save_CallLists);
+   SET_CallList(table, save_CallList);
+   SET_CallLists(table, save_CallLists);
    SET_Clear(table, save_Clear);
    SET_ClearAccum(table, save_ClearAccum);
    SET_ClearColor(table, save_ClearColor);
@@ -8457,8 +8552,8 @@ _mesa_init_dlist_table(struct _glapi_table *table)
    SET_DrawPixels(table, save_DrawPixels);
    SET_Enable(table, save_Enable);
    SET_EndList(table, _mesa_EndList);
-   SET_EvalMesh1(table, _mesa_save_EvalMesh1);
-   SET_EvalMesh2(table, _mesa_save_EvalMesh2);
+   SET_EvalMesh1(table, save_EvalMesh1);
+   SET_EvalMesh2(table, save_EvalMesh2);
    SET_Finish(table, exec_Finish);
    SET_Flush(table, exec_Flush);
    SET_Fogf(table, save_Fogf);
@@ -8930,7 +9025,7 @@ _mesa_init_dlist_table(struct _glapi_table *table)
    SET_UnmapBufferARB(table, _mesa_UnmapBufferARB);
 #endif
 
-#if FEATURE_ARB_occlusion_query
+#if FEATURE_queryobj
    SET_BeginQueryARB(table, save_BeginQueryARB);
    SET_EndQueryARB(table, save_EndQueryARB);
    SET_GenQueriesARB(table, _mesa_GenQueriesARB);
@@ -9036,13 +9131,10 @@ print_list(GLcontext *ctx, GLuint list)
 
    done = n ? GL_FALSE : GL_TRUE;
    while (!done) {
-      OpCode opcode = n[0].opcode;
-      GLint i = (GLint) n[0].opcode - (GLint) OPCODE_EXT_0;
+      const OpCode opcode = n[0].opcode;
 
-      if (i >= 0 && i < (GLint) ctx->ListExt.NumOpcodes) {
-         /* this is a driver-extended opcode */
-         ctx->ListExt.Opcode[i].Print(ctx, &n[1]);
-         n += ctx->ListExt.Opcode[i].Size;
+      if (is_ext_opcode(opcode)) {
+         n += ext_opcode_print(ctx, n);
       }
       else {
          switch (opcode) {
@@ -9297,22 +9389,21 @@ mesa_print_display_list(GLuint list)
 void
 _mesa_save_vtxfmt_init(GLvertexformat * vfmt)
 {
-   vfmt->ArrayElement = _ae_loopback_array_elt; /* generic helper */
+   _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_);
+
    vfmt->Begin = save_Begin;
-   vfmt->CallList = _mesa_save_CallList;
-   vfmt->CallLists = _mesa_save_CallLists;
+
+   _MESA_INIT_DLIST_VTXFMT(vfmt, save_);
+
    vfmt->Color3f = save_Color3f;
    vfmt->Color3fv = save_Color3fv;
    vfmt->Color4f = save_Color4f;
    vfmt->Color4fv = save_Color4fv;
    vfmt->EdgeFlag = save_EdgeFlag;
    vfmt->End = save_End;
-   vfmt->EvalCoord1f = save_EvalCoord1f;
-   vfmt->EvalCoord1fv = save_EvalCoord1fv;
-   vfmt->EvalCoord2f = save_EvalCoord2f;
-   vfmt->EvalCoord2fv = save_EvalCoord2fv;
-   vfmt->EvalPoint1 = save_EvalPoint1;
-   vfmt->EvalPoint2 = save_EvalPoint2;
+
+   _MESA_INIT_EVAL_VTXFMT(vfmt, save_);
+
    vfmt->FogCoordfEXT = save_FogCoordfEXT;
    vfmt->FogCoordfvEXT = save_FogCoordfvEXT;
    vfmt->Indexf = save_Indexf;
@@ -9361,8 +9452,6 @@ _mesa_save_vtxfmt_init(GLvertexformat * vfmt)
    vfmt->VertexAttrib4fARB = save_VertexAttrib4fARB;
    vfmt->VertexAttrib4fvARB = save_VertexAttrib4fvARB;
 
-   vfmt->EvalMesh1 = _mesa_save_EvalMesh1;
-   vfmt->EvalMesh2 = _mesa_save_EvalMesh2;
    vfmt->Rectf = save_Rectf;
 
    /* The driver is required to implement these as
@@ -9376,10 +9465,39 @@ _mesa_save_vtxfmt_init(GLvertexformat * vfmt)
    vfmt->DrawElements = 0;
    vfmt->DrawRangeElements = 0;
    vfmt->MultiDrawElemementsEXT = 0;
+   vfmt->DrawElementsBaseVertex = 0;
+   vfmt->DrawRangeElementsBaseVertex = 0;
+   vfmt->MultiDrawElemementsBaseVertex = 0;
 #endif
 }
 
 
+void
+_mesa_install_dlist_vtxfmt(struct _glapi_table *disp,
+                           const GLvertexformat *vfmt)
+{
+   SET_CallList(disp, vfmt->CallList);
+   SET_CallLists(disp, vfmt->CallLists);
+}
+
+
+void _mesa_init_dlist_dispatch(struct _glapi_table *disp)
+{
+   SET_CallList(disp, _mesa_CallList);
+   SET_CallLists(disp, _mesa_CallLists);
+
+   SET_DeleteLists(disp, _mesa_DeleteLists);
+   SET_EndList(disp, _mesa_EndList);
+   SET_GenLists(disp, _mesa_GenLists);
+   SET_IsList(disp, _mesa_IsList);
+   SET_ListBase(disp, _mesa_ListBase);
+   SET_NewList(disp, _mesa_NewList);
+}
+
+
+#endif /* FEATURE_dlist */
+
+
 /**
  * Initialize display list state for given context.
  */
@@ -9394,6 +9512,9 @@ _mesa_init_display_list(GLcontext *ctx)
       tableInitialized = GL_TRUE;
    }
 
+   /* extension info */
+   ctx->ListExt = CALLOC_STRUCT(gl_list_extensions);
+
    /* Display list */
    ctx->ListState.CallDepth = 0;
    ctx->ExecuteFlag = GL_TRUE;
@@ -9404,5 +9525,15 @@ _mesa_init_display_list(GLcontext *ctx)
    /* Display List group */
    ctx->List.ListBase = 0;
 
+#if FEATURE_dlist
    _mesa_save_vtxfmt_init(&ctx->ListState.ListVtxfmt);
+#endif
+}
+
+
+void
+_mesa_free_display_list_data(GLcontext *ctx)
+{
+   free(ctx->ListExt);
+   ctx->ListExt = NULL;
 }
diff --git a/src/mesa/main/dlist.h b/src/mesa/main/dlist.h
index ab7ec2c8db..f37a93a7f4 100644
--- a/src/mesa/main/dlist.h
+++ b/src/mesa/main/dlist.h
@@ -33,64 +33,80 @@
 #define DLIST_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 
-#if _HAVE_FULL_GL
+#if FEATURE_dlist
 
-extern void
-_mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist);
+#define _MESA_INIT_DLIST_FUNCTIONS(driver, impl)               \
+   do {                                                        \
+      (driver)->NewList           = impl ## NewList;           \
+      (driver)->EndList           = impl ## EndList;           \
+      (driver)->BeginCallList     = impl ## BeginCallList;     \
+      (driver)->EndCallList       = impl ## EndCallList;       \
+      (driver)->SaveFlushVertices = impl ## SaveFlushVertices; \
+      (driver)->NotifySaveBegin   = impl ## NotifyBegin;       \
+   } while (0)
+
+#define _MESA_INIT_DLIST_VTXFMT(vfmt, impl)  \
+   do {                                      \
+      (vfmt)->CallList  = impl ## CallList;  \
+      (vfmt)->CallLists = impl ## CallLists; \
+   } while (0)
 
 extern void GLAPIENTRY _mesa_CallList( GLuint list );
 
 extern void GLAPIENTRY _mesa_CallLists( GLsizei n, GLenum type, const GLvoid *lists );
 
-extern void GLAPIENTRY _mesa_DeleteLists( GLuint list, GLsizei range );
 
-extern void GLAPIENTRY _mesa_EndList( void );
+extern void _mesa_compile_error( GLcontext *ctx, GLenum error, const char *s );
 
-extern GLuint GLAPIENTRY _mesa_GenLists( GLsizei range );
+extern void *_mesa_dlist_alloc(GLcontext *ctx, GLuint opcode, GLuint sz);
 
-extern GLboolean GLAPIENTRY _mesa_IsList( GLuint list );
+extern GLint _mesa_dlist_alloc_opcode( GLcontext *ctx, GLuint sz,
+                                       void (*execute)( GLcontext *, void * ),
+                                       void (*destroy)( GLcontext *, void * ),
+                                       void (*print)( GLcontext *, void * ) );
 
-extern void GLAPIENTRY _mesa_ListBase( GLuint base );
+extern void _mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist);
 
-extern void GLAPIENTRY _mesa_NewList( GLuint list, GLenum mode );
+extern void _mesa_save_vtxfmt_init( GLvertexformat *vfmt );
 
-extern void GLAPIENTRY _mesa_save_CallLists( GLsizei n, GLenum type, const GLvoid *lists );
+extern void _mesa_init_save_table( struct _glapi_table *table );
 
-extern void GLAPIENTRY _mesa_save_CallList( GLuint list );
+extern void _mesa_install_dlist_vtxfmt(struct _glapi_table *disp,
+                                       const GLvertexformat *vfmt);
 
+extern void _mesa_init_dlist_dispatch(struct _glapi_table *disp);
 
+#else /* FEATURE_dlist */
 
-extern void _mesa_init_dlist_table( struct _glapi_table *table );
+#define _MESA_INIT_DLIST_FUNCTIONS(driver, impl) do { } while (0)
+#define _MESA_INIT_DLIST_VTXFMT(vfmt, impl) do { } while (0)
 
-extern void _mesa_compile_error( GLcontext *ctx, GLenum error, const char *s );
+static INLINE void
+_mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist)
+{
+   /* there should be no list to delete */
+   ASSERT_NO_FEATURE();
+}
 
+static INLINE void
+_mesa_install_dlist_vtxfmt(struct _glapi_table *disp,
+                           const GLvertexformat *vfmt)
+{
+}
 
-extern void *_mesa_alloc_instruction(GLcontext *ctx, GLuint opcode, GLuint sz);
+static INLINE void
+_mesa_init_dlist_dispatch(struct _glapi_table *disp)
+{
+}
 
-extern GLint _mesa_alloc_opcode( GLcontext *ctx, GLuint sz,
-                                 void (*execute)( GLcontext *, void * ),
-                                 void (*destroy)( GLcontext *, void * ),
-                                 void (*print)( GLcontext *, void * ) );
+#endif /* FEATURE_dlist */
 
 extern void _mesa_init_display_list( GLcontext * ctx );
 
-extern void _mesa_save_vtxfmt_init( GLvertexformat *vfmt );
-
-
-#else
-
-/** No-op */
-#define _mesa_init_dlist_table(t,ts) ((void)0)
-
-/** No-op */
-#define _mesa_init_display_list(c) ((void)0)
-
-/** No-op */
-#define _mesa_save_vtxfmt_init(v) ((void)0)
+extern void _mesa_free_display_list_data(GLcontext *ctx);
 
-#endif
 
-#endif
+#endif /* DLIST_H */
diff --git a/src/mesa/main/drawpix.c b/src/mesa/main/drawpix.c
index aef6585641..5d4b53af4c 100644
--- a/src/mesa/main/drawpix.c
+++ b/src/mesa/main/drawpix.c
@@ -33,8 +33,11 @@
 #include "image.h"
 #include "readpix.h"
 #include "state.h"
+#include "glapi/dispatch.h"
 
 
+#if FEATURE_drawpix
+
 
 /**
  * If a fragment program is enabled, check that it's valid.
@@ -47,12 +50,10 @@ valid_fragment_program(GLcontext *ctx)
 }
 
 
-#if _HAVE_FULL_GL
-
 /*
  * Execute glDrawPixels
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_DrawPixels( GLsizei width, GLsizei height,
                   GLenum format, GLenum type, const GLvoid *pixels )
 {
@@ -140,7 +141,7 @@ end:
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height,
                   GLenum type )
 {
@@ -225,11 +226,8 @@ end:
    _mesa_set_vp_override(ctx, GL_FALSE);
 }
 
-#endif /* _HAVE_FULL_GL */
 
-
-
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Bitmap( GLsizei width, GLsizei height,
               GLfloat xorig, GLfloat yorig, GLfloat xmove, GLfloat ymove,
               const GLubyte *bitmap )
@@ -309,3 +307,15 @@ _mesa_Bitmap( GLsizei width, GLsizei height,
    ctx->Current.RasterPos[0] += xmove;
    ctx->Current.RasterPos[1] += ymove;
 }
+
+
+void
+_mesa_init_drawpix_dispatch(struct _glapi_table *disp)
+{
+   SET_Bitmap(disp, _mesa_Bitmap);
+   SET_CopyPixels(disp, _mesa_CopyPixels);
+   SET_DrawPixels(disp, _mesa_DrawPixels);
+}
+
+
+#endif /* FEATURE_drawpix */
diff --git a/src/mesa/main/drawpix.h b/src/mesa/main/drawpix.h
index 6177adad6d..8ffb1a6d88 100644
--- a/src/mesa/main/drawpix.h
+++ b/src/mesa/main/drawpix.h
@@ -22,28 +22,35 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
+#ifndef DRAWPIX_H
+#define DRAWPIX_H
 
-#ifndef DRAWPIXELS_H
-#define DRAWPIXELS_H
 
+#include "main/mtypes.h"
 
-#include "main/glheader.h"
 
+#if FEATURE_drawpix
 
-extern void GLAPIENTRY
-_mesa_DrawPixels( GLsizei width, GLsizei height,
-                  GLenum format, GLenum type, const GLvoid *pixels );
+#define _MESA_INIT_DRAWPIX_FUNCTIONS(driver, impl) \
+   do {                                            \
+      (driver)->DrawPixels = impl ## DrawPixels;   \
+      (driver)->CopyPixels = impl ## CopyPixels;   \
+      (driver)->Bitmap     = impl ## Bitmap;       \
+   } while (0)
 
+extern void
+_mesa_init_drawpix_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY
-_mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height,
-                  GLenum type );
+#else /* FEATURE_drawpix */
 
+#define _MESA_INIT_DRAWPIX_FUNCTIONS(driver, impl) do { } while (0)
 
-extern void GLAPIENTRY
-_mesa_Bitmap( GLsizei width, GLsizei height,
-              GLfloat xorig, GLfloat yorig, GLfloat xmove, GLfloat ymove,
-              const GLubyte *bitmap );
+static INLINE void
+_mesa_init_drawpix_dispatch(struct _glapi_table *disp)
+{
+}
 
+#endif /* FEATURE_drawpix */
 
-#endif
+
+#endif /* DRAWPIX_H */
diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c
index 4a00440ab1..4383aed669 100644
--- a/src/mesa/main/enable.c
+++ b/src/mesa/main/enable.c
@@ -36,8 +36,8 @@
 #include "simple_list.h"
 #include "mtypes.h"
 #include "enums.h"
-#include "math/m_matrix.h"
 #include "api_arrayelt.h"
+#include "texstate.h"
 
 
 
@@ -229,8 +229,7 @@ get_texcoord_unit(GLcontext *ctx)
 static GLboolean
 enable_texture(GLcontext *ctx, GLboolean state, GLbitfield texBit)
 {
-   const GLuint curr = ctx->Texture.CurrentUnit;
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[curr];
+   struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
    const GLbitfield newenabled = state
       ? (texUnit->Enabled | texBit) : (texUnit->Enabled & ~texBit);
 
@@ -942,6 +941,14 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
          ctx->Depth.BoundsTest = state;
          break;
 
+      case GL_DEPTH_CLAMP:
+         if (ctx->Transform.DepthClamp == state)
+            return;
+	 CHECK_EXTENSION(ARB_depth_clamp, cap);
+         FLUSH_VERTICES(ctx, _NEW_TRANSFORM);
+	 ctx->Transform.DepthClamp = state;
+	 break;
+
 #if FEATURE_ATI_fragment_shader
       case GL_FRAGMENT_SHADER_ATI:
         CHECK_EXTENSION(ATI_fragment_shader, cap);
@@ -1390,6 +1397,11 @@ _mesa_IsEnabled( GLenum cap )
          CHECK_EXTENSION(EXT_depth_bounds_test);
          return ctx->Depth.BoundsTest;
 
+      /* GL_ARB_depth_clamp */
+      case GL_DEPTH_CLAMP:
+         CHECK_EXTENSION(ARB_depth_clamp);
+         return ctx->Transform.DepthClamp;
+
 #if FEATURE_ATI_fragment_shader
       case GL_FRAGMENT_SHADER_ATI:
 	 CHECK_EXTENSION(ATI_fragment_shader);
diff --git a/src/mesa/main/enums.c b/src/mesa/main/enums.c
index 9f650dadd3..606d50c59a 100644
--- a/src/mesa/main/enums.c
+++ b/src/mesa/main/enums.c
@@ -391,6 +391,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_DEPTH_BOUNDS_EXT\0"
    "GL_DEPTH_BOUNDS_TEST_EXT\0"
    "GL_DEPTH_BUFFER_BIT\0"
+   "GL_DEPTH_CLAMP\0"
    "GL_DEPTH_CLAMP_NV\0"
    "GL_DEPTH_CLEAR_VALUE\0"
    "GL_DEPTH_COMPONENT\0"
@@ -518,6 +519,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_FEEDBACK_BUFFER_SIZE\0"
    "GL_FEEDBACK_BUFFER_TYPE\0"
    "GL_FILL\0"
+   "GL_FIRST_VERTEX_CONVENTION\0"
    "GL_FIRST_VERTEX_CONVENTION_EXT\0"
    "GL_FLAT\0"
    "GL_FLOAT\0"
@@ -701,6 +703,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_INVERSE_TRANSPOSE_NV\0"
    "GL_INVERT\0"
    "GL_KEEP\0"
+   "GL_LAST_VERTEX_CONVENTION\0"
    "GL_LAST_VERTEX_CONVENTION_EXT\0"
    "GL_LEFT\0"
    "GL_LEQUAL\0"
@@ -1287,6 +1290,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_PROJECTION\0"
    "GL_PROJECTION_MATRIX\0"
    "GL_PROJECTION_STACK_DEPTH\0"
+   "GL_PROVOKING_VERTEX\0"
    "GL_PROVOKING_VERTEX_EXT\0"
    "GL_PROXY_COLOR_TABLE\0"
    "GL_PROXY_HISTOGRAM\0"
@@ -1308,6 +1312,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_Q\0"
    "GL_QUADRATIC_ATTENUATION\0"
    "GL_QUADS\0"
+   "GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION\0"
    "GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT\0"
    "GL_QUAD_MESH_SUN\0"
    "GL_QUAD_STRIP\0"
@@ -1895,7 +1900,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_ZOOM_Y\0"
    ;
 
-static const enum_elt all_enums[1857] =
+static const enum_elt all_enums[1862] =
 {
    {     0, 0x00000600 }, /* GL_2D */
    {     6, 0x00001407 }, /* GL_2_BYTES */
@@ -2252,1588 +2257,1593 @@ static const enum_elt all_enums[1857] =
    {  7262, 0x00008891 }, /* GL_DEPTH_BOUNDS_EXT */
    {  7282, 0x00008890 }, /* GL_DEPTH_BOUNDS_TEST_EXT */
    {  7307, 0x00000100 }, /* GL_DEPTH_BUFFER_BIT */
-   {  7327, 0x0000864F }, /* GL_DEPTH_CLAMP_NV */
-   {  7345, 0x00000B73 }, /* GL_DEPTH_CLEAR_VALUE */
-   {  7366, 0x00001902 }, /* GL_DEPTH_COMPONENT */
-   {  7385, 0x000081A5 }, /* GL_DEPTH_COMPONENT16 */
-   {  7406, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_ARB */
-   {  7431, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_SGIX */
-   {  7457, 0x000081A6 }, /* GL_DEPTH_COMPONENT24 */
-   {  7478, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_ARB */
-   {  7503, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_SGIX */
-   {  7529, 0x000081A7 }, /* GL_DEPTH_COMPONENT32 */
-   {  7550, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_ARB */
-   {  7575, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_SGIX */
-   {  7601, 0x00000B74 }, /* GL_DEPTH_FUNC */
-   {  7615, 0x00000B70 }, /* GL_DEPTH_RANGE */
-   {  7630, 0x00000D1E }, /* GL_DEPTH_SCALE */
-   {  7645, 0x000084F9 }, /* GL_DEPTH_STENCIL */
-   {  7662, 0x0000821A }, /* GL_DEPTH_STENCIL_ATTACHMENT */
-   {  7690, 0x000084F9 }, /* GL_DEPTH_STENCIL_NV */
-   {  7710, 0x0000886F }, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
-   {  7738, 0x0000886E }, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
-   {  7766, 0x00000B71 }, /* GL_DEPTH_TEST */
-   {  7780, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE */
-   {  7802, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE_ARB */
-   {  7828, 0x00000B72 }, /* GL_DEPTH_WRITEMASK */
-   {  7847, 0x00001201 }, /* GL_DIFFUSE */
-   {  7858, 0x00000BD0 }, /* GL_DITHER */
-   {  7868, 0x00000A02 }, /* GL_DOMAIN */
-   {  7878, 0x00001100 }, /* GL_DONT_CARE */
-   {  7891, 0x000086AE }, /* GL_DOT3_RGB */
-   {  7903, 0x000086AF }, /* GL_DOT3_RGBA */
-   {  7916, 0x000086AF }, /* GL_DOT3_RGBA_ARB */
-   {  7933, 0x00008741 }, /* GL_DOT3_RGBA_EXT */
-   {  7950, 0x000086AE }, /* GL_DOT3_RGB_ARB */
-   {  7966, 0x00008740 }, /* GL_DOT3_RGB_EXT */
-   {  7982, 0x0000140A }, /* GL_DOUBLE */
-   {  7992, 0x00000C32 }, /* GL_DOUBLEBUFFER */
-   {  8008, 0x00000C01 }, /* GL_DRAW_BUFFER */
-   {  8023, 0x00008825 }, /* GL_DRAW_BUFFER0 */
-   {  8039, 0x00008825 }, /* GL_DRAW_BUFFER0_ARB */
-   {  8059, 0x00008825 }, /* GL_DRAW_BUFFER0_ATI */
-   {  8079, 0x00008826 }, /* GL_DRAW_BUFFER1 */
-   {  8095, 0x0000882F }, /* GL_DRAW_BUFFER10 */
-   {  8112, 0x0000882F }, /* GL_DRAW_BUFFER10_ARB */
-   {  8133, 0x0000882F }, /* GL_DRAW_BUFFER10_ATI */
-   {  8154, 0x00008830 }, /* GL_DRAW_BUFFER11 */
-   {  8171, 0x00008830 }, /* GL_DRAW_BUFFER11_ARB */
-   {  8192, 0x00008830 }, /* GL_DRAW_BUFFER11_ATI */
-   {  8213, 0x00008831 }, /* GL_DRAW_BUFFER12 */
-   {  8230, 0x00008831 }, /* GL_DRAW_BUFFER12_ARB */
-   {  8251, 0x00008831 }, /* GL_DRAW_BUFFER12_ATI */
-   {  8272, 0x00008832 }, /* GL_DRAW_BUFFER13 */
-   {  8289, 0x00008832 }, /* GL_DRAW_BUFFER13_ARB */
-   {  8310, 0x00008832 }, /* GL_DRAW_BUFFER13_ATI */
-   {  8331, 0x00008833 }, /* GL_DRAW_BUFFER14 */
-   {  8348, 0x00008833 }, /* GL_DRAW_BUFFER14_ARB */
-   {  8369, 0x00008833 }, /* GL_DRAW_BUFFER14_ATI */
-   {  8390, 0x00008834 }, /* GL_DRAW_BUFFER15 */
-   {  8407, 0x00008834 }, /* GL_DRAW_BUFFER15_ARB */
-   {  8428, 0x00008834 }, /* GL_DRAW_BUFFER15_ATI */
-   {  8449, 0x00008826 }, /* GL_DRAW_BUFFER1_ARB */
-   {  8469, 0x00008826 }, /* GL_DRAW_BUFFER1_ATI */
-   {  8489, 0x00008827 }, /* GL_DRAW_BUFFER2 */
-   {  8505, 0x00008827 }, /* GL_DRAW_BUFFER2_ARB */
-   {  8525, 0x00008827 }, /* GL_DRAW_BUFFER2_ATI */
-   {  8545, 0x00008828 }, /* GL_DRAW_BUFFER3 */
-   {  8561, 0x00008828 }, /* GL_DRAW_BUFFER3_ARB */
-   {  8581, 0x00008828 }, /* GL_DRAW_BUFFER3_ATI */
-   {  8601, 0x00008829 }, /* GL_DRAW_BUFFER4 */
-   {  8617, 0x00008829 }, /* GL_DRAW_BUFFER4_ARB */
-   {  8637, 0x00008829 }, /* GL_DRAW_BUFFER4_ATI */
-   {  8657, 0x0000882A }, /* GL_DRAW_BUFFER5 */
-   {  8673, 0x0000882A }, /* GL_DRAW_BUFFER5_ARB */
-   {  8693, 0x0000882A }, /* GL_DRAW_BUFFER5_ATI */
-   {  8713, 0x0000882B }, /* GL_DRAW_BUFFER6 */
-   {  8729, 0x0000882B }, /* GL_DRAW_BUFFER6_ARB */
-   {  8749, 0x0000882B }, /* GL_DRAW_BUFFER6_ATI */
-   {  8769, 0x0000882C }, /* GL_DRAW_BUFFER7 */
-   {  8785, 0x0000882C }, /* GL_DRAW_BUFFER7_ARB */
-   {  8805, 0x0000882C }, /* GL_DRAW_BUFFER7_ATI */
-   {  8825, 0x0000882D }, /* GL_DRAW_BUFFER8 */
-   {  8841, 0x0000882D }, /* GL_DRAW_BUFFER8_ARB */
-   {  8861, 0x0000882D }, /* GL_DRAW_BUFFER8_ATI */
-   {  8881, 0x0000882E }, /* GL_DRAW_BUFFER9 */
-   {  8897, 0x0000882E }, /* GL_DRAW_BUFFER9_ARB */
-   {  8917, 0x0000882E }, /* GL_DRAW_BUFFER9_ATI */
-   {  8937, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER */
-   {  8957, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
-   {  8989, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER_EXT */
-   {  9013, 0x00000705 }, /* GL_DRAW_PIXEL_TOKEN */
-   {  9033, 0x00000304 }, /* GL_DST_ALPHA */
-   {  9046, 0x00000306 }, /* GL_DST_COLOR */
-   {  9059, 0x0000877A }, /* GL_DU8DV8_ATI */
-   {  9073, 0x00008779 }, /* GL_DUDV_ATI */
-   {  9085, 0x000088EA }, /* GL_DYNAMIC_COPY */
-   {  9101, 0x000088EA }, /* GL_DYNAMIC_COPY_ARB */
-   {  9121, 0x000088E8 }, /* GL_DYNAMIC_DRAW */
-   {  9137, 0x000088E8 }, /* GL_DYNAMIC_DRAW_ARB */
-   {  9157, 0x000088E9 }, /* GL_DYNAMIC_READ */
-   {  9173, 0x000088E9 }, /* GL_DYNAMIC_READ_ARB */
-   {  9193, 0x00000B43 }, /* GL_EDGE_FLAG */
-   {  9206, 0x00008079 }, /* GL_EDGE_FLAG_ARRAY */
-   {  9225, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
-   {  9259, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB */
-   {  9297, 0x00008093 }, /* GL_EDGE_FLAG_ARRAY_POINTER */
-   {  9324, 0x0000808C }, /* GL_EDGE_FLAG_ARRAY_STRIDE */
-   {  9350, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER */
-   {  9374, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
-   {  9406, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB */
-   {  9442, 0x00001600 }, /* GL_EMISSION */
-   {  9454, 0x00002000 }, /* GL_ENABLE_BIT */
-   {  9468, 0x00000202 }, /* GL_EQUAL */
-   {  9477, 0x00001509 }, /* GL_EQUIV */
-   {  9486, 0x00010000 }, /* GL_EVAL_BIT */
-   {  9498, 0x00000800 }, /* GL_EXP */
-   {  9505, 0x00000801 }, /* GL_EXP2 */
-   {  9513, 0x00001F03 }, /* GL_EXTENSIONS */
-   {  9527, 0x00002400 }, /* GL_EYE_LINEAR */
-   {  9541, 0x00002502 }, /* GL_EYE_PLANE */
-   {  9554, 0x0000855C }, /* GL_EYE_PLANE_ABSOLUTE_NV */
-   {  9579, 0x0000855B }, /* GL_EYE_RADIAL_NV */
-   {  9596, 0x00000000 }, /* GL_FALSE */
-   {  9605, 0x00001101 }, /* GL_FASTEST */
-   {  9616, 0x00001C01 }, /* GL_FEEDBACK */
-   {  9628, 0x00000DF0 }, /* GL_FEEDBACK_BUFFER_POINTER */
-   {  9655, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */
-   {  9679, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */
-   {  9703, 0x00001B02 }, /* GL_FILL */
-   {  9711, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION_EXT */
-   {  9742, 0x00001D00 }, /* GL_FLAT */
-   {  9750, 0x00001406 }, /* GL_FLOAT */
-   {  9759, 0x00008B5A }, /* GL_FLOAT_MAT2 */
-   {  9773, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */
-   {  9791, 0x00008B65 }, /* GL_FLOAT_MAT2x3 */
-   {  9807, 0x00008B66 }, /* GL_FLOAT_MAT2x4 */
-   {  9823, 0x00008B5B }, /* GL_FLOAT_MAT3 */
-   {  9837, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */
-   {  9855, 0x00008B67 }, /* GL_FLOAT_MAT3x2 */
-   {  9871, 0x00008B68 }, /* GL_FLOAT_MAT3x4 */
-   {  9887, 0x00008B5C }, /* GL_FLOAT_MAT4 */
-   {  9901, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */
-   {  9919, 0x00008B69 }, /* GL_FLOAT_MAT4x2 */
-   {  9935, 0x00008B6A }, /* GL_FLOAT_MAT4x3 */
-   {  9951, 0x00008B50 }, /* GL_FLOAT_VEC2 */
-   {  9965, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */
-   {  9983, 0x00008B51 }, /* GL_FLOAT_VEC3 */
-   {  9997, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */
-   { 10015, 0x00008B52 }, /* GL_FLOAT_VEC4 */
-   { 10029, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */
-   { 10047, 0x00000B60 }, /* GL_FOG */
-   { 10054, 0x00000080 }, /* GL_FOG_BIT */
-   { 10065, 0x00000B66 }, /* GL_FOG_COLOR */
-   { 10078, 0x00008451 }, /* GL_FOG_COORD */
-   { 10091, 0x00008451 }, /* GL_FOG_COORDINATE */
-   { 10109, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */
-   { 10133, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
-   { 10172, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */
-   { 10215, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */
-   { 10247, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
-   { 10278, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */
-   { 10307, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */
-   { 10332, 0x00008457 }, /* GL_FOG_COORD_ARRAY */
-   { 10351, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */
-   { 10385, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */
-   { 10412, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */
-   { 10438, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */
-   { 10462, 0x00008450 }, /* GL_FOG_COORD_SRC */
-   { 10479, 0x00000B62 }, /* GL_FOG_DENSITY */
-   { 10494, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */
-   { 10518, 0x00000B64 }, /* GL_FOG_END */
-   { 10529, 0x00000C54 }, /* GL_FOG_HINT */
-   { 10541, 0x00000B61 }, /* GL_FOG_INDEX */
-   { 10554, 0x00000B65 }, /* GL_FOG_MODE */
-   { 10566, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */
-   { 10585, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */
-   { 10610, 0x00000B63 }, /* GL_FOG_START */
-   { 10623, 0x00008452 }, /* GL_FRAGMENT_DEPTH */
-   { 10641, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */
-   { 10665, 0x00008B30 }, /* GL_FRAGMENT_SHADER */
-   { 10684, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */
-   { 10707, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
-   { 10742, 0x00008D40 }, /* GL_FRAMEBUFFER */
-   { 10757, 0x00008215 }, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
-   { 10794, 0x00008214 }, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
-   { 10830, 0x00008210 }, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
-   { 10871, 0x00008211 }, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
-   { 10912, 0x00008216 }, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
-   { 10949, 0x00008213 }, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
-   { 10986, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
-   { 11024, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
-   { 11066, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
-   { 11104, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
-   { 11146, 0x00008212 }, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
-   { 11181, 0x00008217 }, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
-   { 11220, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
-   { 11269, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
-   { 11317, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
-   { 11369, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
-   { 11409, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */
-   { 11453, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
-   { 11493, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
-   { 11537, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */
-   { 11564, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE */
-   { 11588, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */
-   { 11616, 0x00008218 }, /* GL_FRAMEBUFFER_DEFAULT */
-   { 11639, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */
-   { 11658, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
-   { 11695, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
-   { 11736, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
-   { 11777, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
-   { 11819, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
-   { 11870, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
-   { 11908, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
-   { 11953, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
-   { 12002, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
-   { 12040, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
-   { 12082, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
-   { 12114, 0x00008219 }, /* GL_FRAMEBUFFER_UNDEFINED */
-   { 12139, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED */
-   { 12166, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
-   { 12197, 0x00000404 }, /* GL_FRONT */
-   { 12206, 0x00000408 }, /* GL_FRONT_AND_BACK */
-   { 12224, 0x00000B46 }, /* GL_FRONT_FACE */
-   { 12238, 0x00000400 }, /* GL_FRONT_LEFT */
-   { 12252, 0x00000401 }, /* GL_FRONT_RIGHT */
-   { 12267, 0x00008006 }, /* GL_FUNC_ADD */
-   { 12279, 0x00008006 }, /* GL_FUNC_ADD_EXT */
-   { 12295, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */
-   { 12320, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */
-   { 12349, 0x0000800A }, /* GL_FUNC_SUBTRACT */
-   { 12366, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */
-   { 12387, 0x00008191 }, /* GL_GENERATE_MIPMAP */
-   { 12406, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */
-   { 12430, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */
-   { 12459, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */
-   { 12483, 0x00000206 }, /* GL_GEQUAL */
-   { 12493, 0x00000204 }, /* GL_GREATER */
-   { 12504, 0x00001904 }, /* GL_GREEN */
-   { 12513, 0x00000D19 }, /* GL_GREEN_BIAS */
-   { 12527, 0x00000D53 }, /* GL_GREEN_BITS */
-   { 12541, 0x00000D18 }, /* GL_GREEN_SCALE */
-   { 12556, 0x00008000 }, /* GL_HINT_BIT */
-   { 12568, 0x00008024 }, /* GL_HISTOGRAM */
-   { 12581, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */
-   { 12605, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */
-   { 12633, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */
-   { 12656, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */
-   { 12683, 0x00008024 }, /* GL_HISTOGRAM_EXT */
-   { 12700, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */
-   { 12720, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */
-   { 12744, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */
-   { 12768, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */
-   { 12796, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */
-   { 12824, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */
-   { 12856, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */
-   { 12878, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */
-   { 12904, 0x0000802D }, /* GL_HISTOGRAM_SINK */
-   { 12922, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */
-   { 12944, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */
-   { 12963, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */
-   { 12986, 0x0000862A }, /* GL_IDENTITY_NV */
-   { 13001, 0x00008150 }, /* GL_IGNORE_BORDER_HP */
-   { 13021, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
-   { 13061, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
-   { 13099, 0x00001E02 }, /* GL_INCR */
-   { 13107, 0x00008507 }, /* GL_INCR_WRAP */
-   { 13120, 0x00008507 }, /* GL_INCR_WRAP_EXT */
-   { 13137, 0x00008222 }, /* GL_INDEX */
-   { 13146, 0x00008077 }, /* GL_INDEX_ARRAY */
-   { 13161, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */
-   { 13191, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */
-   { 13225, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */
-   { 13248, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */
-   { 13270, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */
-   { 13290, 0x00000D51 }, /* GL_INDEX_BITS */
-   { 13304, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */
-   { 13325, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */
-   { 13343, 0x00000C30 }, /* GL_INDEX_MODE */
-   { 13357, 0x00000D13 }, /* GL_INDEX_OFFSET */
-   { 13373, 0x00000D12 }, /* GL_INDEX_SHIFT */
-   { 13388, 0x00000C21 }, /* GL_INDEX_WRITEMASK */
-   { 13407, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */
-   { 13426, 0x00001404 }, /* GL_INT */
-   { 13433, 0x00008049 }, /* GL_INTENSITY */
-   { 13446, 0x0000804C }, /* GL_INTENSITY12 */
-   { 13461, 0x0000804C }, /* GL_INTENSITY12_EXT */
-   { 13480, 0x0000804D }, /* GL_INTENSITY16 */
-   { 13495, 0x0000804D }, /* GL_INTENSITY16_EXT */
-   { 13514, 0x0000804A }, /* GL_INTENSITY4 */
-   { 13528, 0x0000804A }, /* GL_INTENSITY4_EXT */
-   { 13546, 0x0000804B }, /* GL_INTENSITY8 */
-   { 13560, 0x0000804B }, /* GL_INTENSITY8_EXT */
-   { 13578, 0x00008049 }, /* GL_INTENSITY_EXT */
-   { 13595, 0x00008575 }, /* GL_INTERPOLATE */
-   { 13610, 0x00008575 }, /* GL_INTERPOLATE_ARB */
-   { 13629, 0x00008575 }, /* GL_INTERPOLATE_EXT */
-   { 13648, 0x00008B53 }, /* GL_INT_VEC2 */
-   { 13660, 0x00008B53 }, /* GL_INT_VEC2_ARB */
-   { 13676, 0x00008B54 }, /* GL_INT_VEC3 */
-   { 13688, 0x00008B54 }, /* GL_INT_VEC3_ARB */
-   { 13704, 0x00008B55 }, /* GL_INT_VEC4 */
-   { 13716, 0x00008B55 }, /* GL_INT_VEC4_ARB */
-   { 13732, 0x00000500 }, /* GL_INVALID_ENUM */
-   { 13748, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION */
-   { 13781, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
-   { 13818, 0x00000502 }, /* GL_INVALID_OPERATION */
-   { 13839, 0x00000501 }, /* GL_INVALID_VALUE */
-   { 13856, 0x0000862B }, /* GL_INVERSE_NV */
-   { 13870, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */
-   { 13894, 0x0000150A }, /* GL_INVERT */
-   { 13904, 0x00001E00 }, /* GL_KEEP */
-   { 13912, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION_EXT */
-   { 13942, 0x00000406 }, /* GL_LEFT */
-   { 13950, 0x00000203 }, /* GL_LEQUAL */
-   { 13960, 0x00000201 }, /* GL_LESS */
-   { 13968, 0x00004000 }, /* GL_LIGHT0 */
-   { 13978, 0x00004001 }, /* GL_LIGHT1 */
-   { 13988, 0x00004002 }, /* GL_LIGHT2 */
-   { 13998, 0x00004003 }, /* GL_LIGHT3 */
-   { 14008, 0x00004004 }, /* GL_LIGHT4 */
-   { 14018, 0x00004005 }, /* GL_LIGHT5 */
-   { 14028, 0x00004006 }, /* GL_LIGHT6 */
-   { 14038, 0x00004007 }, /* GL_LIGHT7 */
-   { 14048, 0x00000B50 }, /* GL_LIGHTING */
-   { 14060, 0x00000040 }, /* GL_LIGHTING_BIT */
-   { 14076, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */
-   { 14099, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */
-   { 14128, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */
-   { 14161, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
-   { 14189, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */
-   { 14213, 0x00001B01 }, /* GL_LINE */
-   { 14221, 0x00002601 }, /* GL_LINEAR */
-   { 14231, 0x00001208 }, /* GL_LINEAR_ATTENUATION */
-   { 14253, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
-   { 14283, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
-   { 14314, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */
-   { 14338, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */
-   { 14363, 0x00000001 }, /* GL_LINES */
-   { 14372, 0x00000004 }, /* GL_LINE_BIT */
-   { 14384, 0x00000002 }, /* GL_LINE_LOOP */
-   { 14397, 0x00000707 }, /* GL_LINE_RESET_TOKEN */
-   { 14417, 0x00000B20 }, /* GL_LINE_SMOOTH */
-   { 14432, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */
-   { 14452, 0x00000B24 }, /* GL_LINE_STIPPLE */
-   { 14468, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */
-   { 14492, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */
-   { 14515, 0x00000003 }, /* GL_LINE_STRIP */
-   { 14529, 0x00000702 }, /* GL_LINE_TOKEN */
-   { 14543, 0x00000B21 }, /* GL_LINE_WIDTH */
-   { 14557, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */
-   { 14583, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */
-   { 14603, 0x00008B82 }, /* GL_LINK_STATUS */
-   { 14618, 0x00000B32 }, /* GL_LIST_BASE */
-   { 14631, 0x00020000 }, /* GL_LIST_BIT */
-   { 14643, 0x00000B33 }, /* GL_LIST_INDEX */
-   { 14657, 0x00000B30 }, /* GL_LIST_MODE */
-   { 14670, 0x00000101 }, /* GL_LOAD */
-   { 14678, 0x00000BF1 }, /* GL_LOGIC_OP */
-   { 14690, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */
-   { 14707, 0x00008CA1 }, /* GL_LOWER_LEFT */
-   { 14721, 0x00001909 }, /* GL_LUMINANCE */
-   { 14734, 0x00008041 }, /* GL_LUMINANCE12 */
-   { 14749, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */
-   { 14772, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */
-   { 14799, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */
-   { 14821, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */
-   { 14847, 0x00008041 }, /* GL_LUMINANCE12_EXT */
-   { 14866, 0x00008042 }, /* GL_LUMINANCE16 */
-   { 14881, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */
-   { 14904, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */
-   { 14931, 0x00008042 }, /* GL_LUMINANCE16_EXT */
-   { 14950, 0x0000803F }, /* GL_LUMINANCE4 */
-   { 14964, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */
-   { 14985, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */
-   { 15010, 0x0000803F }, /* GL_LUMINANCE4_EXT */
-   { 15028, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */
-   { 15049, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */
-   { 15074, 0x00008040 }, /* GL_LUMINANCE8 */
-   { 15088, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */
-   { 15109, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */
-   { 15134, 0x00008040 }, /* GL_LUMINANCE8_EXT */
-   { 15152, 0x0000190A }, /* GL_LUMINANCE_ALPHA */
-   { 15171, 0x00000D90 }, /* GL_MAP1_COLOR_4 */
-   { 15187, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */
-   { 15207, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */
-   { 15229, 0x00000D91 }, /* GL_MAP1_INDEX */
-   { 15243, 0x00000D92 }, /* GL_MAP1_NORMAL */
-   { 15258, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */
-   { 15282, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */
-   { 15306, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */
-   { 15330, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */
-   { 15354, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */
-   { 15371, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */
-   { 15388, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
-   { 15416, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
-   { 15445, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
-   { 15474, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
-   { 15503, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
-   { 15532, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
-   { 15561, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
-   { 15590, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
-   { 15618, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
-   { 15646, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
-   { 15674, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
-   { 15702, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
-   { 15730, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
-   { 15758, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
-   { 15786, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
-   { 15814, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
-   { 15842, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */
-   { 15858, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */
-   { 15878, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */
-   { 15900, 0x00000DB1 }, /* GL_MAP2_INDEX */
-   { 15914, 0x00000DB2 }, /* GL_MAP2_NORMAL */
-   { 15929, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */
-   { 15953, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */
-   { 15977, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */
-   { 16001, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */
-   { 16025, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */
-   { 16042, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */
-   { 16059, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
-   { 16087, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
-   { 16116, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
-   { 16145, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
-   { 16174, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
-   { 16203, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
-   { 16232, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
-   { 16261, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
-   { 16289, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
-   { 16317, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
-   { 16345, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
-   { 16373, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
-   { 16401, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
-   { 16429, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */
-   { 16457, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
-   { 16485, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
-   { 16513, 0x00000D10 }, /* GL_MAP_COLOR */
-   { 16526, 0x00000010 }, /* GL_MAP_FLUSH_EXPLICIT_BIT */
-   { 16552, 0x00000008 }, /* GL_MAP_INVALIDATE_BUFFER_BIT */
-   { 16581, 0x00000004 }, /* GL_MAP_INVALIDATE_RANGE_BIT */
-   { 16609, 0x00000001 }, /* GL_MAP_READ_BIT */
-   { 16625, 0x00000D11 }, /* GL_MAP_STENCIL */
-   { 16640, 0x00000020 }, /* GL_MAP_UNSYNCHRONIZED_BIT */
-   { 16666, 0x00000002 }, /* GL_MAP_WRITE_BIT */
-   { 16683, 0x000088C0 }, /* GL_MATRIX0_ARB */
-   { 16698, 0x00008630 }, /* GL_MATRIX0_NV */
-   { 16712, 0x000088CA }, /* GL_MATRIX10_ARB */
-   { 16728, 0x000088CB }, /* GL_MATRIX11_ARB */
-   { 16744, 0x000088CC }, /* GL_MATRIX12_ARB */
-   { 16760, 0x000088CD }, /* GL_MATRIX13_ARB */
-   { 16776, 0x000088CE }, /* GL_MATRIX14_ARB */
-   { 16792, 0x000088CF }, /* GL_MATRIX15_ARB */
-   { 16808, 0x000088D0 }, /* GL_MATRIX16_ARB */
-   { 16824, 0x000088D1 }, /* GL_MATRIX17_ARB */
-   { 16840, 0x000088D2 }, /* GL_MATRIX18_ARB */
-   { 16856, 0x000088D3 }, /* GL_MATRIX19_ARB */
-   { 16872, 0x000088C1 }, /* GL_MATRIX1_ARB */
-   { 16887, 0x00008631 }, /* GL_MATRIX1_NV */
-   { 16901, 0x000088D4 }, /* GL_MATRIX20_ARB */
-   { 16917, 0x000088D5 }, /* GL_MATRIX21_ARB */
-   { 16933, 0x000088D6 }, /* GL_MATRIX22_ARB */
-   { 16949, 0x000088D7 }, /* GL_MATRIX23_ARB */
-   { 16965, 0x000088D8 }, /* GL_MATRIX24_ARB */
-   { 16981, 0x000088D9 }, /* GL_MATRIX25_ARB */
-   { 16997, 0x000088DA }, /* GL_MATRIX26_ARB */
-   { 17013, 0x000088DB }, /* GL_MATRIX27_ARB */
-   { 17029, 0x000088DC }, /* GL_MATRIX28_ARB */
-   { 17045, 0x000088DD }, /* GL_MATRIX29_ARB */
-   { 17061, 0x000088C2 }, /* GL_MATRIX2_ARB */
-   { 17076, 0x00008632 }, /* GL_MATRIX2_NV */
-   { 17090, 0x000088DE }, /* GL_MATRIX30_ARB */
-   { 17106, 0x000088DF }, /* GL_MATRIX31_ARB */
-   { 17122, 0x000088C3 }, /* GL_MATRIX3_ARB */
-   { 17137, 0x00008633 }, /* GL_MATRIX3_NV */
-   { 17151, 0x000088C4 }, /* GL_MATRIX4_ARB */
-   { 17166, 0x00008634 }, /* GL_MATRIX4_NV */
-   { 17180, 0x000088C5 }, /* GL_MATRIX5_ARB */
-   { 17195, 0x00008635 }, /* GL_MATRIX5_NV */
-   { 17209, 0x000088C6 }, /* GL_MATRIX6_ARB */
-   { 17224, 0x00008636 }, /* GL_MATRIX6_NV */
-   { 17238, 0x000088C7 }, /* GL_MATRIX7_ARB */
-   { 17253, 0x00008637 }, /* GL_MATRIX7_NV */
-   { 17267, 0x000088C8 }, /* GL_MATRIX8_ARB */
-   { 17282, 0x000088C9 }, /* GL_MATRIX9_ARB */
-   { 17297, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */
-   { 17323, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
-   { 17357, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
-   { 17388, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
-   { 17421, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
-   { 17452, 0x00000BA0 }, /* GL_MATRIX_MODE */
-   { 17467, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */
-   { 17489, 0x00008008 }, /* GL_MAX */
-   { 17496, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */
-   { 17519, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
-   { 17551, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */
-   { 17577, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
-   { 17610, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
-   { 17636, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-   { 17670, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */
-   { 17689, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
-   { 17718, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
-   { 17750, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */
-   { 17786, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
-   { 17822, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */
-   { 17862, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */
-   { 17888, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */
-   { 17918, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */
-   { 17943, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */
-   { 17972, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
-   { 18001, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */
-   { 18034, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */
-   { 18054, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */
-   { 18078, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */
-   { 18102, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */
-   { 18126, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */
-   { 18151, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */
-   { 18169, 0x00008008 }, /* GL_MAX_EXT */
-   { 18180, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
-   { 18215, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */
-   { 18254, 0x00000D31 }, /* GL_MAX_LIGHTS */
-   { 18268, 0x00000B31 }, /* GL_MAX_LIST_NESTING */
-   { 18288, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
-   { 18326, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */
-   { 18355, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */
-   { 18379, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */
-   { 18407, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */
-   { 18430, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
-   { 18467, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
-   { 18503, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
-   { 18530, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
-   { 18559, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
-   { 18593, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
-   { 18629, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
-   { 18656, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
-   { 18688, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
-   { 18724, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
-   { 18753, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
-   { 18782, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */
-   { 18810, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
-   { 18848, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-   { 18892, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-   { 18935, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
-   { 18969, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-   { 19008, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
-   { 19045, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
-   { 19083, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-   { 19126, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-   { 19169, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
-   { 19199, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
-   { 19230, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
-   { 19266, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
-   { 19302, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */
-   { 19332, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
-   { 19366, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */
-   { 19399, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
-   { 19428, 0x00008D57 }, /* GL_MAX_SAMPLES */
-   { 19443, 0x00009111 }, /* GL_MAX_SERVER_WAIT_TIMEOUT */
-   { 19470, 0x00008504 }, /* GL_MAX_SHININESS_NV */
-   { 19490, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */
-   { 19514, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */
-   { 19536, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */
-   { 19562, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */
-   { 19589, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */
-   { 19620, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */
-   { 19644, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
-   { 19678, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */
-   { 19698, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */
-   { 19725, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */
-   { 19746, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */
-   { 19771, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */
-   { 19796, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */
-   { 19831, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */
-   { 19853, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */
-   { 19879, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */
-   { 19901, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */
-   { 19927, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
-   { 19961, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */
-   { 19999, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
-   { 20032, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */
-   { 20069, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */
-   { 20093, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */
-   { 20114, 0x00008007 }, /* GL_MIN */
-   { 20121, 0x0000802E }, /* GL_MINMAX */
-   { 20131, 0x0000802E }, /* GL_MINMAX_EXT */
-   { 20145, 0x0000802F }, /* GL_MINMAX_FORMAT */
-   { 20162, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */
-   { 20183, 0x00008030 }, /* GL_MINMAX_SINK */
-   { 20198, 0x00008030 }, /* GL_MINMAX_SINK_EXT */
-   { 20217, 0x00008007 }, /* GL_MIN_EXT */
-   { 20228, 0x00008370 }, /* GL_MIRRORED_REPEAT */
-   { 20247, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */
-   { 20270, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */
-   { 20293, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */
-   { 20313, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */
-   { 20333, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
-   { 20363, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */
-   { 20391, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
-   { 20419, 0x00001700 }, /* GL_MODELVIEW */
-   { 20432, 0x00001700 }, /* GL_MODELVIEW0_ARB */
-   { 20450, 0x0000872A }, /* GL_MODELVIEW10_ARB */
-   { 20469, 0x0000872B }, /* GL_MODELVIEW11_ARB */
-   { 20488, 0x0000872C }, /* GL_MODELVIEW12_ARB */
-   { 20507, 0x0000872D }, /* GL_MODELVIEW13_ARB */
-   { 20526, 0x0000872E }, /* GL_MODELVIEW14_ARB */
-   { 20545, 0x0000872F }, /* GL_MODELVIEW15_ARB */
-   { 20564, 0x00008730 }, /* GL_MODELVIEW16_ARB */
-   { 20583, 0x00008731 }, /* GL_MODELVIEW17_ARB */
-   { 20602, 0x00008732 }, /* GL_MODELVIEW18_ARB */
-   { 20621, 0x00008733 }, /* GL_MODELVIEW19_ARB */
-   { 20640, 0x0000850A }, /* GL_MODELVIEW1_ARB */
-   { 20658, 0x00008734 }, /* GL_MODELVIEW20_ARB */
-   { 20677, 0x00008735 }, /* GL_MODELVIEW21_ARB */
-   { 20696, 0x00008736 }, /* GL_MODELVIEW22_ARB */
-   { 20715, 0x00008737 }, /* GL_MODELVIEW23_ARB */
-   { 20734, 0x00008738 }, /* GL_MODELVIEW24_ARB */
-   { 20753, 0x00008739 }, /* GL_MODELVIEW25_ARB */
-   { 20772, 0x0000873A }, /* GL_MODELVIEW26_ARB */
-   { 20791, 0x0000873B }, /* GL_MODELVIEW27_ARB */
-   { 20810, 0x0000873C }, /* GL_MODELVIEW28_ARB */
-   { 20829, 0x0000873D }, /* GL_MODELVIEW29_ARB */
-   { 20848, 0x00008722 }, /* GL_MODELVIEW2_ARB */
-   { 20866, 0x0000873E }, /* GL_MODELVIEW30_ARB */
-   { 20885, 0x0000873F }, /* GL_MODELVIEW31_ARB */
-   { 20904, 0x00008723 }, /* GL_MODELVIEW3_ARB */
-   { 20922, 0x00008724 }, /* GL_MODELVIEW4_ARB */
-   { 20940, 0x00008725 }, /* GL_MODELVIEW5_ARB */
-   { 20958, 0x00008726 }, /* GL_MODELVIEW6_ARB */
-   { 20976, 0x00008727 }, /* GL_MODELVIEW7_ARB */
-   { 20994, 0x00008728 }, /* GL_MODELVIEW8_ARB */
-   { 21012, 0x00008729 }, /* GL_MODELVIEW9_ARB */
-   { 21030, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */
-   { 21050, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */
-   { 21077, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */
-   { 21102, 0x00002100 }, /* GL_MODULATE */
-   { 21114, 0x00008744 }, /* GL_MODULATE_ADD_ATI */
-   { 21134, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */
-   { 21161, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */
-   { 21186, 0x00000103 }, /* GL_MULT */
-   { 21194, 0x0000809D }, /* GL_MULTISAMPLE */
-   { 21209, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */
-   { 21229, 0x0000809D }, /* GL_MULTISAMPLE_ARB */
-   { 21248, 0x20000000 }, /* GL_MULTISAMPLE_BIT */
-   { 21267, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */
-   { 21291, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */
-   { 21314, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */
-   { 21344, 0x00002A25 }, /* GL_N3F_V3F */
-   { 21355, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */
-   { 21375, 0x0000150E }, /* GL_NAND */
-   { 21383, 0x00002600 }, /* GL_NEAREST */
-   { 21394, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
-   { 21425, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
-   { 21457, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */
-   { 21482, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */
-   { 21508, 0x00000200 }, /* GL_NEVER */
-   { 21517, 0x00001102 }, /* GL_NICEST */
-   { 21527, 0x00000000 }, /* GL_NONE */
-   { 21535, 0x00001505 }, /* GL_NOOP */
-   { 21543, 0x00001508 }, /* GL_NOR */
-   { 21550, 0x00000BA1 }, /* GL_NORMALIZE */
-   { 21563, 0x00008075 }, /* GL_NORMAL_ARRAY */
-   { 21579, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
-   { 21610, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */
-   { 21645, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */
-   { 21669, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */
-   { 21692, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */
-   { 21713, 0x00008511 }, /* GL_NORMAL_MAP */
-   { 21727, 0x00008511 }, /* GL_NORMAL_MAP_ARB */
-   { 21745, 0x00008511 }, /* GL_NORMAL_MAP_NV */
-   { 21762, 0x00000205 }, /* GL_NOTEQUAL */
-   { 21774, 0x00000000 }, /* GL_NO_ERROR */
-   { 21786, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
-   { 21820, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */
-   { 21858, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */
-   { 21890, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */
-   { 21932, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */
-   { 21962, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */
-   { 22002, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */
-   { 22033, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */
-   { 22062, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */
-   { 22090, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */
-   { 22120, 0x00002401 }, /* GL_OBJECT_LINEAR */
-   { 22137, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */
-   { 22163, 0x00002501 }, /* GL_OBJECT_PLANE */
-   { 22179, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */
-   { 22214, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */
-   { 22236, 0x00009112 }, /* GL_OBJECT_TYPE */
-   { 22251, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */
-   { 22270, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */
-   { 22300, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */
-   { 22321, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */
-   { 22349, 0x00000001 }, /* GL_ONE */
-   { 22356, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */
-   { 22384, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */
-   { 22416, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */
-   { 22444, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */
-   { 22476, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */
-   { 22499, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */
-   { 22522, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */
-   { 22545, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */
-   { 22568, 0x00008598 }, /* GL_OPERAND0_ALPHA */
-   { 22586, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */
-   { 22608, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */
-   { 22630, 0x00008590 }, /* GL_OPERAND0_RGB */
-   { 22646, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */
-   { 22666, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */
-   { 22686, 0x00008599 }, /* GL_OPERAND1_ALPHA */
-   { 22704, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */
-   { 22726, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */
-   { 22748, 0x00008591 }, /* GL_OPERAND1_RGB */
-   { 22764, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */
-   { 22784, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */
-   { 22804, 0x0000859A }, /* GL_OPERAND2_ALPHA */
-   { 22822, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */
-   { 22844, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */
-   { 22866, 0x00008592 }, /* GL_OPERAND2_RGB */
-   { 22882, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */
-   { 22902, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */
-   { 22922, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */
-   { 22943, 0x00008593 }, /* GL_OPERAND3_RGB_NV */
-   { 22962, 0x00001507 }, /* GL_OR */
-   { 22968, 0x00000A01 }, /* GL_ORDER */
-   { 22977, 0x0000150D }, /* GL_OR_INVERTED */
-   { 22992, 0x0000150B }, /* GL_OR_REVERSE */
-   { 23006, 0x00000505 }, /* GL_OUT_OF_MEMORY */
-   { 23023, 0x00000D05 }, /* GL_PACK_ALIGNMENT */
-   { 23041, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */
-   { 23062, 0x00008758 }, /* GL_PACK_INVERT_MESA */
-   { 23082, 0x00000D01 }, /* GL_PACK_LSB_FIRST */
-   { 23100, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */
-   { 23119, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */
-   { 23139, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */
-   { 23159, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */
-   { 23177, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */
-   { 23196, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */
-   { 23221, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */
-   { 23245, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */
-   { 23266, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */
-   { 23288, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */
-   { 23310, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */
-   { 23335, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */
-   { 23359, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */
-   { 23380, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */
-   { 23402, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */
-   { 23424, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */
-   { 23446, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */
-   { 23477, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */
-   { 23497, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */
-   { 23522, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */
-   { 23542, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */
-   { 23567, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */
-   { 23587, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */
-   { 23612, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */
-   { 23632, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */
-   { 23657, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */
-   { 23677, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */
-   { 23702, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */
-   { 23722, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */
-   { 23747, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */
-   { 23767, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */
-   { 23792, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */
-   { 23812, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */
-   { 23837, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */
-   { 23857, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */
-   { 23882, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */
-   { 23902, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */
-   { 23927, 0x00000020 }, /* GL_PIXEL_MODE_BIT */
-   { 23945, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER */
-   { 23966, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING */
-   { 23995, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */
-   { 24028, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */
-   { 24053, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER */
-   { 24076, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
-   { 24107, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */
-   { 24142, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */
-   { 24169, 0x00001B00 }, /* GL_POINT */
-   { 24178, 0x00000000 }, /* GL_POINTS */
-   { 24188, 0x00000002 }, /* GL_POINT_BIT */
-   { 24201, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */
-   { 24231, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */
-   { 24265, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */
-   { 24299, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */
-   { 24334, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */
-   { 24363, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */
-   { 24396, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */
-   { 24429, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */
-   { 24463, 0x00000B11 }, /* GL_POINT_SIZE */
-   { 24477, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */
-   { 24503, 0x00008127 }, /* GL_POINT_SIZE_MAX */
-   { 24521, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */
-   { 24543, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */
-   { 24565, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */
-   { 24588, 0x00008126 }, /* GL_POINT_SIZE_MIN */
-   { 24606, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */
-   { 24628, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */
-   { 24650, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */
-   { 24673, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */
-   { 24693, 0x00000B10 }, /* GL_POINT_SMOOTH */
-   { 24709, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */
-   { 24730, 0x00008861 }, /* GL_POINT_SPRITE */
-   { 24746, 0x00008861 }, /* GL_POINT_SPRITE_ARB */
-   { 24766, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */
-   { 24795, 0x00008861 }, /* GL_POINT_SPRITE_NV */
-   { 24814, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */
-   { 24840, 0x00000701 }, /* GL_POINT_TOKEN */
-   { 24855, 0x00000009 }, /* GL_POLYGON */
-   { 24866, 0x00000008 }, /* GL_POLYGON_BIT */
-   { 24881, 0x00000B40 }, /* GL_POLYGON_MODE */
-   { 24897, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */
-   { 24920, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */
-   { 24945, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */
-   { 24968, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */
-   { 24991, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */
-   { 25015, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */
-   { 25039, 0x00000B41 }, /* GL_POLYGON_SMOOTH */
-   { 25057, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */
-   { 25080, 0x00000B42 }, /* GL_POLYGON_STIPPLE */
-   { 25099, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */
-   { 25122, 0x00000703 }, /* GL_POLYGON_TOKEN */
-   { 25139, 0x00001203 }, /* GL_POSITION */
-   { 25151, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
-   { 25183, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */
-   { 25219, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
-   { 25252, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */
-   { 25289, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
-   { 25320, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */
-   { 25355, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
-   { 25387, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */
-   { 25423, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
-   { 25456, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
-   { 25488, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */
-   { 25524, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
-   { 25557, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */
-   { 25594, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */
-   { 25624, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */
-   { 25658, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */
-   { 25689, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */
-   { 25724, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
-   { 25755, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */
-   { 25790, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
-   { 25822, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */
-   { 25858, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */
-   { 25888, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */
-   { 25922, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */
-   { 25953, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */
-   { 25988, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */
-   { 26020, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */
-   { 26051, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */
-   { 26086, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */
-   { 26118, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */
-   { 26154, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */
-   { 26183, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */
-   { 26216, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */
-   { 26246, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */
-   { 26280, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
-   { 26319, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
-   { 26352, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
-   { 26392, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
-   { 26426, 0x00008578 }, /* GL_PREVIOUS */
-   { 26438, 0x00008578 }, /* GL_PREVIOUS_ARB */
-   { 26454, 0x00008578 }, /* GL_PREVIOUS_EXT */
-   { 26470, 0x00008577 }, /* GL_PRIMARY_COLOR */
-   { 26487, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */
-   { 26508, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */
-   { 26529, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
-   { 26562, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
-   { 26594, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */
-   { 26617, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */
-   { 26640, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */
-   { 26670, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */
-   { 26699, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */
-   { 26727, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */
-   { 26749, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */
-   { 26777, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */
-   { 26805, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */
-   { 26827, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */
-   { 26848, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-   { 26888, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-   { 26927, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
-   { 26957, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-   { 26992, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
-   { 27025, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
-   { 27059, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-   { 27098, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-   { 27137, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */
-   { 27159, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */
-   { 27185, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */
-   { 27209, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */
-   { 27232, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */
-   { 27254, 0x00008628 }, /* GL_PROGRAM_STRING_NV */
-   { 27275, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */
-   { 27296, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */
-   { 27323, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
-   { 27355, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
-   { 27387, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
-   { 27422, 0x00001701 }, /* GL_PROJECTION */
-   { 27436, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */
-   { 27457, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */
-   { 27483, 0x00008E4F }, /* GL_PROVOKING_VERTEX_EXT */
-   { 27507, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */
-   { 27528, 0x00008025 }, /* GL_PROXY_HISTOGRAM */
-   { 27547, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */
-   { 27570, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
-   { 27609, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
-   { 27647, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */
-   { 27667, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
-   { 27697, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */
-   { 27721, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */
-   { 27741, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
-   { 27771, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */
-   { 27795, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */
-   { 27815, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
-   { 27848, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */
-   { 27874, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */
-   { 27904, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
-   { 27935, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */
-   { 27965, 0x00002003 }, /* GL_Q */
-   { 27970, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */
-   { 27995, 0x00000007 }, /* GL_QUADS */
-   { 28004, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */
-   { 28052, 0x00008614 }, /* GL_QUAD_MESH_SUN */
-   { 28069, 0x00000008 }, /* GL_QUAD_STRIP */
-   { 28083, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */
-   { 28105, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */
-   { 28131, 0x00008866 }, /* GL_QUERY_RESULT */
-   { 28147, 0x00008866 }, /* GL_QUERY_RESULT_ARB */
-   { 28167, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */
-   { 28193, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */
-   { 28223, 0x00002002 }, /* GL_R */
-   { 28228, 0x00002A10 }, /* GL_R3_G3_B2 */
-   { 28240, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
-   { 28273, 0x00000C02 }, /* GL_READ_BUFFER */
-   { 28288, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER */
-   { 28308, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
-   { 28340, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */
-   { 28364, 0x000088B8 }, /* GL_READ_ONLY */
-   { 28377, 0x000088B8 }, /* GL_READ_ONLY_ARB */
-   { 28394, 0x000088BA }, /* GL_READ_WRITE */
-   { 28408, 0x000088BA }, /* GL_READ_WRITE_ARB */
-   { 28426, 0x00001903 }, /* GL_RED */
-   { 28433, 0x00008016 }, /* GL_REDUCE */
-   { 28443, 0x00008016 }, /* GL_REDUCE_EXT */
-   { 28457, 0x00000D15 }, /* GL_RED_BIAS */
-   { 28469, 0x00000D52 }, /* GL_RED_BITS */
-   { 28481, 0x00000D14 }, /* GL_RED_SCALE */
-   { 28494, 0x00008512 }, /* GL_REFLECTION_MAP */
-   { 28512, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */
-   { 28534, 0x00008512 }, /* GL_REFLECTION_MAP_NV */
-   { 28555, 0x00001C00 }, /* GL_RENDER */
-   { 28565, 0x00008D41 }, /* GL_RENDERBUFFER */
-   { 28581, 0x00008D53 }, /* GL_RENDERBUFFER_ALPHA_SIZE */
-   { 28608, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */
-   { 28636, 0x00008D52 }, /* GL_RENDERBUFFER_BLUE_SIZE */
-   { 28662, 0x00008D54 }, /* GL_RENDERBUFFER_DEPTH_SIZE */
-   { 28689, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */
-   { 28709, 0x00008D51 }, /* GL_RENDERBUFFER_GREEN_SIZE */
-   { 28736, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT */
-   { 28759, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */
-   { 28786, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
-   { 28818, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
-   { 28854, 0x00008D50 }, /* GL_RENDERBUFFER_RED_SIZE */
-   { 28879, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES */
-   { 28903, 0x00008D55 }, /* GL_RENDERBUFFER_STENCIL_SIZE */
-   { 28932, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH */
-   { 28954, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */
-   { 28980, 0x00001F01 }, /* GL_RENDERER */
-   { 28992, 0x00000C40 }, /* GL_RENDER_MODE */
-   { 29007, 0x00002901 }, /* GL_REPEAT */
-   { 29017, 0x00001E01 }, /* GL_REPLACE */
-   { 29028, 0x00008062 }, /* GL_REPLACE_EXT */
-   { 29043, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */
-   { 29066, 0x0000803A }, /* GL_RESCALE_NORMAL */
-   { 29084, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */
-   { 29106, 0x00000102 }, /* GL_RETURN */
-   { 29116, 0x00001907 }, /* GL_RGB */
-   { 29123, 0x00008052 }, /* GL_RGB10 */
-   { 29132, 0x00008059 }, /* GL_RGB10_A2 */
-   { 29144, 0x00008059 }, /* GL_RGB10_A2_EXT */
-   { 29160, 0x00008052 }, /* GL_RGB10_EXT */
-   { 29173, 0x00008053 }, /* GL_RGB12 */
-   { 29182, 0x00008053 }, /* GL_RGB12_EXT */
-   { 29195, 0x00008054 }, /* GL_RGB16 */
-   { 29204, 0x00008054 }, /* GL_RGB16_EXT */
-   { 29217, 0x0000804E }, /* GL_RGB2_EXT */
-   { 29229, 0x0000804F }, /* GL_RGB4 */
-   { 29237, 0x0000804F }, /* GL_RGB4_EXT */
-   { 29249, 0x000083A1 }, /* GL_RGB4_S3TC */
-   { 29262, 0x00008050 }, /* GL_RGB5 */
-   { 29270, 0x00008057 }, /* GL_RGB5_A1 */
-   { 29281, 0x00008057 }, /* GL_RGB5_A1_EXT */
-   { 29296, 0x00008050 }, /* GL_RGB5_EXT */
-   { 29308, 0x00008051 }, /* GL_RGB8 */
-   { 29316, 0x00008051 }, /* GL_RGB8_EXT */
-   { 29328, 0x00001908 }, /* GL_RGBA */
-   { 29336, 0x0000805A }, /* GL_RGBA12 */
-   { 29346, 0x0000805A }, /* GL_RGBA12_EXT */
-   { 29360, 0x0000805B }, /* GL_RGBA16 */
-   { 29370, 0x0000805B }, /* GL_RGBA16_EXT */
-   { 29384, 0x00008055 }, /* GL_RGBA2 */
-   { 29393, 0x00008055 }, /* GL_RGBA2_EXT */
-   { 29406, 0x00008056 }, /* GL_RGBA4 */
-   { 29415, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */
-   { 29434, 0x00008056 }, /* GL_RGBA4_EXT */
-   { 29447, 0x000083A3 }, /* GL_RGBA4_S3TC */
-   { 29461, 0x00008058 }, /* GL_RGBA8 */
-   { 29470, 0x00008058 }, /* GL_RGBA8_EXT */
-   { 29483, 0x00008F97 }, /* GL_RGBA8_SNORM */
-   { 29498, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */
-   { 29516, 0x00000C31 }, /* GL_RGBA_MODE */
-   { 29529, 0x000083A2 }, /* GL_RGBA_S3TC */
-   { 29542, 0x00008F93 }, /* GL_RGBA_SNORM */
-   { 29556, 0x000083A0 }, /* GL_RGB_S3TC */
-   { 29568, 0x00008573 }, /* GL_RGB_SCALE */
-   { 29581, 0x00008573 }, /* GL_RGB_SCALE_ARB */
-   { 29598, 0x00008573 }, /* GL_RGB_SCALE_EXT */
-   { 29615, 0x00000407 }, /* GL_RIGHT */
-   { 29624, 0x00002000 }, /* GL_S */
-   { 29629, 0x00008B5D }, /* GL_SAMPLER_1D */
-   { 29643, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */
-   { 29664, 0x00008B5E }, /* GL_SAMPLER_2D */
-   { 29678, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */
-   { 29699, 0x00008B5F }, /* GL_SAMPLER_3D */
-   { 29713, 0x00008B60 }, /* GL_SAMPLER_CUBE */
-   { 29729, 0x000080A9 }, /* GL_SAMPLES */
-   { 29740, 0x000086B4 }, /* GL_SAMPLES_3DFX */
-   { 29756, 0x000080A9 }, /* GL_SAMPLES_ARB */
-   { 29771, 0x00008914 }, /* GL_SAMPLES_PASSED */
-   { 29789, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */
-   { 29811, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
-   { 29839, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */
-   { 29871, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */
-   { 29894, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */
-   { 29921, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */
-   { 29939, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */
-   { 29962, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */
-   { 29984, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */
-   { 30003, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */
-   { 30026, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */
-   { 30052, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */
-   { 30082, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */
-   { 30107, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */
-   { 30136, 0x00080000 }, /* GL_SCISSOR_BIT */
-   { 30151, 0x00000C10 }, /* GL_SCISSOR_BOX */
-   { 30166, 0x00000C11 }, /* GL_SCISSOR_TEST */
-   { 30182, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */
-   { 30207, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
-   { 30247, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */
-   { 30291, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
-   { 30324, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
-   { 30354, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
-   { 30386, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
-   { 30416, 0x00001C02 }, /* GL_SELECT */
-   { 30426, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */
-   { 30454, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */
-   { 30479, 0x00008012 }, /* GL_SEPARABLE_2D */
-   { 30495, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */
-   { 30522, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */
-   { 30553, 0x0000150F }, /* GL_SET */
-   { 30560, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */
-   { 30581, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */
-   { 30605, 0x00008B4F }, /* GL_SHADER_TYPE */
-   { 30620, 0x00000B54 }, /* GL_SHADE_MODEL */
-   { 30635, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */
-   { 30663, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */
-   { 30686, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */
-   { 30716, 0x00001601 }, /* GL_SHININESS */
-   { 30729, 0x00001402 }, /* GL_SHORT */
-   { 30738, 0x00009119 }, /* GL_SIGNALED */
-   { 30750, 0x00008F9C }, /* GL_SIGNED_NORMALIZED */
-   { 30771, 0x000081F9 }, /* GL_SINGLE_COLOR */
-   { 30787, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */
-   { 30807, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */
-   { 30826, 0x00008C46 }, /* GL_SLUMINANCE */
-   { 30840, 0x00008C47 }, /* GL_SLUMINANCE8 */
-   { 30855, 0x00008C45 }, /* GL_SLUMINANCE8_ALPHA8 */
-   { 30877, 0x00008C44 }, /* GL_SLUMINANCE_ALPHA */
-   { 30897, 0x00001D01 }, /* GL_SMOOTH */
-   { 30907, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */
-   { 30940, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */
-   { 30967, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */
-   { 31000, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */
-   { 31027, 0x00008588 }, /* GL_SOURCE0_ALPHA */
-   { 31044, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */
-   { 31065, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */
-   { 31086, 0x00008580 }, /* GL_SOURCE0_RGB */
-   { 31101, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */
-   { 31120, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */
-   { 31139, 0x00008589 }, /* GL_SOURCE1_ALPHA */
-   { 31156, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */
-   { 31177, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */
-   { 31198, 0x00008581 }, /* GL_SOURCE1_RGB */
-   { 31213, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */
-   { 31232, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */
-   { 31251, 0x0000858A }, /* GL_SOURCE2_ALPHA */
-   { 31268, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */
-   { 31289, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */
-   { 31310, 0x00008582 }, /* GL_SOURCE2_RGB */
-   { 31325, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */
-   { 31344, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */
-   { 31363, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */
-   { 31383, 0x00008583 }, /* GL_SOURCE3_RGB_NV */
-   { 31401, 0x00001202 }, /* GL_SPECULAR */
-   { 31413, 0x00002402 }, /* GL_SPHERE_MAP */
-   { 31427, 0x00001206 }, /* GL_SPOT_CUTOFF */
-   { 31442, 0x00001204 }, /* GL_SPOT_DIRECTION */
-   { 31460, 0x00001205 }, /* GL_SPOT_EXPONENT */
-   { 31477, 0x00008588 }, /* GL_SRC0_ALPHA */
-   { 31491, 0x00008580 }, /* GL_SRC0_RGB */
-   { 31503, 0x00008589 }, /* GL_SRC1_ALPHA */
-   { 31517, 0x00008581 }, /* GL_SRC1_RGB */
-   { 31529, 0x0000858A }, /* GL_SRC2_ALPHA */
-   { 31543, 0x00008582 }, /* GL_SRC2_RGB */
-   { 31555, 0x00000302 }, /* GL_SRC_ALPHA */
-   { 31568, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */
-   { 31590, 0x00000300 }, /* GL_SRC_COLOR */
-   { 31603, 0x00008C40 }, /* GL_SRGB */
-   { 31611, 0x00008C41 }, /* GL_SRGB8 */
-   { 31620, 0x00008C43 }, /* GL_SRGB8_ALPHA8 */
-   { 31636, 0x00008C42 }, /* GL_SRGB_ALPHA */
-   { 31650, 0x00000503 }, /* GL_STACK_OVERFLOW */
-   { 31668, 0x00000504 }, /* GL_STACK_UNDERFLOW */
-   { 31687, 0x000088E6 }, /* GL_STATIC_COPY */
-   { 31702, 0x000088E6 }, /* GL_STATIC_COPY_ARB */
-   { 31721, 0x000088E4 }, /* GL_STATIC_DRAW */
-   { 31736, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */
-   { 31755, 0x000088E5 }, /* GL_STATIC_READ */
-   { 31770, 0x000088E5 }, /* GL_STATIC_READ_ARB */
-   { 31789, 0x00001802 }, /* GL_STENCIL */
-   { 31800, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT */
-   { 31822, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */
-   { 31848, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */
-   { 31869, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */
-   { 31894, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */
-   { 31915, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */
-   { 31940, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
-   { 31972, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */
-   { 32008, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
-   { 32040, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */
-   { 32076, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */
-   { 32096, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */
-   { 32123, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */
-   { 32149, 0x00000D57 }, /* GL_STENCIL_BITS */
-   { 32165, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */
-   { 32187, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */
-   { 32210, 0x00000B94 }, /* GL_STENCIL_FAIL */
-   { 32226, 0x00000B92 }, /* GL_STENCIL_FUNC */
-   { 32242, 0x00001901 }, /* GL_STENCIL_INDEX */
-   { 32259, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */
-   { 32282, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */
-   { 32304, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */
-   { 32326, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */
-   { 32348, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */
-   { 32369, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */
-   { 32396, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */
-   { 32423, 0x00000B97 }, /* GL_STENCIL_REF */
-   { 32438, 0x00000B90 }, /* GL_STENCIL_TEST */
-   { 32454, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
-   { 32483, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */
-   { 32505, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */
-   { 32526, 0x00000C33 }, /* GL_STEREO */
-   { 32536, 0x000085BE }, /* GL_STORAGE_CACHED_APPLE */
-   { 32560, 0x000085BD }, /* GL_STORAGE_PRIVATE_APPLE */
-   { 32585, 0x000085BF }, /* GL_STORAGE_SHARED_APPLE */
-   { 32609, 0x000088E2 }, /* GL_STREAM_COPY */
-   { 32624, 0x000088E2 }, /* GL_STREAM_COPY_ARB */
-   { 32643, 0x000088E0 }, /* GL_STREAM_DRAW */
-   { 32658, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */
-   { 32677, 0x000088E1 }, /* GL_STREAM_READ */
-   { 32692, 0x000088E1 }, /* GL_STREAM_READ_ARB */
-   { 32711, 0x00000D50 }, /* GL_SUBPIXEL_BITS */
-   { 32728, 0x000084E7 }, /* GL_SUBTRACT */
-   { 32740, 0x000084E7 }, /* GL_SUBTRACT_ARB */
-   { 32756, 0x00009113 }, /* GL_SYNC_CONDITION */
-   { 32774, 0x00009116 }, /* GL_SYNC_FENCE */
-   { 32788, 0x00009115 }, /* GL_SYNC_FLAGS */
-   { 32802, 0x00000001 }, /* GL_SYNC_FLUSH_COMMANDS_BIT */
-   { 32829, 0x00009117 }, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
-   { 32859, 0x00009114 }, /* GL_SYNC_STATUS */
-   { 32874, 0x00002001 }, /* GL_T */
-   { 32879, 0x00002A2A }, /* GL_T2F_C3F_V3F */
-   { 32894, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */
-   { 32913, 0x00002A29 }, /* GL_T2F_C4UB_V3F */
-   { 32929, 0x00002A2B }, /* GL_T2F_N3F_V3F */
-   { 32944, 0x00002A27 }, /* GL_T2F_V3F */
-   { 32955, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */
-   { 32974, 0x00002A28 }, /* GL_T4F_V4F */
-   { 32985, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */
-   { 33008, 0x00001702 }, /* GL_TEXTURE */
-   { 33019, 0x000084C0 }, /* GL_TEXTURE0 */
-   { 33031, 0x000084C0 }, /* GL_TEXTURE0_ARB */
-   { 33047, 0x000084C1 }, /* GL_TEXTURE1 */
-   { 33059, 0x000084CA }, /* GL_TEXTURE10 */
-   { 33072, 0x000084CA }, /* GL_TEXTURE10_ARB */
-   { 33089, 0x000084CB }, /* GL_TEXTURE11 */
-   { 33102, 0x000084CB }, /* GL_TEXTURE11_ARB */
-   { 33119, 0x000084CC }, /* GL_TEXTURE12 */
-   { 33132, 0x000084CC }, /* GL_TEXTURE12_ARB */
-   { 33149, 0x000084CD }, /* GL_TEXTURE13 */
-   { 33162, 0x000084CD }, /* GL_TEXTURE13_ARB */
-   { 33179, 0x000084CE }, /* GL_TEXTURE14 */
-   { 33192, 0x000084CE }, /* GL_TEXTURE14_ARB */
-   { 33209, 0x000084CF }, /* GL_TEXTURE15 */
-   { 33222, 0x000084CF }, /* GL_TEXTURE15_ARB */
-   { 33239, 0x000084D0 }, /* GL_TEXTURE16 */
-   { 33252, 0x000084D0 }, /* GL_TEXTURE16_ARB */
-   { 33269, 0x000084D1 }, /* GL_TEXTURE17 */
-   { 33282, 0x000084D1 }, /* GL_TEXTURE17_ARB */
-   { 33299, 0x000084D2 }, /* GL_TEXTURE18 */
-   { 33312, 0x000084D2 }, /* GL_TEXTURE18_ARB */
-   { 33329, 0x000084D3 }, /* GL_TEXTURE19 */
-   { 33342, 0x000084D3 }, /* GL_TEXTURE19_ARB */
-   { 33359, 0x000084C1 }, /* GL_TEXTURE1_ARB */
-   { 33375, 0x000084C2 }, /* GL_TEXTURE2 */
-   { 33387, 0x000084D4 }, /* GL_TEXTURE20 */
-   { 33400, 0x000084D4 }, /* GL_TEXTURE20_ARB */
-   { 33417, 0x000084D5 }, /* GL_TEXTURE21 */
-   { 33430, 0x000084D5 }, /* GL_TEXTURE21_ARB */
-   { 33447, 0x000084D6 }, /* GL_TEXTURE22 */
-   { 33460, 0x000084D6 }, /* GL_TEXTURE22_ARB */
-   { 33477, 0x000084D7 }, /* GL_TEXTURE23 */
-   { 33490, 0x000084D7 }, /* GL_TEXTURE23_ARB */
-   { 33507, 0x000084D8 }, /* GL_TEXTURE24 */
-   { 33520, 0x000084D8 }, /* GL_TEXTURE24_ARB */
-   { 33537, 0x000084D9 }, /* GL_TEXTURE25 */
-   { 33550, 0x000084D9 }, /* GL_TEXTURE25_ARB */
-   { 33567, 0x000084DA }, /* GL_TEXTURE26 */
-   { 33580, 0x000084DA }, /* GL_TEXTURE26_ARB */
-   { 33597, 0x000084DB }, /* GL_TEXTURE27 */
-   { 33610, 0x000084DB }, /* GL_TEXTURE27_ARB */
-   { 33627, 0x000084DC }, /* GL_TEXTURE28 */
-   { 33640, 0x000084DC }, /* GL_TEXTURE28_ARB */
-   { 33657, 0x000084DD }, /* GL_TEXTURE29 */
-   { 33670, 0x000084DD }, /* GL_TEXTURE29_ARB */
-   { 33687, 0x000084C2 }, /* GL_TEXTURE2_ARB */
-   { 33703, 0x000084C3 }, /* GL_TEXTURE3 */
-   { 33715, 0x000084DE }, /* GL_TEXTURE30 */
-   { 33728, 0x000084DE }, /* GL_TEXTURE30_ARB */
-   { 33745, 0x000084DF }, /* GL_TEXTURE31 */
-   { 33758, 0x000084DF }, /* GL_TEXTURE31_ARB */
-   { 33775, 0x000084C3 }, /* GL_TEXTURE3_ARB */
-   { 33791, 0x000084C4 }, /* GL_TEXTURE4 */
-   { 33803, 0x000084C4 }, /* GL_TEXTURE4_ARB */
-   { 33819, 0x000084C5 }, /* GL_TEXTURE5 */
-   { 33831, 0x000084C5 }, /* GL_TEXTURE5_ARB */
-   { 33847, 0x000084C6 }, /* GL_TEXTURE6 */
-   { 33859, 0x000084C6 }, /* GL_TEXTURE6_ARB */
-   { 33875, 0x000084C7 }, /* GL_TEXTURE7 */
-   { 33887, 0x000084C7 }, /* GL_TEXTURE7_ARB */
-   { 33903, 0x000084C8 }, /* GL_TEXTURE8 */
-   { 33915, 0x000084C8 }, /* GL_TEXTURE8_ARB */
-   { 33931, 0x000084C9 }, /* GL_TEXTURE9 */
-   { 33943, 0x000084C9 }, /* GL_TEXTURE9_ARB */
-   { 33959, 0x00000DE0 }, /* GL_TEXTURE_1D */
-   { 33973, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */
-   { 33997, 0x00000DE1 }, /* GL_TEXTURE_2D */
-   { 34011, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */
-   { 34035, 0x0000806F }, /* GL_TEXTURE_3D */
-   { 34049, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */
-   { 34071, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */
-   { 34097, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */
-   { 34119, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */
-   { 34141, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
-   { 34173, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */
-   { 34195, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
-   { 34227, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */
-   { 34249, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */
-   { 34277, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */
-   { 34309, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
-   { 34342, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */
-   { 34374, 0x00040000 }, /* GL_TEXTURE_BIT */
-   { 34389, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */
-   { 34410, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */
-   { 34435, 0x00001005 }, /* GL_TEXTURE_BORDER */
-   { 34453, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */
-   { 34477, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
-   { 34508, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
-   { 34538, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
-   { 34568, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
-   { 34603, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
-   { 34634, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-   { 34672, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */
-   { 34699, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
-   { 34731, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
-   { 34765, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */
-   { 34789, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */
-   { 34817, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */
-   { 34841, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */
-   { 34869, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
-   { 34902, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */
-   { 34926, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */
-   { 34948, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */
-   { 34970, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */
-   { 34996, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */
-   { 35030, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
-   { 35063, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */
-   { 35100, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */
-   { 35128, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */
-   { 35160, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */
-   { 35183, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
-   { 35221, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */
-   { 35263, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */
-   { 35294, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */
-   { 35322, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
-   { 35352, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */
-   { 35380, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */
-   { 35400, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */
-   { 35424, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
-   { 35455, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */
-   { 35490, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
-   { 35521, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */
-   { 35556, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
-   { 35587, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */
-   { 35622, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
-   { 35653, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */
-   { 35688, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
-   { 35719, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */
-   { 35754, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
-   { 35785, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */
-   { 35820, 0x000088F4 }, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
-   { 35849, 0x00008071 }, /* GL_TEXTURE_DEPTH */
-   { 35866, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */
-   { 35888, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */
-   { 35914, 0x00002300 }, /* GL_TEXTURE_ENV */
-   { 35929, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */
-   { 35950, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */
-   { 35970, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */
-   { 35996, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */
-   { 36016, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */
-   { 36033, 0x00000C62 }, /* GL_TEXTURE_GEN_R */
-   { 36050, 0x00000C60 }, /* GL_TEXTURE_GEN_S */
-   { 36067, 0x00000C61 }, /* GL_TEXTURE_GEN_T */
-   { 36084, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */
-   { 36109, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */
-   { 36131, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */
-   { 36157, 0x00001001 }, /* GL_TEXTURE_HEIGHT */
-   { 36175, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */
-   { 36201, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */
-   { 36227, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */
-   { 36257, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */
-   { 36284, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */
-   { 36309, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */
-   { 36329, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */
-   { 36353, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
-   { 36380, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
-   { 36407, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
-   { 36434, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */
-   { 36460, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */
-   { 36490, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */
-   { 36512, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */
-   { 36530, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
-   { 36560, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
-   { 36588, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
-   { 36616, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
-   { 36644, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */
-   { 36665, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */
-   { 36684, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */
-   { 36706, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */
-   { 36725, 0x00008066 }, /* GL_TEXTURE_PRIORITY */
-   { 36745, 0x000085B7 }, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
-   { 36775, 0x000085B8 }, /* GL_TEXTURE_RANGE_POINTER_APPLE */
-   { 36806, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */
-   { 36831, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */
-   { 36855, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */
-   { 36875, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */
-   { 36899, 0x00008067 }, /* GL_TEXTURE_RESIDENT */
-   { 36919, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */
-   { 36942, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE */
-   { 36966, 0x000085BC }, /* GL_TEXTURE_STORAGE_HINT_APPLE */
-   { 36996, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */
-   { 37021, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
-   { 37055, 0x00001000 }, /* GL_TEXTURE_WIDTH */
-   { 37072, 0x00008072 }, /* GL_TEXTURE_WRAP_R */
-   { 37090, 0x00002802 }, /* GL_TEXTURE_WRAP_S */
-   { 37108, 0x00002803 }, /* GL_TEXTURE_WRAP_T */
-   { 37126, 0x0000911B }, /* GL_TIMEOUT_EXPIRED */
-   { 37145, 0xFFFFFFFFFFFFFFFF }, /* GL_TIMEOUT_IGNORED */
-   { 37164, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */
-   { 37184, 0x00008648 }, /* GL_TRACK_MATRIX_NV */
-   { 37203, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */
-   { 37232, 0x00001000 }, /* GL_TRANSFORM_BIT */
-   { 37249, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */
-   { 37275, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */
-   { 37305, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
-   { 37337, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
-   { 37367, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */
-   { 37401, 0x0000862C }, /* GL_TRANSPOSE_NV */
-   { 37417, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */
-   { 37448, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */
-   { 37483, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */
-   { 37511, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */
-   { 37543, 0x00000004 }, /* GL_TRIANGLES */
-   { 37556, 0x00000006 }, /* GL_TRIANGLE_FAN */
-   { 37572, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */
-   { 37593, 0x00000005 }, /* GL_TRIANGLE_STRIP */
-   { 37611, 0x00000001 }, /* GL_TRUE */
-   { 37619, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */
-   { 37639, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */
-   { 37662, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */
-   { 37682, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */
-   { 37703, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */
-   { 37725, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */
-   { 37747, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */
-   { 37767, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */
-   { 37788, 0x00009118 }, /* GL_UNSIGNALED */
-   { 37802, 0x00001401 }, /* GL_UNSIGNED_BYTE */
-   { 37819, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */
-   { 37846, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */
-   { 37869, 0x00001405 }, /* GL_UNSIGNED_INT */
-   { 37885, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */
-   { 37912, 0x000084FA }, /* GL_UNSIGNED_INT_24_8 */
-   { 37933, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */
-   { 37957, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */
-   { 37988, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */
-   { 38012, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */
-   { 38040, 0x00008C17 }, /* GL_UNSIGNED_NORMALIZED */
-   { 38063, 0x00001403 }, /* GL_UNSIGNED_SHORT */
-   { 38081, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
-   { 38111, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */
-   { 38137, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
-   { 38167, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */
-   { 38193, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */
-   { 38217, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */
-   { 38245, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */
-   { 38273, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */
-   { 38300, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
-   { 38332, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */
-   { 38363, 0x00008CA2 }, /* GL_UPPER_LEFT */
-   { 38377, 0x00002A20 }, /* GL_V2F */
-   { 38384, 0x00002A21 }, /* GL_V3F */
-   { 38391, 0x00008B83 }, /* GL_VALIDATE_STATUS */
-   { 38410, 0x00001F00 }, /* GL_VENDOR */
-   { 38420, 0x00001F02 }, /* GL_VERSION */
-   { 38431, 0x00008074 }, /* GL_VERTEX_ARRAY */
-   { 38447, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING */
-   { 38471, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */
-   { 38501, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
-   { 38532, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */
-   { 38567, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */
-   { 38591, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */
-   { 38612, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */
-   { 38635, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */
-   { 38656, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
-   { 38683, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
-   { 38711, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
-   { 38739, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
-   { 38767, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
-   { 38795, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
-   { 38823, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
-   { 38851, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
-   { 38878, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
-   { 38905, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
-   { 38932, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
-   { 38959, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
-   { 38986, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
-   { 39013, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
-   { 39040, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
-   { 39067, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
-   { 39094, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
-   { 39132, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */
-   { 39174, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
-   { 39205, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */
-   { 39240, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
-   { 39274, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */
-   { 39312, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
-   { 39343, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */
-   { 39378, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
-   { 39406, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */
-   { 39438, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
-   { 39468, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */
-   { 39502, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
-   { 39530, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */
-   { 39562, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */
-   { 39582, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */
-   { 39604, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */
-   { 39633, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */
-   { 39654, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */
-   { 39683, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */
-   { 39716, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */
-   { 39748, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */
-   { 39775, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */
-   { 39806, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */
-   { 39836, 0x00008B31 }, /* GL_VERTEX_SHADER */
-   { 39853, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */
-   { 39874, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */
-   { 39901, 0x00000BA2 }, /* GL_VIEWPORT */
-   { 39913, 0x00000800 }, /* GL_VIEWPORT_BIT */
-   { 39929, 0x0000911D }, /* GL_WAIT_FAILED */
-   { 39944, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */
-   { 39964, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
-   { 39995, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */
-   { 40030, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */
-   { 40058, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */
-   { 40083, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
-   { 40110, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */
-   { 40135, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */
-   { 40159, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */
-   { 40178, 0x000088B9 }, /* GL_WRITE_ONLY */
-   { 40192, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */
-   { 40210, 0x00001506 }, /* GL_XOR */
-   { 40217, 0x000085B9 }, /* GL_YCBCR_422_APPLE */
-   { 40236, 0x00008757 }, /* GL_YCBCR_MESA */
-   { 40250, 0x00000000 }, /* GL_ZERO */
-   { 40258, 0x00000D16 }, /* GL_ZOOM_X */
-   { 40268, 0x00000D17 }, /* GL_ZOOM_Y */
+   {  7327, 0x0000864F }, /* GL_DEPTH_CLAMP */
+   {  7342, 0x0000864F }, /* GL_DEPTH_CLAMP_NV */
+   {  7360, 0x00000B73 }, /* GL_DEPTH_CLEAR_VALUE */
+   {  7381, 0x00001902 }, /* GL_DEPTH_COMPONENT */
+   {  7400, 0x000081A5 }, /* GL_DEPTH_COMPONENT16 */
+   {  7421, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_ARB */
+   {  7446, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_SGIX */
+   {  7472, 0x000081A6 }, /* GL_DEPTH_COMPONENT24 */
+   {  7493, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_ARB */
+   {  7518, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_SGIX */
+   {  7544, 0x000081A7 }, /* GL_DEPTH_COMPONENT32 */
+   {  7565, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_ARB */
+   {  7590, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_SGIX */
+   {  7616, 0x00000B74 }, /* GL_DEPTH_FUNC */
+   {  7630, 0x00000B70 }, /* GL_DEPTH_RANGE */
+   {  7645, 0x00000D1E }, /* GL_DEPTH_SCALE */
+   {  7660, 0x000084F9 }, /* GL_DEPTH_STENCIL */
+   {  7677, 0x0000821A }, /* GL_DEPTH_STENCIL_ATTACHMENT */
+   {  7705, 0x000084F9 }, /* GL_DEPTH_STENCIL_NV */
+   {  7725, 0x0000886F }, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
+   {  7753, 0x0000886E }, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
+   {  7781, 0x00000B71 }, /* GL_DEPTH_TEST */
+   {  7795, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE */
+   {  7817, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE_ARB */
+   {  7843, 0x00000B72 }, /* GL_DEPTH_WRITEMASK */
+   {  7862, 0x00001201 }, /* GL_DIFFUSE */
+   {  7873, 0x00000BD0 }, /* GL_DITHER */
+   {  7883, 0x00000A02 }, /* GL_DOMAIN */
+   {  7893, 0x00001100 }, /* GL_DONT_CARE */
+   {  7906, 0x000086AE }, /* GL_DOT3_RGB */
+   {  7918, 0x000086AF }, /* GL_DOT3_RGBA */
+   {  7931, 0x000086AF }, /* GL_DOT3_RGBA_ARB */
+   {  7948, 0x00008741 }, /* GL_DOT3_RGBA_EXT */
+   {  7965, 0x000086AE }, /* GL_DOT3_RGB_ARB */
+   {  7981, 0x00008740 }, /* GL_DOT3_RGB_EXT */
+   {  7997, 0x0000140A }, /* GL_DOUBLE */
+   {  8007, 0x00000C32 }, /* GL_DOUBLEBUFFER */
+   {  8023, 0x00000C01 }, /* GL_DRAW_BUFFER */
+   {  8038, 0x00008825 }, /* GL_DRAW_BUFFER0 */
+   {  8054, 0x00008825 }, /* GL_DRAW_BUFFER0_ARB */
+   {  8074, 0x00008825 }, /* GL_DRAW_BUFFER0_ATI */
+   {  8094, 0x00008826 }, /* GL_DRAW_BUFFER1 */
+   {  8110, 0x0000882F }, /* GL_DRAW_BUFFER10 */
+   {  8127, 0x0000882F }, /* GL_DRAW_BUFFER10_ARB */
+   {  8148, 0x0000882F }, /* GL_DRAW_BUFFER10_ATI */
+   {  8169, 0x00008830 }, /* GL_DRAW_BUFFER11 */
+   {  8186, 0x00008830 }, /* GL_DRAW_BUFFER11_ARB */
+   {  8207, 0x00008830 }, /* GL_DRAW_BUFFER11_ATI */
+   {  8228, 0x00008831 }, /* GL_DRAW_BUFFER12 */
+   {  8245, 0x00008831 }, /* GL_DRAW_BUFFER12_ARB */
+   {  8266, 0x00008831 }, /* GL_DRAW_BUFFER12_ATI */
+   {  8287, 0x00008832 }, /* GL_DRAW_BUFFER13 */
+   {  8304, 0x00008832 }, /* GL_DRAW_BUFFER13_ARB */
+   {  8325, 0x00008832 }, /* GL_DRAW_BUFFER13_ATI */
+   {  8346, 0x00008833 }, /* GL_DRAW_BUFFER14 */
+   {  8363, 0x00008833 }, /* GL_DRAW_BUFFER14_ARB */
+   {  8384, 0x00008833 }, /* GL_DRAW_BUFFER14_ATI */
+   {  8405, 0x00008834 }, /* GL_DRAW_BUFFER15 */
+   {  8422, 0x00008834 }, /* GL_DRAW_BUFFER15_ARB */
+   {  8443, 0x00008834 }, /* GL_DRAW_BUFFER15_ATI */
+   {  8464, 0x00008826 }, /* GL_DRAW_BUFFER1_ARB */
+   {  8484, 0x00008826 }, /* GL_DRAW_BUFFER1_ATI */
+   {  8504, 0x00008827 }, /* GL_DRAW_BUFFER2 */
+   {  8520, 0x00008827 }, /* GL_DRAW_BUFFER2_ARB */
+   {  8540, 0x00008827 }, /* GL_DRAW_BUFFER2_ATI */
+   {  8560, 0x00008828 }, /* GL_DRAW_BUFFER3 */
+   {  8576, 0x00008828 }, /* GL_DRAW_BUFFER3_ARB */
+   {  8596, 0x00008828 }, /* GL_DRAW_BUFFER3_ATI */
+   {  8616, 0x00008829 }, /* GL_DRAW_BUFFER4 */
+   {  8632, 0x00008829 }, /* GL_DRAW_BUFFER4_ARB */
+   {  8652, 0x00008829 }, /* GL_DRAW_BUFFER4_ATI */
+   {  8672, 0x0000882A }, /* GL_DRAW_BUFFER5 */
+   {  8688, 0x0000882A }, /* GL_DRAW_BUFFER5_ARB */
+   {  8708, 0x0000882A }, /* GL_DRAW_BUFFER5_ATI */
+   {  8728, 0x0000882B }, /* GL_DRAW_BUFFER6 */
+   {  8744, 0x0000882B }, /* GL_DRAW_BUFFER6_ARB */
+   {  8764, 0x0000882B }, /* GL_DRAW_BUFFER6_ATI */
+   {  8784, 0x0000882C }, /* GL_DRAW_BUFFER7 */
+   {  8800, 0x0000882C }, /* GL_DRAW_BUFFER7_ARB */
+   {  8820, 0x0000882C }, /* GL_DRAW_BUFFER7_ATI */
+   {  8840, 0x0000882D }, /* GL_DRAW_BUFFER8 */
+   {  8856, 0x0000882D }, /* GL_DRAW_BUFFER8_ARB */
+   {  8876, 0x0000882D }, /* GL_DRAW_BUFFER8_ATI */
+   {  8896, 0x0000882E }, /* GL_DRAW_BUFFER9 */
+   {  8912, 0x0000882E }, /* GL_DRAW_BUFFER9_ARB */
+   {  8932, 0x0000882E }, /* GL_DRAW_BUFFER9_ATI */
+   {  8952, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER */
+   {  8972, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
+   {  9004, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER_EXT */
+   {  9028, 0x00000705 }, /* GL_DRAW_PIXEL_TOKEN */
+   {  9048, 0x00000304 }, /* GL_DST_ALPHA */
+   {  9061, 0x00000306 }, /* GL_DST_COLOR */
+   {  9074, 0x0000877A }, /* GL_DU8DV8_ATI */
+   {  9088, 0x00008779 }, /* GL_DUDV_ATI */
+   {  9100, 0x000088EA }, /* GL_DYNAMIC_COPY */
+   {  9116, 0x000088EA }, /* GL_DYNAMIC_COPY_ARB */
+   {  9136, 0x000088E8 }, /* GL_DYNAMIC_DRAW */
+   {  9152, 0x000088E8 }, /* GL_DYNAMIC_DRAW_ARB */
+   {  9172, 0x000088E9 }, /* GL_DYNAMIC_READ */
+   {  9188, 0x000088E9 }, /* GL_DYNAMIC_READ_ARB */
+   {  9208, 0x00000B43 }, /* GL_EDGE_FLAG */
+   {  9221, 0x00008079 }, /* GL_EDGE_FLAG_ARRAY */
+   {  9240, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
+   {  9274, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB */
+   {  9312, 0x00008093 }, /* GL_EDGE_FLAG_ARRAY_POINTER */
+   {  9339, 0x0000808C }, /* GL_EDGE_FLAG_ARRAY_STRIDE */
+   {  9365, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER */
+   {  9389, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
+   {  9421, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB */
+   {  9457, 0x00001600 }, /* GL_EMISSION */
+   {  9469, 0x00002000 }, /* GL_ENABLE_BIT */
+   {  9483, 0x00000202 }, /* GL_EQUAL */
+   {  9492, 0x00001509 }, /* GL_EQUIV */
+   {  9501, 0x00010000 }, /* GL_EVAL_BIT */
+   {  9513, 0x00000800 }, /* GL_EXP */
+   {  9520, 0x00000801 }, /* GL_EXP2 */
+   {  9528, 0x00001F03 }, /* GL_EXTENSIONS */
+   {  9542, 0x00002400 }, /* GL_EYE_LINEAR */
+   {  9556, 0x00002502 }, /* GL_EYE_PLANE */
+   {  9569, 0x0000855C }, /* GL_EYE_PLANE_ABSOLUTE_NV */
+   {  9594, 0x0000855B }, /* GL_EYE_RADIAL_NV */
+   {  9611, 0x00000000 }, /* GL_FALSE */
+   {  9620, 0x00001101 }, /* GL_FASTEST */
+   {  9631, 0x00001C01 }, /* GL_FEEDBACK */
+   {  9643, 0x00000DF0 }, /* GL_FEEDBACK_BUFFER_POINTER */
+   {  9670, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */
+   {  9694, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */
+   {  9718, 0x00001B02 }, /* GL_FILL */
+   {  9726, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION */
+   {  9753, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION_EXT */
+   {  9784, 0x00001D00 }, /* GL_FLAT */
+   {  9792, 0x00001406 }, /* GL_FLOAT */
+   {  9801, 0x00008B5A }, /* GL_FLOAT_MAT2 */
+   {  9815, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */
+   {  9833, 0x00008B65 }, /* GL_FLOAT_MAT2x3 */
+   {  9849, 0x00008B66 }, /* GL_FLOAT_MAT2x4 */
+   {  9865, 0x00008B5B }, /* GL_FLOAT_MAT3 */
+   {  9879, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */
+   {  9897, 0x00008B67 }, /* GL_FLOAT_MAT3x2 */
+   {  9913, 0x00008B68 }, /* GL_FLOAT_MAT3x4 */
+   {  9929, 0x00008B5C }, /* GL_FLOAT_MAT4 */
+   {  9943, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */
+   {  9961, 0x00008B69 }, /* GL_FLOAT_MAT4x2 */
+   {  9977, 0x00008B6A }, /* GL_FLOAT_MAT4x3 */
+   {  9993, 0x00008B50 }, /* GL_FLOAT_VEC2 */
+   { 10007, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */
+   { 10025, 0x00008B51 }, /* GL_FLOAT_VEC3 */
+   { 10039, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */
+   { 10057, 0x00008B52 }, /* GL_FLOAT_VEC4 */
+   { 10071, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */
+   { 10089, 0x00000B60 }, /* GL_FOG */
+   { 10096, 0x00000080 }, /* GL_FOG_BIT */
+   { 10107, 0x00000B66 }, /* GL_FOG_COLOR */
+   { 10120, 0x00008451 }, /* GL_FOG_COORD */
+   { 10133, 0x00008451 }, /* GL_FOG_COORDINATE */
+   { 10151, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */
+   { 10175, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
+   { 10214, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */
+   { 10257, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */
+   { 10289, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
+   { 10320, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */
+   { 10349, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */
+   { 10374, 0x00008457 }, /* GL_FOG_COORD_ARRAY */
+   { 10393, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */
+   { 10427, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */
+   { 10454, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */
+   { 10480, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */
+   { 10504, 0x00008450 }, /* GL_FOG_COORD_SRC */
+   { 10521, 0x00000B62 }, /* GL_FOG_DENSITY */
+   { 10536, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */
+   { 10560, 0x00000B64 }, /* GL_FOG_END */
+   { 10571, 0x00000C54 }, /* GL_FOG_HINT */
+   { 10583, 0x00000B61 }, /* GL_FOG_INDEX */
+   { 10596, 0x00000B65 }, /* GL_FOG_MODE */
+   { 10608, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */
+   { 10627, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */
+   { 10652, 0x00000B63 }, /* GL_FOG_START */
+   { 10665, 0x00008452 }, /* GL_FRAGMENT_DEPTH */
+   { 10683, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */
+   { 10707, 0x00008B30 }, /* GL_FRAGMENT_SHADER */
+   { 10726, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */
+   { 10749, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
+   { 10784, 0x00008D40 }, /* GL_FRAMEBUFFER */
+   { 10799, 0x00008215 }, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
+   { 10836, 0x00008214 }, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
+   { 10872, 0x00008210 }, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
+   { 10913, 0x00008211 }, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
+   { 10954, 0x00008216 }, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
+   { 10991, 0x00008213 }, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
+   { 11028, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
+   { 11066, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
+   { 11108, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
+   { 11146, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
+   { 11188, 0x00008212 }, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
+   { 11223, 0x00008217 }, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
+   { 11262, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
+   { 11311, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
+   { 11359, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
+   { 11411, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
+   { 11451, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */
+   { 11495, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
+   { 11535, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
+   { 11579, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */
+   { 11606, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE */
+   { 11630, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */
+   { 11658, 0x00008218 }, /* GL_FRAMEBUFFER_DEFAULT */
+   { 11681, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */
+   { 11700, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
+   { 11737, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
+   { 11778, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
+   { 11819, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
+   { 11861, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
+   { 11912, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
+   { 11950, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
+   { 11995, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
+   { 12044, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
+   { 12082, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
+   { 12124, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
+   { 12156, 0x00008219 }, /* GL_FRAMEBUFFER_UNDEFINED */
+   { 12181, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED */
+   { 12208, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
+   { 12239, 0x00000404 }, /* GL_FRONT */
+   { 12248, 0x00000408 }, /* GL_FRONT_AND_BACK */
+   { 12266, 0x00000B46 }, /* GL_FRONT_FACE */
+   { 12280, 0x00000400 }, /* GL_FRONT_LEFT */
+   { 12294, 0x00000401 }, /* GL_FRONT_RIGHT */
+   { 12309, 0x00008006 }, /* GL_FUNC_ADD */
+   { 12321, 0x00008006 }, /* GL_FUNC_ADD_EXT */
+   { 12337, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */
+   { 12362, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */
+   { 12391, 0x0000800A }, /* GL_FUNC_SUBTRACT */
+   { 12408, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */
+   { 12429, 0x00008191 }, /* GL_GENERATE_MIPMAP */
+   { 12448, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */
+   { 12472, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */
+   { 12501, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */
+   { 12525, 0x00000206 }, /* GL_GEQUAL */
+   { 12535, 0x00000204 }, /* GL_GREATER */
+   { 12546, 0x00001904 }, /* GL_GREEN */
+   { 12555, 0x00000D19 }, /* GL_GREEN_BIAS */
+   { 12569, 0x00000D53 }, /* GL_GREEN_BITS */
+   { 12583, 0x00000D18 }, /* GL_GREEN_SCALE */
+   { 12598, 0x00008000 }, /* GL_HINT_BIT */
+   { 12610, 0x00008024 }, /* GL_HISTOGRAM */
+   { 12623, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */
+   { 12647, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */
+   { 12675, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */
+   { 12698, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */
+   { 12725, 0x00008024 }, /* GL_HISTOGRAM_EXT */
+   { 12742, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */
+   { 12762, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */
+   { 12786, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */
+   { 12810, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */
+   { 12838, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */
+   { 12866, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */
+   { 12898, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */
+   { 12920, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */
+   { 12946, 0x0000802D }, /* GL_HISTOGRAM_SINK */
+   { 12964, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */
+   { 12986, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */
+   { 13005, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */
+   { 13028, 0x0000862A }, /* GL_IDENTITY_NV */
+   { 13043, 0x00008150 }, /* GL_IGNORE_BORDER_HP */
+   { 13063, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
+   { 13103, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
+   { 13141, 0x00001E02 }, /* GL_INCR */
+   { 13149, 0x00008507 }, /* GL_INCR_WRAP */
+   { 13162, 0x00008507 }, /* GL_INCR_WRAP_EXT */
+   { 13179, 0x00008222 }, /* GL_INDEX */
+   { 13188, 0x00008077 }, /* GL_INDEX_ARRAY */
+   { 13203, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */
+   { 13233, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */
+   { 13267, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */
+   { 13290, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */
+   { 13312, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */
+   { 13332, 0x00000D51 }, /* GL_INDEX_BITS */
+   { 13346, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */
+   { 13367, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */
+   { 13385, 0x00000C30 }, /* GL_INDEX_MODE */
+   { 13399, 0x00000D13 }, /* GL_INDEX_OFFSET */
+   { 13415, 0x00000D12 }, /* GL_INDEX_SHIFT */
+   { 13430, 0x00000C21 }, /* GL_INDEX_WRITEMASK */
+   { 13449, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */
+   { 13468, 0x00001404 }, /* GL_INT */
+   { 13475, 0x00008049 }, /* GL_INTENSITY */
+   { 13488, 0x0000804C }, /* GL_INTENSITY12 */
+   { 13503, 0x0000804C }, /* GL_INTENSITY12_EXT */
+   { 13522, 0x0000804D }, /* GL_INTENSITY16 */
+   { 13537, 0x0000804D }, /* GL_INTENSITY16_EXT */
+   { 13556, 0x0000804A }, /* GL_INTENSITY4 */
+   { 13570, 0x0000804A }, /* GL_INTENSITY4_EXT */
+   { 13588, 0x0000804B }, /* GL_INTENSITY8 */
+   { 13602, 0x0000804B }, /* GL_INTENSITY8_EXT */
+   { 13620, 0x00008049 }, /* GL_INTENSITY_EXT */
+   { 13637, 0x00008575 }, /* GL_INTERPOLATE */
+   { 13652, 0x00008575 }, /* GL_INTERPOLATE_ARB */
+   { 13671, 0x00008575 }, /* GL_INTERPOLATE_EXT */
+   { 13690, 0x00008B53 }, /* GL_INT_VEC2 */
+   { 13702, 0x00008B53 }, /* GL_INT_VEC2_ARB */
+   { 13718, 0x00008B54 }, /* GL_INT_VEC3 */
+   { 13730, 0x00008B54 }, /* GL_INT_VEC3_ARB */
+   { 13746, 0x00008B55 }, /* GL_INT_VEC4 */
+   { 13758, 0x00008B55 }, /* GL_INT_VEC4_ARB */
+   { 13774, 0x00000500 }, /* GL_INVALID_ENUM */
+   { 13790, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION */
+   { 13823, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
+   { 13860, 0x00000502 }, /* GL_INVALID_OPERATION */
+   { 13881, 0x00000501 }, /* GL_INVALID_VALUE */
+   { 13898, 0x0000862B }, /* GL_INVERSE_NV */
+   { 13912, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */
+   { 13936, 0x0000150A }, /* GL_INVERT */
+   { 13946, 0x00001E00 }, /* GL_KEEP */
+   { 13954, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION */
+   { 13980, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION_EXT */
+   { 14010, 0x00000406 }, /* GL_LEFT */
+   { 14018, 0x00000203 }, /* GL_LEQUAL */
+   { 14028, 0x00000201 }, /* GL_LESS */
+   { 14036, 0x00004000 }, /* GL_LIGHT0 */
+   { 14046, 0x00004001 }, /* GL_LIGHT1 */
+   { 14056, 0x00004002 }, /* GL_LIGHT2 */
+   { 14066, 0x00004003 }, /* GL_LIGHT3 */
+   { 14076, 0x00004004 }, /* GL_LIGHT4 */
+   { 14086, 0x00004005 }, /* GL_LIGHT5 */
+   { 14096, 0x00004006 }, /* GL_LIGHT6 */
+   { 14106, 0x00004007 }, /* GL_LIGHT7 */
+   { 14116, 0x00000B50 }, /* GL_LIGHTING */
+   { 14128, 0x00000040 }, /* GL_LIGHTING_BIT */
+   { 14144, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */
+   { 14167, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */
+   { 14196, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */
+   { 14229, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
+   { 14257, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */
+   { 14281, 0x00001B01 }, /* GL_LINE */
+   { 14289, 0x00002601 }, /* GL_LINEAR */
+   { 14299, 0x00001208 }, /* GL_LINEAR_ATTENUATION */
+   { 14321, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
+   { 14351, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
+   { 14382, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */
+   { 14406, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */
+   { 14431, 0x00000001 }, /* GL_LINES */
+   { 14440, 0x00000004 }, /* GL_LINE_BIT */
+   { 14452, 0x00000002 }, /* GL_LINE_LOOP */
+   { 14465, 0x00000707 }, /* GL_LINE_RESET_TOKEN */
+   { 14485, 0x00000B20 }, /* GL_LINE_SMOOTH */
+   { 14500, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */
+   { 14520, 0x00000B24 }, /* GL_LINE_STIPPLE */
+   { 14536, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */
+   { 14560, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */
+   { 14583, 0x00000003 }, /* GL_LINE_STRIP */
+   { 14597, 0x00000702 }, /* GL_LINE_TOKEN */
+   { 14611, 0x00000B21 }, /* GL_LINE_WIDTH */
+   { 14625, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */
+   { 14651, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */
+   { 14671, 0x00008B82 }, /* GL_LINK_STATUS */
+   { 14686, 0x00000B32 }, /* GL_LIST_BASE */
+   { 14699, 0x00020000 }, /* GL_LIST_BIT */
+   { 14711, 0x00000B33 }, /* GL_LIST_INDEX */
+   { 14725, 0x00000B30 }, /* GL_LIST_MODE */
+   { 14738, 0x00000101 }, /* GL_LOAD */
+   { 14746, 0x00000BF1 }, /* GL_LOGIC_OP */
+   { 14758, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */
+   { 14775, 0x00008CA1 }, /* GL_LOWER_LEFT */
+   { 14789, 0x00001909 }, /* GL_LUMINANCE */
+   { 14802, 0x00008041 }, /* GL_LUMINANCE12 */
+   { 14817, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */
+   { 14840, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */
+   { 14867, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */
+   { 14889, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */
+   { 14915, 0x00008041 }, /* GL_LUMINANCE12_EXT */
+   { 14934, 0x00008042 }, /* GL_LUMINANCE16 */
+   { 14949, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */
+   { 14972, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */
+   { 14999, 0x00008042 }, /* GL_LUMINANCE16_EXT */
+   { 15018, 0x0000803F }, /* GL_LUMINANCE4 */
+   { 15032, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */
+   { 15053, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */
+   { 15078, 0x0000803F }, /* GL_LUMINANCE4_EXT */
+   { 15096, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */
+   { 15117, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */
+   { 15142, 0x00008040 }, /* GL_LUMINANCE8 */
+   { 15156, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */
+   { 15177, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */
+   { 15202, 0x00008040 }, /* GL_LUMINANCE8_EXT */
+   { 15220, 0x0000190A }, /* GL_LUMINANCE_ALPHA */
+   { 15239, 0x00000D90 }, /* GL_MAP1_COLOR_4 */
+   { 15255, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */
+   { 15275, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */
+   { 15297, 0x00000D91 }, /* GL_MAP1_INDEX */
+   { 15311, 0x00000D92 }, /* GL_MAP1_NORMAL */
+   { 15326, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */
+   { 15350, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */
+   { 15374, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */
+   { 15398, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */
+   { 15422, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */
+   { 15439, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */
+   { 15456, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
+   { 15484, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
+   { 15513, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
+   { 15542, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
+   { 15571, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
+   { 15600, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
+   { 15629, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
+   { 15658, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
+   { 15686, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
+   { 15714, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
+   { 15742, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
+   { 15770, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
+   { 15798, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
+   { 15826, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
+   { 15854, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
+   { 15882, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
+   { 15910, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */
+   { 15926, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */
+   { 15946, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */
+   { 15968, 0x00000DB1 }, /* GL_MAP2_INDEX */
+   { 15982, 0x00000DB2 }, /* GL_MAP2_NORMAL */
+   { 15997, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */
+   { 16021, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */
+   { 16045, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */
+   { 16069, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */
+   { 16093, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */
+   { 16110, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */
+   { 16127, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
+   { 16155, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
+   { 16184, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
+   { 16213, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
+   { 16242, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
+   { 16271, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
+   { 16300, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
+   { 16329, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
+   { 16357, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
+   { 16385, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
+   { 16413, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
+   { 16441, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
+   { 16469, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
+   { 16497, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */
+   { 16525, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
+   { 16553, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
+   { 16581, 0x00000D10 }, /* GL_MAP_COLOR */
+   { 16594, 0x00000010 }, /* GL_MAP_FLUSH_EXPLICIT_BIT */
+   { 16620, 0x00000008 }, /* GL_MAP_INVALIDATE_BUFFER_BIT */
+   { 16649, 0x00000004 }, /* GL_MAP_INVALIDATE_RANGE_BIT */
+   { 16677, 0x00000001 }, /* GL_MAP_READ_BIT */
+   { 16693, 0x00000D11 }, /* GL_MAP_STENCIL */
+   { 16708, 0x00000020 }, /* GL_MAP_UNSYNCHRONIZED_BIT */
+   { 16734, 0x00000002 }, /* GL_MAP_WRITE_BIT */
+   { 16751, 0x000088C0 }, /* GL_MATRIX0_ARB */
+   { 16766, 0x00008630 }, /* GL_MATRIX0_NV */
+   { 16780, 0x000088CA }, /* GL_MATRIX10_ARB */
+   { 16796, 0x000088CB }, /* GL_MATRIX11_ARB */
+   { 16812, 0x000088CC }, /* GL_MATRIX12_ARB */
+   { 16828, 0x000088CD }, /* GL_MATRIX13_ARB */
+   { 16844, 0x000088CE }, /* GL_MATRIX14_ARB */
+   { 16860, 0x000088CF }, /* GL_MATRIX15_ARB */
+   { 16876, 0x000088D0 }, /* GL_MATRIX16_ARB */
+   { 16892, 0x000088D1 }, /* GL_MATRIX17_ARB */
+   { 16908, 0x000088D2 }, /* GL_MATRIX18_ARB */
+   { 16924, 0x000088D3 }, /* GL_MATRIX19_ARB */
+   { 16940, 0x000088C1 }, /* GL_MATRIX1_ARB */
+   { 16955, 0x00008631 }, /* GL_MATRIX1_NV */
+   { 16969, 0x000088D4 }, /* GL_MATRIX20_ARB */
+   { 16985, 0x000088D5 }, /* GL_MATRIX21_ARB */
+   { 17001, 0x000088D6 }, /* GL_MATRIX22_ARB */
+   { 17017, 0x000088D7 }, /* GL_MATRIX23_ARB */
+   { 17033, 0x000088D8 }, /* GL_MATRIX24_ARB */
+   { 17049, 0x000088D9 }, /* GL_MATRIX25_ARB */
+   { 17065, 0x000088DA }, /* GL_MATRIX26_ARB */
+   { 17081, 0x000088DB }, /* GL_MATRIX27_ARB */
+   { 17097, 0x000088DC }, /* GL_MATRIX28_ARB */
+   { 17113, 0x000088DD }, /* GL_MATRIX29_ARB */
+   { 17129, 0x000088C2 }, /* GL_MATRIX2_ARB */
+   { 17144, 0x00008632 }, /* GL_MATRIX2_NV */
+   { 17158, 0x000088DE }, /* GL_MATRIX30_ARB */
+   { 17174, 0x000088DF }, /* GL_MATRIX31_ARB */
+   { 17190, 0x000088C3 }, /* GL_MATRIX3_ARB */
+   { 17205, 0x00008633 }, /* GL_MATRIX3_NV */
+   { 17219, 0x000088C4 }, /* GL_MATRIX4_ARB */
+   { 17234, 0x00008634 }, /* GL_MATRIX4_NV */
+   { 17248, 0x000088C5 }, /* GL_MATRIX5_ARB */
+   { 17263, 0x00008635 }, /* GL_MATRIX5_NV */
+   { 17277, 0x000088C6 }, /* GL_MATRIX6_ARB */
+   { 17292, 0x00008636 }, /* GL_MATRIX6_NV */
+   { 17306, 0x000088C7 }, /* GL_MATRIX7_ARB */
+   { 17321, 0x00008637 }, /* GL_MATRIX7_NV */
+   { 17335, 0x000088C8 }, /* GL_MATRIX8_ARB */
+   { 17350, 0x000088C9 }, /* GL_MATRIX9_ARB */
+   { 17365, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */
+   { 17391, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
+   { 17425, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
+   { 17456, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
+   { 17489, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
+   { 17520, 0x00000BA0 }, /* GL_MATRIX_MODE */
+   { 17535, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */
+   { 17557, 0x00008008 }, /* GL_MAX */
+   { 17564, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */
+   { 17587, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
+   { 17619, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */
+   { 17645, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
+   { 17678, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
+   { 17704, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+   { 17738, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */
+   { 17757, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
+   { 17786, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
+   { 17818, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */
+   { 17854, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
+   { 17890, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */
+   { 17930, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */
+   { 17956, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */
+   { 17986, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */
+   { 18011, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */
+   { 18040, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
+   { 18069, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */
+   { 18102, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */
+   { 18122, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */
+   { 18146, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */
+   { 18170, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */
+   { 18194, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */
+   { 18219, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */
+   { 18237, 0x00008008 }, /* GL_MAX_EXT */
+   { 18248, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
+   { 18283, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */
+   { 18322, 0x00000D31 }, /* GL_MAX_LIGHTS */
+   { 18336, 0x00000B31 }, /* GL_MAX_LIST_NESTING */
+   { 18356, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
+   { 18394, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */
+   { 18423, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */
+   { 18447, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */
+   { 18475, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */
+   { 18498, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
+   { 18535, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
+   { 18571, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
+   { 18598, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
+   { 18627, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
+   { 18661, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
+   { 18697, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
+   { 18724, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
+   { 18756, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
+   { 18792, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
+   { 18821, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
+   { 18850, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */
+   { 18878, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
+   { 18916, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+   { 18960, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+   { 19003, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
+   { 19037, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+   { 19076, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
+   { 19113, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
+   { 19151, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+   { 19194, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+   { 19237, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
+   { 19267, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
+   { 19298, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
+   { 19334, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
+   { 19370, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */
+   { 19400, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
+   { 19434, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */
+   { 19467, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
+   { 19496, 0x00008D57 }, /* GL_MAX_SAMPLES */
+   { 19511, 0x00009111 }, /* GL_MAX_SERVER_WAIT_TIMEOUT */
+   { 19538, 0x00008504 }, /* GL_MAX_SHININESS_NV */
+   { 19558, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */
+   { 19582, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */
+   { 19604, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */
+   { 19630, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */
+   { 19657, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */
+   { 19688, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */
+   { 19712, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
+   { 19746, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */
+   { 19766, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */
+   { 19793, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */
+   { 19814, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */
+   { 19839, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */
+   { 19864, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */
+   { 19899, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */
+   { 19921, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */
+   { 19947, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */
+   { 19969, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */
+   { 19995, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
+   { 20029, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */
+   { 20067, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
+   { 20100, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */
+   { 20137, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */
+   { 20161, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */
+   { 20182, 0x00008007 }, /* GL_MIN */
+   { 20189, 0x0000802E }, /* GL_MINMAX */
+   { 20199, 0x0000802E }, /* GL_MINMAX_EXT */
+   { 20213, 0x0000802F }, /* GL_MINMAX_FORMAT */
+   { 20230, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */
+   { 20251, 0x00008030 }, /* GL_MINMAX_SINK */
+   { 20266, 0x00008030 }, /* GL_MINMAX_SINK_EXT */
+   { 20285, 0x00008007 }, /* GL_MIN_EXT */
+   { 20296, 0x00008370 }, /* GL_MIRRORED_REPEAT */
+   { 20315, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */
+   { 20338, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */
+   { 20361, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */
+   { 20381, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */
+   { 20401, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
+   { 20431, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */
+   { 20459, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
+   { 20487, 0x00001700 }, /* GL_MODELVIEW */
+   { 20500, 0x00001700 }, /* GL_MODELVIEW0_ARB */
+   { 20518, 0x0000872A }, /* GL_MODELVIEW10_ARB */
+   { 20537, 0x0000872B }, /* GL_MODELVIEW11_ARB */
+   { 20556, 0x0000872C }, /* GL_MODELVIEW12_ARB */
+   { 20575, 0x0000872D }, /* GL_MODELVIEW13_ARB */
+   { 20594, 0x0000872E }, /* GL_MODELVIEW14_ARB */
+   { 20613, 0x0000872F }, /* GL_MODELVIEW15_ARB */
+   { 20632, 0x00008730 }, /* GL_MODELVIEW16_ARB */
+   { 20651, 0x00008731 }, /* GL_MODELVIEW17_ARB */
+   { 20670, 0x00008732 }, /* GL_MODELVIEW18_ARB */
+   { 20689, 0x00008733 }, /* GL_MODELVIEW19_ARB */
+   { 20708, 0x0000850A }, /* GL_MODELVIEW1_ARB */
+   { 20726, 0x00008734 }, /* GL_MODELVIEW20_ARB */
+   { 20745, 0x00008735 }, /* GL_MODELVIEW21_ARB */
+   { 20764, 0x00008736 }, /* GL_MODELVIEW22_ARB */
+   { 20783, 0x00008737 }, /* GL_MODELVIEW23_ARB */
+   { 20802, 0x00008738 }, /* GL_MODELVIEW24_ARB */
+   { 20821, 0x00008739 }, /* GL_MODELVIEW25_ARB */
+   { 20840, 0x0000873A }, /* GL_MODELVIEW26_ARB */
+   { 20859, 0x0000873B }, /* GL_MODELVIEW27_ARB */
+   { 20878, 0x0000873C }, /* GL_MODELVIEW28_ARB */
+   { 20897, 0x0000873D }, /* GL_MODELVIEW29_ARB */
+   { 20916, 0x00008722 }, /* GL_MODELVIEW2_ARB */
+   { 20934, 0x0000873E }, /* GL_MODELVIEW30_ARB */
+   { 20953, 0x0000873F }, /* GL_MODELVIEW31_ARB */
+   { 20972, 0x00008723 }, /* GL_MODELVIEW3_ARB */
+   { 20990, 0x00008724 }, /* GL_MODELVIEW4_ARB */
+   { 21008, 0x00008725 }, /* GL_MODELVIEW5_ARB */
+   { 21026, 0x00008726 }, /* GL_MODELVIEW6_ARB */
+   { 21044, 0x00008727 }, /* GL_MODELVIEW7_ARB */
+   { 21062, 0x00008728 }, /* GL_MODELVIEW8_ARB */
+   { 21080, 0x00008729 }, /* GL_MODELVIEW9_ARB */
+   { 21098, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */
+   { 21118, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */
+   { 21145, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */
+   { 21170, 0x00002100 }, /* GL_MODULATE */
+   { 21182, 0x00008744 }, /* GL_MODULATE_ADD_ATI */
+   { 21202, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */
+   { 21229, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */
+   { 21254, 0x00000103 }, /* GL_MULT */
+   { 21262, 0x0000809D }, /* GL_MULTISAMPLE */
+   { 21277, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */
+   { 21297, 0x0000809D }, /* GL_MULTISAMPLE_ARB */
+   { 21316, 0x20000000 }, /* GL_MULTISAMPLE_BIT */
+   { 21335, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */
+   { 21359, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */
+   { 21382, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */
+   { 21412, 0x00002A25 }, /* GL_N3F_V3F */
+   { 21423, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */
+   { 21443, 0x0000150E }, /* GL_NAND */
+   { 21451, 0x00002600 }, /* GL_NEAREST */
+   { 21462, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
+   { 21493, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
+   { 21525, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */
+   { 21550, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */
+   { 21576, 0x00000200 }, /* GL_NEVER */
+   { 21585, 0x00001102 }, /* GL_NICEST */
+   { 21595, 0x00000000 }, /* GL_NONE */
+   { 21603, 0x00001505 }, /* GL_NOOP */
+   { 21611, 0x00001508 }, /* GL_NOR */
+   { 21618, 0x00000BA1 }, /* GL_NORMALIZE */
+   { 21631, 0x00008075 }, /* GL_NORMAL_ARRAY */
+   { 21647, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
+   { 21678, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */
+   { 21713, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */
+   { 21737, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */
+   { 21760, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */
+   { 21781, 0x00008511 }, /* GL_NORMAL_MAP */
+   { 21795, 0x00008511 }, /* GL_NORMAL_MAP_ARB */
+   { 21813, 0x00008511 }, /* GL_NORMAL_MAP_NV */
+   { 21830, 0x00000205 }, /* GL_NOTEQUAL */
+   { 21842, 0x00000000 }, /* GL_NO_ERROR */
+   { 21854, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
+   { 21888, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */
+   { 21926, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */
+   { 21958, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */
+   { 22000, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */
+   { 22030, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */
+   { 22070, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */
+   { 22101, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */
+   { 22130, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */
+   { 22158, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */
+   { 22188, 0x00002401 }, /* GL_OBJECT_LINEAR */
+   { 22205, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */
+   { 22231, 0x00002501 }, /* GL_OBJECT_PLANE */
+   { 22247, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */
+   { 22282, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */
+   { 22304, 0x00009112 }, /* GL_OBJECT_TYPE */
+   { 22319, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */
+   { 22338, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */
+   { 22368, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */
+   { 22389, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */
+   { 22417, 0x00000001 }, /* GL_ONE */
+   { 22424, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */
+   { 22452, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */
+   { 22484, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */
+   { 22512, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */
+   { 22544, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */
+   { 22567, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */
+   { 22590, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */
+   { 22613, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */
+   { 22636, 0x00008598 }, /* GL_OPERAND0_ALPHA */
+   { 22654, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */
+   { 22676, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */
+   { 22698, 0x00008590 }, /* GL_OPERAND0_RGB */
+   { 22714, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */
+   { 22734, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */
+   { 22754, 0x00008599 }, /* GL_OPERAND1_ALPHA */
+   { 22772, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */
+   { 22794, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */
+   { 22816, 0x00008591 }, /* GL_OPERAND1_RGB */
+   { 22832, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */
+   { 22852, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */
+   { 22872, 0x0000859A }, /* GL_OPERAND2_ALPHA */
+   { 22890, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */
+   { 22912, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */
+   { 22934, 0x00008592 }, /* GL_OPERAND2_RGB */
+   { 22950, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */
+   { 22970, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */
+   { 22990, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */
+   { 23011, 0x00008593 }, /* GL_OPERAND3_RGB_NV */
+   { 23030, 0x00001507 }, /* GL_OR */
+   { 23036, 0x00000A01 }, /* GL_ORDER */
+   { 23045, 0x0000150D }, /* GL_OR_INVERTED */
+   { 23060, 0x0000150B }, /* GL_OR_REVERSE */
+   { 23074, 0x00000505 }, /* GL_OUT_OF_MEMORY */
+   { 23091, 0x00000D05 }, /* GL_PACK_ALIGNMENT */
+   { 23109, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */
+   { 23130, 0x00008758 }, /* GL_PACK_INVERT_MESA */
+   { 23150, 0x00000D01 }, /* GL_PACK_LSB_FIRST */
+   { 23168, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */
+   { 23187, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */
+   { 23207, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */
+   { 23227, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */
+   { 23245, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */
+   { 23264, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */
+   { 23289, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */
+   { 23313, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */
+   { 23334, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */
+   { 23356, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */
+   { 23378, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */
+   { 23403, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */
+   { 23427, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */
+   { 23448, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */
+   { 23470, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */
+   { 23492, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */
+   { 23514, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */
+   { 23545, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */
+   { 23565, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */
+   { 23590, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */
+   { 23610, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */
+   { 23635, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */
+   { 23655, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */
+   { 23680, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */
+   { 23700, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */
+   { 23725, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */
+   { 23745, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */
+   { 23770, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */
+   { 23790, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */
+   { 23815, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */
+   { 23835, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */
+   { 23860, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */
+   { 23880, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */
+   { 23905, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */
+   { 23925, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */
+   { 23950, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */
+   { 23970, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */
+   { 23995, 0x00000020 }, /* GL_PIXEL_MODE_BIT */
+   { 24013, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER */
+   { 24034, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING */
+   { 24063, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */
+   { 24096, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */
+   { 24121, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER */
+   { 24144, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
+   { 24175, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */
+   { 24210, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */
+   { 24237, 0x00001B00 }, /* GL_POINT */
+   { 24246, 0x00000000 }, /* GL_POINTS */
+   { 24256, 0x00000002 }, /* GL_POINT_BIT */
+   { 24269, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */
+   { 24299, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */
+   { 24333, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */
+   { 24367, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */
+   { 24402, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */
+   { 24431, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */
+   { 24464, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */
+   { 24497, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */
+   { 24531, 0x00000B11 }, /* GL_POINT_SIZE */
+   { 24545, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */
+   { 24571, 0x00008127 }, /* GL_POINT_SIZE_MAX */
+   { 24589, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */
+   { 24611, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */
+   { 24633, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */
+   { 24656, 0x00008126 }, /* GL_POINT_SIZE_MIN */
+   { 24674, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */
+   { 24696, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */
+   { 24718, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */
+   { 24741, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */
+   { 24761, 0x00000B10 }, /* GL_POINT_SMOOTH */
+   { 24777, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */
+   { 24798, 0x00008861 }, /* GL_POINT_SPRITE */
+   { 24814, 0x00008861 }, /* GL_POINT_SPRITE_ARB */
+   { 24834, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */
+   { 24863, 0x00008861 }, /* GL_POINT_SPRITE_NV */
+   { 24882, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */
+   { 24908, 0x00000701 }, /* GL_POINT_TOKEN */
+   { 24923, 0x00000009 }, /* GL_POLYGON */
+   { 24934, 0x00000008 }, /* GL_POLYGON_BIT */
+   { 24949, 0x00000B40 }, /* GL_POLYGON_MODE */
+   { 24965, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */
+   { 24988, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */
+   { 25013, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */
+   { 25036, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */
+   { 25059, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */
+   { 25083, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */
+   { 25107, 0x00000B41 }, /* GL_POLYGON_SMOOTH */
+   { 25125, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */
+   { 25148, 0x00000B42 }, /* GL_POLYGON_STIPPLE */
+   { 25167, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */
+   { 25190, 0x00000703 }, /* GL_POLYGON_TOKEN */
+   { 25207, 0x00001203 }, /* GL_POSITION */
+   { 25219, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
+   { 25251, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */
+   { 25287, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
+   { 25320, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */
+   { 25357, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
+   { 25388, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */
+   { 25423, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
+   { 25455, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */
+   { 25491, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
+   { 25524, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
+   { 25556, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */
+   { 25592, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
+   { 25625, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */
+   { 25662, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */
+   { 25692, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */
+   { 25726, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */
+   { 25757, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */
+   { 25792, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
+   { 25823, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */
+   { 25858, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
+   { 25890, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */
+   { 25926, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */
+   { 25956, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */
+   { 25990, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */
+   { 26021, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */
+   { 26056, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */
+   { 26088, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */
+   { 26119, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */
+   { 26154, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */
+   { 26186, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */
+   { 26222, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */
+   { 26251, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */
+   { 26284, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */
+   { 26314, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */
+   { 26348, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
+   { 26387, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
+   { 26420, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
+   { 26460, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
+   { 26494, 0x00008578 }, /* GL_PREVIOUS */
+   { 26506, 0x00008578 }, /* GL_PREVIOUS_ARB */
+   { 26522, 0x00008578 }, /* GL_PREVIOUS_EXT */
+   { 26538, 0x00008577 }, /* GL_PRIMARY_COLOR */
+   { 26555, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */
+   { 26576, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */
+   { 26597, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
+   { 26630, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
+   { 26662, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */
+   { 26685, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */
+   { 26708, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */
+   { 26738, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */
+   { 26767, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */
+   { 26795, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */
+   { 26817, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */
+   { 26845, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */
+   { 26873, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */
+   { 26895, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */
+   { 26916, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+   { 26956, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+   { 26995, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
+   { 27025, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+   { 27060, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
+   { 27093, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
+   { 27127, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+   { 27166, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+   { 27205, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */
+   { 27227, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */
+   { 27253, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */
+   { 27277, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */
+   { 27300, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */
+   { 27322, 0x00008628 }, /* GL_PROGRAM_STRING_NV */
+   { 27343, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */
+   { 27364, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */
+   { 27391, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
+   { 27423, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
+   { 27455, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
+   { 27490, 0x00001701 }, /* GL_PROJECTION */
+   { 27504, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */
+   { 27525, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */
+   { 27551, 0x00008E4F }, /* GL_PROVOKING_VERTEX */
+   { 27571, 0x00008E4F }, /* GL_PROVOKING_VERTEX_EXT */
+   { 27595, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */
+   { 27616, 0x00008025 }, /* GL_PROXY_HISTOGRAM */
+   { 27635, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */
+   { 27658, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
+   { 27697, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
+   { 27735, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */
+   { 27755, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
+   { 27785, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */
+   { 27809, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */
+   { 27829, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
+   { 27859, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */
+   { 27883, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */
+   { 27903, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
+   { 27936, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */
+   { 27962, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */
+   { 27992, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
+   { 28023, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */
+   { 28053, 0x00002003 }, /* GL_Q */
+   { 28058, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */
+   { 28083, 0x00000007 }, /* GL_QUADS */
+   { 28092, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
+   { 28136, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */
+   { 28184, 0x00008614 }, /* GL_QUAD_MESH_SUN */
+   { 28201, 0x00000008 }, /* GL_QUAD_STRIP */
+   { 28215, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */
+   { 28237, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */
+   { 28263, 0x00008866 }, /* GL_QUERY_RESULT */
+   { 28279, 0x00008866 }, /* GL_QUERY_RESULT_ARB */
+   { 28299, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */
+   { 28325, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */
+   { 28355, 0x00002002 }, /* GL_R */
+   { 28360, 0x00002A10 }, /* GL_R3_G3_B2 */
+   { 28372, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
+   { 28405, 0x00000C02 }, /* GL_READ_BUFFER */
+   { 28420, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER */
+   { 28440, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
+   { 28472, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */
+   { 28496, 0x000088B8 }, /* GL_READ_ONLY */
+   { 28509, 0x000088B8 }, /* GL_READ_ONLY_ARB */
+   { 28526, 0x000088BA }, /* GL_READ_WRITE */
+   { 28540, 0x000088BA }, /* GL_READ_WRITE_ARB */
+   { 28558, 0x00001903 }, /* GL_RED */
+   { 28565, 0x00008016 }, /* GL_REDUCE */
+   { 28575, 0x00008016 }, /* GL_REDUCE_EXT */
+   { 28589, 0x00000D15 }, /* GL_RED_BIAS */
+   { 28601, 0x00000D52 }, /* GL_RED_BITS */
+   { 28613, 0x00000D14 }, /* GL_RED_SCALE */
+   { 28626, 0x00008512 }, /* GL_REFLECTION_MAP */
+   { 28644, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */
+   { 28666, 0x00008512 }, /* GL_REFLECTION_MAP_NV */
+   { 28687, 0x00001C00 }, /* GL_RENDER */
+   { 28697, 0x00008D41 }, /* GL_RENDERBUFFER */
+   { 28713, 0x00008D53 }, /* GL_RENDERBUFFER_ALPHA_SIZE */
+   { 28740, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */
+   { 28768, 0x00008D52 }, /* GL_RENDERBUFFER_BLUE_SIZE */
+   { 28794, 0x00008D54 }, /* GL_RENDERBUFFER_DEPTH_SIZE */
+   { 28821, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */
+   { 28841, 0x00008D51 }, /* GL_RENDERBUFFER_GREEN_SIZE */
+   { 28868, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT */
+   { 28891, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */
+   { 28918, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
+   { 28950, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
+   { 28986, 0x00008D50 }, /* GL_RENDERBUFFER_RED_SIZE */
+   { 29011, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES */
+   { 29035, 0x00008D55 }, /* GL_RENDERBUFFER_STENCIL_SIZE */
+   { 29064, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH */
+   { 29086, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */
+   { 29112, 0x00001F01 }, /* GL_RENDERER */
+   { 29124, 0x00000C40 }, /* GL_RENDER_MODE */
+   { 29139, 0x00002901 }, /* GL_REPEAT */
+   { 29149, 0x00001E01 }, /* GL_REPLACE */
+   { 29160, 0x00008062 }, /* GL_REPLACE_EXT */
+   { 29175, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */
+   { 29198, 0x0000803A }, /* GL_RESCALE_NORMAL */
+   { 29216, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */
+   { 29238, 0x00000102 }, /* GL_RETURN */
+   { 29248, 0x00001907 }, /* GL_RGB */
+   { 29255, 0x00008052 }, /* GL_RGB10 */
+   { 29264, 0x00008059 }, /* GL_RGB10_A2 */
+   { 29276, 0x00008059 }, /* GL_RGB10_A2_EXT */
+   { 29292, 0x00008052 }, /* GL_RGB10_EXT */
+   { 29305, 0x00008053 }, /* GL_RGB12 */
+   { 29314, 0x00008053 }, /* GL_RGB12_EXT */
+   { 29327, 0x00008054 }, /* GL_RGB16 */
+   { 29336, 0x00008054 }, /* GL_RGB16_EXT */
+   { 29349, 0x0000804E }, /* GL_RGB2_EXT */
+   { 29361, 0x0000804F }, /* GL_RGB4 */
+   { 29369, 0x0000804F }, /* GL_RGB4_EXT */
+   { 29381, 0x000083A1 }, /* GL_RGB4_S3TC */
+   { 29394, 0x00008050 }, /* GL_RGB5 */
+   { 29402, 0x00008057 }, /* GL_RGB5_A1 */
+   { 29413, 0x00008057 }, /* GL_RGB5_A1_EXT */
+   { 29428, 0x00008050 }, /* GL_RGB5_EXT */
+   { 29440, 0x00008051 }, /* GL_RGB8 */
+   { 29448, 0x00008051 }, /* GL_RGB8_EXT */
+   { 29460, 0x00001908 }, /* GL_RGBA */
+   { 29468, 0x0000805A }, /* GL_RGBA12 */
+   { 29478, 0x0000805A }, /* GL_RGBA12_EXT */
+   { 29492, 0x0000805B }, /* GL_RGBA16 */
+   { 29502, 0x0000805B }, /* GL_RGBA16_EXT */
+   { 29516, 0x00008055 }, /* GL_RGBA2 */
+   { 29525, 0x00008055 }, /* GL_RGBA2_EXT */
+   { 29538, 0x00008056 }, /* GL_RGBA4 */
+   { 29547, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */
+   { 29566, 0x00008056 }, /* GL_RGBA4_EXT */
+   { 29579, 0x000083A3 }, /* GL_RGBA4_S3TC */
+   { 29593, 0x00008058 }, /* GL_RGBA8 */
+   { 29602, 0x00008058 }, /* GL_RGBA8_EXT */
+   { 29615, 0x00008F97 }, /* GL_RGBA8_SNORM */
+   { 29630, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */
+   { 29648, 0x00000C31 }, /* GL_RGBA_MODE */
+   { 29661, 0x000083A2 }, /* GL_RGBA_S3TC */
+   { 29674, 0x00008F93 }, /* GL_RGBA_SNORM */
+   { 29688, 0x000083A0 }, /* GL_RGB_S3TC */
+   { 29700, 0x00008573 }, /* GL_RGB_SCALE */
+   { 29713, 0x00008573 }, /* GL_RGB_SCALE_ARB */
+   { 29730, 0x00008573 }, /* GL_RGB_SCALE_EXT */
+   { 29747, 0x00000407 }, /* GL_RIGHT */
+   { 29756, 0x00002000 }, /* GL_S */
+   { 29761, 0x00008B5D }, /* GL_SAMPLER_1D */
+   { 29775, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */
+   { 29796, 0x00008B5E }, /* GL_SAMPLER_2D */
+   { 29810, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */
+   { 29831, 0x00008B5F }, /* GL_SAMPLER_3D */
+   { 29845, 0x00008B60 }, /* GL_SAMPLER_CUBE */
+   { 29861, 0x000080A9 }, /* GL_SAMPLES */
+   { 29872, 0x000086B4 }, /* GL_SAMPLES_3DFX */
+   { 29888, 0x000080A9 }, /* GL_SAMPLES_ARB */
+   { 29903, 0x00008914 }, /* GL_SAMPLES_PASSED */
+   { 29921, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */
+   { 29943, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
+   { 29971, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */
+   { 30003, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */
+   { 30026, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */
+   { 30053, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */
+   { 30071, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */
+   { 30094, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */
+   { 30116, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */
+   { 30135, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */
+   { 30158, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */
+   { 30184, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */
+   { 30214, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */
+   { 30239, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */
+   { 30268, 0x00080000 }, /* GL_SCISSOR_BIT */
+   { 30283, 0x00000C10 }, /* GL_SCISSOR_BOX */
+   { 30298, 0x00000C11 }, /* GL_SCISSOR_TEST */
+   { 30314, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */
+   { 30339, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
+   { 30379, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */
+   { 30423, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
+   { 30456, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
+   { 30486, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
+   { 30518, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
+   { 30548, 0x00001C02 }, /* GL_SELECT */
+   { 30558, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */
+   { 30586, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */
+   { 30611, 0x00008012 }, /* GL_SEPARABLE_2D */
+   { 30627, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */
+   { 30654, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */
+   { 30685, 0x0000150F }, /* GL_SET */
+   { 30692, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */
+   { 30713, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */
+   { 30737, 0x00008B4F }, /* GL_SHADER_TYPE */
+   { 30752, 0x00000B54 }, /* GL_SHADE_MODEL */
+   { 30767, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */
+   { 30795, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */
+   { 30818, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */
+   { 30848, 0x00001601 }, /* GL_SHININESS */
+   { 30861, 0x00001402 }, /* GL_SHORT */
+   { 30870, 0x00009119 }, /* GL_SIGNALED */
+   { 30882, 0x00008F9C }, /* GL_SIGNED_NORMALIZED */
+   { 30903, 0x000081F9 }, /* GL_SINGLE_COLOR */
+   { 30919, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */
+   { 30939, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */
+   { 30958, 0x00008C46 }, /* GL_SLUMINANCE */
+   { 30972, 0x00008C47 }, /* GL_SLUMINANCE8 */
+   { 30987, 0x00008C45 }, /* GL_SLUMINANCE8_ALPHA8 */
+   { 31009, 0x00008C44 }, /* GL_SLUMINANCE_ALPHA */
+   { 31029, 0x00001D01 }, /* GL_SMOOTH */
+   { 31039, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */
+   { 31072, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */
+   { 31099, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */
+   { 31132, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */
+   { 31159, 0x00008588 }, /* GL_SOURCE0_ALPHA */
+   { 31176, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */
+   { 31197, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */
+   { 31218, 0x00008580 }, /* GL_SOURCE0_RGB */
+   { 31233, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */
+   { 31252, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */
+   { 31271, 0x00008589 }, /* GL_SOURCE1_ALPHA */
+   { 31288, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */
+   { 31309, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */
+   { 31330, 0x00008581 }, /* GL_SOURCE1_RGB */
+   { 31345, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */
+   { 31364, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */
+   { 31383, 0x0000858A }, /* GL_SOURCE2_ALPHA */
+   { 31400, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */
+   { 31421, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */
+   { 31442, 0x00008582 }, /* GL_SOURCE2_RGB */
+   { 31457, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */
+   { 31476, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */
+   { 31495, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */
+   { 31515, 0x00008583 }, /* GL_SOURCE3_RGB_NV */
+   { 31533, 0x00001202 }, /* GL_SPECULAR */
+   { 31545, 0x00002402 }, /* GL_SPHERE_MAP */
+   { 31559, 0x00001206 }, /* GL_SPOT_CUTOFF */
+   { 31574, 0x00001204 }, /* GL_SPOT_DIRECTION */
+   { 31592, 0x00001205 }, /* GL_SPOT_EXPONENT */
+   { 31609, 0x00008588 }, /* GL_SRC0_ALPHA */
+   { 31623, 0x00008580 }, /* GL_SRC0_RGB */
+   { 31635, 0x00008589 }, /* GL_SRC1_ALPHA */
+   { 31649, 0x00008581 }, /* GL_SRC1_RGB */
+   { 31661, 0x0000858A }, /* GL_SRC2_ALPHA */
+   { 31675, 0x00008582 }, /* GL_SRC2_RGB */
+   { 31687, 0x00000302 }, /* GL_SRC_ALPHA */
+   { 31700, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */
+   { 31722, 0x00000300 }, /* GL_SRC_COLOR */
+   { 31735, 0x00008C40 }, /* GL_SRGB */
+   { 31743, 0x00008C41 }, /* GL_SRGB8 */
+   { 31752, 0x00008C43 }, /* GL_SRGB8_ALPHA8 */
+   { 31768, 0x00008C42 }, /* GL_SRGB_ALPHA */
+   { 31782, 0x00000503 }, /* GL_STACK_OVERFLOW */
+   { 31800, 0x00000504 }, /* GL_STACK_UNDERFLOW */
+   { 31819, 0x000088E6 }, /* GL_STATIC_COPY */
+   { 31834, 0x000088E6 }, /* GL_STATIC_COPY_ARB */
+   { 31853, 0x000088E4 }, /* GL_STATIC_DRAW */
+   { 31868, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */
+   { 31887, 0x000088E5 }, /* GL_STATIC_READ */
+   { 31902, 0x000088E5 }, /* GL_STATIC_READ_ARB */
+   { 31921, 0x00001802 }, /* GL_STENCIL */
+   { 31932, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT */
+   { 31954, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */
+   { 31980, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */
+   { 32001, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */
+   { 32026, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */
+   { 32047, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */
+   { 32072, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
+   { 32104, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */
+   { 32140, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
+   { 32172, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */
+   { 32208, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */
+   { 32228, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */
+   { 32255, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */
+   { 32281, 0x00000D57 }, /* GL_STENCIL_BITS */
+   { 32297, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */
+   { 32319, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */
+   { 32342, 0x00000B94 }, /* GL_STENCIL_FAIL */
+   { 32358, 0x00000B92 }, /* GL_STENCIL_FUNC */
+   { 32374, 0x00001901 }, /* GL_STENCIL_INDEX */
+   { 32391, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */
+   { 32414, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */
+   { 32436, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */
+   { 32458, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */
+   { 32480, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */
+   { 32501, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */
+   { 32528, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */
+   { 32555, 0x00000B97 }, /* GL_STENCIL_REF */
+   { 32570, 0x00000B90 }, /* GL_STENCIL_TEST */
+   { 32586, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
+   { 32615, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */
+   { 32637, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */
+   { 32658, 0x00000C33 }, /* GL_STEREO */
+   { 32668, 0x000085BE }, /* GL_STORAGE_CACHED_APPLE */
+   { 32692, 0x000085BD }, /* GL_STORAGE_PRIVATE_APPLE */
+   { 32717, 0x000085BF }, /* GL_STORAGE_SHARED_APPLE */
+   { 32741, 0x000088E2 }, /* GL_STREAM_COPY */
+   { 32756, 0x000088E2 }, /* GL_STREAM_COPY_ARB */
+   { 32775, 0x000088E0 }, /* GL_STREAM_DRAW */
+   { 32790, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */
+   { 32809, 0x000088E1 }, /* GL_STREAM_READ */
+   { 32824, 0x000088E1 }, /* GL_STREAM_READ_ARB */
+   { 32843, 0x00000D50 }, /* GL_SUBPIXEL_BITS */
+   { 32860, 0x000084E7 }, /* GL_SUBTRACT */
+   { 32872, 0x000084E7 }, /* GL_SUBTRACT_ARB */
+   { 32888, 0x00009113 }, /* GL_SYNC_CONDITION */
+   { 32906, 0x00009116 }, /* GL_SYNC_FENCE */
+   { 32920, 0x00009115 }, /* GL_SYNC_FLAGS */
+   { 32934, 0x00000001 }, /* GL_SYNC_FLUSH_COMMANDS_BIT */
+   { 32961, 0x00009117 }, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
+   { 32991, 0x00009114 }, /* GL_SYNC_STATUS */
+   { 33006, 0x00002001 }, /* GL_T */
+   { 33011, 0x00002A2A }, /* GL_T2F_C3F_V3F */
+   { 33026, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */
+   { 33045, 0x00002A29 }, /* GL_T2F_C4UB_V3F */
+   { 33061, 0x00002A2B }, /* GL_T2F_N3F_V3F */
+   { 33076, 0x00002A27 }, /* GL_T2F_V3F */
+   { 33087, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */
+   { 33106, 0x00002A28 }, /* GL_T4F_V4F */
+   { 33117, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */
+   { 33140, 0x00001702 }, /* GL_TEXTURE */
+   { 33151, 0x000084C0 }, /* GL_TEXTURE0 */
+   { 33163, 0x000084C0 }, /* GL_TEXTURE0_ARB */
+   { 33179, 0x000084C1 }, /* GL_TEXTURE1 */
+   { 33191, 0x000084CA }, /* GL_TEXTURE10 */
+   { 33204, 0x000084CA }, /* GL_TEXTURE10_ARB */
+   { 33221, 0x000084CB }, /* GL_TEXTURE11 */
+   { 33234, 0x000084CB }, /* GL_TEXTURE11_ARB */
+   { 33251, 0x000084CC }, /* GL_TEXTURE12 */
+   { 33264, 0x000084CC }, /* GL_TEXTURE12_ARB */
+   { 33281, 0x000084CD }, /* GL_TEXTURE13 */
+   { 33294, 0x000084CD }, /* GL_TEXTURE13_ARB */
+   { 33311, 0x000084CE }, /* GL_TEXTURE14 */
+   { 33324, 0x000084CE }, /* GL_TEXTURE14_ARB */
+   { 33341, 0x000084CF }, /* GL_TEXTURE15 */
+   { 33354, 0x000084CF }, /* GL_TEXTURE15_ARB */
+   { 33371, 0x000084D0 }, /* GL_TEXTURE16 */
+   { 33384, 0x000084D0 }, /* GL_TEXTURE16_ARB */
+   { 33401, 0x000084D1 }, /* GL_TEXTURE17 */
+   { 33414, 0x000084D1 }, /* GL_TEXTURE17_ARB */
+   { 33431, 0x000084D2 }, /* GL_TEXTURE18 */
+   { 33444, 0x000084D2 }, /* GL_TEXTURE18_ARB */
+   { 33461, 0x000084D3 }, /* GL_TEXTURE19 */
+   { 33474, 0x000084D3 }, /* GL_TEXTURE19_ARB */
+   { 33491, 0x000084C1 }, /* GL_TEXTURE1_ARB */
+   { 33507, 0x000084C2 }, /* GL_TEXTURE2 */
+   { 33519, 0x000084D4 }, /* GL_TEXTURE20 */
+   { 33532, 0x000084D4 }, /* GL_TEXTURE20_ARB */
+   { 33549, 0x000084D5 }, /* GL_TEXTURE21 */
+   { 33562, 0x000084D5 }, /* GL_TEXTURE21_ARB */
+   { 33579, 0x000084D6 }, /* GL_TEXTURE22 */
+   { 33592, 0x000084D6 }, /* GL_TEXTURE22_ARB */
+   { 33609, 0x000084D7 }, /* GL_TEXTURE23 */
+   { 33622, 0x000084D7 }, /* GL_TEXTURE23_ARB */
+   { 33639, 0x000084D8 }, /* GL_TEXTURE24 */
+   { 33652, 0x000084D8 }, /* GL_TEXTURE24_ARB */
+   { 33669, 0x000084D9 }, /* GL_TEXTURE25 */
+   { 33682, 0x000084D9 }, /* GL_TEXTURE25_ARB */
+   { 33699, 0x000084DA }, /* GL_TEXTURE26 */
+   { 33712, 0x000084DA }, /* GL_TEXTURE26_ARB */
+   { 33729, 0x000084DB }, /* GL_TEXTURE27 */
+   { 33742, 0x000084DB }, /* GL_TEXTURE27_ARB */
+   { 33759, 0x000084DC }, /* GL_TEXTURE28 */
+   { 33772, 0x000084DC }, /* GL_TEXTURE28_ARB */
+   { 33789, 0x000084DD }, /* GL_TEXTURE29 */
+   { 33802, 0x000084DD }, /* GL_TEXTURE29_ARB */
+   { 33819, 0x000084C2 }, /* GL_TEXTURE2_ARB */
+   { 33835, 0x000084C3 }, /* GL_TEXTURE3 */
+   { 33847, 0x000084DE }, /* GL_TEXTURE30 */
+   { 33860, 0x000084DE }, /* GL_TEXTURE30_ARB */
+   { 33877, 0x000084DF }, /* GL_TEXTURE31 */
+   { 33890, 0x000084DF }, /* GL_TEXTURE31_ARB */
+   { 33907, 0x000084C3 }, /* GL_TEXTURE3_ARB */
+   { 33923, 0x000084C4 }, /* GL_TEXTURE4 */
+   { 33935, 0x000084C4 }, /* GL_TEXTURE4_ARB */
+   { 33951, 0x000084C5 }, /* GL_TEXTURE5 */
+   { 33963, 0x000084C5 }, /* GL_TEXTURE5_ARB */
+   { 33979, 0x000084C6 }, /* GL_TEXTURE6 */
+   { 33991, 0x000084C6 }, /* GL_TEXTURE6_ARB */
+   { 34007, 0x000084C7 }, /* GL_TEXTURE7 */
+   { 34019, 0x000084C7 }, /* GL_TEXTURE7_ARB */
+   { 34035, 0x000084C8 }, /* GL_TEXTURE8 */
+   { 34047, 0x000084C8 }, /* GL_TEXTURE8_ARB */
+   { 34063, 0x000084C9 }, /* GL_TEXTURE9 */
+   { 34075, 0x000084C9 }, /* GL_TEXTURE9_ARB */
+   { 34091, 0x00000DE0 }, /* GL_TEXTURE_1D */
+   { 34105, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */
+   { 34129, 0x00000DE1 }, /* GL_TEXTURE_2D */
+   { 34143, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */
+   { 34167, 0x0000806F }, /* GL_TEXTURE_3D */
+   { 34181, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */
+   { 34203, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */
+   { 34229, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */
+   { 34251, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */
+   { 34273, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
+   { 34305, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */
+   { 34327, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
+   { 34359, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */
+   { 34381, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */
+   { 34409, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */
+   { 34441, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
+   { 34474, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */
+   { 34506, 0x00040000 }, /* GL_TEXTURE_BIT */
+   { 34521, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */
+   { 34542, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */
+   { 34567, 0x00001005 }, /* GL_TEXTURE_BORDER */
+   { 34585, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */
+   { 34609, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
+   { 34640, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
+   { 34670, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
+   { 34700, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
+   { 34735, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
+   { 34766, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+   { 34804, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */
+   { 34831, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
+   { 34863, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
+   { 34897, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */
+   { 34921, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */
+   { 34949, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */
+   { 34973, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */
+   { 35001, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
+   { 35034, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */
+   { 35058, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */
+   { 35080, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */
+   { 35102, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */
+   { 35128, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */
+   { 35162, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
+   { 35195, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */
+   { 35232, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */
+   { 35260, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */
+   { 35292, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */
+   { 35315, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
+   { 35353, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */
+   { 35395, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */
+   { 35426, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */
+   { 35454, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
+   { 35484, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */
+   { 35512, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */
+   { 35532, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */
+   { 35556, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
+   { 35587, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */
+   { 35622, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
+   { 35653, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */
+   { 35688, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
+   { 35719, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */
+   { 35754, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
+   { 35785, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */
+   { 35820, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
+   { 35851, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */
+   { 35886, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
+   { 35917, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */
+   { 35952, 0x000088F4 }, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
+   { 35981, 0x00008071 }, /* GL_TEXTURE_DEPTH */
+   { 35998, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */
+   { 36020, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */
+   { 36046, 0x00002300 }, /* GL_TEXTURE_ENV */
+   { 36061, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */
+   { 36082, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */
+   { 36102, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */
+   { 36128, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */
+   { 36148, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */
+   { 36165, 0x00000C62 }, /* GL_TEXTURE_GEN_R */
+   { 36182, 0x00000C60 }, /* GL_TEXTURE_GEN_S */
+   { 36199, 0x00000C61 }, /* GL_TEXTURE_GEN_T */
+   { 36216, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */
+   { 36241, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */
+   { 36263, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */
+   { 36289, 0x00001001 }, /* GL_TEXTURE_HEIGHT */
+   { 36307, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */
+   { 36333, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */
+   { 36359, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */
+   { 36389, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */
+   { 36416, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */
+   { 36441, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */
+   { 36461, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */
+   { 36485, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
+   { 36512, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
+   { 36539, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
+   { 36566, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */
+   { 36592, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */
+   { 36622, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */
+   { 36644, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */
+   { 36662, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
+   { 36692, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
+   { 36720, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
+   { 36748, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
+   { 36776, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */
+   { 36797, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */
+   { 36816, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */
+   { 36838, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */
+   { 36857, 0x00008066 }, /* GL_TEXTURE_PRIORITY */
+   { 36877, 0x000085B7 }, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
+   { 36907, 0x000085B8 }, /* GL_TEXTURE_RANGE_POINTER_APPLE */
+   { 36938, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */
+   { 36963, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */
+   { 36987, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */
+   { 37007, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */
+   { 37031, 0x00008067 }, /* GL_TEXTURE_RESIDENT */
+   { 37051, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */
+   { 37074, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE */
+   { 37098, 0x000085BC }, /* GL_TEXTURE_STORAGE_HINT_APPLE */
+   { 37128, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */
+   { 37153, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
+   { 37187, 0x00001000 }, /* GL_TEXTURE_WIDTH */
+   { 37204, 0x00008072 }, /* GL_TEXTURE_WRAP_R */
+   { 37222, 0x00002802 }, /* GL_TEXTURE_WRAP_S */
+   { 37240, 0x00002803 }, /* GL_TEXTURE_WRAP_T */
+   { 37258, 0x0000911B }, /* GL_TIMEOUT_EXPIRED */
+   { 37277, 0xFFFFFFFFFFFFFFFF }, /* GL_TIMEOUT_IGNORED */
+   { 37296, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */
+   { 37316, 0x00008648 }, /* GL_TRACK_MATRIX_NV */
+   { 37335, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */
+   { 37364, 0x00001000 }, /* GL_TRANSFORM_BIT */
+   { 37381, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */
+   { 37407, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */
+   { 37437, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
+   { 37469, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
+   { 37499, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */
+   { 37533, 0x0000862C }, /* GL_TRANSPOSE_NV */
+   { 37549, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */
+   { 37580, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */
+   { 37615, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */
+   { 37643, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */
+   { 37675, 0x00000004 }, /* GL_TRIANGLES */
+   { 37688, 0x00000006 }, /* GL_TRIANGLE_FAN */
+   { 37704, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */
+   { 37725, 0x00000005 }, /* GL_TRIANGLE_STRIP */
+   { 37743, 0x00000001 }, /* GL_TRUE */
+   { 37751, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */
+   { 37771, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */
+   { 37794, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */
+   { 37814, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */
+   { 37835, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */
+   { 37857, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */
+   { 37879, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */
+   { 37899, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */
+   { 37920, 0x00009118 }, /* GL_UNSIGNALED */
+   { 37934, 0x00001401 }, /* GL_UNSIGNED_BYTE */
+   { 37951, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */
+   { 37978, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */
+   { 38001, 0x00001405 }, /* GL_UNSIGNED_INT */
+   { 38017, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */
+   { 38044, 0x000084FA }, /* GL_UNSIGNED_INT_24_8 */
+   { 38065, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */
+   { 38089, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */
+   { 38120, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */
+   { 38144, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */
+   { 38172, 0x00008C17 }, /* GL_UNSIGNED_NORMALIZED */
+   { 38195, 0x00001403 }, /* GL_UNSIGNED_SHORT */
+   { 38213, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
+   { 38243, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */
+   { 38269, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
+   { 38299, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */
+   { 38325, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */
+   { 38349, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */
+   { 38377, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */
+   { 38405, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */
+   { 38432, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
+   { 38464, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */
+   { 38495, 0x00008CA2 }, /* GL_UPPER_LEFT */
+   { 38509, 0x00002A20 }, /* GL_V2F */
+   { 38516, 0x00002A21 }, /* GL_V3F */
+   { 38523, 0x00008B83 }, /* GL_VALIDATE_STATUS */
+   { 38542, 0x00001F00 }, /* GL_VENDOR */
+   { 38552, 0x00001F02 }, /* GL_VERSION */
+   { 38563, 0x00008074 }, /* GL_VERTEX_ARRAY */
+   { 38579, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING */
+   { 38603, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */
+   { 38633, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
+   { 38664, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */
+   { 38699, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */
+   { 38723, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */
+   { 38744, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */
+   { 38767, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */
+   { 38788, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
+   { 38815, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
+   { 38843, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
+   { 38871, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
+   { 38899, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
+   { 38927, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
+   { 38955, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
+   { 38983, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
+   { 39010, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
+   { 39037, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
+   { 39064, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
+   { 39091, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
+   { 39118, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
+   { 39145, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
+   { 39172, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
+   { 39199, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
+   { 39226, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
+   { 39264, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */
+   { 39306, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
+   { 39337, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */
+   { 39372, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
+   { 39406, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */
+   { 39444, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
+   { 39475, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */
+   { 39510, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
+   { 39538, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */
+   { 39570, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
+   { 39600, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */
+   { 39634, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
+   { 39662, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */
+   { 39694, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */
+   { 39714, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */
+   { 39736, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */
+   { 39765, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */
+   { 39786, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */
+   { 39815, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */
+   { 39848, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */
+   { 39880, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */
+   { 39907, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */
+   { 39938, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */
+   { 39968, 0x00008B31 }, /* GL_VERTEX_SHADER */
+   { 39985, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */
+   { 40006, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */
+   { 40033, 0x00000BA2 }, /* GL_VIEWPORT */
+   { 40045, 0x00000800 }, /* GL_VIEWPORT_BIT */
+   { 40061, 0x0000911D }, /* GL_WAIT_FAILED */
+   { 40076, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */
+   { 40096, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
+   { 40127, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */
+   { 40162, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */
+   { 40190, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */
+   { 40215, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
+   { 40242, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */
+   { 40267, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */
+   { 40291, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */
+   { 40310, 0x000088B9 }, /* GL_WRITE_ONLY */
+   { 40324, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */
+   { 40342, 0x00001506 }, /* GL_XOR */
+   { 40349, 0x000085B9 }, /* GL_YCBCR_422_APPLE */
+   { 40368, 0x00008757 }, /* GL_YCBCR_MESA */
+   { 40382, 0x00000000 }, /* GL_ZERO */
+   { 40390, 0x00000D16 }, /* GL_ZOOM_X */
+   { 40400, 0x00000D17 }, /* GL_ZOOM_Y */
 };
 
 static const unsigned reduced_enums[1347] =
 {
-       475, /* GL_FALSE */
-       691, /* GL_LINES */
-       693, /* GL_LINE_LOOP */
-       700, /* GL_LINE_STRIP */
-      1743, /* GL_TRIANGLES */
-      1746, /* GL_TRIANGLE_STRIP */
-      1744, /* GL_TRIANGLE_FAN */
-      1271, /* GL_QUADS */
-      1274, /* GL_QUAD_STRIP */
-      1158, /* GL_POLYGON */
-      1170, /* GL_POLYGON_STIPPLE_BIT */
-      1119, /* GL_PIXEL_MODE_BIT */
-       678, /* GL_LIGHTING_BIT */
-       504, /* GL_FOG_BIT */
+       476, /* GL_FALSE */
+       694, /* GL_LINES */
+       696, /* GL_LINE_LOOP */
+       703, /* GL_LINE_STRIP */
+      1748, /* GL_TRIANGLES */
+      1751, /* GL_TRIANGLE_STRIP */
+      1749, /* GL_TRIANGLE_FAN */
+      1275, /* GL_QUADS */
+      1279, /* GL_QUAD_STRIP */
+      1161, /* GL_POLYGON */
+      1173, /* GL_POLYGON_STIPPLE_BIT */
+      1122, /* GL_PIXEL_MODE_BIT */
+       681, /* GL_LIGHTING_BIT */
+       506, /* GL_FOG_BIT */
          8, /* GL_ACCUM */
-       710, /* GL_LOAD */
-      1326, /* GL_RETURN */
-       991, /* GL_MULT */
+       713, /* GL_LOAD */
+      1331, /* GL_RETURN */
+       994, /* GL_MULT */
         23, /* GL_ADD */
-      1007, /* GL_NEVER */
-       668, /* GL_LESS */
-       465, /* GL_EQUAL */
-       667, /* GL_LEQUAL */
-       590, /* GL_GREATER */
-      1022, /* GL_NOTEQUAL */
-       589, /* GL_GEQUAL */
+      1010, /* GL_NEVER */
+       671, /* GL_LESS */
+       466, /* GL_EQUAL */
+       670, /* GL_LEQUAL */
+       592, /* GL_GREATER */
+      1025, /* GL_NOTEQUAL */
+       591, /* GL_GEQUAL */
         47, /* GL_ALWAYS */
-      1467, /* GL_SRC_COLOR */
-      1052, /* GL_ONE_MINUS_SRC_COLOR */
-      1465, /* GL_SRC_ALPHA */
-      1051, /* GL_ONE_MINUS_SRC_ALPHA */
-       444, /* GL_DST_ALPHA */
-      1049, /* GL_ONE_MINUS_DST_ALPHA */
-       445, /* GL_DST_COLOR */
-      1050, /* GL_ONE_MINUS_DST_COLOR */
-      1466, /* GL_SRC_ALPHA_SATURATE */
-       577, /* GL_FRONT_LEFT */
-       578, /* GL_FRONT_RIGHT */
+      1472, /* GL_SRC_COLOR */
+      1055, /* GL_ONE_MINUS_SRC_COLOR */
+      1470, /* GL_SRC_ALPHA */
+      1054, /* GL_ONE_MINUS_SRC_ALPHA */
+       445, /* GL_DST_ALPHA */
+      1052, /* GL_ONE_MINUS_DST_ALPHA */
+       446, /* GL_DST_COLOR */
+      1053, /* GL_ONE_MINUS_DST_COLOR */
+      1471, /* GL_SRC_ALPHA_SATURATE */
+       579, /* GL_FRONT_LEFT */
+       580, /* GL_FRONT_RIGHT */
         69, /* GL_BACK_LEFT */
         70, /* GL_BACK_RIGHT */
-       574, /* GL_FRONT */
+       576, /* GL_FRONT */
         68, /* GL_BACK */
-       666, /* GL_LEFT */
-      1368, /* GL_RIGHT */
-       575, /* GL_FRONT_AND_BACK */
+       669, /* GL_LEFT */
+      1373, /* GL_RIGHT */
+       577, /* GL_FRONT_AND_BACK */
         63, /* GL_AUX0 */
         64, /* GL_AUX1 */
         65, /* GL_AUX2 */
         66, /* GL_AUX3 */
-       656, /* GL_INVALID_ENUM */
-       660, /* GL_INVALID_VALUE */
-       659, /* GL_INVALID_OPERATION */
-      1472, /* GL_STACK_OVERFLOW */
-      1473, /* GL_STACK_UNDERFLOW */
-      1077, /* GL_OUT_OF_MEMORY */
-       657, /* GL_INVALID_FRAMEBUFFER_OPERATION */
+       658, /* GL_INVALID_ENUM */
+       662, /* GL_INVALID_VALUE */
+       661, /* GL_INVALID_OPERATION */
+      1477, /* GL_STACK_OVERFLOW */
+      1478, /* GL_STACK_UNDERFLOW */
+      1080, /* GL_OUT_OF_MEMORY */
+       659, /* GL_INVALID_FRAMEBUFFER_OPERATION */
          0, /* GL_2D */
          2, /* GL_3D */
          3, /* GL_3D_COLOR */
          4, /* GL_3D_COLOR_TEXTURE */
          6, /* GL_4D_COLOR_TEXTURE */
-      1097, /* GL_PASS_THROUGH_TOKEN */
-      1157, /* GL_POINT_TOKEN */
-       701, /* GL_LINE_TOKEN */
-      1171, /* GL_POLYGON_TOKEN */
+      1100, /* GL_PASS_THROUGH_TOKEN */
+      1160, /* GL_POINT_TOKEN */
+       704, /* GL_LINE_TOKEN */
+      1174, /* GL_POLYGON_TOKEN */
         74, /* GL_BITMAP_TOKEN */
-       443, /* GL_DRAW_PIXEL_TOKEN */
+       444, /* GL_DRAW_PIXEL_TOKEN */
        301, /* GL_COPY_PIXEL_TOKEN */
-       694, /* GL_LINE_RESET_TOKEN */
-       468, /* GL_EXP */
-       469, /* GL_EXP2 */
+       697, /* GL_LINE_RESET_TOKEN */
+       469, /* GL_EXP */
+       470, /* GL_EXP2 */
        337, /* GL_CW */
        125, /* GL_CCW */
        146, /* GL_COEFF */
-      1074, /* GL_ORDER */
-       381, /* GL_DOMAIN */
+      1077, /* GL_ORDER */
+       382, /* GL_DOMAIN */
        311, /* GL_CURRENT_COLOR */
        314, /* GL_CURRENT_INDEX */
        320, /* GL_CURRENT_NORMAL */
@@ -3844,519 +3854,519 @@ static const unsigned reduced_enums[1347] =
        328, /* GL_CURRENT_RASTER_POSITION */
        329, /* GL_CURRENT_RASTER_POSITION_VALID */
        326, /* GL_CURRENT_RASTER_DISTANCE */
-      1150, /* GL_POINT_SMOOTH */
-      1139, /* GL_POINT_SIZE */
-      1149, /* GL_POINT_SIZE_RANGE */
-      1140, /* GL_POINT_SIZE_GRANULARITY */
-       695, /* GL_LINE_SMOOTH */
-       702, /* GL_LINE_WIDTH */
-       704, /* GL_LINE_WIDTH_RANGE */
-       703, /* GL_LINE_WIDTH_GRANULARITY */
-       697, /* GL_LINE_STIPPLE */
-       698, /* GL_LINE_STIPPLE_PATTERN */
-       699, /* GL_LINE_STIPPLE_REPEAT */
-       709, /* GL_LIST_MODE */
-       874, /* GL_MAX_LIST_NESTING */
-       706, /* GL_LIST_BASE */
-       708, /* GL_LIST_INDEX */
-      1160, /* GL_POLYGON_MODE */
-      1167, /* GL_POLYGON_SMOOTH */
-      1169, /* GL_POLYGON_STIPPLE */
-       454, /* GL_EDGE_FLAG */
+      1153, /* GL_POINT_SMOOTH */
+      1142, /* GL_POINT_SIZE */
+      1152, /* GL_POINT_SIZE_RANGE */
+      1143, /* GL_POINT_SIZE_GRANULARITY */
+       698, /* GL_LINE_SMOOTH */
+       705, /* GL_LINE_WIDTH */
+       707, /* GL_LINE_WIDTH_RANGE */
+       706, /* GL_LINE_WIDTH_GRANULARITY */
+       700, /* GL_LINE_STIPPLE */
+       701, /* GL_LINE_STIPPLE_PATTERN */
+       702, /* GL_LINE_STIPPLE_REPEAT */
+       712, /* GL_LIST_MODE */
+       877, /* GL_MAX_LIST_NESTING */
+       709, /* GL_LIST_BASE */
+       711, /* GL_LIST_INDEX */
+      1163, /* GL_POLYGON_MODE */
+      1170, /* GL_POLYGON_SMOOTH */
+      1172, /* GL_POLYGON_STIPPLE */
+       455, /* GL_EDGE_FLAG */
        304, /* GL_CULL_FACE */
        305, /* GL_CULL_FACE_MODE */
-       576, /* GL_FRONT_FACE */
-       677, /* GL_LIGHTING */
-       682, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
-       683, /* GL_LIGHT_MODEL_TWO_SIDE */
-       679, /* GL_LIGHT_MODEL_AMBIENT */
-      1414, /* GL_SHADE_MODEL */
+       578, /* GL_FRONT_FACE */
+       680, /* GL_LIGHTING */
+       685, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
+       686, /* GL_LIGHT_MODEL_TWO_SIDE */
+       682, /* GL_LIGHT_MODEL_AMBIENT */
+      1419, /* GL_SHADE_MODEL */
        193, /* GL_COLOR_MATERIAL_FACE */
        194, /* GL_COLOR_MATERIAL_PARAMETER */
        192, /* GL_COLOR_MATERIAL */
-       503, /* GL_FOG */
-       525, /* GL_FOG_INDEX */
-       521, /* GL_FOG_DENSITY */
-       529, /* GL_FOG_START */
-       523, /* GL_FOG_END */
-       526, /* GL_FOG_MODE */
-       505, /* GL_FOG_COLOR */
-       368, /* GL_DEPTH_RANGE */
-       375, /* GL_DEPTH_TEST */
-       378, /* GL_DEPTH_WRITEMASK */
-       356, /* GL_DEPTH_CLEAR_VALUE */
-       367, /* GL_DEPTH_FUNC */
+       505, /* GL_FOG */
+       527, /* GL_FOG_INDEX */
+       523, /* GL_FOG_DENSITY */
+       531, /* GL_FOG_START */
+       525, /* GL_FOG_END */
+       528, /* GL_FOG_MODE */
+       507, /* GL_FOG_COLOR */
+       369, /* GL_DEPTH_RANGE */
+       376, /* GL_DEPTH_TEST */
+       379, /* GL_DEPTH_WRITEMASK */
+       357, /* GL_DEPTH_CLEAR_VALUE */
+       368, /* GL_DEPTH_FUNC */
         12, /* GL_ACCUM_CLEAR_VALUE */
-      1508, /* GL_STENCIL_TEST */
-      1496, /* GL_STENCIL_CLEAR_VALUE */
-      1498, /* GL_STENCIL_FUNC */
-      1510, /* GL_STENCIL_VALUE_MASK */
-      1497, /* GL_STENCIL_FAIL */
-      1505, /* GL_STENCIL_PASS_DEPTH_FAIL */
-      1506, /* GL_STENCIL_PASS_DEPTH_PASS */
-      1507, /* GL_STENCIL_REF */
-      1511, /* GL_STENCIL_WRITEMASK */
-       843, /* GL_MATRIX_MODE */
-      1012, /* GL_NORMALIZE */
-      1837, /* GL_VIEWPORT */
-       986, /* GL_MODELVIEW_STACK_DEPTH */
-      1250, /* GL_PROJECTION_STACK_DEPTH */
-      1718, /* GL_TEXTURE_STACK_DEPTH */
-       984, /* GL_MODELVIEW_MATRIX */
-      1249, /* GL_PROJECTION_MATRIX */
-      1701, /* GL_TEXTURE_MATRIX */
+      1513, /* GL_STENCIL_TEST */
+      1501, /* GL_STENCIL_CLEAR_VALUE */
+      1503, /* GL_STENCIL_FUNC */
+      1515, /* GL_STENCIL_VALUE_MASK */
+      1502, /* GL_STENCIL_FAIL */
+      1510, /* GL_STENCIL_PASS_DEPTH_FAIL */
+      1511, /* GL_STENCIL_PASS_DEPTH_PASS */
+      1512, /* GL_STENCIL_REF */
+      1516, /* GL_STENCIL_WRITEMASK */
+       846, /* GL_MATRIX_MODE */
+      1015, /* GL_NORMALIZE */
+      1842, /* GL_VIEWPORT */
+       989, /* GL_MODELVIEW_STACK_DEPTH */
+      1253, /* GL_PROJECTION_STACK_DEPTH */
+      1723, /* GL_TEXTURE_STACK_DEPTH */
+       987, /* GL_MODELVIEW_MATRIX */
+      1252, /* GL_PROJECTION_MATRIX */
+      1706, /* GL_TEXTURE_MATRIX */
         61, /* GL_ATTRIB_STACK_DEPTH */
        136, /* GL_CLIENT_ATTRIB_STACK_DEPTH */
         43, /* GL_ALPHA_TEST */
         44, /* GL_ALPHA_TEST_FUNC */
         45, /* GL_ALPHA_TEST_REF */
-       380, /* GL_DITHER */
+       381, /* GL_DITHER */
         78, /* GL_BLEND_DST */
         87, /* GL_BLEND_SRC */
         75, /* GL_BLEND */
-       712, /* GL_LOGIC_OP_MODE */
-       630, /* GL_INDEX_LOGIC_OP */
+       715, /* GL_LOGIC_OP_MODE */
+       632, /* GL_INDEX_LOGIC_OP */
        191, /* GL_COLOR_LOGIC_OP */
         67, /* GL_AUX_BUFFERS */
-       391, /* GL_DRAW_BUFFER */
-      1284, /* GL_READ_BUFFER */
-      1395, /* GL_SCISSOR_BOX */
-      1396, /* GL_SCISSOR_TEST */
-       629, /* GL_INDEX_CLEAR_VALUE */
-       634, /* GL_INDEX_WRITEMASK */
+       392, /* GL_DRAW_BUFFER */
+      1289, /* GL_READ_BUFFER */
+      1400, /* GL_SCISSOR_BOX */
+      1401, /* GL_SCISSOR_TEST */
+       631, /* GL_INDEX_CLEAR_VALUE */
+       636, /* GL_INDEX_WRITEMASK */
        188, /* GL_COLOR_CLEAR_VALUE */
        230, /* GL_COLOR_WRITEMASK */
-       631, /* GL_INDEX_MODE */
-      1361, /* GL_RGBA_MODE */
-       390, /* GL_DOUBLEBUFFER */
-      1512, /* GL_STEREO */
-      1319, /* GL_RENDER_MODE */
-      1098, /* GL_PERSPECTIVE_CORRECTION_HINT */
-      1151, /* GL_POINT_SMOOTH_HINT */
-       696, /* GL_LINE_SMOOTH_HINT */
-      1168, /* GL_POLYGON_SMOOTH_HINT */
-       524, /* GL_FOG_HINT */
-      1682, /* GL_TEXTURE_GEN_S */
-      1683, /* GL_TEXTURE_GEN_T */
-      1681, /* GL_TEXTURE_GEN_R */
-      1680, /* GL_TEXTURE_GEN_Q */
-      1111, /* GL_PIXEL_MAP_I_TO_I */
-      1117, /* GL_PIXEL_MAP_S_TO_S */
-      1113, /* GL_PIXEL_MAP_I_TO_R */
-      1109, /* GL_PIXEL_MAP_I_TO_G */
-      1107, /* GL_PIXEL_MAP_I_TO_B */
-      1105, /* GL_PIXEL_MAP_I_TO_A */
-      1115, /* GL_PIXEL_MAP_R_TO_R */
-      1103, /* GL_PIXEL_MAP_G_TO_G */
-      1101, /* GL_PIXEL_MAP_B_TO_B */
-      1099, /* GL_PIXEL_MAP_A_TO_A */
-      1112, /* GL_PIXEL_MAP_I_TO_I_SIZE */
-      1118, /* GL_PIXEL_MAP_S_TO_S_SIZE */
-      1114, /* GL_PIXEL_MAP_I_TO_R_SIZE */
-      1110, /* GL_PIXEL_MAP_I_TO_G_SIZE */
-      1108, /* GL_PIXEL_MAP_I_TO_B_SIZE */
-      1106, /* GL_PIXEL_MAP_I_TO_A_SIZE */
-      1116, /* GL_PIXEL_MAP_R_TO_R_SIZE */
-      1104, /* GL_PIXEL_MAP_G_TO_G_SIZE */
-      1102, /* GL_PIXEL_MAP_B_TO_B_SIZE */
-      1100, /* GL_PIXEL_MAP_A_TO_A_SIZE */
-      1755, /* GL_UNPACK_SWAP_BYTES */
-      1750, /* GL_UNPACK_LSB_FIRST */
-      1751, /* GL_UNPACK_ROW_LENGTH */
-      1754, /* GL_UNPACK_SKIP_ROWS */
-      1753, /* GL_UNPACK_SKIP_PIXELS */
-      1748, /* GL_UNPACK_ALIGNMENT */
-      1086, /* GL_PACK_SWAP_BYTES */
-      1081, /* GL_PACK_LSB_FIRST */
-      1082, /* GL_PACK_ROW_LENGTH */
-      1085, /* GL_PACK_SKIP_ROWS */
-      1084, /* GL_PACK_SKIP_PIXELS */
-      1078, /* GL_PACK_ALIGNMENT */
-       790, /* GL_MAP_COLOR */
-       795, /* GL_MAP_STENCIL */
-       633, /* GL_INDEX_SHIFT */
-       632, /* GL_INDEX_OFFSET */
-      1297, /* GL_RED_SCALE */
-      1295, /* GL_RED_BIAS */
-      1855, /* GL_ZOOM_X */
-      1856, /* GL_ZOOM_Y */
-       594, /* GL_GREEN_SCALE */
-       592, /* GL_GREEN_BIAS */
+       633, /* GL_INDEX_MODE */
+      1366, /* GL_RGBA_MODE */
+       391, /* GL_DOUBLEBUFFER */
+      1517, /* GL_STEREO */
+      1324, /* GL_RENDER_MODE */
+      1101, /* GL_PERSPECTIVE_CORRECTION_HINT */
+      1154, /* GL_POINT_SMOOTH_HINT */
+       699, /* GL_LINE_SMOOTH_HINT */
+      1171, /* GL_POLYGON_SMOOTH_HINT */
+       526, /* GL_FOG_HINT */
+      1687, /* GL_TEXTURE_GEN_S */
+      1688, /* GL_TEXTURE_GEN_T */
+      1686, /* GL_TEXTURE_GEN_R */
+      1685, /* GL_TEXTURE_GEN_Q */
+      1114, /* GL_PIXEL_MAP_I_TO_I */
+      1120, /* GL_PIXEL_MAP_S_TO_S */
+      1116, /* GL_PIXEL_MAP_I_TO_R */
+      1112, /* GL_PIXEL_MAP_I_TO_G */
+      1110, /* GL_PIXEL_MAP_I_TO_B */
+      1108, /* GL_PIXEL_MAP_I_TO_A */
+      1118, /* GL_PIXEL_MAP_R_TO_R */
+      1106, /* GL_PIXEL_MAP_G_TO_G */
+      1104, /* GL_PIXEL_MAP_B_TO_B */
+      1102, /* GL_PIXEL_MAP_A_TO_A */
+      1115, /* GL_PIXEL_MAP_I_TO_I_SIZE */
+      1121, /* GL_PIXEL_MAP_S_TO_S_SIZE */
+      1117, /* GL_PIXEL_MAP_I_TO_R_SIZE */
+      1113, /* GL_PIXEL_MAP_I_TO_G_SIZE */
+      1111, /* GL_PIXEL_MAP_I_TO_B_SIZE */
+      1109, /* GL_PIXEL_MAP_I_TO_A_SIZE */
+      1119, /* GL_PIXEL_MAP_R_TO_R_SIZE */
+      1107, /* GL_PIXEL_MAP_G_TO_G_SIZE */
+      1105, /* GL_PIXEL_MAP_B_TO_B_SIZE */
+      1103, /* GL_PIXEL_MAP_A_TO_A_SIZE */
+      1760, /* GL_UNPACK_SWAP_BYTES */
+      1755, /* GL_UNPACK_LSB_FIRST */
+      1756, /* GL_UNPACK_ROW_LENGTH */
+      1759, /* GL_UNPACK_SKIP_ROWS */
+      1758, /* GL_UNPACK_SKIP_PIXELS */
+      1753, /* GL_UNPACK_ALIGNMENT */
+      1089, /* GL_PACK_SWAP_BYTES */
+      1084, /* GL_PACK_LSB_FIRST */
+      1085, /* GL_PACK_ROW_LENGTH */
+      1088, /* GL_PACK_SKIP_ROWS */
+      1087, /* GL_PACK_SKIP_PIXELS */
+      1081, /* GL_PACK_ALIGNMENT */
+       793, /* GL_MAP_COLOR */
+       798, /* GL_MAP_STENCIL */
+       635, /* GL_INDEX_SHIFT */
+       634, /* GL_INDEX_OFFSET */
+      1302, /* GL_RED_SCALE */
+      1300, /* GL_RED_BIAS */
+      1860, /* GL_ZOOM_X */
+      1861, /* GL_ZOOM_Y */
+       596, /* GL_GREEN_SCALE */
+       594, /* GL_GREEN_BIAS */
         93, /* GL_BLUE_SCALE */
         91, /* GL_BLUE_BIAS */
         42, /* GL_ALPHA_SCALE */
         40, /* GL_ALPHA_BIAS */
-       369, /* GL_DEPTH_SCALE */
+       370, /* GL_DEPTH_SCALE */
        350, /* GL_DEPTH_BIAS */
-       869, /* GL_MAX_EVAL_ORDER */
-       873, /* GL_MAX_LIGHTS */
-       852, /* GL_MAX_CLIP_PLANES */
-       919, /* GL_MAX_TEXTURE_SIZE */
-       879, /* GL_MAX_PIXEL_MAP_TABLE */
-       848, /* GL_MAX_ATTRIB_STACK_DEPTH */
-       876, /* GL_MAX_MODELVIEW_STACK_DEPTH */
-       877, /* GL_MAX_NAME_STACK_DEPTH */
-       905, /* GL_MAX_PROJECTION_STACK_DEPTH */
-       920, /* GL_MAX_TEXTURE_STACK_DEPTH */
-       934, /* GL_MAX_VIEWPORT_DIMS */
-       849, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
-      1522, /* GL_SUBPIXEL_BITS */
-       628, /* GL_INDEX_BITS */
-      1296, /* GL_RED_BITS */
-       593, /* GL_GREEN_BITS */
+       872, /* GL_MAX_EVAL_ORDER */
+       876, /* GL_MAX_LIGHTS */
+       855, /* GL_MAX_CLIP_PLANES */
+       922, /* GL_MAX_TEXTURE_SIZE */
+       882, /* GL_MAX_PIXEL_MAP_TABLE */
+       851, /* GL_MAX_ATTRIB_STACK_DEPTH */
+       879, /* GL_MAX_MODELVIEW_STACK_DEPTH */
+       880, /* GL_MAX_NAME_STACK_DEPTH */
+       908, /* GL_MAX_PROJECTION_STACK_DEPTH */
+       923, /* GL_MAX_TEXTURE_STACK_DEPTH */
+       937, /* GL_MAX_VIEWPORT_DIMS */
+       852, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
+      1527, /* GL_SUBPIXEL_BITS */
+       630, /* GL_INDEX_BITS */
+      1301, /* GL_RED_BITS */
+       595, /* GL_GREEN_BITS */
         92, /* GL_BLUE_BITS */
         41, /* GL_ALPHA_BITS */
        351, /* GL_DEPTH_BITS */
-      1494, /* GL_STENCIL_BITS */
+      1499, /* GL_STENCIL_BITS */
         14, /* GL_ACCUM_RED_BITS */
         13, /* GL_ACCUM_GREEN_BITS */
         10, /* GL_ACCUM_BLUE_BITS */
          9, /* GL_ACCUM_ALPHA_BITS */
-      1000, /* GL_NAME_STACK_DEPTH */
+      1003, /* GL_NAME_STACK_DEPTH */
         62, /* GL_AUTO_NORMAL */
-       736, /* GL_MAP1_COLOR_4 */
-       739, /* GL_MAP1_INDEX */
-       740, /* GL_MAP1_NORMAL */
-       741, /* GL_MAP1_TEXTURE_COORD_1 */
-       742, /* GL_MAP1_TEXTURE_COORD_2 */
-       743, /* GL_MAP1_TEXTURE_COORD_3 */
-       744, /* GL_MAP1_TEXTURE_COORD_4 */
-       745, /* GL_MAP1_VERTEX_3 */
-       746, /* GL_MAP1_VERTEX_4 */
-       763, /* GL_MAP2_COLOR_4 */
-       766, /* GL_MAP2_INDEX */
-       767, /* GL_MAP2_NORMAL */
-       768, /* GL_MAP2_TEXTURE_COORD_1 */
-       769, /* GL_MAP2_TEXTURE_COORD_2 */
-       770, /* GL_MAP2_TEXTURE_COORD_3 */
-       771, /* GL_MAP2_TEXTURE_COORD_4 */
-       772, /* GL_MAP2_VERTEX_3 */
-       773, /* GL_MAP2_VERTEX_4 */
-       737, /* GL_MAP1_GRID_DOMAIN */
-       738, /* GL_MAP1_GRID_SEGMENTS */
-       764, /* GL_MAP2_GRID_DOMAIN */
-       765, /* GL_MAP2_GRID_SEGMENTS */
-      1605, /* GL_TEXTURE_1D */
-      1607, /* GL_TEXTURE_2D */
-       478, /* GL_FEEDBACK_BUFFER_POINTER */
-       479, /* GL_FEEDBACK_BUFFER_SIZE */
-       480, /* GL_FEEDBACK_BUFFER_TYPE */
-      1405, /* GL_SELECTION_BUFFER_POINTER */
-      1406, /* GL_SELECTION_BUFFER_SIZE */
-      1723, /* GL_TEXTURE_WIDTH */
-      1687, /* GL_TEXTURE_HEIGHT */
-      1642, /* GL_TEXTURE_COMPONENTS */
-      1626, /* GL_TEXTURE_BORDER_COLOR */
-      1625, /* GL_TEXTURE_BORDER */
-       382, /* GL_DONT_CARE */
-       476, /* GL_FASTEST */
-      1008, /* GL_NICEST */
+       739, /* GL_MAP1_COLOR_4 */
+       742, /* GL_MAP1_INDEX */
+       743, /* GL_MAP1_NORMAL */
+       744, /* GL_MAP1_TEXTURE_COORD_1 */
+       745, /* GL_MAP1_TEXTURE_COORD_2 */
+       746, /* GL_MAP1_TEXTURE_COORD_3 */
+       747, /* GL_MAP1_TEXTURE_COORD_4 */
+       748, /* GL_MAP1_VERTEX_3 */
+       749, /* GL_MAP1_VERTEX_4 */
+       766, /* GL_MAP2_COLOR_4 */
+       769, /* GL_MAP2_INDEX */
+       770, /* GL_MAP2_NORMAL */
+       771, /* GL_MAP2_TEXTURE_COORD_1 */
+       772, /* GL_MAP2_TEXTURE_COORD_2 */
+       773, /* GL_MAP2_TEXTURE_COORD_3 */
+       774, /* GL_MAP2_TEXTURE_COORD_4 */
+       775, /* GL_MAP2_VERTEX_3 */
+       776, /* GL_MAP2_VERTEX_4 */
+       740, /* GL_MAP1_GRID_DOMAIN */
+       741, /* GL_MAP1_GRID_SEGMENTS */
+       767, /* GL_MAP2_GRID_DOMAIN */
+       768, /* GL_MAP2_GRID_SEGMENTS */
+      1610, /* GL_TEXTURE_1D */
+      1612, /* GL_TEXTURE_2D */
+       479, /* GL_FEEDBACK_BUFFER_POINTER */
+       480, /* GL_FEEDBACK_BUFFER_SIZE */
+       481, /* GL_FEEDBACK_BUFFER_TYPE */
+      1410, /* GL_SELECTION_BUFFER_POINTER */
+      1411, /* GL_SELECTION_BUFFER_SIZE */
+      1728, /* GL_TEXTURE_WIDTH */
+      1692, /* GL_TEXTURE_HEIGHT */
+      1647, /* GL_TEXTURE_COMPONENTS */
+      1631, /* GL_TEXTURE_BORDER_COLOR */
+      1630, /* GL_TEXTURE_BORDER */
+       383, /* GL_DONT_CARE */
+       477, /* GL_FASTEST */
+      1011, /* GL_NICEST */
         48, /* GL_AMBIENT */
-       379, /* GL_DIFFUSE */
-      1454, /* GL_SPECULAR */
-      1172, /* GL_POSITION */
-      1457, /* GL_SPOT_DIRECTION */
-      1458, /* GL_SPOT_EXPONENT */
-      1456, /* GL_SPOT_CUTOFF */
+       380, /* GL_DIFFUSE */
+      1459, /* GL_SPECULAR */
+      1175, /* GL_POSITION */
+      1462, /* GL_SPOT_DIRECTION */
+      1463, /* GL_SPOT_EXPONENT */
+      1461, /* GL_SPOT_CUTOFF */
        275, /* GL_CONSTANT_ATTENUATION */
-       686, /* GL_LINEAR_ATTENUATION */
-      1270, /* GL_QUADRATIC_ATTENUATION */
+       689, /* GL_LINEAR_ATTENUATION */
+      1274, /* GL_QUADRATIC_ATTENUATION */
        244, /* GL_COMPILE */
        245, /* GL_COMPILE_AND_EXECUTE */
        120, /* GL_BYTE */
-      1757, /* GL_UNSIGNED_BYTE */
-      1419, /* GL_SHORT */
-      1768, /* GL_UNSIGNED_SHORT */
-       636, /* GL_INT */
-      1760, /* GL_UNSIGNED_INT */
-       484, /* GL_FLOAT */
+      1762, /* GL_UNSIGNED_BYTE */
+      1424, /* GL_SHORT */
+      1773, /* GL_UNSIGNED_SHORT */
+       638, /* GL_INT */
+      1765, /* GL_UNSIGNED_INT */
+       486, /* GL_FLOAT */
          1, /* GL_2_BYTES */
          5, /* GL_3_BYTES */
          7, /* GL_4_BYTES */
-       389, /* GL_DOUBLE */
+       390, /* GL_DOUBLE */
        132, /* GL_CLEAR */
         50, /* GL_AND */
         52, /* GL_AND_REVERSE */
        299, /* GL_COPY */
         51, /* GL_AND_INVERTED */
-      1010, /* GL_NOOP */
-      1851, /* GL_XOR */
-      1073, /* GL_OR */
-      1011, /* GL_NOR */
-       466, /* GL_EQUIV */
-       663, /* GL_INVERT */
-      1076, /* GL_OR_REVERSE */
+      1013, /* GL_NOOP */
+      1856, /* GL_XOR */
+      1076, /* GL_OR */
+      1014, /* GL_NOR */
+       467, /* GL_EQUIV */
+       665, /* GL_INVERT */
+      1079, /* GL_OR_REVERSE */
        300, /* GL_COPY_INVERTED */
-      1075, /* GL_OR_INVERTED */
-      1001, /* GL_NAND */
-      1410, /* GL_SET */
-       463, /* GL_EMISSION */
-      1418, /* GL_SHININESS */
+      1078, /* GL_OR_INVERTED */
+      1004, /* GL_NAND */
+      1415, /* GL_SET */
+       464, /* GL_EMISSION */
+      1423, /* GL_SHININESS */
         49, /* GL_AMBIENT_AND_DIFFUSE */
        190, /* GL_COLOR_INDEXES */
-       951, /* GL_MODELVIEW */
-      1248, /* GL_PROJECTION */
-      1540, /* GL_TEXTURE */
+       954, /* GL_MODELVIEW */
+      1251, /* GL_PROJECTION */
+      1545, /* GL_TEXTURE */
        147, /* GL_COLOR */
        346, /* GL_DEPTH */
-      1480, /* GL_STENCIL */
+      1485, /* GL_STENCIL */
        189, /* GL_COLOR_INDEX */
-      1499, /* GL_STENCIL_INDEX */
-       357, /* GL_DEPTH_COMPONENT */
-      1292, /* GL_RED */
-       591, /* GL_GREEN */
+      1504, /* GL_STENCIL_INDEX */
+       358, /* GL_DEPTH_COMPONENT */
+      1297, /* GL_RED */
+       593, /* GL_GREEN */
         90, /* GL_BLUE */
         31, /* GL_ALPHA */
-      1327, /* GL_RGB */
-      1346, /* GL_RGBA */
-       714, /* GL_LUMINANCE */
-       735, /* GL_LUMINANCE_ALPHA */
+      1332, /* GL_RGB */
+      1351, /* GL_RGBA */
+       717, /* GL_LUMINANCE */
+       738, /* GL_LUMINANCE_ALPHA */
         73, /* GL_BITMAP */
-      1128, /* GL_POINT */
-       684, /* GL_LINE */
-       481, /* GL_FILL */
-      1301, /* GL_RENDER */
-       477, /* GL_FEEDBACK */
-      1404, /* GL_SELECT */
-       483, /* GL_FLAT */
-      1429, /* GL_SMOOTH */
-       664, /* GL_KEEP */
-      1321, /* GL_REPLACE */
-       618, /* GL_INCR */
+      1131, /* GL_POINT */
+       687, /* GL_LINE */
+       482, /* GL_FILL */
+      1306, /* GL_RENDER */
+       478, /* GL_FEEDBACK */
+      1409, /* GL_SELECT */
+       485, /* GL_FLAT */
+      1434, /* GL_SMOOTH */
+       666, /* GL_KEEP */
+      1326, /* GL_REPLACE */
+       620, /* GL_INCR */
        342, /* GL_DECR */
-      1783, /* GL_VENDOR */
-      1318, /* GL_RENDERER */
-      1784, /* GL_VERSION */
-       470, /* GL_EXTENSIONS */
-      1369, /* GL_S */
-      1531, /* GL_T */
-      1281, /* GL_R */
-      1269, /* GL_Q */
-       987, /* GL_MODULATE */
+      1788, /* GL_VENDOR */
+      1323, /* GL_RENDERER */
+      1789, /* GL_VERSION */
+       471, /* GL_EXTENSIONS */
+      1374, /* GL_S */
+      1536, /* GL_T */
+      1286, /* GL_R */
+      1273, /* GL_Q */
+       990, /* GL_MODULATE */
        341, /* GL_DECAL */
-      1677, /* GL_TEXTURE_ENV_MODE */
-      1676, /* GL_TEXTURE_ENV_COLOR */
-      1675, /* GL_TEXTURE_ENV */
-       471, /* GL_EYE_LINEAR */
-      1034, /* GL_OBJECT_LINEAR */
-      1455, /* GL_SPHERE_MAP */
-      1679, /* GL_TEXTURE_GEN_MODE */
-      1036, /* GL_OBJECT_PLANE */
-       472, /* GL_EYE_PLANE */
-      1002, /* GL_NEAREST */
-       685, /* GL_LINEAR */
-      1006, /* GL_NEAREST_MIPMAP_NEAREST */
-       690, /* GL_LINEAR_MIPMAP_NEAREST */
-      1005, /* GL_NEAREST_MIPMAP_LINEAR */
-       689, /* GL_LINEAR_MIPMAP_LINEAR */
-      1700, /* GL_TEXTURE_MAG_FILTER */
-      1708, /* GL_TEXTURE_MIN_FILTER */
-      1725, /* GL_TEXTURE_WRAP_S */
-      1726, /* GL_TEXTURE_WRAP_T */
+      1682, /* GL_TEXTURE_ENV_MODE */
+      1681, /* GL_TEXTURE_ENV_COLOR */
+      1680, /* GL_TEXTURE_ENV */
+       472, /* GL_EYE_LINEAR */
+      1037, /* GL_OBJECT_LINEAR */
+      1460, /* GL_SPHERE_MAP */
+      1684, /* GL_TEXTURE_GEN_MODE */
+      1039, /* GL_OBJECT_PLANE */
+       473, /* GL_EYE_PLANE */
+      1005, /* GL_NEAREST */
+       688, /* GL_LINEAR */
+      1009, /* GL_NEAREST_MIPMAP_NEAREST */
+       693, /* GL_LINEAR_MIPMAP_NEAREST */
+      1008, /* GL_NEAREST_MIPMAP_LINEAR */
+       692, /* GL_LINEAR_MIPMAP_LINEAR */
+      1705, /* GL_TEXTURE_MAG_FILTER */
+      1713, /* GL_TEXTURE_MIN_FILTER */
+      1730, /* GL_TEXTURE_WRAP_S */
+      1731, /* GL_TEXTURE_WRAP_T */
        126, /* GL_CLAMP */
-      1320, /* GL_REPEAT */
-      1166, /* GL_POLYGON_OFFSET_UNITS */
-      1165, /* GL_POLYGON_OFFSET_POINT */
-      1164, /* GL_POLYGON_OFFSET_LINE */
-      1282, /* GL_R3_G3_B2 */
-      1780, /* GL_V2F */
-      1781, /* GL_V3F */
+      1325, /* GL_REPEAT */
+      1169, /* GL_POLYGON_OFFSET_UNITS */
+      1168, /* GL_POLYGON_OFFSET_POINT */
+      1167, /* GL_POLYGON_OFFSET_LINE */
+      1287, /* GL_R3_G3_B2 */
+      1785, /* GL_V2F */
+      1786, /* GL_V3F */
        123, /* GL_C4UB_V2F */
        124, /* GL_C4UB_V3F */
        121, /* GL_C3F_V3F */
-       999, /* GL_N3F_V3F */
+      1002, /* GL_N3F_V3F */
        122, /* GL_C4F_N3F_V3F */
-      1536, /* GL_T2F_V3F */
-      1538, /* GL_T4F_V4F */
-      1534, /* GL_T2F_C4UB_V3F */
-      1532, /* GL_T2F_C3F_V3F */
-      1535, /* GL_T2F_N3F_V3F */
-      1533, /* GL_T2F_C4F_N3F_V3F */
-      1537, /* GL_T4F_C4F_N3F_V4F */
+      1541, /* GL_T2F_V3F */
+      1543, /* GL_T4F_V4F */
+      1539, /* GL_T2F_C4UB_V3F */
+      1537, /* GL_T2F_C3F_V3F */
+      1540, /* GL_T2F_N3F_V3F */
+      1538, /* GL_T2F_C4F_N3F_V3F */
+      1542, /* GL_T4F_C4F_N3F_V4F */
        139, /* GL_CLIP_PLANE0 */
        140, /* GL_CLIP_PLANE1 */
        141, /* GL_CLIP_PLANE2 */
        142, /* GL_CLIP_PLANE3 */
        143, /* GL_CLIP_PLANE4 */
        144, /* GL_CLIP_PLANE5 */
-       669, /* GL_LIGHT0 */
-       670, /* GL_LIGHT1 */
-       671, /* GL_LIGHT2 */
-       672, /* GL_LIGHT3 */
-       673, /* GL_LIGHT4 */
-       674, /* GL_LIGHT5 */
-       675, /* GL_LIGHT6 */
-       676, /* GL_LIGHT7 */
-       595, /* GL_HINT_BIT */
+       672, /* GL_LIGHT0 */
+       673, /* GL_LIGHT1 */
+       674, /* GL_LIGHT2 */
+       675, /* GL_LIGHT3 */
+       676, /* GL_LIGHT4 */
+       677, /* GL_LIGHT5 */
+       678, /* GL_LIGHT6 */
+       679, /* GL_LIGHT7 */
+       597, /* GL_HINT_BIT */
        277, /* GL_CONSTANT_COLOR */
-      1047, /* GL_ONE_MINUS_CONSTANT_COLOR */
+      1050, /* GL_ONE_MINUS_CONSTANT_COLOR */
        272, /* GL_CONSTANT_ALPHA */
-      1045, /* GL_ONE_MINUS_CONSTANT_ALPHA */
+      1048, /* GL_ONE_MINUS_CONSTANT_ALPHA */
         76, /* GL_BLEND_COLOR */
-       579, /* GL_FUNC_ADD */
-       935, /* GL_MIN */
-       845, /* GL_MAX */
+       581, /* GL_FUNC_ADD */
+       938, /* GL_MIN */
+       848, /* GL_MAX */
         81, /* GL_BLEND_EQUATION */
-       583, /* GL_FUNC_SUBTRACT */
-       581, /* GL_FUNC_REVERSE_SUBTRACT */
+       585, /* GL_FUNC_SUBTRACT */
+       583, /* GL_FUNC_REVERSE_SUBTRACT */
        280, /* GL_CONVOLUTION_1D */
        281, /* GL_CONVOLUTION_2D */
-      1407, /* GL_SEPARABLE_2D */
+      1412, /* GL_SEPARABLE_2D */
        284, /* GL_CONVOLUTION_BORDER_MODE */
        288, /* GL_CONVOLUTION_FILTER_SCALE */
        286, /* GL_CONVOLUTION_FILTER_BIAS */
-      1293, /* GL_REDUCE */
+      1298, /* GL_REDUCE */
        290, /* GL_CONVOLUTION_FORMAT */
        294, /* GL_CONVOLUTION_WIDTH */
        292, /* GL_CONVOLUTION_HEIGHT */
-       860, /* GL_MAX_CONVOLUTION_WIDTH */
-       858, /* GL_MAX_CONVOLUTION_HEIGHT */
-      1205, /* GL_POST_CONVOLUTION_RED_SCALE */
-      1201, /* GL_POST_CONVOLUTION_GREEN_SCALE */
-      1196, /* GL_POST_CONVOLUTION_BLUE_SCALE */
-      1192, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
-      1203, /* GL_POST_CONVOLUTION_RED_BIAS */
-      1199, /* GL_POST_CONVOLUTION_GREEN_BIAS */
-      1194, /* GL_POST_CONVOLUTION_BLUE_BIAS */
-      1190, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
-       596, /* GL_HISTOGRAM */
-      1253, /* GL_PROXY_HISTOGRAM */
-       612, /* GL_HISTOGRAM_WIDTH */
-       602, /* GL_HISTOGRAM_FORMAT */
-       608, /* GL_HISTOGRAM_RED_SIZE */
-       604, /* GL_HISTOGRAM_GREEN_SIZE */
-       599, /* GL_HISTOGRAM_BLUE_SIZE */
-       597, /* GL_HISTOGRAM_ALPHA_SIZE */
-       606, /* GL_HISTOGRAM_LUMINANCE_SIZE */
-       610, /* GL_HISTOGRAM_SINK */
-       936, /* GL_MINMAX */
-       938, /* GL_MINMAX_FORMAT */
-       940, /* GL_MINMAX_SINK */
-      1539, /* GL_TABLE_TOO_LARGE_EXT */
-      1759, /* GL_UNSIGNED_BYTE_3_3_2 */
-      1770, /* GL_UNSIGNED_SHORT_4_4_4_4 */
-      1772, /* GL_UNSIGNED_SHORT_5_5_5_1 */
-      1765, /* GL_UNSIGNED_INT_8_8_8_8 */
-      1761, /* GL_UNSIGNED_INT_10_10_10_2 */
-      1163, /* GL_POLYGON_OFFSET_FILL */
-      1162, /* GL_POLYGON_OFFSET_FACTOR */
-      1161, /* GL_POLYGON_OFFSET_BIAS */
-      1324, /* GL_RESCALE_NORMAL */
+       863, /* GL_MAX_CONVOLUTION_WIDTH */
+       861, /* GL_MAX_CONVOLUTION_HEIGHT */
+      1208, /* GL_POST_CONVOLUTION_RED_SCALE */
+      1204, /* GL_POST_CONVOLUTION_GREEN_SCALE */
+      1199, /* GL_POST_CONVOLUTION_BLUE_SCALE */
+      1195, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
+      1206, /* GL_POST_CONVOLUTION_RED_BIAS */
+      1202, /* GL_POST_CONVOLUTION_GREEN_BIAS */
+      1197, /* GL_POST_CONVOLUTION_BLUE_BIAS */
+      1193, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
+       598, /* GL_HISTOGRAM */
+      1257, /* GL_PROXY_HISTOGRAM */
+       614, /* GL_HISTOGRAM_WIDTH */
+       604, /* GL_HISTOGRAM_FORMAT */
+       610, /* GL_HISTOGRAM_RED_SIZE */
+       606, /* GL_HISTOGRAM_GREEN_SIZE */
+       601, /* GL_HISTOGRAM_BLUE_SIZE */
+       599, /* GL_HISTOGRAM_ALPHA_SIZE */
+       608, /* GL_HISTOGRAM_LUMINANCE_SIZE */
+       612, /* GL_HISTOGRAM_SINK */
+       939, /* GL_MINMAX */
+       941, /* GL_MINMAX_FORMAT */
+       943, /* GL_MINMAX_SINK */
+      1544, /* GL_TABLE_TOO_LARGE_EXT */
+      1764, /* GL_UNSIGNED_BYTE_3_3_2 */
+      1775, /* GL_UNSIGNED_SHORT_4_4_4_4 */
+      1777, /* GL_UNSIGNED_SHORT_5_5_5_1 */
+      1770, /* GL_UNSIGNED_INT_8_8_8_8 */
+      1766, /* GL_UNSIGNED_INT_10_10_10_2 */
+      1166, /* GL_POLYGON_OFFSET_FILL */
+      1165, /* GL_POLYGON_OFFSET_FACTOR */
+      1164, /* GL_POLYGON_OFFSET_BIAS */
+      1329, /* GL_RESCALE_NORMAL */
         36, /* GL_ALPHA4 */
         38, /* GL_ALPHA8 */
         32, /* GL_ALPHA12 */
         34, /* GL_ALPHA16 */
-       725, /* GL_LUMINANCE4 */
-       731, /* GL_LUMINANCE8 */
-       715, /* GL_LUMINANCE12 */
-       721, /* GL_LUMINANCE16 */
-       726, /* GL_LUMINANCE4_ALPHA4 */
-       729, /* GL_LUMINANCE6_ALPHA2 */
-       732, /* GL_LUMINANCE8_ALPHA8 */
-       718, /* GL_LUMINANCE12_ALPHA4 */
-       716, /* GL_LUMINANCE12_ALPHA12 */
-       722, /* GL_LUMINANCE16_ALPHA16 */
-       637, /* GL_INTENSITY */
-       642, /* GL_INTENSITY4 */
-       644, /* GL_INTENSITY8 */
-       638, /* GL_INTENSITY12 */
-       640, /* GL_INTENSITY16 */
-      1336, /* GL_RGB2_EXT */
-      1337, /* GL_RGB4 */
-      1340, /* GL_RGB5 */
-      1344, /* GL_RGB8 */
-      1328, /* GL_RGB10 */
-      1332, /* GL_RGB12 */
-      1334, /* GL_RGB16 */
-      1351, /* GL_RGBA2 */
-      1353, /* GL_RGBA4 */
-      1341, /* GL_RGB5_A1 */
-      1357, /* GL_RGBA8 */
-      1329, /* GL_RGB10_A2 */
-      1347, /* GL_RGBA12 */
-      1349, /* GL_RGBA16 */
-      1715, /* GL_TEXTURE_RED_SIZE */
-      1685, /* GL_TEXTURE_GREEN_SIZE */
-      1623, /* GL_TEXTURE_BLUE_SIZE */
-      1610, /* GL_TEXTURE_ALPHA_SIZE */
-      1698, /* GL_TEXTURE_LUMINANCE_SIZE */
-      1689, /* GL_TEXTURE_INTENSITY_SIZE */
-      1322, /* GL_REPLACE_EXT */
-      1257, /* GL_PROXY_TEXTURE_1D */
-      1260, /* GL_PROXY_TEXTURE_2D */
-      1721, /* GL_TEXTURE_TOO_LARGE_EXT */
-      1710, /* GL_TEXTURE_PRIORITY */
-      1717, /* GL_TEXTURE_RESIDENT */
-      1613, /* GL_TEXTURE_BINDING_1D */
-      1615, /* GL_TEXTURE_BINDING_2D */
-      1617, /* GL_TEXTURE_BINDING_3D */
-      1083, /* GL_PACK_SKIP_IMAGES */
-      1079, /* GL_PACK_IMAGE_HEIGHT */
-      1752, /* GL_UNPACK_SKIP_IMAGES */
-      1749, /* GL_UNPACK_IMAGE_HEIGHT */
-      1609, /* GL_TEXTURE_3D */
-      1263, /* GL_PROXY_TEXTURE_3D */
-      1672, /* GL_TEXTURE_DEPTH */
-      1724, /* GL_TEXTURE_WRAP_R */
-       846, /* GL_MAX_3D_TEXTURE_SIZE */
-      1785, /* GL_VERTEX_ARRAY */
-      1013, /* GL_NORMAL_ARRAY */
+       728, /* GL_LUMINANCE4 */
+       734, /* GL_LUMINANCE8 */
+       718, /* GL_LUMINANCE12 */
+       724, /* GL_LUMINANCE16 */
+       729, /* GL_LUMINANCE4_ALPHA4 */
+       732, /* GL_LUMINANCE6_ALPHA2 */
+       735, /* GL_LUMINANCE8_ALPHA8 */
+       721, /* GL_LUMINANCE12_ALPHA4 */
+       719, /* GL_LUMINANCE12_ALPHA12 */
+       725, /* GL_LUMINANCE16_ALPHA16 */
+       639, /* GL_INTENSITY */
+       644, /* GL_INTENSITY4 */
+       646, /* GL_INTENSITY8 */
+       640, /* GL_INTENSITY12 */
+       642, /* GL_INTENSITY16 */
+      1341, /* GL_RGB2_EXT */
+      1342, /* GL_RGB4 */
+      1345, /* GL_RGB5 */
+      1349, /* GL_RGB8 */
+      1333, /* GL_RGB10 */
+      1337, /* GL_RGB12 */
+      1339, /* GL_RGB16 */
+      1356, /* GL_RGBA2 */
+      1358, /* GL_RGBA4 */
+      1346, /* GL_RGB5_A1 */
+      1362, /* GL_RGBA8 */
+      1334, /* GL_RGB10_A2 */
+      1352, /* GL_RGBA12 */
+      1354, /* GL_RGBA16 */
+      1720, /* GL_TEXTURE_RED_SIZE */
+      1690, /* GL_TEXTURE_GREEN_SIZE */
+      1628, /* GL_TEXTURE_BLUE_SIZE */
+      1615, /* GL_TEXTURE_ALPHA_SIZE */
+      1703, /* GL_TEXTURE_LUMINANCE_SIZE */
+      1694, /* GL_TEXTURE_INTENSITY_SIZE */
+      1327, /* GL_REPLACE_EXT */
+      1261, /* GL_PROXY_TEXTURE_1D */
+      1264, /* GL_PROXY_TEXTURE_2D */
+      1726, /* GL_TEXTURE_TOO_LARGE_EXT */
+      1715, /* GL_TEXTURE_PRIORITY */
+      1722, /* GL_TEXTURE_RESIDENT */
+      1618, /* GL_TEXTURE_BINDING_1D */
+      1620, /* GL_TEXTURE_BINDING_2D */
+      1622, /* GL_TEXTURE_BINDING_3D */
+      1086, /* GL_PACK_SKIP_IMAGES */
+      1082, /* GL_PACK_IMAGE_HEIGHT */
+      1757, /* GL_UNPACK_SKIP_IMAGES */
+      1754, /* GL_UNPACK_IMAGE_HEIGHT */
+      1614, /* GL_TEXTURE_3D */
+      1267, /* GL_PROXY_TEXTURE_3D */
+      1677, /* GL_TEXTURE_DEPTH */
+      1729, /* GL_TEXTURE_WRAP_R */
+       849, /* GL_MAX_3D_TEXTURE_SIZE */
+      1790, /* GL_VERTEX_ARRAY */
+      1016, /* GL_NORMAL_ARRAY */
        148, /* GL_COLOR_ARRAY */
-       622, /* GL_INDEX_ARRAY */
-      1650, /* GL_TEXTURE_COORD_ARRAY */
-       455, /* GL_EDGE_FLAG_ARRAY */
-      1791, /* GL_VERTEX_ARRAY_SIZE */
-      1793, /* GL_VERTEX_ARRAY_TYPE */
-      1792, /* GL_VERTEX_ARRAY_STRIDE */
-      1018, /* GL_NORMAL_ARRAY_TYPE */
-      1017, /* GL_NORMAL_ARRAY_STRIDE */
+       624, /* GL_INDEX_ARRAY */
+      1655, /* GL_TEXTURE_COORD_ARRAY */
+       456, /* GL_EDGE_FLAG_ARRAY */
+      1796, /* GL_VERTEX_ARRAY_SIZE */
+      1798, /* GL_VERTEX_ARRAY_TYPE */
+      1797, /* GL_VERTEX_ARRAY_STRIDE */
+      1021, /* GL_NORMAL_ARRAY_TYPE */
+      1020, /* GL_NORMAL_ARRAY_STRIDE */
        152, /* GL_COLOR_ARRAY_SIZE */
        154, /* GL_COLOR_ARRAY_TYPE */
        153, /* GL_COLOR_ARRAY_STRIDE */
-       627, /* GL_INDEX_ARRAY_TYPE */
-       626, /* GL_INDEX_ARRAY_STRIDE */
-      1654, /* GL_TEXTURE_COORD_ARRAY_SIZE */
-      1656, /* GL_TEXTURE_COORD_ARRAY_TYPE */
-      1655, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
-       459, /* GL_EDGE_FLAG_ARRAY_STRIDE */
-      1790, /* GL_VERTEX_ARRAY_POINTER */
-      1016, /* GL_NORMAL_ARRAY_POINTER */
+       629, /* GL_INDEX_ARRAY_TYPE */
+       628, /* GL_INDEX_ARRAY_STRIDE */
+      1659, /* GL_TEXTURE_COORD_ARRAY_SIZE */
+      1661, /* GL_TEXTURE_COORD_ARRAY_TYPE */
+      1660, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
+       460, /* GL_EDGE_FLAG_ARRAY_STRIDE */
+      1795, /* GL_VERTEX_ARRAY_POINTER */
+      1019, /* GL_NORMAL_ARRAY_POINTER */
        151, /* GL_COLOR_ARRAY_POINTER */
-       625, /* GL_INDEX_ARRAY_POINTER */
-      1653, /* GL_TEXTURE_COORD_ARRAY_POINTER */
-       458, /* GL_EDGE_FLAG_ARRAY_POINTER */
-       992, /* GL_MULTISAMPLE */
-      1381, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
-      1383, /* GL_SAMPLE_ALPHA_TO_ONE */
-      1388, /* GL_SAMPLE_COVERAGE */
-      1385, /* GL_SAMPLE_BUFFERS */
-      1376, /* GL_SAMPLES */
-      1392, /* GL_SAMPLE_COVERAGE_VALUE */
-      1390, /* GL_SAMPLE_COVERAGE_INVERT */
+       627, /* GL_INDEX_ARRAY_POINTER */
+      1658, /* GL_TEXTURE_COORD_ARRAY_POINTER */
+       459, /* GL_EDGE_FLAG_ARRAY_POINTER */
+       995, /* GL_MULTISAMPLE */
+      1386, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
+      1388, /* GL_SAMPLE_ALPHA_TO_ONE */
+      1393, /* GL_SAMPLE_COVERAGE */
+      1390, /* GL_SAMPLE_BUFFERS */
+      1381, /* GL_SAMPLES */
+      1397, /* GL_SAMPLE_COVERAGE_VALUE */
+      1395, /* GL_SAMPLE_COVERAGE_INVERT */
        195, /* GL_COLOR_MATRIX */
        197, /* GL_COLOR_MATRIX_STACK_DEPTH */
-       854, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
-      1188, /* GL_POST_COLOR_MATRIX_RED_SCALE */
-      1184, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
-      1179, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
-      1175, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
-      1186, /* GL_POST_COLOR_MATRIX_RED_BIAS */
-      1182, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
-      1177, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
-      1173, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
-      1633, /* GL_TEXTURE_COLOR_TABLE_SGI */
-      1264, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
-      1635, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
+       857, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
+      1191, /* GL_POST_COLOR_MATRIX_RED_SCALE */
+      1187, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
+      1182, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
+      1178, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
+      1189, /* GL_POST_COLOR_MATRIX_RED_BIAS */
+      1185, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
+      1180, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
+      1176, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
+      1638, /* GL_TEXTURE_COLOR_TABLE_SGI */
+      1268, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
+      1640, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
         80, /* GL_BLEND_DST_RGB */
         89, /* GL_BLEND_SRC_RGB */
         79, /* GL_BLEND_DST_ALPHA */
         88, /* GL_BLEND_SRC_ALPHA */
        201, /* GL_COLOR_TABLE */
-      1198, /* GL_POST_CONVOLUTION_COLOR_TABLE */
-      1181, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
-      1252, /* GL_PROXY_COLOR_TABLE */
-      1256, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
-      1255, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
+      1201, /* GL_POST_CONVOLUTION_COLOR_TABLE */
+      1184, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
+      1256, /* GL_PROXY_COLOR_TABLE */
+      1260, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
+      1259, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
        225, /* GL_COLOR_TABLE_SCALE */
        205, /* GL_COLOR_TABLE_BIAS */
        210, /* GL_COLOR_TABLE_FORMAT */
@@ -4369,380 +4379,380 @@ static const unsigned reduced_enums[1347] =
        216, /* GL_COLOR_TABLE_INTENSITY_SIZE */
         71, /* GL_BGR */
         72, /* GL_BGRA */
-       868, /* GL_MAX_ELEMENTS_VERTICES */
-       867, /* GL_MAX_ELEMENTS_INDICES */
-      1688, /* GL_TEXTURE_INDEX_SIZE_EXT */
+       871, /* GL_MAX_ELEMENTS_VERTICES */
+       870, /* GL_MAX_ELEMENTS_INDICES */
+      1693, /* GL_TEXTURE_INDEX_SIZE_EXT */
        145, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */
-      1145, /* GL_POINT_SIZE_MIN */
-      1141, /* GL_POINT_SIZE_MAX */
-      1135, /* GL_POINT_FADE_THRESHOLD_SIZE */
-      1131, /* GL_POINT_DISTANCE_ATTENUATION */
+      1148, /* GL_POINT_SIZE_MIN */
+      1144, /* GL_POINT_SIZE_MAX */
+      1138, /* GL_POINT_FADE_THRESHOLD_SIZE */
+      1134, /* GL_POINT_DISTANCE_ATTENUATION */
        127, /* GL_CLAMP_TO_BORDER */
        130, /* GL_CLAMP_TO_EDGE */
-      1709, /* GL_TEXTURE_MIN_LOD */
-      1707, /* GL_TEXTURE_MAX_LOD */
-      1612, /* GL_TEXTURE_BASE_LEVEL */
-      1706, /* GL_TEXTURE_MAX_LEVEL */
-       615, /* GL_IGNORE_BORDER_HP */
+      1714, /* GL_TEXTURE_MIN_LOD */
+      1712, /* GL_TEXTURE_MAX_LOD */
+      1617, /* GL_TEXTURE_BASE_LEVEL */
+      1711, /* GL_TEXTURE_MAX_LEVEL */
+       617, /* GL_IGNORE_BORDER_HP */
        276, /* GL_CONSTANT_BORDER_HP */
-      1323, /* GL_REPLICATE_BORDER_HP */
+      1328, /* GL_REPLICATE_BORDER_HP */
        282, /* GL_CONVOLUTION_BORDER_COLOR */
-      1042, /* GL_OCCLUSION_TEST_HP */
-      1043, /* GL_OCCLUSION_TEST_RESULT_HP */
-       687, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
-      1627, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
-      1629, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
-      1631, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
-      1632, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-      1630, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
-      1628, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
-       850, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
-       851, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-      1208, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
-      1210, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
-      1207, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
-      1209, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
-      1696, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
-      1697, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
-      1695, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
-       585, /* GL_GENERATE_MIPMAP */
-       586, /* GL_GENERATE_MIPMAP_HINT */
-       527, /* GL_FOG_OFFSET_SGIX */
-       528, /* GL_FOG_OFFSET_VALUE_SGIX */
-      1641, /* GL_TEXTURE_COMPARE_SGIX */
-      1640, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
-      1692, /* GL_TEXTURE_LEQUAL_R_SGIX */
-      1684, /* GL_TEXTURE_GEQUAL_R_SGIX */
-       358, /* GL_DEPTH_COMPONENT16 */
-       361, /* GL_DEPTH_COMPONENT24 */
-       364, /* GL_DEPTH_COMPONENT32 */
+      1045, /* GL_OCCLUSION_TEST_HP */
+      1046, /* GL_OCCLUSION_TEST_RESULT_HP */
+       690, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
+      1632, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
+      1634, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
+      1636, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
+      1637, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+      1635, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
+      1633, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
+       853, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
+       854, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+      1211, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
+      1213, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
+      1210, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
+      1212, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
+      1701, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
+      1702, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
+      1700, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
+       587, /* GL_GENERATE_MIPMAP */
+       588, /* GL_GENERATE_MIPMAP_HINT */
+       529, /* GL_FOG_OFFSET_SGIX */
+       530, /* GL_FOG_OFFSET_VALUE_SGIX */
+      1646, /* GL_TEXTURE_COMPARE_SGIX */
+      1645, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
+      1697, /* GL_TEXTURE_LEQUAL_R_SGIX */
+      1689, /* GL_TEXTURE_GEQUAL_R_SGIX */
+       359, /* GL_DEPTH_COMPONENT16 */
+       362, /* GL_DEPTH_COMPONENT24 */
+       365, /* GL_DEPTH_COMPONENT32 */
        306, /* GL_CULL_VERTEX_EXT */
        308, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */
        307, /* GL_CULL_VERTEX_EYE_POSITION_EXT */
-      1848, /* GL_WRAP_BORDER_SUN */
-      1634, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
-       680, /* GL_LIGHT_MODEL_COLOR_CONTROL */
-      1422, /* GL_SINGLE_COLOR */
-      1408, /* GL_SEPARATE_SPECULAR_COLOR */
-      1417, /* GL_SHARED_TEXTURE_PALETTE_EXT */
-       538, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
-       539, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
-       546, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
-       541, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
-       537, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
-       536, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
-       540, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
-       547, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
-       558, /* GL_FRAMEBUFFER_DEFAULT */
-       571, /* GL_FRAMEBUFFER_UNDEFINED */
-       371, /* GL_DEPTH_STENCIL_ATTACHMENT */
-       621, /* GL_INDEX */
-      1758, /* GL_UNSIGNED_BYTE_2_3_3_REV */
-      1773, /* GL_UNSIGNED_SHORT_5_6_5 */
-      1774, /* GL_UNSIGNED_SHORT_5_6_5_REV */
-      1771, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
-      1769, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
-      1766, /* GL_UNSIGNED_INT_8_8_8_8_REV */
-      1764, /* GL_UNSIGNED_INT_2_10_10_10_REV */
-      1704, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
-      1705, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
-      1703, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
-       943, /* GL_MIRRORED_REPEAT */
-      1364, /* GL_RGB_S3TC */
-      1339, /* GL_RGB4_S3TC */
-      1362, /* GL_RGBA_S3TC */
-      1356, /* GL_RGBA4_S3TC */
-      1360, /* GL_RGBA_DXT5_S3TC */
-      1354, /* GL_RGBA4_DXT5_S3TC */
+      1853, /* GL_WRAP_BORDER_SUN */
+      1639, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
+       683, /* GL_LIGHT_MODEL_COLOR_CONTROL */
+      1427, /* GL_SINGLE_COLOR */
+      1413, /* GL_SEPARATE_SPECULAR_COLOR */
+      1422, /* GL_SHARED_TEXTURE_PALETTE_EXT */
+       540, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
+       541, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
+       548, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
+       543, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
+       539, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
+       538, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
+       542, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
+       549, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
+       560, /* GL_FRAMEBUFFER_DEFAULT */
+       573, /* GL_FRAMEBUFFER_UNDEFINED */
+       372, /* GL_DEPTH_STENCIL_ATTACHMENT */
+       623, /* GL_INDEX */
+      1763, /* GL_UNSIGNED_BYTE_2_3_3_REV */
+      1778, /* GL_UNSIGNED_SHORT_5_6_5 */
+      1779, /* GL_UNSIGNED_SHORT_5_6_5_REV */
+      1776, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
+      1774, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
+      1771, /* GL_UNSIGNED_INT_8_8_8_8_REV */
+      1769, /* GL_UNSIGNED_INT_2_10_10_10_REV */
+      1709, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
+      1710, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
+      1708, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
+       946, /* GL_MIRRORED_REPEAT */
+      1369, /* GL_RGB_S3TC */
+      1344, /* GL_RGB4_S3TC */
+      1367, /* GL_RGBA_S3TC */
+      1361, /* GL_RGBA4_S3TC */
+      1365, /* GL_RGBA_DXT5_S3TC */
+      1359, /* GL_RGBA4_DXT5_S3TC */
        264, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */
        259, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */
        260, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */
        261, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */
-      1004, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
-      1003, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
-       688, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
-       514, /* GL_FOG_COORDINATE_SOURCE */
-       506, /* GL_FOG_COORD */
-       530, /* GL_FRAGMENT_DEPTH */
+      1007, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
+      1006, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
+       691, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
+       516, /* GL_FOG_COORDINATE_SOURCE */
+       508, /* GL_FOG_COORD */
+       532, /* GL_FRAGMENT_DEPTH */
        312, /* GL_CURRENT_FOG_COORD */
-       513, /* GL_FOG_COORDINATE_ARRAY_TYPE */
-       512, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
-       511, /* GL_FOG_COORDINATE_ARRAY_POINTER */
-       508, /* GL_FOG_COORDINATE_ARRAY */
+       515, /* GL_FOG_COORDINATE_ARRAY_TYPE */
+       514, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
+       513, /* GL_FOG_COORDINATE_ARRAY_POINTER */
+       510, /* GL_FOG_COORDINATE_ARRAY */
        199, /* GL_COLOR_SUM */
        332, /* GL_CURRENT_SECONDARY_COLOR */
-      1401, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
-      1403, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
-      1402, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
-      1400, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
-      1397, /* GL_SECONDARY_COLOR_ARRAY */
+      1406, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
+      1408, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
+      1407, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
+      1405, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
+      1402, /* GL_SECONDARY_COLOR_ARRAY */
        330, /* GL_CURRENT_RASTER_SECONDARY_COLOR */
         28, /* GL_ALIASED_POINT_SIZE_RANGE */
         27, /* GL_ALIASED_LINE_WIDTH_RANGE */
-      1541, /* GL_TEXTURE0 */
-      1543, /* GL_TEXTURE1 */
-      1565, /* GL_TEXTURE2 */
-      1587, /* GL_TEXTURE3 */
-      1593, /* GL_TEXTURE4 */
-      1595, /* GL_TEXTURE5 */
-      1597, /* GL_TEXTURE6 */
-      1599, /* GL_TEXTURE7 */
-      1601, /* GL_TEXTURE8 */
-      1603, /* GL_TEXTURE9 */
-      1544, /* GL_TEXTURE10 */
-      1546, /* GL_TEXTURE11 */
-      1548, /* GL_TEXTURE12 */
-      1550, /* GL_TEXTURE13 */
-      1552, /* GL_TEXTURE14 */
-      1554, /* GL_TEXTURE15 */
-      1556, /* GL_TEXTURE16 */
-      1558, /* GL_TEXTURE17 */
-      1560, /* GL_TEXTURE18 */
-      1562, /* GL_TEXTURE19 */
-      1566, /* GL_TEXTURE20 */
-      1568, /* GL_TEXTURE21 */
-      1570, /* GL_TEXTURE22 */
-      1572, /* GL_TEXTURE23 */
-      1574, /* GL_TEXTURE24 */
-      1576, /* GL_TEXTURE25 */
-      1578, /* GL_TEXTURE26 */
-      1580, /* GL_TEXTURE27 */
-      1582, /* GL_TEXTURE28 */
-      1584, /* GL_TEXTURE29 */
-      1588, /* GL_TEXTURE30 */
-      1590, /* GL_TEXTURE31 */
+      1546, /* GL_TEXTURE0 */
+      1548, /* GL_TEXTURE1 */
+      1570, /* GL_TEXTURE2 */
+      1592, /* GL_TEXTURE3 */
+      1598, /* GL_TEXTURE4 */
+      1600, /* GL_TEXTURE5 */
+      1602, /* GL_TEXTURE6 */
+      1604, /* GL_TEXTURE7 */
+      1606, /* GL_TEXTURE8 */
+      1608, /* GL_TEXTURE9 */
+      1549, /* GL_TEXTURE10 */
+      1551, /* GL_TEXTURE11 */
+      1553, /* GL_TEXTURE12 */
+      1555, /* GL_TEXTURE13 */
+      1557, /* GL_TEXTURE14 */
+      1559, /* GL_TEXTURE15 */
+      1561, /* GL_TEXTURE16 */
+      1563, /* GL_TEXTURE17 */
+      1565, /* GL_TEXTURE18 */
+      1567, /* GL_TEXTURE19 */
+      1571, /* GL_TEXTURE20 */
+      1573, /* GL_TEXTURE21 */
+      1575, /* GL_TEXTURE22 */
+      1577, /* GL_TEXTURE23 */
+      1579, /* GL_TEXTURE24 */
+      1581, /* GL_TEXTURE25 */
+      1583, /* GL_TEXTURE26 */
+      1585, /* GL_TEXTURE27 */
+      1587, /* GL_TEXTURE28 */
+      1589, /* GL_TEXTURE29 */
+      1593, /* GL_TEXTURE30 */
+      1595, /* GL_TEXTURE31 */
         18, /* GL_ACTIVE_TEXTURE */
        133, /* GL_CLIENT_ACTIVE_TEXTURE */
-       921, /* GL_MAX_TEXTURE_UNITS */
-      1736, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
-      1739, /* GL_TRANSPOSE_PROJECTION_MATRIX */
-      1741, /* GL_TRANSPOSE_TEXTURE_MATRIX */
-      1733, /* GL_TRANSPOSE_COLOR_MATRIX */
-      1523, /* GL_SUBTRACT */
-       908, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
+       924, /* GL_MAX_TEXTURE_UNITS */
+      1741, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
+      1744, /* GL_TRANSPOSE_PROJECTION_MATRIX */
+      1746, /* GL_TRANSPOSE_TEXTURE_MATRIX */
+      1738, /* GL_TRANSPOSE_COLOR_MATRIX */
+      1528, /* GL_SUBTRACT */
+       911, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
        247, /* GL_COMPRESSED_ALPHA */
        251, /* GL_COMPRESSED_LUMINANCE */
        252, /* GL_COMPRESSED_LUMINANCE_ALPHA */
        249, /* GL_COMPRESSED_INTENSITY */
        255, /* GL_COMPRESSED_RGB */
        256, /* GL_COMPRESSED_RGBA */
-      1648, /* GL_TEXTURE_COMPRESSION_HINT */
-      1713, /* GL_TEXTURE_RECTANGLE_ARB */
-      1620, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
-      1267, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
-       906, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
-       370, /* GL_DEPTH_STENCIL */
-      1762, /* GL_UNSIGNED_INT_24_8 */
-       917, /* GL_MAX_TEXTURE_LOD_BIAS */
-      1702, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
-       918, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
-      1678, /* GL_TEXTURE_FILTER_CONTROL */
-      1693, /* GL_TEXTURE_LOD_BIAS */
+      1653, /* GL_TEXTURE_COMPRESSION_HINT */
+      1718, /* GL_TEXTURE_RECTANGLE_ARB */
+      1625, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
+      1271, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
+       909, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
+       371, /* GL_DEPTH_STENCIL */
+      1767, /* GL_UNSIGNED_INT_24_8 */
+       920, /* GL_MAX_TEXTURE_LOD_BIAS */
+      1707, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
+       921, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
+      1683, /* GL_TEXTURE_FILTER_CONTROL */
+      1698, /* GL_TEXTURE_LOD_BIAS */
        232, /* GL_COMBINE4 */
-       911, /* GL_MAX_SHININESS_NV */
-       912, /* GL_MAX_SPOT_EXPONENT_NV */
-       619, /* GL_INCR_WRAP */
+       914, /* GL_MAX_SHININESS_NV */
+       915, /* GL_MAX_SPOT_EXPONENT_NV */
+       621, /* GL_INCR_WRAP */
        343, /* GL_DECR_WRAP */
-       963, /* GL_MODELVIEW1_ARB */
-      1019, /* GL_NORMAL_MAP */
-      1298, /* GL_REFLECTION_MAP */
-      1657, /* GL_TEXTURE_CUBE_MAP */
-      1618, /* GL_TEXTURE_BINDING_CUBE_MAP */
-      1665, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
-      1659, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
-      1667, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
-      1661, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
-      1669, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
-      1663, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
-      1265, /* GL_PROXY_TEXTURE_CUBE_MAP */
-       862, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
-       998, /* GL_MULTISAMPLE_FILTER_HINT_NV */
-       522, /* GL_FOG_DISTANCE_MODE_NV */
-       474, /* GL_EYE_RADIAL_NV */
-       473, /* GL_EYE_PLANE_ABSOLUTE_NV */
+       966, /* GL_MODELVIEW1_ARB */
+      1022, /* GL_NORMAL_MAP */
+      1303, /* GL_REFLECTION_MAP */
+      1662, /* GL_TEXTURE_CUBE_MAP */
+      1623, /* GL_TEXTURE_BINDING_CUBE_MAP */
+      1670, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
+      1664, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
+      1672, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
+      1666, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
+      1674, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
+      1668, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
+      1269, /* GL_PROXY_TEXTURE_CUBE_MAP */
+       865, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
+      1001, /* GL_MULTISAMPLE_FILTER_HINT_NV */
+       524, /* GL_FOG_DISTANCE_MODE_NV */
+       475, /* GL_EYE_RADIAL_NV */
+       474, /* GL_EYE_PLANE_ABSOLUTE_NV */
        231, /* GL_COMBINE */
        238, /* GL_COMBINE_RGB */
        233, /* GL_COMBINE_ALPHA */
-      1365, /* GL_RGB_SCALE */
+      1370, /* GL_RGB_SCALE */
         24, /* GL_ADD_SIGNED */
-       647, /* GL_INTERPOLATE */
+       649, /* GL_INTERPOLATE */
        271, /* GL_CONSTANT */
-      1214, /* GL_PRIMARY_COLOR */
-      1211, /* GL_PREVIOUS */
-      1437, /* GL_SOURCE0_RGB */
-      1443, /* GL_SOURCE1_RGB */
-      1449, /* GL_SOURCE2_RGB */
-      1453, /* GL_SOURCE3_RGB_NV */
-      1434, /* GL_SOURCE0_ALPHA */
-      1440, /* GL_SOURCE1_ALPHA */
-      1446, /* GL_SOURCE2_ALPHA */
-      1452, /* GL_SOURCE3_ALPHA_NV */
-      1056, /* GL_OPERAND0_RGB */
-      1062, /* GL_OPERAND1_RGB */
-      1068, /* GL_OPERAND2_RGB */
-      1072, /* GL_OPERAND3_RGB_NV */
-      1053, /* GL_OPERAND0_ALPHA */
-      1059, /* GL_OPERAND1_ALPHA */
-      1065, /* GL_OPERAND2_ALPHA */
-      1071, /* GL_OPERAND3_ALPHA_NV */
-      1786, /* GL_VERTEX_ARRAY_BINDING */
-      1711, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
-      1712, /* GL_TEXTURE_RANGE_POINTER_APPLE */
-      1852, /* GL_YCBCR_422_APPLE */
-      1775, /* GL_UNSIGNED_SHORT_8_8_APPLE */
-      1777, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
-      1720, /* GL_TEXTURE_STORAGE_HINT_APPLE */
-      1514, /* GL_STORAGE_PRIVATE_APPLE */
-      1513, /* GL_STORAGE_CACHED_APPLE */
-      1515, /* GL_STORAGE_SHARED_APPLE */
-      1424, /* GL_SLICE_ACCUM_SUN */
-      1273, /* GL_QUAD_MESH_SUN */
-      1745, /* GL_TRIANGLE_MESH_SUN */
-      1825, /* GL_VERTEX_PROGRAM_ARB */
-      1836, /* GL_VERTEX_STATE_PROGRAM_NV */
-      1812, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
-      1818, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
-      1820, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
-      1822, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
+      1217, /* GL_PRIMARY_COLOR */
+      1214, /* GL_PREVIOUS */
+      1442, /* GL_SOURCE0_RGB */
+      1448, /* GL_SOURCE1_RGB */
+      1454, /* GL_SOURCE2_RGB */
+      1458, /* GL_SOURCE3_RGB_NV */
+      1439, /* GL_SOURCE0_ALPHA */
+      1445, /* GL_SOURCE1_ALPHA */
+      1451, /* GL_SOURCE2_ALPHA */
+      1457, /* GL_SOURCE3_ALPHA_NV */
+      1059, /* GL_OPERAND0_RGB */
+      1065, /* GL_OPERAND1_RGB */
+      1071, /* GL_OPERAND2_RGB */
+      1075, /* GL_OPERAND3_RGB_NV */
+      1056, /* GL_OPERAND0_ALPHA */
+      1062, /* GL_OPERAND1_ALPHA */
+      1068, /* GL_OPERAND2_ALPHA */
+      1074, /* GL_OPERAND3_ALPHA_NV */
+      1791, /* GL_VERTEX_ARRAY_BINDING */
+      1716, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
+      1717, /* GL_TEXTURE_RANGE_POINTER_APPLE */
+      1857, /* GL_YCBCR_422_APPLE */
+      1780, /* GL_UNSIGNED_SHORT_8_8_APPLE */
+      1782, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
+      1725, /* GL_TEXTURE_STORAGE_HINT_APPLE */
+      1519, /* GL_STORAGE_PRIVATE_APPLE */
+      1518, /* GL_STORAGE_CACHED_APPLE */
+      1520, /* GL_STORAGE_SHARED_APPLE */
+      1429, /* GL_SLICE_ACCUM_SUN */
+      1278, /* GL_QUAD_MESH_SUN */
+      1750, /* GL_TRIANGLE_MESH_SUN */
+      1830, /* GL_VERTEX_PROGRAM_ARB */
+      1841, /* GL_VERTEX_STATE_PROGRAM_NV */
+      1817, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
+      1823, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
+      1825, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
+      1827, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
        334, /* GL_CURRENT_VERTEX_ATTRIB */
-      1227, /* GL_PROGRAM_LENGTH_ARB */
-      1241, /* GL_PROGRAM_STRING_ARB */
-       985, /* GL_MODELVIEW_PROJECTION_NV */
-       614, /* GL_IDENTITY_NV */
-       661, /* GL_INVERSE_NV */
-      1738, /* GL_TRANSPOSE_NV */
-       662, /* GL_INVERSE_TRANSPOSE_NV */
-       892, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
-       891, /* GL_MAX_PROGRAM_MATRICES_ARB */
-       799, /* GL_MATRIX0_NV */
-       811, /* GL_MATRIX1_NV */
-       823, /* GL_MATRIX2_NV */
-       827, /* GL_MATRIX3_NV */
-       829, /* GL_MATRIX4_NV */
-       831, /* GL_MATRIX5_NV */
-       833, /* GL_MATRIX6_NV */
-       835, /* GL_MATRIX7_NV */
+      1230, /* GL_PROGRAM_LENGTH_ARB */
+      1244, /* GL_PROGRAM_STRING_ARB */
+       988, /* GL_MODELVIEW_PROJECTION_NV */
+       616, /* GL_IDENTITY_NV */
+       663, /* GL_INVERSE_NV */
+      1743, /* GL_TRANSPOSE_NV */
+       664, /* GL_INVERSE_TRANSPOSE_NV */
+       895, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
+       894, /* GL_MAX_PROGRAM_MATRICES_ARB */
+       802, /* GL_MATRIX0_NV */
+       814, /* GL_MATRIX1_NV */
+       826, /* GL_MATRIX2_NV */
+       830, /* GL_MATRIX3_NV */
+       832, /* GL_MATRIX4_NV */
+       834, /* GL_MATRIX5_NV */
+       836, /* GL_MATRIX6_NV */
+       838, /* GL_MATRIX7_NV */
        318, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */
        315, /* GL_CURRENT_MATRIX_ARB */
-      1828, /* GL_VERTEX_PROGRAM_POINT_SIZE */
-      1831, /* GL_VERTEX_PROGRAM_TWO_SIDE */
-      1239, /* GL_PROGRAM_PARAMETER_NV */
-      1816, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
-      1243, /* GL_PROGRAM_TARGET_NV */
-      1240, /* GL_PROGRAM_RESIDENT_NV */
-      1730, /* GL_TRACK_MATRIX_NV */
-      1731, /* GL_TRACK_MATRIX_TRANSFORM_NV */
-      1826, /* GL_VERTEX_PROGRAM_BINDING_NV */
-      1221, /* GL_PROGRAM_ERROR_POSITION_ARB */
-       355, /* GL_DEPTH_CLAMP_NV */
-      1794, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
-      1801, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
-      1802, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
-      1803, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
-      1804, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
-      1805, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
-      1806, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
-      1807, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
-      1808, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
-      1809, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
-      1795, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
-      1796, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
-      1797, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
-      1798, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
-      1799, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
-      1800, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
-       747, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
-       754, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
-       755, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
-       756, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
-       757, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
-       758, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
-       759, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
-       760, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
-       761, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
-       762, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
-       748, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
-       749, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
-       750, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
-       751, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
-       752, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
-       753, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
-       774, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
-       781, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
-       782, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
-       783, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
-       784, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
-       785, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
-       786, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
-      1220, /* GL_PROGRAM_BINDING_ARB */
-       788, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
-       789, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
-       775, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
-       776, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
-       777, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
-       778, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
-       779, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
-       780, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
-      1646, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
-      1643, /* GL_TEXTURE_COMPRESSED */
-      1024, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
+      1833, /* GL_VERTEX_PROGRAM_POINT_SIZE */
+      1836, /* GL_VERTEX_PROGRAM_TWO_SIDE */
+      1242, /* GL_PROGRAM_PARAMETER_NV */
+      1821, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
+      1246, /* GL_PROGRAM_TARGET_NV */
+      1243, /* GL_PROGRAM_RESIDENT_NV */
+      1735, /* GL_TRACK_MATRIX_NV */
+      1736, /* GL_TRACK_MATRIX_TRANSFORM_NV */
+      1831, /* GL_VERTEX_PROGRAM_BINDING_NV */
+      1224, /* GL_PROGRAM_ERROR_POSITION_ARB */
+       355, /* GL_DEPTH_CLAMP */
+      1799, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
+      1806, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
+      1807, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
+      1808, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
+      1809, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
+      1810, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
+      1811, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
+      1812, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
+      1813, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
+      1814, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
+      1800, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
+      1801, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
+      1802, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
+      1803, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
+      1804, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
+      1805, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
+       750, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
+       757, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
+       758, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
+       759, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
+       760, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
+       761, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
+       762, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
+       763, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
+       764, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
+       765, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
+       751, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
+       752, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
+       753, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
+       754, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
+       755, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
+       756, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
+       777, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
+       784, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
+       785, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
+       786, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
+       787, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
+       788, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
+       789, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
+      1223, /* GL_PROGRAM_BINDING_ARB */
+       791, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
+       792, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
+       778, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
+       779, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
+       780, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
+       781, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
+       782, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
+       783, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
+      1651, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
+      1648, /* GL_TEXTURE_COMPRESSED */
+      1027, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
        269, /* GL_COMPRESSED_TEXTURE_FORMATS */
-       933, /* GL_MAX_VERTEX_UNITS_ARB */
+       936, /* GL_MAX_VERTEX_UNITS_ARB */
         22, /* GL_ACTIVE_VERTEX_UNITS_ARB */
-      1847, /* GL_WEIGHT_SUM_UNITY_ARB */
-      1824, /* GL_VERTEX_BLEND_ARB */
+      1852, /* GL_WEIGHT_SUM_UNITY_ARB */
+      1829, /* GL_VERTEX_BLEND_ARB */
        336, /* GL_CURRENT_WEIGHT_ARB */
-      1846, /* GL_WEIGHT_ARRAY_TYPE_ARB */
-      1845, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
-      1844, /* GL_WEIGHT_ARRAY_SIZE_ARB */
-      1843, /* GL_WEIGHT_ARRAY_POINTER_ARB */
-      1840, /* GL_WEIGHT_ARRAY_ARB */
-       383, /* GL_DOT3_RGB */
-       384, /* GL_DOT3_RGBA */
+      1851, /* GL_WEIGHT_ARRAY_TYPE_ARB */
+      1850, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
+      1849, /* GL_WEIGHT_ARRAY_SIZE_ARB */
+      1848, /* GL_WEIGHT_ARRAY_POINTER_ARB */
+      1845, /* GL_WEIGHT_ARRAY_ARB */
+       384, /* GL_DOT3_RGB */
+       385, /* GL_DOT3_RGBA */
        263, /* GL_COMPRESSED_RGB_FXT1_3DFX */
        258, /* GL_COMPRESSED_RGBA_FXT1_3DFX */
-       993, /* GL_MULTISAMPLE_3DFX */
-      1386, /* GL_SAMPLE_BUFFERS_3DFX */
-      1377, /* GL_SAMPLES_3DFX */
-       974, /* GL_MODELVIEW2_ARB */
-       977, /* GL_MODELVIEW3_ARB */
-       978, /* GL_MODELVIEW4_ARB */
-       979, /* GL_MODELVIEW5_ARB */
-       980, /* GL_MODELVIEW6_ARB */
-       981, /* GL_MODELVIEW7_ARB */
-       982, /* GL_MODELVIEW8_ARB */
-       983, /* GL_MODELVIEW9_ARB */
-       953, /* GL_MODELVIEW10_ARB */
-       954, /* GL_MODELVIEW11_ARB */
-       955, /* GL_MODELVIEW12_ARB */
-       956, /* GL_MODELVIEW13_ARB */
-       957, /* GL_MODELVIEW14_ARB */
-       958, /* GL_MODELVIEW15_ARB */
-       959, /* GL_MODELVIEW16_ARB */
-       960, /* GL_MODELVIEW17_ARB */
-       961, /* GL_MODELVIEW18_ARB */
-       962, /* GL_MODELVIEW19_ARB */
-       964, /* GL_MODELVIEW20_ARB */
-       965, /* GL_MODELVIEW21_ARB */
-       966, /* GL_MODELVIEW22_ARB */
-       967, /* GL_MODELVIEW23_ARB */
-       968, /* GL_MODELVIEW24_ARB */
-       969, /* GL_MODELVIEW25_ARB */
-       970, /* GL_MODELVIEW26_ARB */
-       971, /* GL_MODELVIEW27_ARB */
-       972, /* GL_MODELVIEW28_ARB */
-       973, /* GL_MODELVIEW29_ARB */
-       975, /* GL_MODELVIEW30_ARB */
-       976, /* GL_MODELVIEW31_ARB */
-       388, /* GL_DOT3_RGB_EXT */
-       386, /* GL_DOT3_RGBA_EXT */
-       947, /* GL_MIRROR_CLAMP_EXT */
-       950, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
-       988, /* GL_MODULATE_ADD_ATI */
-       989, /* GL_MODULATE_SIGNED_ADD_ATI */
-       990, /* GL_MODULATE_SUBTRACT_ATI */
-      1853, /* GL_YCBCR_MESA */
-      1080, /* GL_PACK_INVERT_MESA */
+       996, /* GL_MULTISAMPLE_3DFX */
+      1391, /* GL_SAMPLE_BUFFERS_3DFX */
+      1382, /* GL_SAMPLES_3DFX */
+       977, /* GL_MODELVIEW2_ARB */
+       980, /* GL_MODELVIEW3_ARB */
+       981, /* GL_MODELVIEW4_ARB */
+       982, /* GL_MODELVIEW5_ARB */
+       983, /* GL_MODELVIEW6_ARB */
+       984, /* GL_MODELVIEW7_ARB */
+       985, /* GL_MODELVIEW8_ARB */
+       986, /* GL_MODELVIEW9_ARB */
+       956, /* GL_MODELVIEW10_ARB */
+       957, /* GL_MODELVIEW11_ARB */
+       958, /* GL_MODELVIEW12_ARB */
+       959, /* GL_MODELVIEW13_ARB */
+       960, /* GL_MODELVIEW14_ARB */
+       961, /* GL_MODELVIEW15_ARB */
+       962, /* GL_MODELVIEW16_ARB */
+       963, /* GL_MODELVIEW17_ARB */
+       964, /* GL_MODELVIEW18_ARB */
+       965, /* GL_MODELVIEW19_ARB */
+       967, /* GL_MODELVIEW20_ARB */
+       968, /* GL_MODELVIEW21_ARB */
+       969, /* GL_MODELVIEW22_ARB */
+       970, /* GL_MODELVIEW23_ARB */
+       971, /* GL_MODELVIEW24_ARB */
+       972, /* GL_MODELVIEW25_ARB */
+       973, /* GL_MODELVIEW26_ARB */
+       974, /* GL_MODELVIEW27_ARB */
+       975, /* GL_MODELVIEW28_ARB */
+       976, /* GL_MODELVIEW29_ARB */
+       978, /* GL_MODELVIEW30_ARB */
+       979, /* GL_MODELVIEW31_ARB */
+       389, /* GL_DOT3_RGB_EXT */
+       387, /* GL_DOT3_RGBA_EXT */
+       950, /* GL_MIRROR_CLAMP_EXT */
+       953, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
+       991, /* GL_MODULATE_ADD_ATI */
+       992, /* GL_MODULATE_SIGNED_ADD_ATI */
+       993, /* GL_MODULATE_SUBTRACT_ATI */
+      1858, /* GL_YCBCR_MESA */
+      1083, /* GL_PACK_INVERT_MESA */
        339, /* GL_DEBUG_OBJECT_MESA */
        340, /* GL_DEBUG_PRINT_MESA */
        338, /* GL_DEBUG_ASSERT_MESA */
@@ -4752,292 +4762,292 @@ static const unsigned reduced_enums[1347] =
        117, /* GL_BUMP_ROT_MATRIX_SIZE_ATI */
        115, /* GL_BUMP_NUM_TEX_UNITS_ATI */
        119, /* GL_BUMP_TEX_UNITS_ATI */
-       447, /* GL_DUDV_ATI */
-       446, /* GL_DU8DV8_ATI */
+       448, /* GL_DUDV_ATI */
+       447, /* GL_DU8DV8_ATI */
        114, /* GL_BUMP_ENVMAP_ATI */
        118, /* GL_BUMP_TARGET_ATI */
-      1485, /* GL_STENCIL_BACK_FUNC */
-      1483, /* GL_STENCIL_BACK_FAIL */
-      1487, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
-      1489, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
-       531, /* GL_FRAGMENT_PROGRAM_ARB */
-      1218, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
-      1246, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
-      1245, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
-      1230, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-      1236, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-      1235, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-       881, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
-       904, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
-       903, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
-       894, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-       900, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-       899, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-       864, /* GL_MAX_DRAW_BUFFERS */
-       392, /* GL_DRAW_BUFFER0 */
-       395, /* GL_DRAW_BUFFER1 */
-       416, /* GL_DRAW_BUFFER2 */
-       419, /* GL_DRAW_BUFFER3 */
-       422, /* GL_DRAW_BUFFER4 */
-       425, /* GL_DRAW_BUFFER5 */
-       428, /* GL_DRAW_BUFFER6 */
-       431, /* GL_DRAW_BUFFER7 */
-       434, /* GL_DRAW_BUFFER8 */
-       437, /* GL_DRAW_BUFFER9 */
-       396, /* GL_DRAW_BUFFER10 */
-       399, /* GL_DRAW_BUFFER11 */
-       402, /* GL_DRAW_BUFFER12 */
-       405, /* GL_DRAW_BUFFER13 */
-       408, /* GL_DRAW_BUFFER14 */
-       411, /* GL_DRAW_BUFFER15 */
+      1490, /* GL_STENCIL_BACK_FUNC */
+      1488, /* GL_STENCIL_BACK_FAIL */
+      1492, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
+      1494, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
+       533, /* GL_FRAGMENT_PROGRAM_ARB */
+      1221, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
+      1249, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
+      1248, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
+      1233, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+      1239, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+      1238, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+       884, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
+       907, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
+       906, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
+       897, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+       903, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+       902, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+       867, /* GL_MAX_DRAW_BUFFERS */
+       393, /* GL_DRAW_BUFFER0 */
+       396, /* GL_DRAW_BUFFER1 */
+       417, /* GL_DRAW_BUFFER2 */
+       420, /* GL_DRAW_BUFFER3 */
+       423, /* GL_DRAW_BUFFER4 */
+       426, /* GL_DRAW_BUFFER5 */
+       429, /* GL_DRAW_BUFFER6 */
+       432, /* GL_DRAW_BUFFER7 */
+       435, /* GL_DRAW_BUFFER8 */
+       438, /* GL_DRAW_BUFFER9 */
+       397, /* GL_DRAW_BUFFER10 */
+       400, /* GL_DRAW_BUFFER11 */
+       403, /* GL_DRAW_BUFFER12 */
+       406, /* GL_DRAW_BUFFER13 */
+       409, /* GL_DRAW_BUFFER14 */
+       412, /* GL_DRAW_BUFFER15 */
         82, /* GL_BLEND_EQUATION_ALPHA */
-       844, /* GL_MATRIX_PALETTE_ARB */
-       875, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
-       878, /* GL_MAX_PALETTE_MATRICES_ARB */
+       847, /* GL_MATRIX_PALETTE_ARB */
+       878, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
+       881, /* GL_MAX_PALETTE_MATRICES_ARB */
        321, /* GL_CURRENT_PALETTE_MATRIX_ARB */
-       838, /* GL_MATRIX_INDEX_ARRAY_ARB */
+       841, /* GL_MATRIX_INDEX_ARRAY_ARB */
        316, /* GL_CURRENT_MATRIX_INDEX_ARB */
-       840, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
-       842, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
-       841, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
-       839, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
-      1673, /* GL_TEXTURE_DEPTH_SIZE */
-       376, /* GL_DEPTH_TEXTURE_MODE */
-      1638, /* GL_TEXTURE_COMPARE_MODE */
-      1636, /* GL_TEXTURE_COMPARE_FUNC */
+       843, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
+       845, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
+       844, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
+       842, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
+      1678, /* GL_TEXTURE_DEPTH_SIZE */
+       377, /* GL_DEPTH_TEXTURE_MODE */
+      1643, /* GL_TEXTURE_COMPARE_MODE */
+      1641, /* GL_TEXTURE_COMPARE_FUNC */
        242, /* GL_COMPARE_R_TO_TEXTURE */
-      1152, /* GL_POINT_SPRITE */
+      1155, /* GL_POINT_SPRITE */
        296, /* GL_COORD_REPLACE */
-      1156, /* GL_POINT_SPRITE_R_MODE_NV */
-      1275, /* GL_QUERY_COUNTER_BITS */
+      1159, /* GL_POINT_SPRITE_R_MODE_NV */
+      1280, /* GL_QUERY_COUNTER_BITS */
        323, /* GL_CURRENT_QUERY */
-      1277, /* GL_QUERY_RESULT */
-      1279, /* GL_QUERY_RESULT_AVAILABLE */
-       927, /* GL_MAX_VERTEX_ATTRIBS */
-      1814, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
-       374, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
-       373, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
-       913, /* GL_MAX_TEXTURE_COORDS */
-       915, /* GL_MAX_TEXTURE_IMAGE_UNITS */
-      1223, /* GL_PROGRAM_ERROR_STRING_ARB */
-      1225, /* GL_PROGRAM_FORMAT_ASCII_ARB */
-      1224, /* GL_PROGRAM_FORMAT_ARB */
-      1722, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
+      1282, /* GL_QUERY_RESULT */
+      1284, /* GL_QUERY_RESULT_AVAILABLE */
+       930, /* GL_MAX_VERTEX_ATTRIBS */
+      1819, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
+       375, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
+       374, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
+       916, /* GL_MAX_TEXTURE_COORDS */
+       918, /* GL_MAX_TEXTURE_IMAGE_UNITS */
+      1226, /* GL_PROGRAM_ERROR_STRING_ARB */
+      1228, /* GL_PROGRAM_FORMAT_ASCII_ARB */
+      1227, /* GL_PROGRAM_FORMAT_ARB */
+      1727, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
        353, /* GL_DEPTH_BOUNDS_TEST_EXT */
        352, /* GL_DEPTH_BOUNDS_EXT */
         53, /* GL_ARRAY_BUFFER */
-       460, /* GL_ELEMENT_ARRAY_BUFFER */
+       461, /* GL_ELEMENT_ARRAY_BUFFER */
         54, /* GL_ARRAY_BUFFER_BINDING */
-       461, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
-      1788, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
-      1014, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
+       462, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
+      1793, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
+      1017, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
        149, /* GL_COLOR_ARRAY_BUFFER_BINDING */
-       623, /* GL_INDEX_ARRAY_BUFFER_BINDING */
-      1651, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
-       456, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
-      1398, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
-       509, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
-      1841, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
-      1810, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
-      1226, /* GL_PROGRAM_INSTRUCTIONS_ARB */
-       887, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
-      1232, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-       896, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-      1244, /* GL_PROGRAM_TEMPORARIES_ARB */
-       902, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
-      1234, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
-       898, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
-      1238, /* GL_PROGRAM_PARAMETERS_ARB */
-       901, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
-      1233, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
-       897, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
-      1219, /* GL_PROGRAM_ATTRIBS_ARB */
-       882, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
-      1231, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
-       895, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
-      1217, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
-       880, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
-      1229, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-       893, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-       888, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
-       884, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
-      1247, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
-      1735, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
-      1288, /* GL_READ_ONLY */
-      1849, /* GL_WRITE_ONLY */
-      1290, /* GL_READ_WRITE */
+       625, /* GL_INDEX_ARRAY_BUFFER_BINDING */
+      1656, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
+       457, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
+      1403, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
+       511, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
+      1846, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
+      1815, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
+      1229, /* GL_PROGRAM_INSTRUCTIONS_ARB */
+       890, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
+      1235, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+       899, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+      1247, /* GL_PROGRAM_TEMPORARIES_ARB */
+       905, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
+      1237, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
+       901, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
+      1241, /* GL_PROGRAM_PARAMETERS_ARB */
+       904, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
+      1236, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
+       900, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
+      1222, /* GL_PROGRAM_ATTRIBS_ARB */
+       885, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
+      1234, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
+       898, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
+      1220, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
+       883, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
+      1232, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+       896, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+       891, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
+       887, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
+      1250, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
+      1740, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
+      1293, /* GL_READ_ONLY */
+      1854, /* GL_WRITE_ONLY */
+      1295, /* GL_READ_WRITE */
        102, /* GL_BUFFER_ACCESS */
        105, /* GL_BUFFER_MAPPED */
        107, /* GL_BUFFER_MAP_POINTER */
-      1729, /* GL_TIME_ELAPSED_EXT */
-       798, /* GL_MATRIX0_ARB */
-       810, /* GL_MATRIX1_ARB */
-       822, /* GL_MATRIX2_ARB */
-       826, /* GL_MATRIX3_ARB */
-       828, /* GL_MATRIX4_ARB */
-       830, /* GL_MATRIX5_ARB */
-       832, /* GL_MATRIX6_ARB */
-       834, /* GL_MATRIX7_ARB */
-       836, /* GL_MATRIX8_ARB */
-       837, /* GL_MATRIX9_ARB */
-       800, /* GL_MATRIX10_ARB */
-       801, /* GL_MATRIX11_ARB */
-       802, /* GL_MATRIX12_ARB */
-       803, /* GL_MATRIX13_ARB */
-       804, /* GL_MATRIX14_ARB */
-       805, /* GL_MATRIX15_ARB */
-       806, /* GL_MATRIX16_ARB */
-       807, /* GL_MATRIX17_ARB */
-       808, /* GL_MATRIX18_ARB */
-       809, /* GL_MATRIX19_ARB */
-       812, /* GL_MATRIX20_ARB */
-       813, /* GL_MATRIX21_ARB */
-       814, /* GL_MATRIX22_ARB */
-       815, /* GL_MATRIX23_ARB */
-       816, /* GL_MATRIX24_ARB */
-       817, /* GL_MATRIX25_ARB */
-       818, /* GL_MATRIX26_ARB */
-       819, /* GL_MATRIX27_ARB */
-       820, /* GL_MATRIX28_ARB */
-       821, /* GL_MATRIX29_ARB */
-       824, /* GL_MATRIX30_ARB */
-       825, /* GL_MATRIX31_ARB */
-      1518, /* GL_STREAM_DRAW */
-      1520, /* GL_STREAM_READ */
-      1516, /* GL_STREAM_COPY */
-      1476, /* GL_STATIC_DRAW */
-      1478, /* GL_STATIC_READ */
-      1474, /* GL_STATIC_COPY */
-       450, /* GL_DYNAMIC_DRAW */
-       452, /* GL_DYNAMIC_READ */
-       448, /* GL_DYNAMIC_COPY */
-      1120, /* GL_PIXEL_PACK_BUFFER */
-      1124, /* GL_PIXEL_UNPACK_BUFFER */
-      1121, /* GL_PIXEL_PACK_BUFFER_BINDING */
-      1125, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
+      1734, /* GL_TIME_ELAPSED_EXT */
+       801, /* GL_MATRIX0_ARB */
+       813, /* GL_MATRIX1_ARB */
+       825, /* GL_MATRIX2_ARB */
+       829, /* GL_MATRIX3_ARB */
+       831, /* GL_MATRIX4_ARB */
+       833, /* GL_MATRIX5_ARB */
+       835, /* GL_MATRIX6_ARB */
+       837, /* GL_MATRIX7_ARB */
+       839, /* GL_MATRIX8_ARB */
+       840, /* GL_MATRIX9_ARB */
+       803, /* GL_MATRIX10_ARB */
+       804, /* GL_MATRIX11_ARB */
+       805, /* GL_MATRIX12_ARB */
+       806, /* GL_MATRIX13_ARB */
+       807, /* GL_MATRIX14_ARB */
+       808, /* GL_MATRIX15_ARB */
+       809, /* GL_MATRIX16_ARB */
+       810, /* GL_MATRIX17_ARB */
+       811, /* GL_MATRIX18_ARB */
+       812, /* GL_MATRIX19_ARB */
+       815, /* GL_MATRIX20_ARB */
+       816, /* GL_MATRIX21_ARB */
+       817, /* GL_MATRIX22_ARB */
+       818, /* GL_MATRIX23_ARB */
+       819, /* GL_MATRIX24_ARB */
+       820, /* GL_MATRIX25_ARB */
+       821, /* GL_MATRIX26_ARB */
+       822, /* GL_MATRIX27_ARB */
+       823, /* GL_MATRIX28_ARB */
+       824, /* GL_MATRIX29_ARB */
+       827, /* GL_MATRIX30_ARB */
+       828, /* GL_MATRIX31_ARB */
+      1523, /* GL_STREAM_DRAW */
+      1525, /* GL_STREAM_READ */
+      1521, /* GL_STREAM_COPY */
+      1481, /* GL_STATIC_DRAW */
+      1483, /* GL_STATIC_READ */
+      1479, /* GL_STATIC_COPY */
+       451, /* GL_DYNAMIC_DRAW */
+       453, /* GL_DYNAMIC_READ */
+       449, /* GL_DYNAMIC_COPY */
+      1123, /* GL_PIXEL_PACK_BUFFER */
+      1127, /* GL_PIXEL_UNPACK_BUFFER */
+      1124, /* GL_PIXEL_PACK_BUFFER_BINDING */
+      1128, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
        347, /* GL_DEPTH24_STENCIL8 */
-      1719, /* GL_TEXTURE_STENCIL_SIZE */
-      1671, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
-       883, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
-       886, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
-       890, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
-       889, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
-       847, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
-      1509, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
+      1724, /* GL_TEXTURE_STENCIL_SIZE */
+      1676, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
+       886, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
+       889, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
+       893, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
+       892, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
+       850, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
+      1514, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
         17, /* GL_ACTIVE_STENCIL_FACE_EXT */
-       948, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
-      1379, /* GL_SAMPLES_PASSED */
+       951, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
+      1384, /* GL_SAMPLES_PASSED */
        109, /* GL_BUFFER_SERIALIZED_MODIFY_APPLE */
        104, /* GL_BUFFER_FLUSHING_UNMAP_APPLE */
-       532, /* GL_FRAGMENT_SHADER */
-      1834, /* GL_VERTEX_SHADER */
-      1237, /* GL_PROGRAM_OBJECT_ARB */
-      1411, /* GL_SHADER_OBJECT_ARB */
-       871, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
-       931, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
-       925, /* GL_MAX_VARYING_FLOATS */
-       929, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
-       856, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
-      1040, /* GL_OBJECT_TYPE_ARB */
-      1413, /* GL_SHADER_TYPE */
-       497, /* GL_FLOAT_VEC2 */
-       499, /* GL_FLOAT_VEC3 */
-       501, /* GL_FLOAT_VEC4 */
-       650, /* GL_INT_VEC2 */
-       652, /* GL_INT_VEC3 */
-       654, /* GL_INT_VEC4 */
+       534, /* GL_FRAGMENT_SHADER */
+      1839, /* GL_VERTEX_SHADER */
+      1240, /* GL_PROGRAM_OBJECT_ARB */
+      1416, /* GL_SHADER_OBJECT_ARB */
+       874, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
+       934, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
+       928, /* GL_MAX_VARYING_FLOATS */
+       932, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
+       859, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
+      1043, /* GL_OBJECT_TYPE_ARB */
+      1418, /* GL_SHADER_TYPE */
+       499, /* GL_FLOAT_VEC2 */
+       501, /* GL_FLOAT_VEC3 */
+       503, /* GL_FLOAT_VEC4 */
+       652, /* GL_INT_VEC2 */
+       654, /* GL_INT_VEC3 */
+       656, /* GL_INT_VEC4 */
         94, /* GL_BOOL */
         96, /* GL_BOOL_VEC2 */
         98, /* GL_BOOL_VEC3 */
        100, /* GL_BOOL_VEC4 */
-       485, /* GL_FLOAT_MAT2 */
-       489, /* GL_FLOAT_MAT3 */
-       493, /* GL_FLOAT_MAT4 */
-      1370, /* GL_SAMPLER_1D */
-      1372, /* GL_SAMPLER_2D */
-      1374, /* GL_SAMPLER_3D */
-      1375, /* GL_SAMPLER_CUBE */
-      1371, /* GL_SAMPLER_1D_SHADOW */
-      1373, /* GL_SAMPLER_2D_SHADOW */
-       487, /* GL_FLOAT_MAT2x3 */
-       488, /* GL_FLOAT_MAT2x4 */
-       491, /* GL_FLOAT_MAT3x2 */
-       492, /* GL_FLOAT_MAT3x4 */
-       495, /* GL_FLOAT_MAT4x2 */
-       496, /* GL_FLOAT_MAT4x3 */
+       487, /* GL_FLOAT_MAT2 */
+       491, /* GL_FLOAT_MAT3 */
+       495, /* GL_FLOAT_MAT4 */
+      1375, /* GL_SAMPLER_1D */
+      1377, /* GL_SAMPLER_2D */
+      1379, /* GL_SAMPLER_3D */
+      1380, /* GL_SAMPLER_CUBE */
+      1376, /* GL_SAMPLER_1D_SHADOW */
+      1378, /* GL_SAMPLER_2D_SHADOW */
+       489, /* GL_FLOAT_MAT2x3 */
+       490, /* GL_FLOAT_MAT2x4 */
+       493, /* GL_FLOAT_MAT3x2 */
+       494, /* GL_FLOAT_MAT3x4 */
+       497, /* GL_FLOAT_MAT4x2 */
+       498, /* GL_FLOAT_MAT4x3 */
        345, /* GL_DELETE_STATUS */
        246, /* GL_COMPILE_STATUS */
-       705, /* GL_LINK_STATUS */
-      1782, /* GL_VALIDATE_STATUS */
-       635, /* GL_INFO_LOG_LENGTH */
+       708, /* GL_LINK_STATUS */
+      1787, /* GL_VALIDATE_STATUS */
+       637, /* GL_INFO_LOG_LENGTH */
         56, /* GL_ATTACHED_SHADERS */
         20, /* GL_ACTIVE_UNIFORMS */
         21, /* GL_ACTIVE_UNIFORM_MAX_LENGTH */
-      1412, /* GL_SHADER_SOURCE_LENGTH */
+      1417, /* GL_SHADER_SOURCE_LENGTH */
         15, /* GL_ACTIVE_ATTRIBUTES */
         16, /* GL_ACTIVE_ATTRIBUTE_MAX_LENGTH */
-       534, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
-      1415, /* GL_SHADING_LANGUAGE_VERSION */
+       536, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
+      1420, /* GL_SHADING_LANGUAGE_VERSION */
        322, /* GL_CURRENT_PROGRAM */
-      1089, /* GL_PALETTE4_RGB8_OES */
-      1091, /* GL_PALETTE4_RGBA8_OES */
-      1087, /* GL_PALETTE4_R5_G6_B5_OES */
-      1090, /* GL_PALETTE4_RGBA4_OES */
-      1088, /* GL_PALETTE4_RGB5_A1_OES */
-      1094, /* GL_PALETTE8_RGB8_OES */
-      1096, /* GL_PALETTE8_RGBA8_OES */
-      1092, /* GL_PALETTE8_R5_G6_B5_OES */
-      1095, /* GL_PALETTE8_RGBA4_OES */
-      1093, /* GL_PALETTE8_RGB5_A1_OES */
-       617, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
-       616, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
-      1767, /* GL_UNSIGNED_NORMALIZED */
-      1606, /* GL_TEXTURE_1D_ARRAY_EXT */
-      1258, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
-      1608, /* GL_TEXTURE_2D_ARRAY_EXT */
-      1261, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
-      1614, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
-      1616, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
-      1468, /* GL_SRGB */
-      1469, /* GL_SRGB8 */
-      1471, /* GL_SRGB_ALPHA */
-      1470, /* GL_SRGB8_ALPHA8 */
-      1428, /* GL_SLUMINANCE_ALPHA */
-      1427, /* GL_SLUMINANCE8_ALPHA8 */
-      1425, /* GL_SLUMINANCE */
-      1426, /* GL_SLUMINANCE8 */
+      1092, /* GL_PALETTE4_RGB8_OES */
+      1094, /* GL_PALETTE4_RGBA8_OES */
+      1090, /* GL_PALETTE4_R5_G6_B5_OES */
+      1093, /* GL_PALETTE4_RGBA4_OES */
+      1091, /* GL_PALETTE4_RGB5_A1_OES */
+      1097, /* GL_PALETTE8_RGB8_OES */
+      1099, /* GL_PALETTE8_RGBA8_OES */
+      1095, /* GL_PALETTE8_R5_G6_B5_OES */
+      1098, /* GL_PALETTE8_RGBA4_OES */
+      1096, /* GL_PALETTE8_RGB5_A1_OES */
+       619, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
+       618, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
+      1772, /* GL_UNSIGNED_NORMALIZED */
+      1611, /* GL_TEXTURE_1D_ARRAY_EXT */
+      1262, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
+      1613, /* GL_TEXTURE_2D_ARRAY_EXT */
+      1265, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
+      1619, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
+      1621, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
+      1473, /* GL_SRGB */
+      1474, /* GL_SRGB8 */
+      1476, /* GL_SRGB_ALPHA */
+      1475, /* GL_SRGB8_ALPHA8 */
+      1433, /* GL_SLUMINANCE_ALPHA */
+      1432, /* GL_SLUMINANCE8_ALPHA8 */
+      1430, /* GL_SLUMINANCE */
+      1431, /* GL_SLUMINANCE8 */
        267, /* GL_COMPRESSED_SRGB */
        268, /* GL_COMPRESSED_SRGB_ALPHA */
        265, /* GL_COMPRESSED_SLUMINANCE */
        266, /* GL_COMPRESSED_SLUMINANCE_ALPHA */
-      1154, /* GL_POINT_SPRITE_COORD_ORIGIN */
-       713, /* GL_LOWER_LEFT */
-      1779, /* GL_UPPER_LEFT */
-      1491, /* GL_STENCIL_BACK_REF */
-      1492, /* GL_STENCIL_BACK_VALUE_MASK */
-      1493, /* GL_STENCIL_BACK_WRITEMASK */
-       441, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
-      1304, /* GL_RENDERBUFFER_BINDING_EXT */
-      1285, /* GL_READ_FRAMEBUFFER */
-       440, /* GL_DRAW_FRAMEBUFFER */
-      1286, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
-      1314, /* GL_RENDERBUFFER_SAMPLES */
-       544, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
-       542, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
-       553, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
-       549, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
-       551, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
-       556, /* GL_FRAMEBUFFER_COMPLETE */
-       560, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
-       566, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
-       564, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
-       562, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
-       565, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
-       563, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
-       569, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
-       572, /* GL_FRAMEBUFFER_UNSUPPORTED */
-       570, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
-       853, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
+      1157, /* GL_POINT_SPRITE_COORD_ORIGIN */
+       716, /* GL_LOWER_LEFT */
+      1784, /* GL_UPPER_LEFT */
+      1496, /* GL_STENCIL_BACK_REF */
+      1497, /* GL_STENCIL_BACK_VALUE_MASK */
+      1498, /* GL_STENCIL_BACK_WRITEMASK */
+       442, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
+      1309, /* GL_RENDERBUFFER_BINDING_EXT */
+      1290, /* GL_READ_FRAMEBUFFER */
+       441, /* GL_DRAW_FRAMEBUFFER */
+      1291, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
+      1319, /* GL_RENDERBUFFER_SAMPLES */
+       546, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
+       544, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
+       555, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
+       551, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
+       553, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
+       558, /* GL_FRAMEBUFFER_COMPLETE */
+       562, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
+       568, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
+       566, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
+       564, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
+       567, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
+       565, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
+       571, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
+       574, /* GL_FRAMEBUFFER_UNSUPPORTED */
+       572, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
+       856, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
        155, /* GL_COLOR_ATTACHMENT0 */
        157, /* GL_COLOR_ATTACHMENT1 */
        171, /* GL_COLOR_ATTACHMENT2 */
@@ -5055,56 +5065,56 @@ static const unsigned reduced_enums[1347] =
        166, /* GL_COLOR_ATTACHMENT14 */
        168, /* GL_COLOR_ATTACHMENT15 */
        348, /* GL_DEPTH_ATTACHMENT */
-      1481, /* GL_STENCIL_ATTACHMENT */
-       535, /* GL_FRAMEBUFFER */
-      1302, /* GL_RENDERBUFFER */
-      1316, /* GL_RENDERBUFFER_WIDTH */
-      1309, /* GL_RENDERBUFFER_HEIGHT */
-      1311, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
-      1504, /* GL_STENCIL_INDEX_EXT */
-      1501, /* GL_STENCIL_INDEX1_EXT */
-      1502, /* GL_STENCIL_INDEX4_EXT */
-      1503, /* GL_STENCIL_INDEX8_EXT */
-      1500, /* GL_STENCIL_INDEX16_EXT */
-      1313, /* GL_RENDERBUFFER_RED_SIZE */
-      1308, /* GL_RENDERBUFFER_GREEN_SIZE */
-      1305, /* GL_RENDERBUFFER_BLUE_SIZE */
-      1303, /* GL_RENDERBUFFER_ALPHA_SIZE */
-      1306, /* GL_RENDERBUFFER_DEPTH_SIZE */
-      1315, /* GL_RENDERBUFFER_STENCIL_SIZE */
-       568, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
-       909, /* GL_MAX_SAMPLES */
-      1272, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */
-       482, /* GL_FIRST_VERTEX_CONVENTION_EXT */
-       665, /* GL_LAST_VERTEX_CONVENTION_EXT */
-      1251, /* GL_PROVOKING_VERTEX_EXT */
+      1486, /* GL_STENCIL_ATTACHMENT */
+       537, /* GL_FRAMEBUFFER */
+      1307, /* GL_RENDERBUFFER */
+      1321, /* GL_RENDERBUFFER_WIDTH */
+      1314, /* GL_RENDERBUFFER_HEIGHT */
+      1316, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
+      1509, /* GL_STENCIL_INDEX_EXT */
+      1506, /* GL_STENCIL_INDEX1_EXT */
+      1507, /* GL_STENCIL_INDEX4_EXT */
+      1508, /* GL_STENCIL_INDEX8_EXT */
+      1505, /* GL_STENCIL_INDEX16_EXT */
+      1318, /* GL_RENDERBUFFER_RED_SIZE */
+      1313, /* GL_RENDERBUFFER_GREEN_SIZE */
+      1310, /* GL_RENDERBUFFER_BLUE_SIZE */
+      1308, /* GL_RENDERBUFFER_ALPHA_SIZE */
+      1311, /* GL_RENDERBUFFER_DEPTH_SIZE */
+      1320, /* GL_RENDERBUFFER_STENCIL_SIZE */
+       570, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
+       912, /* GL_MAX_SAMPLES */
+      1276, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
+       483, /* GL_FIRST_VERTEX_CONVENTION */
+       667, /* GL_LAST_VERTEX_CONVENTION */
+      1254, /* GL_PROVOKING_VERTEX */
        302, /* GL_COPY_READ_BUFFER */
        303, /* GL_COPY_WRITE_BUFFER */
-      1363, /* GL_RGBA_SNORM */
-      1359, /* GL_RGBA8_SNORM */
-      1421, /* GL_SIGNED_NORMALIZED */
-       910, /* GL_MAX_SERVER_WAIT_TIMEOUT */
-      1039, /* GL_OBJECT_TYPE */
-      1525, /* GL_SYNC_CONDITION */
-      1530, /* GL_SYNC_STATUS */
-      1527, /* GL_SYNC_FLAGS */
-      1526, /* GL_SYNC_FENCE */
-      1529, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
-      1756, /* GL_UNSIGNALED */
-      1420, /* GL_SIGNALED */
+      1368, /* GL_RGBA_SNORM */
+      1364, /* GL_RGBA8_SNORM */
+      1426, /* GL_SIGNED_NORMALIZED */
+       913, /* GL_MAX_SERVER_WAIT_TIMEOUT */
+      1042, /* GL_OBJECT_TYPE */
+      1530, /* GL_SYNC_CONDITION */
+      1535, /* GL_SYNC_STATUS */
+      1532, /* GL_SYNC_FLAGS */
+      1531, /* GL_SYNC_FENCE */
+      1534, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
+      1761, /* GL_UNSIGNALED */
+      1425, /* GL_SIGNALED */
         46, /* GL_ALREADY_SIGNALED */
-      1727, /* GL_TIMEOUT_EXPIRED */
+      1732, /* GL_TIMEOUT_EXPIRED */
        270, /* GL_CONDITION_SATISFIED */
-      1839, /* GL_WAIT_FAILED */
-       467, /* GL_EVAL_BIT */
-      1283, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
-       707, /* GL_LIST_BIT */
-      1622, /* GL_TEXTURE_BIT */
-      1394, /* GL_SCISSOR_BIT */
+      1844, /* GL_WAIT_FAILED */
+       468, /* GL_EVAL_BIT */
+      1288, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
+       710, /* GL_LIST_BIT */
+      1627, /* GL_TEXTURE_BIT */
+      1399, /* GL_SCISSOR_BIT */
         29, /* GL_ALL_ATTRIB_BITS */
-       995, /* GL_MULTISAMPLE_BIT */
+       998, /* GL_MULTISAMPLE_BIT */
         30, /* GL_ALL_CLIENT_ATTRIB_BITS */
-      1728, /* GL_TIMEOUT_IGNORED */
+      1733, /* GL_TIMEOUT_IGNORED */
 };
 
 typedef int (*cfunc)(const void *, const void *);
diff --git a/src/mesa/main/eval.c b/src/mesa/main/eval.c
index 3f89f9c1ea..95d6e23187 100644
--- a/src/mesa/main/eval.c
+++ b/src/mesa/main/eval.c
@@ -44,6 +44,10 @@
 #include "eval.h"
 #include "macros.h"
 #include "mtypes.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_evaluators
 
 
 /*
@@ -417,7 +421,7 @@ map1(GLenum target, GLfloat u1, GLfloat u2, GLint ustride,
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Map1f( GLenum target, GLfloat u1, GLfloat u2, GLint stride,
              GLint order, const GLfloat *points )
 {
@@ -425,7 +429,7 @@ _mesa_Map1f( GLenum target, GLfloat u1, GLfloat u2, GLint stride,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Map1d( GLenum target, GLdouble u1, GLdouble u2, GLint stride,
              GLint order, const GLdouble *points )
 {
@@ -516,7 +520,7 @@ map2( GLenum target, GLfloat u1, GLfloat u2, GLint ustride, GLint uorder,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Map2f( GLenum target,
              GLfloat u1, GLfloat u2, GLint ustride, GLint uorder,
              GLfloat v1, GLfloat v2, GLint vstride, GLint vorder,
@@ -527,7 +531,7 @@ _mesa_Map2f( GLenum target,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Map2d( GLenum target,
              GLdouble u1, GLdouble u2, GLint ustride, GLint uorder,
              GLdouble v1, GLdouble v2, GLint vstride, GLint vorder,
@@ -539,7 +543,7 @@ _mesa_Map2d( GLenum target,
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetMapdv( GLenum target, GLenum query, GLdouble *v )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -604,7 +608,7 @@ _mesa_GetMapdv( GLenum target, GLenum query, GLdouble *v )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetMapfv( GLenum target, GLenum query, GLfloat *v )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -669,7 +673,7 @@ _mesa_GetMapfv( GLenum target, GLenum query, GLfloat *v )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetMapiv( GLenum target, GLenum query, GLint *v )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -735,7 +739,7 @@ _mesa_GetMapiv( GLenum target, GLenum query, GLint *v )
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_MapGrid1f( GLint un, GLfloat u1, GLfloat u2 )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -753,14 +757,14 @@ _mesa_MapGrid1f( GLint un, GLfloat u1, GLfloat u2 )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_MapGrid1d( GLint un, GLdouble u1, GLdouble u2 )
 {
    _mesa_MapGrid1f( un, (GLfloat) u1, (GLfloat) u2 );
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_MapGrid2f( GLint un, GLfloat u1, GLfloat u2,
                  GLint vn, GLfloat v1, GLfloat v2 )
 {
@@ -788,7 +792,7 @@ _mesa_MapGrid2f( GLint un, GLfloat u1, GLfloat u2,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_MapGrid2d( GLint un, GLdouble u1, GLdouble u2,
                  GLint vn, GLdouble v1, GLdouble v2 )
 {
@@ -797,6 +801,41 @@ _mesa_MapGrid2d( GLint un, GLdouble u1, GLdouble u2,
 }
 
 
+void
+_mesa_install_eval_vtxfmt(struct _glapi_table *disp,
+                          const GLvertexformat *vfmt)
+{
+   SET_EvalCoord1f(disp, vfmt->EvalCoord1f);
+   SET_EvalCoord1fv(disp, vfmt->EvalCoord1fv);
+   SET_EvalCoord2f(disp, vfmt->EvalCoord2f);
+   SET_EvalCoord2fv(disp, vfmt->EvalCoord2fv);
+   SET_EvalPoint1(disp, vfmt->EvalPoint1);
+   SET_EvalPoint2(disp, vfmt->EvalPoint2);
+
+   SET_EvalMesh1(disp, vfmt->EvalMesh1);
+   SET_EvalMesh2(disp, vfmt->EvalMesh2);
+}
+
+
+void
+_mesa_init_eval_dispatch(struct _glapi_table *disp)
+{
+   SET_GetMapdv(disp, _mesa_GetMapdv);
+   SET_GetMapfv(disp, _mesa_GetMapfv);
+   SET_GetMapiv(disp, _mesa_GetMapiv);
+   SET_Map1d(disp, _mesa_Map1d);
+   SET_Map1f(disp, _mesa_Map1f);
+   SET_Map2d(disp, _mesa_Map2d);
+   SET_Map2f(disp, _mesa_Map2f);
+   SET_MapGrid1d(disp, _mesa_MapGrid1d);
+   SET_MapGrid1f(disp, _mesa_MapGrid1f);
+   SET_MapGrid2d(disp, _mesa_MapGrid2d);
+   SET_MapGrid2f(disp, _mesa_MapGrid2f);
+}
+
+
+#endif /* FEATURE_evaluators */
+
 
 /**********************************************************************/
 /*****                      Initialization                        *****/
diff --git a/src/mesa/main/eval.h b/src/mesa/main/eval.h
index b3ff0a96f8..ffd1bab76d 100644
--- a/src/mesa/main/eval.h
+++ b/src/mesa/main/eval.h
@@ -37,13 +37,22 @@
 #define EVAL_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
-#if _HAVE_FULL_GL
 
-extern void _mesa_init_eval( GLcontext *ctx );
-extern void _mesa_free_eval_data( GLcontext *ctx );
+#if FEATURE_evaluators
 
+#define _MESA_INIT_EVAL_VTXFMT(vfmt, impl)         \
+   do {                                            \
+      (vfmt)->EvalCoord1f  = impl ## EvalCoord1f;  \
+      (vfmt)->EvalCoord1fv = impl ## EvalCoord1fv; \
+      (vfmt)->EvalCoord2f  = impl ## EvalCoord2f;  \
+      (vfmt)->EvalCoord2fv = impl ## EvalCoord2fv; \
+      (vfmt)->EvalPoint1   = impl ## EvalPoint1;   \
+      (vfmt)->EvalPoint2   = impl ## EvalPoint2;   \
+      (vfmt)->EvalMesh1    = impl ## EvalMesh1;    \
+      (vfmt)->EvalMesh2    = impl ## EvalMesh2;    \
+   } while (0)
 
 extern GLuint _mesa_evaluator_components( GLenum target );
 
@@ -70,59 +79,32 @@ extern GLfloat *_mesa_copy_map_points2d(GLenum target,
                                      GLint vstride, GLint vorder,
                                      const GLdouble *points );
 
+extern void
+_mesa_install_eval_vtxfmt(struct _glapi_table *disp,
+                          const GLvertexformat *vfmt);
 
+extern void
+_mesa_init_eval_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY
-_mesa_Map1f( GLenum target, GLfloat u1, GLfloat u2, GLint stride,
-             GLint order, const GLfloat *points );
-
-extern void GLAPIENTRY
-_mesa_Map2f( GLenum target,
-             GLfloat u1, GLfloat u2, GLint ustride, GLint uorder,
-             GLfloat v1, GLfloat v2, GLint vstride, GLint vorder,
-             const GLfloat *points );
-
-extern void GLAPIENTRY
-_mesa_Map1d( GLenum target, GLdouble u1, GLdouble u2, GLint stride,
-             GLint order, const GLdouble *points );
-
-extern void GLAPIENTRY
-_mesa_Map2d( GLenum target,
-             GLdouble u1, GLdouble u2, GLint ustride, GLint uorder,
-             GLdouble v1, GLdouble v2, GLint vstride, GLint vorder,
-             const GLdouble *points );
-
-extern void GLAPIENTRY
-_mesa_MapGrid1f( GLint un, GLfloat u1, GLfloat u2 );
+#else /* FEATURE_evaluators */
 
-extern void GLAPIENTRY
-_mesa_MapGrid1d( GLint un, GLdouble u1, GLdouble u2 );
+#define _MESA_INIT_EVAL_VTXFMT(vfmt, impl) do { } while (0)
 
-extern void GLAPIENTRY
-_mesa_MapGrid2f( GLint un, GLfloat u1, GLfloat u2,
-                 GLint vn, GLfloat v1, GLfloat v2 );
+static INLINE void
+_mesa_install_eval_vtxfmt(struct _glapi_table *disp,
+                          const GLvertexformat *vfmt)
+{
+}
 
-extern void GLAPIENTRY
-_mesa_MapGrid2d( GLint un, GLdouble u1, GLdouble u2,
-                 GLint vn, GLdouble v1, GLdouble v2 );
+static INLINE void
+_mesa_init_eval_dispatch(struct _glapi_table *disp)
+{
+}
 
-extern void GLAPIENTRY
-_mesa_GetMapdv( GLenum target, GLenum query, GLdouble *v );
+#endif /* FEATURE_evaluators */
 
-extern void GLAPIENTRY
-_mesa_GetMapfv( GLenum target, GLenum query, GLfloat *v );
-
-extern void GLAPIENTRY
-_mesa_GetMapiv( GLenum target, GLenum query, GLint *v );
-
-#else
-
-/** No-op */
-#define _mesa_init_eval( c ) ((void)0)
-
-/** No-op */
-#define _mesa_free_eval_data( c ) ((void)0)
+extern void _mesa_init_eval( GLcontext *ctx );
+extern void _mesa_free_eval_data( GLcontext *ctx );
 
-#endif
 
-#endif
+#endif /* EVAL_H */
diff --git a/src/mesa/main/execmem.c b/src/mesa/main/execmem.c
index 57c1e117c8..4c6139985f 100644
--- a/src/mesa/main/execmem.c
+++ b/src/mesa/main/execmem.c
@@ -80,11 +80,10 @@ init_heap(void)
       exec_heap = mmInit( 0, EXEC_HEAP_SIZE );
    
    if (!exec_mem)
-      exec_mem = (unsigned char *) mmap(0, EXEC_HEAP_SIZE, 
-					PROT_EXEC | PROT_READ | PROT_WRITE, 
-					MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+      exec_mem = mmap(NULL, EXEC_HEAP_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE,
+		      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
 
-   return (exec_mem != NULL);
+   return (exec_mem != MAP_FAILED);
 }
 
 
diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c
index e3070b1547..54cf37c5f4 100644
--- a/src/mesa/main/extensions.c
+++ b/src/mesa/main/extensions.c
@@ -47,7 +47,9 @@ static const struct {
 } default_extensions[] = {
    { OFF, "GL_ARB_copy_buffer",                F(ARB_copy_buffer) },
    { OFF, "GL_ARB_depth_texture",              F(ARB_depth_texture) },
+   { OFF, "GL_ARB_depth_clamp",                F(ARB_depth_clamp) },
    { ON,  "GL_ARB_draw_buffers",               F(ARB_draw_buffers) },
+   { OFF, "GL_ARB_draw_elements_base_vertex",  F(ARB_draw_elements_base_vertex) },
    { OFF, "GL_ARB_fragment_program",           F(ARB_fragment_program) },
    { OFF, "GL_ARB_fragment_program_shadow",    F(ARB_fragment_program_shadow) },
    { OFF, "GL_ARB_fragment_shader",            F(ARB_fragment_shader) },
@@ -61,6 +63,7 @@ static const struct {
    { OFF, "GL_ARB_pixel_buffer_object",        F(EXT_pixel_buffer_object) },
    { OFF, "GL_ARB_point_parameters",           F(EXT_point_parameters) },
    { OFF, "GL_ARB_point_sprite",               F(ARB_point_sprite) },
+   { OFF, "GL_ARB_provoking_vertex",           F(EXT_provoking_vertex) },
    { OFF, "GL_ARB_seamless_cube_map",          F(ARB_seamless_cube_map) },
    { OFF, "GL_ARB_shader_objects",             F(ARB_shader_objects) },
    { OFF, "GL_ARB_shading_language_100",       F(ARB_shading_language_100) },
@@ -162,7 +165,9 @@ static const struct {
    { OFF, "GL_MESA_ycbcr_texture",             F(MESA_ycbcr_texture) },
    { ON,  "GL_MESA_window_pos",                F(ARB_window_pos) },
    { OFF, "GL_NV_blend_square",                F(NV_blend_square) },
+   { OFF, "GL_NV_depth_clamp",                 F(ARB_depth_clamp) },
    { OFF, "GL_NV_fragment_program",            F(NV_fragment_program) },
+   { OFF, "GL_NV_fragment_program_option",     F(NV_fragment_program_option) },
    { ON,  "GL_NV_light_max_exponent",          F(NV_light_max_exponent) },
    { OFF, "GL_NV_point_sprite",                F(NV_point_sprite) },
    { OFF, "GL_NV_texture_env_combine4",        F(NV_texture_env_combine4) },
@@ -192,8 +197,10 @@ void
 _mesa_enable_sw_extensions(GLcontext *ctx)
 {
    ctx->Extensions.ARB_copy_buffer = GL_TRUE;
+   ctx->Extensions.ARB_depth_clamp = GL_TRUE;
    ctx->Extensions.ARB_depth_texture = GL_TRUE;
    /*ctx->Extensions.ARB_draw_buffers = GL_TRUE;*/
+   ctx->Extensions.ARB_draw_elements_base_vertex = GL_TRUE;
 #if FEATURE_ARB_fragment_program
    ctx->Extensions.ARB_fragment_program = GL_TRUE;
    ctx->Extensions.ARB_fragment_program_shadow = GL_TRUE;
@@ -208,7 +215,7 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
    ctx->Extensions.ARB_imaging = GL_TRUE;
    ctx->Extensions.ARB_map_buffer_range = GL_TRUE;
    ctx->Extensions.ARB_multitexture = GL_TRUE;
-#if FEATURE_ARB_occlusion_query
+#if FEATURE_queryobj
    ctx->Extensions.ARB_occlusion_query = GL_TRUE;
 #endif
    ctx->Extensions.ARB_point_sprite = GL_TRUE;
@@ -309,6 +316,9 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
 #if FEATURE_NV_fragment_program
    ctx->Extensions.NV_fragment_program = GL_TRUE;
 #endif
+#if FEATURE_NV_fragment_program && FEATURE_ARB_fragment_program
+   ctx->Extensions.NV_fragment_program_option = GL_TRUE;
+#endif
    ctx->Extensions.SGI_color_matrix = GL_TRUE;
    ctx->Extensions.SGI_color_table = GL_TRUE;
    ctx->Extensions.SGI_texture_color_table = GL_TRUE;
diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c
index 825a23090b..87061aeb0d 100644
--- a/src/mesa/main/fbobject.c
+++ b/src/mesa/main/fbobject.c
@@ -44,6 +44,11 @@
 #include "teximage.h"
 #include "texobj.h"
 #include "texstore.h"
+#include "texstate.h"
+
+
+/** Set this to 1 to help debug FBO incompleteness problems */
+#define DEBUG_FBO 0
 
 
 /**
@@ -308,8 +313,8 @@ _mesa_framebuffer_renderbuffer(GLcontext *ctx, struct gl_framebuffer *fb,
 static void
 att_incomplete(const char *msg)
 {
-#if 0
-   _mesa_printf("attachment incomplete: %s\n", msg);
+#if DEBUG_FBO
+   _mesa_debug(NULL, "attachment incomplete: %s\n", msg);
 #else
    (void) msg;
 #endif
@@ -317,6 +322,23 @@ att_incomplete(const char *msg)
 
 
 /**
+ * For debug only.
+ */
+static void
+fbo_incomplete(const char *msg, int index)
+{
+#if DEBUG_FBO
+   _mesa_debug(NULL, "FBO Incomplete: %s [%d]\n", msg, index);
+#else
+   (void) msg;
+   (void) index;
+#endif
+}
+
+
+
+
+/**
  * Test if an attachment point is complete and update its Complete field.
  * \param format if GL_COLOR, this is a color attachment point,
  *               if GL_DEPTH, this is a depth component attachment point,
@@ -468,20 +490,6 @@ test_attachment_completeness(const GLcontext *ctx, GLenum format,
 
 
 /**
- * Helpful for debugging
- */
-static void
-fbo_incomplete(const char *msg, int index)
-{
-   (void) msg;
-   (void) index;
-   /*
-   _mesa_debug(NULL, "FBO Incomplete: %s [%d]\n", msg, index);
-   */
-}
-
-
-/**
  * Test if the given framebuffer object is complete and update its
  * Status field with the results.
  * Calls the ctx->Driver.ValidateFramebuffer() function to allow the
@@ -710,12 +718,6 @@ _mesa_BindRenderbufferEXT(GLenum target, GLuint renderbuffer)
    }
 
    FLUSH_CURRENT(ctx, _NEW_BUFFERS);
-   /* The above doesn't fully flush the drivers in the way that a
-    * glFlush does, but that is required here:
-    */
-   if (ctx->Driver.Flush)
-      ctx->Driver.Flush(ctx);
-
 
    if (renderbuffer) {
       newRb = _mesa_lookup_renderbuffer(ctx, renderbuffer);
@@ -1127,7 +1129,7 @@ check_begin_texture_render(GLcontext *ctx, struct gl_framebuffer *fb)
       struct gl_renderbuffer_attachment *att = fb->Attachment + i;
       struct gl_texture_object *texObj = att->Texture;
       if (texObj
-          && att->Texture->Image[att->CubeMapFace][att->TextureLevel]) {
+          && texObj->Image[att->CubeMapFace][att->TextureLevel]) {
          ctx->Driver.RenderTexture(ctx, fb, att);
       }
    }
@@ -1205,9 +1207,6 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
    }
 
    FLUSH_CURRENT(ctx, _NEW_BUFFERS);
-   if (ctx->Driver.Flush) {  
-      ctx->Driver.Flush(ctx);
-   }
 
    if (framebuffer) {
       /* Binding a user-created framebuffer object */
@@ -1286,11 +1285,6 @@ _mesa_DeleteFramebuffersEXT(GLsizei n, const GLuint *framebuffers)
 
    ASSERT_OUTSIDE_BEGIN_END(ctx);
    FLUSH_CURRENT(ctx, _NEW_BUFFERS);
-   /* The above doesn't fully flush the drivers in the way that a
-    * glFlush does, but that is required here:
-    */
-   if (ctx->Driver.Flush)
-      ctx->Driver.Flush(ctx);
 
    for (i = 0; i < n; i++) {
       if (framebuffers[i] > 0) {
@@ -1524,11 +1518,6 @@ framebuffer_texture(GLcontext *ctx, const char *caller, GLenum target,
    }
 
    FLUSH_CURRENT(ctx, _NEW_BUFFERS);
-   /* The above doesn't fully flush the drivers in the way that a
-    * glFlush does, but that is required here:
-    */
-   if (ctx->Driver.Flush)
-      ctx->Driver.Flush(ctx);
 
    _glthread_LOCK_MUTEX(fb->Mutex);
    if (texObj) {
@@ -1705,11 +1694,6 @@ _mesa_FramebufferRenderbufferEXT(GLenum target, GLenum attachment,
 
 
    FLUSH_CURRENT(ctx, _NEW_BUFFERS);
-   /* The above doesn't fully flush the drivers in the way that a
-    * glFlush does, but that is required here:
-    */
-   if (ctx->Driver.Flush)
-      ctx->Driver.Flush(ctx);
 
    assert(ctx->Driver.FramebufferRenderbuffer);
    ctx->Driver.FramebufferRenderbuffer(ctx, fb, attachment, rb);
@@ -1786,11 +1770,6 @@ _mesa_GetFramebufferAttachmentParameterivEXT(GLenum target, GLenum attachment,
    }
 
    FLUSH_CURRENT(ctx, _NEW_BUFFERS);
-   /* The above doesn't fully flush the drivers in the way that a
-    * glFlush does, but that is required here:
-    */
-   if (ctx->Driver.Flush)
-      ctx->Driver.Flush(ctx);
 
    switch (pname) {
    case GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT:
@@ -1948,18 +1927,18 @@ _mesa_GenerateMipmapEXT(GLenum target)
       return;
    }
 
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   texUnit = _mesa_get_current_tex_unit(ctx);
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
 
    _mesa_lock_texture(ctx, texObj);
    if (target == GL_TEXTURE_CUBE_MAP) {
-      int face;
-
+      GLuint face;
       for (face = 0; face < 6; face++)
 	 ctx->Driver.GenerateMipmap(ctx,
 				    GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB + face,
 				    texObj);
-   } else {
+   }
+   else {
       ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
    _mesa_unlock_texture(ctx, texObj);
diff --git a/src/mesa/main/feedback.c b/src/mesa/main/feedback.c
index 818a804540..fcdbb75fc4 100644
--- a/src/mesa/main/feedback.c
+++ b/src/mesa/main/feedback.c
@@ -36,9 +36,10 @@
 #include "feedback.h"
 #include "macros.h"
 #include "mtypes.h"
+#include "glapi/dispatch.h"
 
 
-#if _HAVE_FULL_GL
+#if FEATURE_feedback
 
 
 #define FB_3D		0x01
@@ -49,7 +50,7 @@
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_FeedbackBuffer( GLsizei size, GLenum type, GLfloat *buffer )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -103,7 +104,7 @@ _mesa_FeedbackBuffer( GLsizei size, GLenum type, GLfloat *buffer )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PassThrough( GLfloat token )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -153,9 +154,6 @@ _mesa_feedback_vertex(GLcontext *ctx,
 }
 
 
-#endif /* _HAVE_FULL_GL */
-
-
 /**********************************************************************/
 /** \name Selection */
 /*@{*/
@@ -173,7 +171,7 @@ _mesa_feedback_vertex(GLcontext *ctx,
  * Verifies we're not in selection mode, flushes the vertices and initialize
  * the fields in __GLcontextRec::Select with the given buffer.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_SelectBuffer( GLsizei size, GLuint *buffer )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -280,7 +278,7 @@ write_hit_record(GLcontext *ctx)
  * the hit record data in gl_selection. Marks new render mode in
  * __GLcontextRec::NewState.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_InitNames( void )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -311,7 +309,7 @@ _mesa_InitNames( void )
  *
  * sa __GLcontextRec::Select.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_LoadName( GLuint name )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -350,7 +348,7 @@ _mesa_LoadName( GLuint name )
  *
  * sa __GLcontextRec::Select.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PushName( GLuint name )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -381,7 +379,7 @@ _mesa_PushName( GLuint name )
  *
  * sa __GLcontextRec::Select.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PopName( void )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -424,7 +422,7 @@ _mesa_PopName( void )
  * __GLcontextRec::RenderMode and notifies the driver via the
  * dd_function_table::RenderMode callback.
  */
-GLint GLAPIENTRY
+static GLint GLAPIENTRY
 _mesa_RenderMode( GLenum mode )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -507,6 +505,23 @@ _mesa_RenderMode( GLenum mode )
 /*@}*/
 
 
+void
+_mesa_init_feedback_dispatch(struct _glapi_table *disp)
+{
+   SET_InitNames(disp, _mesa_InitNames);
+   SET_FeedbackBuffer(disp, _mesa_FeedbackBuffer);
+   SET_LoadName(disp, _mesa_LoadName);
+   SET_PassThrough(disp, _mesa_PassThrough);
+   SET_PopName(disp, _mesa_PopName);
+   SET_PushName(disp, _mesa_PushName);
+   SET_SelectBuffer(disp, _mesa_SelectBuffer);
+   SET_RenderMode(disp, _mesa_RenderMode);
+}
+
+
+#endif /* FEATURE_feedback */
+
+
 /**********************************************************************/
 /** \name Initialization */
 /*@{*/
diff --git a/src/mesa/main/feedback.h b/src/mesa/main/feedback.h
index 72c2acd5ed..7a648f444f 100644
--- a/src/mesa/main/feedback.h
+++ b/src/mesa/main/feedback.h
@@ -27,11 +27,15 @@
 #define FEEDBACK_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 
-extern void
-_mesa_init_feedback( GLcontext *ctx );
+#if FEATURE_feedback
+
+#define _MESA_INIT_FEEDBACK_FUNCTIONS(driver, impl) \
+   do {                                             \
+      (driver)->RenderMode = impl ## RenderMode;    \
+   } while (0)
 
 extern void
 _mesa_feedback_vertex( GLcontext *ctx,
@@ -55,29 +59,47 @@ extern void
 _mesa_update_hitflag( GLcontext *ctx, GLfloat z );
 
 
-extern void GLAPIENTRY
-_mesa_PassThrough( GLfloat token );
+extern void
+_mesa_init_feedback_dispatch(struct _glapi_table *disp);
+
+#else /* FEATURE_feedback */
 
-extern void GLAPIENTRY
-_mesa_FeedbackBuffer( GLsizei size, GLenum type, GLfloat *buffer );
+#define _MESA_INIT_FEEDBACK_FUNCTIONS(driver, impl) do { } while (0)
 
-extern void GLAPIENTRY
-_mesa_SelectBuffer( GLsizei size, GLuint *buffer );
+static INLINE void
+_mesa_feedback_vertex( GLcontext *ctx,
+                       const GLfloat win[4],
+                       const GLfloat color[4],
+                       GLfloat index,
+                       const GLfloat texcoord[4] )
+{
+   /* render mode is always GL_RENDER */
+   ASSERT_NO_FEATURE();
+}
 
-extern void GLAPIENTRY
-_mesa_InitNames( void );
 
-extern void GLAPIENTRY
-_mesa_LoadName( GLuint name );
+static INLINE void
+_mesa_feedback_token( GLcontext *ctx, GLfloat token )
+{
+   /* render mode is always GL_RENDER */
+   ASSERT_NO_FEATURE();
+}
 
-extern void GLAPIENTRY
-_mesa_PushName( GLuint name );
+static INLINE void
+_mesa_update_hitflag( GLcontext *ctx, GLfloat z )
+{
+   /* render mode is always GL_RENDER */
+   ASSERT_NO_FEATURE();
+}
 
-extern void GLAPIENTRY
-_mesa_PopName( void );
+static INLINE void
+_mesa_init_feedback_dispatch(struct _glapi_table *disp)
+{
+}
 
-extern GLint GLAPIENTRY
-_mesa_RenderMode( GLenum mode );
+#endif /* FEATURE_feedback */
 
+extern void
+_mesa_init_feedback( GLcontext *ctx );
 
-#endif
+#endif /* FEEDBACK_H */
diff --git a/src/mesa/main/get.c b/src/mesa/main/get.c
index 477ed01030..a6e004a816 100644
--- a/src/mesa/main/get.c
+++ b/src/mesa/main/get.c
@@ -1722,6 +1722,10 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params )
          params[0] = FLOAT_TO_BOOLEAN(ctx->Depth.BoundsMin);
          params[1] = FLOAT_TO_BOOLEAN(ctx->Depth.BoundsMax);
          break;
+      case GL_DEPTH_CLAMP:
+         CHECK_EXT1(ARB_depth_clamp, "GetBooleanv");
+         params[0] = ctx->Transform.DepthClamp;
+         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = INT_TO_BOOLEAN(ctx->Const.MaxDrawBuffers);
          break;
@@ -3553,6 +3557,10 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params )
          params[0] = ctx->Depth.BoundsMin;
          params[1] = ctx->Depth.BoundsMax;
          break;
+      case GL_DEPTH_CLAMP:
+         CHECK_EXT1(ARB_depth_clamp, "GetFloatv");
+         params[0] = BOOLEAN_TO_FLOAT(ctx->Transform.DepthClamp);
+         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = (GLfloat)(ctx->Const.MaxDrawBuffers);
          break;
@@ -5384,6 +5392,10 @@ _mesa_GetIntegerv( GLenum pname, GLint *params )
          params[0] = IROUND(ctx->Depth.BoundsMin);
          params[1] = IROUND(ctx->Depth.BoundsMax);
          break;
+      case GL_DEPTH_CLAMP:
+         CHECK_EXT1(ARB_depth_clamp, "GetIntegerv");
+         params[0] = BOOLEAN_TO_INT(ctx->Transform.DepthClamp);
+         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = ctx->Const.MaxDrawBuffers;
          break;
@@ -7216,6 +7228,10 @@ _mesa_GetInteger64v( GLenum pname, GLint64 *params )
          params[0] = IROUND64(ctx->Depth.BoundsMin);
          params[1] = IROUND64(ctx->Depth.BoundsMax);
          break;
+      case GL_DEPTH_CLAMP:
+         CHECK_EXT1(ARB_depth_clamp, "GetInteger64v");
+         params[0] = BOOLEAN_TO_INT64(ctx->Transform.DepthClamp);
+         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = ctx->Const.MaxDrawBuffers;
          break;
diff --git a/src/mesa/main/get_gen.py b/src/mesa/main/get_gen.py
index 2878c1b552..364d8c55c4 100644
--- a/src/mesa/main/get_gen.py
+++ b/src/mesa/main/get_gen.py
@@ -905,6 +905,10 @@ StateVars = [
 	  ["ctx->Depth.BoundsMin", "ctx->Depth.BoundsMax"],
 	  "", ["EXT_depth_bounds_test"] ),
 
+	# GL_ARB_depth_clamp
+	( "GL_DEPTH_CLAMP", GLboolean, ["ctx->Transform.DepthClamp"], "",
+	  ["ARB_depth_clamp"] ),
+
 	# GL_ARB_draw_buffers
 	( "GL_MAX_DRAW_BUFFERS_ARB", GLint,
 	  ["ctx->Const.MaxDrawBuffers"], "", None ),
diff --git a/src/mesa/main/histogram.c b/src/mesa/main/histogram.c
index ceb0d5a6a8..87816d3132 100644
--- a/src/mesa/main/histogram.c
+++ b/src/mesa/main/histogram.c
@@ -29,8 +29,11 @@
 #include "context.h"
 #include "image.h"
 #include "histogram.h"
+#include "glapi/dispatch.h"
 
 
+#if FEATURE_histogram
+
 
 /*
  * XXX the packed pixel formats haven't been tested.
@@ -614,7 +617,11 @@ base_histogram_format( GLenum format )
  */
 
 
-void GLAPIENTRY
+/* this is defined below */
+static void GLAPIENTRY _mesa_ResetMinmax(GLenum target);
+
+
+static void GLAPIENTRY
 _mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -677,7 +684,7 @@ _mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvo
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -737,7 +744,7 @@ _mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, G
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetHistogramParameterfv(GLenum target, GLenum pname, GLfloat *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -784,7 +791,7 @@ _mesa_GetHistogramParameterfv(GLenum target, GLenum pname, GLfloat *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetHistogramParameteriv(GLenum target, GLenum pname, GLint *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -831,7 +838,7 @@ _mesa_GetHistogramParameteriv(GLenum target, GLenum pname, GLint *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetMinmaxParameterfv(GLenum target, GLenum pname, GLfloat *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -857,7 +864,7 @@ _mesa_GetMinmaxParameterfv(GLenum target, GLenum pname, GLfloat *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetMinmaxParameteriv(GLenum target, GLenum pname, GLint *params)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -883,7 +890,7 @@ _mesa_GetMinmaxParameteriv(GLenum target, GLenum pname, GLint *params)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Histogram(GLenum target, GLsizei width, GLenum internalFormat, GLboolean sink)
 {
    GLuint i;
@@ -966,7 +973,7 @@ _mesa_Histogram(GLenum target, GLsizei width, GLenum internalFormat, GLboolean s
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_Minmax(GLenum target, GLenum internalFormat, GLboolean sink)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -994,7 +1001,7 @@ _mesa_Minmax(GLenum target, GLenum internalFormat, GLboolean sink)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ResetHistogram(GLenum target)
 {
    GLuint i;
@@ -1020,7 +1027,7 @@ _mesa_ResetHistogram(GLenum target)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_ResetMinmax(GLenum target)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -1043,6 +1050,24 @@ _mesa_ResetMinmax(GLenum target)
 }
 
 
+void
+_mesa_init_histogram_dispatch(struct _glapi_table *disp)
+{
+   SET_GetHistogram(disp, _mesa_GetHistogram);
+   SET_GetHistogramParameterfv(disp, _mesa_GetHistogramParameterfv);
+   SET_GetHistogramParameteriv(disp, _mesa_GetHistogramParameteriv);
+   SET_GetMinmax(disp, _mesa_GetMinmax);
+   SET_GetMinmaxParameterfv(disp, _mesa_GetMinmaxParameterfv);
+   SET_GetMinmaxParameteriv(disp, _mesa_GetMinmaxParameteriv);
+   SET_Histogram(disp, _mesa_Histogram);
+   SET_Minmax(disp, _mesa_Minmax);
+   SET_ResetHistogram(disp, _mesa_ResetHistogram);
+   SET_ResetMinmax(disp, _mesa_ResetMinmax);
+}
+
+
+#endif /* FEATURE_histogram */
+
 
 /**********************************************************************/
 /*****                      Initialization                        *****/
diff --git a/src/mesa/main/histogram.h b/src/mesa/main/histogram.h
index 367e9b11ba..dbae1bbd06 100644
--- a/src/mesa/main/histogram.h
+++ b/src/mesa/main/histogram.h
@@ -36,48 +36,22 @@
 #ifndef HISTOGRAM_H
 #define HISTOGRAM_H
 
-#include "glheader.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 
-#if _HAVE_FULL_GL
+#if FEATURE_histogram
 
-extern void GLAPIENTRY 
-_mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum types, GLvoid *values);
+extern void
+_mesa_init_histogram_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY 
-_mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values);
+#else /* FEATURE_histogram */
 
-extern void GLAPIENTRY 
-_mesa_GetHistogramParameterfv(GLenum target, GLenum pname, GLfloat *params);
+static INLINE void
+_mesa_init_histogram_dispatch(struct _glapi_table *disp)
+{
+}
 
-extern void GLAPIENTRY 
-_mesa_GetHistogramParameteriv(GLenum target, GLenum pname, GLint *params);
-
-extern void GLAPIENTRY 
-_mesa_GetMinmaxParameterfv(GLenum target, GLenum pname, GLfloat *params);
-
-extern void GLAPIENTRY 
-_mesa_GetMinmaxParameteriv(GLenum target, GLenum pname, GLint *params);
-
-extern void GLAPIENTRY 
-_mesa_Histogram(GLenum target, GLsizei width, GLenum internalformat, GLboolean sink);
-
-extern void GLAPIENTRY 
-_mesa_Minmax(GLenum target, GLenum internalformat, GLboolean sink);
-
-extern void GLAPIENTRY 
-_mesa_ResetHistogram(GLenum target);
-
-extern void GLAPIENTRY 
-_mesa_ResetMinmax(GLenum target);
+#endif /* FEATURE_histogram */
 
 extern void _mesa_init_histogram( GLcontext * ctx );
 
-#else
-
-/** No-op */
-#define _mesa_init_histogram( c ) ((void) 0)
-
-#endif
-
-#endif
+#endif /* HISTOGRAM_H */
diff --git a/src/mesa/main/light.c b/src/mesa/main/light.c
index 10c89f4368..1c8a081e9a 100644
--- a/src/mesa/main/light.c
+++ b/src/mesa/main/light.c
@@ -636,7 +636,7 @@ _mesa_update_material( GLcontext *ctx, GLuint bitmask )
    struct gl_light *light, *list = &ctx->Light.EnabledList;
    GLfloat (*mat)[4] = ctx->Light.Material.Attrib;
 
-   if (MESA_VERBOSE&VERBOSE_IMMEDIATE) 
+   if (MESA_VERBOSE & VERBOSE_MATERIAL) 
       _mesa_debug(ctx, "_mesa_update_material, mask 0x%x\n", bitmask);
 
    if (!bitmask)
diff --git a/src/mesa/main/mfeatures.h b/src/mesa/main/mfeatures.h
index e23cdb1f42..c2fb8404b1 100644
--- a/src/mesa/main/mfeatures.h
+++ b/src/mesa/main/mfeatures.h
@@ -36,12 +36,47 @@
 #define _HAVE_FULL_GL 1
 #endif
 
+/* assert that a feature is disabled and should never be used */
+#define ASSERT_NO_FEATURE() ASSERT(0)
+
+/**
+ * A feature can be anything.  But most of them share certain characteristics.
+ *
+ * When a feature defines driver entries, they can be initialized by
+ *   _MESA_INIT_<FEATURE>_FUNCTIONS
+ *
+ * When a feature defines vtxfmt entries, they can be initialized and
+ * installed by
+ *   _MESA_INIT_<FEATURE>_VTXFMT
+ *   _mesa_install_<feature>_vtxfmt
+ *
+ * When a feature defines dispatch entries, they are initialized by
+ *   _mesa_init_<feature>_dispatch
+ *
+ * When a feature has states, they are initialized and freed by
+ *   _mesa_init_<feature>
+ *   _mesa_free_<feature>_data
+ *
+ * Except for states, the others compile to no-op when a feature is disabled.
+ *
+ * The GLAPIENTRYs and helper functions defined by a feature should also
+ * compile to no-op when it is disabled.  But to save typings and to catch
+ * bugs, some of them may be unavailable, or compile to ASSERT_NO_FEATURE()
+ * when the feature is disabled.
+ *
+ * A feature following the conventions may be used without knowing if it is
+ * enabled or not.
+ */
+
 #define FEATURE_accum  _HAVE_FULL_GL
+#define FEATURE_arrayelt  _HAVE_FULL_GL
 #define FEATURE_attrib_stack  _HAVE_FULL_GL
+/* this disables vtxfmt, api_loopback, and api_noop completely */
+#define FEATURE_beginend  _HAVE_FULL_GL
 #define FEATURE_colortable  _HAVE_FULL_GL
 #define FEATURE_convolve  _HAVE_FULL_GL
 #define FEATURE_dispatch  _HAVE_FULL_GL
-#define FEATURE_dlist  _HAVE_FULL_GL
+#define FEATURE_dlist  (_HAVE_FULL_GL && FEATURE_arrayelt && FEATURE_beginend)
 #define FEATURE_draw_read_buffer  _HAVE_FULL_GL
 #define FEATURE_drawpix  _HAVE_FULL_GL
 #define FEATURE_evaluators  _HAVE_FULL_GL
@@ -50,15 +85,15 @@
 #define FEATURE_histogram  _HAVE_FULL_GL
 #define FEATURE_pixel_transfer  _HAVE_FULL_GL
 #define FEATURE_point_size_array 0
+#define FEATURE_queryobj  _HAVE_FULL_GL
+#define FEATURE_rastpos  _HAVE_FULL_GL
 #define FEATURE_texgen  _HAVE_FULL_GL
 #define FEATURE_texture_fxt1  _HAVE_FULL_GL
 #define FEATURE_texture_s3tc  _HAVE_FULL_GL
 #define FEATURE_userclip  _HAVE_FULL_GL
 #define FEATURE_vertex_array_byte 0
-#define FEATURE_windowpos  _HAVE_FULL_GL
 #define FEATURE_es2_glsl 0
 
-#define FEATURE_ARB_occlusion_query  _HAVE_FULL_GL
 #define FEATURE_ARB_fragment_program  _HAVE_FULL_GL
 #define FEATURE_ARB_framebuffer_object  _HAVE_FULL_GL
 #define FEATURE_ARB_map_buffer_range  _HAVE_FULL_GL
@@ -76,7 +111,6 @@
 #define FEATURE_EXT_framebuffer_object _HAVE_FULL_GL
 #define FEATURE_EXT_pixel_buffer_object  _HAVE_FULL_GL
 #define FEATURE_EXT_texture_sRGB _HAVE_FULL_GL
-#define FEATURE_EXT_timer_query  _HAVE_FULL_GL
 #define FEATURE_ATI_fragment_shader _HAVE_FULL_GL
 #define FEATURE_NV_fence  _HAVE_FULL_GL
 #define FEATURE_NV_fragment_program  _HAVE_FULL_GL
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index f8e4e41583..052fbaaaea 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.5
+ * Version:  7.7
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
@@ -84,6 +84,7 @@
 /*@{*/
 struct _mesa_HashTable;
 struct gl_attrib_node;
+struct gl_list_extensions;
 struct gl_meta_state;
 struct gl_pixelstore_attrib;
 struct gl_program_cache;
@@ -207,7 +208,7 @@ typedef enum
    VERT_RESULT_BFC0 = 13,
    VERT_RESULT_BFC1 = 14,
    VERT_RESULT_EDGE = 15,
-   VERT_RESULT_VAR0 = 16  /**< shader varying */,
+   VERT_RESULT_VAR0 = 16,  /**< shader varying */
    VERT_RESULT_MAX = (VERT_RESULT_VAR0 + MAX_VARYING)
 } gl_vert_result;
 
@@ -820,29 +821,6 @@ struct gl_list_attrib
 
 
 /**
- * Used by device drivers to hook new commands into display lists.
- */
-struct gl_list_instruction
-{
-   GLuint Size;
-   void (*Execute)( GLcontext *ctx, void *data );
-   void (*Destroy)( GLcontext *ctx, void *data );
-   void (*Print)( GLcontext *ctx, void *data );
-};
-
-#define MAX_DLIST_EXT_OPCODES 16
-
-/**
- * Used by device drivers to hook new commands into display lists.
- */
-struct gl_list_extensions
-{
-   struct gl_list_instruction Opcode[MAX_DLIST_EXT_OPCODES];
-   GLuint NumOpcodes;
-};
-
-
-/**
  * Multisample attribute group (GL_MULTISAMPLE_BIT).
  */
 struct gl_multisample_attrib
@@ -1482,6 +1460,7 @@ struct gl_transform_attrib
    GLboolean Normalize;				/**< Normalize all normals? */
    GLboolean RescaleNormals;			/**< GL_EXT_rescale_normal */
    GLboolean RasterPositionUnclipped;           /**< GL_IBM_rasterpos_clip */
+   GLboolean DepthClamp;			/**< GL_ARB_depth_clamp */
 
    GLboolean CullVertexFlag;	/**< True if GL_CULL_VERTEX_EXT is enabled */
    GLfloat CullEyePos[4];
@@ -1833,7 +1812,6 @@ struct gl_vertex_program
    struct gl_program Base;   /**< base class */
    GLboolean IsNVProgram;    /**< is this a GL_NV_vertex_program program? */
    GLboolean IsPositionInvariant;
-   void *TnlData;		/**< should probably use Base.DriverData */
 };
 
 
@@ -1939,10 +1917,10 @@ struct ati_fragment_shader
    struct atifs_instruction *Instructions[2];
    struct atifs_setupinst *SetupInst[2];
    GLfloat Constants[8][4];
-   GLbitfield LocalConstDef;  /** Indicates which constants have been set */
+   GLbitfield LocalConstDef;  /**< Indicates which constants have been set */
    GLubyte numArithInstr[2];
    GLubyte regsAssigned[2];
-   GLubyte NumPasses;         /** 1 or 2 */
+   GLubyte NumPasses;         /**< 1 or 2 */
    GLubyte cur_pass;
    GLubyte last_optype;
    GLboolean interpinp1;
@@ -1956,7 +1934,7 @@ struct ati_fragment_shader
 struct gl_ati_fragment_shader_state
 {
    GLboolean Enabled;
-   GLboolean _Enabled;                      /** enabled and valid shader? */
+   GLboolean _Enabled;                  /**< enabled and valid shader? */
    GLboolean Compiling;
    GLfloat GlobalConstants[8][4];
    struct ati_fragment_shader *Current;
@@ -2066,6 +2044,9 @@ struct gl_shader_program
 #define GLSL_OPT       0x4  /**< Force optimizations (override pragmas) */
 #define GLSL_NO_OPT    0x8  /**< Force no optimizations (override pragmas) */
 #define GLSL_UNIFORMS 0x10  /**< Print glUniform calls */
+#define GLSL_NOP_VERT 0x20  /**< Force no-op vertex shaders */
+#define GLSL_NOP_FRAG 0x40  /**< Force no-op fragment shaders */
+#define GLSL_USE_PROG 0x80  /**< Log glUseProgram calls */
 
 
 /**
@@ -2476,7 +2457,9 @@ struct gl_extensions
    GLboolean dummy;  /* don't remove this! */
    GLboolean ARB_copy_buffer;
    GLboolean ARB_depth_texture;
+   GLboolean ARB_depth_clamp;
    GLboolean ARB_draw_buffers;
+   GLboolean ARB_draw_elements_base_vertex;
    GLboolean ARB_fragment_program;
    GLboolean ARB_fragment_program_shadow;
    GLboolean ARB_fragment_shader;
@@ -2581,6 +2564,7 @@ struct gl_extensions
    GLboolean MESA_texture_signed_rgba;
    GLboolean NV_blend_square;
    GLboolean NV_fragment_program;
+   GLboolean NV_fragment_program_option;
    GLboolean NV_light_max_exponent;
    GLboolean NV_point_sprite;
    GLboolean NV_texgen_reflection;
@@ -3056,7 +3040,7 @@ struct __GLcontextRec
    struct gl_shine_tab *_ShineTabList;  /**< MRU list of inactive shine tables */
    /**@}*/
 
-   struct gl_list_extensions ListExt; /**< driver dlist extensions */
+   struct gl_list_extensions *ListExt; /**< driver dlist extensions */
 
    /** \name For debugging/development only */
    /*@{*/
@@ -3112,11 +3096,12 @@ extern int MESA_DEBUG_FLAGS;
 #endif
 
 
+/** The MESA_VERBOSE var is a bitmask of these flags */
 enum _verbose
 {
    VERBOSE_VARRAY		= 0x0001,
    VERBOSE_TEXTURE		= 0x0002,
-   VERBOSE_IMMEDIATE		= 0x0004,
+   VERBOSE_MATERIAL		= 0x0004,
    VERBOSE_PIPELINE		= 0x0008,
    VERBOSE_DRIVER		= 0x0010,
    VERBOSE_STATE		= 0x0020,
@@ -3126,9 +3111,12 @@ enum _verbose
    VERBOSE_PRIMS		= 0x0400,
    VERBOSE_VERTS		= 0x0800,
    VERBOSE_DISASSEM		= 0x1000,
+   VERBOSE_DRAW                 = 0x2000,
+   VERBOSE_SWAPBUFFERS          = 0x4000
 };
 
 
+/** The MESA_DEBUG_FLAGS var is a bitmask of these flags */
 enum _debug
 {
    DEBUG_ALWAYS_FLUSH		= 0x1
diff --git a/src/mesa/main/pixel.c b/src/mesa/main/pixel.c
index fcef6dfd42..3820ebd889 100644
--- a/src/mesa/main/pixel.c
+++ b/src/mesa/main/pixel.c
@@ -36,13 +36,17 @@
 #include "macros.h"
 #include "pixel.h"
 #include "mtypes.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_pixel_transfer
 
 
 /**********************************************************************/
 /*****                    glPixelZoom                             *****/
 /**********************************************************************/
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PixelZoom( GLfloat xfactor, GLfloat yfactor )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -163,7 +167,7 @@ validate_pbo_access(GLcontext *ctx, struct gl_pixelstore_attrib *pack,
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -205,7 +209,7 @@ _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values )
 {
    GLfloat fvalues[MAX_PIXEL_MAP_TABLE];
@@ -261,7 +265,7 @@ _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values )
 {
    GLfloat fvalues[MAX_PIXEL_MAP_TABLE];
@@ -317,7 +321,7 @@ _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetPixelMapfv( GLenum map, GLfloat *values )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -362,7 +366,7 @@ _mesa_GetPixelMapfv( GLenum map, GLfloat *values )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetPixelMapuiv( GLenum map, GLuint *values )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -406,7 +410,7 @@ _mesa_GetPixelMapuiv( GLenum map, GLuint *values )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetPixelMapusv( GLenum map, GLushort *values )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -468,7 +472,7 @@ _mesa_GetPixelMapusv( GLenum map, GLushort *values )
  * Implements glPixelTransfer[fi] whether called immediately or from a
  * display list.
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PixelTransferf( GLenum pname, GLfloat param )
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -662,7 +666,7 @@ _mesa_PixelTransferf( GLenum pname, GLfloat param )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_PixelTransferi( GLenum pname, GLint param )
 {
    _mesa_PixelTransferf( pname, (GLfloat) param );
@@ -756,6 +760,24 @@ void _mesa_update_pixel( GLcontext *ctx, GLuint new_state )
 }
 
 
+void
+_mesa_init_pixel_dispatch(struct _glapi_table *disp)
+{
+   SET_GetPixelMapfv(disp, _mesa_GetPixelMapfv);
+   SET_GetPixelMapuiv(disp, _mesa_GetPixelMapuiv);
+   SET_GetPixelMapusv(disp, _mesa_GetPixelMapusv);
+   SET_PixelMapfv(disp, _mesa_PixelMapfv);
+   SET_PixelMapuiv(disp, _mesa_PixelMapuiv);
+   SET_PixelMapusv(disp, _mesa_PixelMapusv);
+   SET_PixelTransferf(disp, _mesa_PixelTransferf);
+   SET_PixelTransferi(disp, _mesa_PixelTransferi);
+   SET_PixelZoom(disp, _mesa_PixelZoom);
+}
+
+
+#endif /* FEATURE_pixel_transfer */
+
+
 /**********************************************************************/
 /*****                      Initialization                        *****/
 /**********************************************************************/
diff --git a/src/mesa/main/pixel.h b/src/mesa/main/pixel.h
index cb6c5262a3..f4d3f1efdb 100644
--- a/src/mesa/main/pixel.h
+++ b/src/mesa/main/pixel.h
@@ -33,48 +33,35 @@
 #define PIXEL_H
 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 
-/** \name API functions */
-/*@{*/
+#if FEATURE_pixel_transfer
 
-extern void GLAPIENTRY
-_mesa_GetPixelMapfv( GLenum map, GLfloat *values );
-
-extern void GLAPIENTRY
-_mesa_GetPixelMapuiv( GLenum map, GLuint *values );
-
-extern void GLAPIENTRY
-_mesa_GetPixelMapusv( GLenum map, GLushort *values );
-
-extern void GLAPIENTRY
-_mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values );
-
-extern void GLAPIENTRY
-_mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values );
-
-extern void GLAPIENTRY
-_mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values );
+extern void 
+_mesa_update_pixel( GLcontext *ctx, GLuint newstate );
 
-extern void GLAPIENTRY
-_mesa_PixelTransferf( GLenum pname, GLfloat param );
+extern void
+_mesa_init_pixel_dispatch( struct _glapi_table * disp );
 
-extern void GLAPIENTRY
-_mesa_PixelTransferi( GLenum pname, GLint param );
+#else /* FEATURE_pixel_transfer */
 
-extern void GLAPIENTRY
-_mesa_PixelZoom( GLfloat xfactor, GLfloat yfactor );
+static INLINE void
+_mesa_update_pixel(GLcontext *ctx, GLuint newstate)
+{
+}
 
-/*@}*/
+static INLINE void
+_mesa_init_pixel_dispatch(struct _glapi_table *disp)
+{
+}
 
+#endif /* FEATURE_pixel_transfer */
 
-extern void 
-_mesa_update_pixel( GLcontext *ctx, GLuint newstate );
 
 extern void 
 _mesa_init_pixel( GLcontext * ctx );
 
 /*@}*/
 
-#endif
+#endif /* PIXEL_H */
diff --git a/src/mesa/main/queryobj.c b/src/mesa/main/queryobj.c
index a73c6e0508..f6eb4ee7e1 100644
--- a/src/mesa/main/queryobj.c
+++ b/src/mesa/main/queryobj.c
@@ -29,6 +29,10 @@
 #include "imports.h"
 #include "queryobj.h"
 #include "mtypes.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_queryobj
 
 
 /**
@@ -216,7 +220,7 @@ _mesa_IsQueryARB(GLuint id)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_BeginQueryARB(GLenum target, GLuint id)
 {
    struct gl_query_object *q;
@@ -236,7 +240,6 @@ _mesa_BeginQueryARB(GLenum target, GLuint id)
             return;
          }
          break;
-#if FEATURE_EXT_timer_query
       case GL_TIME_ELAPSED_EXT:
          if (!ctx->Extensions.EXT_timer_query) {
             _mesa_error(ctx, GL_INVALID_ENUM, "glBeginQueryARB(target)");
@@ -247,7 +250,6 @@ _mesa_BeginQueryARB(GLenum target, GLuint id)
             return;
          }
          break;
-#endif
       default:
          _mesa_error(ctx, GL_INVALID_ENUM, "glBeginQueryARB(target)");
          return;
@@ -285,17 +287,15 @@ _mesa_BeginQueryARB(GLenum target, GLuint id)
    if (target == GL_SAMPLES_PASSED_ARB) {
       ctx->Query.CurrentOcclusionObject = q;
    }
-#if FEATURE_EXT_timer_query
    else if (target == GL_TIME_ELAPSED_EXT) {
       ctx->Query.CurrentTimerObject = q;
    }
-#endif
 
    ctx->Driver.BeginQuery(ctx, q);
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_EndQueryARB(GLenum target)
 {
    struct gl_query_object *q;
@@ -313,7 +313,6 @@ _mesa_EndQueryARB(GLenum target)
          q = ctx->Query.CurrentOcclusionObject;
          ctx->Query.CurrentOcclusionObject = NULL;
          break;
-#if FEATURE_EXT_timer_query
       case GL_TIME_ELAPSED_EXT:
          if (!ctx->Extensions.EXT_timer_query) {
             _mesa_error(ctx, GL_INVALID_ENUM, "glEndQueryARB(target)");
@@ -322,7 +321,6 @@ _mesa_EndQueryARB(GLenum target)
          q = ctx->Query.CurrentTimerObject;
          ctx->Query.CurrentTimerObject = NULL;
          break;
-#endif
       default:
          _mesa_error(ctx, GL_INVALID_ENUM, "glEndQueryARB(target)");
          return;
@@ -354,7 +352,6 @@ _mesa_GetQueryivARB(GLenum target, GLenum pname, GLint *params)
          }
          q = ctx->Query.CurrentOcclusionObject;
          break;
-#if FEATURE_EXT_timer_query
       case GL_TIME_ELAPSED_EXT:
          if (!ctx->Extensions.EXT_timer_query) {
             _mesa_error(ctx, GL_INVALID_ENUM, "glEndQueryARB(target)");
@@ -362,7 +359,6 @@ _mesa_GetQueryivARB(GLenum target, GLenum pname, GLint *params)
          }
          q = ctx->Query.CurrentTimerObject;
          break;
-#endif
       default:
          _mesa_error(ctx, GL_INVALID_ENUM, "glGetQueryivARB(target)");
          return;
@@ -462,12 +458,10 @@ _mesa_GetQueryObjectuivARB(GLuint id, GLenum pname, GLuint *params)
 }
 
 
-#if FEATURE_EXT_timer_query
-
 /**
  * New with GL_EXT_timer_query
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetQueryObjecti64vEXT(GLuint id, GLenum pname, GLint64EXT *params)
 {
    struct gl_query_object *q = NULL;
@@ -504,7 +498,7 @@ _mesa_GetQueryObjecti64vEXT(GLuint id, GLenum pname, GLint64EXT *params)
 /**
  * New with GL_EXT_timer_query
  */
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetQueryObjectui64vEXT(GLuint id, GLenum pname, GLuint64EXT *params)
 {
    struct gl_query_object *q = NULL;
@@ -537,19 +531,35 @@ _mesa_GetQueryObjectui64vEXT(GLuint id, GLenum pname, GLuint64EXT *params)
    }
 }
 
-#endif /* FEATURE_EXT_timer_query */
+
+void
+_mesa_init_queryobj_dispatch(struct _glapi_table *disp)
+{
+   SET_GenQueriesARB(disp, _mesa_GenQueriesARB);
+   SET_DeleteQueriesARB(disp, _mesa_DeleteQueriesARB);
+   SET_IsQueryARB(disp, _mesa_IsQueryARB);
+   SET_BeginQueryARB(disp, _mesa_BeginQueryARB);
+   SET_EndQueryARB(disp, _mesa_EndQueryARB);
+   SET_GetQueryivARB(disp, _mesa_GetQueryivARB);
+   SET_GetQueryObjectivARB(disp, _mesa_GetQueryObjectivARB);
+   SET_GetQueryObjectuivARB(disp, _mesa_GetQueryObjectuivARB);
+
+   SET_GetQueryObjecti64vEXT(disp, _mesa_GetQueryObjecti64vEXT);
+   SET_GetQueryObjectui64vEXT(disp, _mesa_GetQueryObjectui64vEXT);
+}
+
+
+#endif /* FEATURE_queryobj */
 
 
 /**
  * Allocate/init the context state related to query objects.
  */
 void
-_mesa_init_query(GLcontext *ctx)
+_mesa_init_queryobj(GLcontext *ctx)
 {
-#if FEATURE_ARB_occlusion_query
    ctx->Query.QueryObjects = _mesa_NewHashTable();
    ctx->Query.CurrentOcclusionObject = NULL;
-#endif
 }
 
 
@@ -569,7 +579,7 @@ delete_queryobj_cb(GLuint id, void *data, void *userData)
  * Free the context state related to query objects.
  */
 void
-_mesa_free_query_data(GLcontext *ctx)
+_mesa_free_queryobj_data(GLcontext *ctx)
 {
    _mesa_HashDeleteAll(ctx->Query.QueryObjects, delete_queryobj_cb, ctx);
    _mesa_DeleteHashTable(ctx->Query.QueryObjects);
diff --git a/src/mesa/main/queryobj.h b/src/mesa/main/queryobj.h
index ee775ef959..6cf3c76d74 100644
--- a/src/mesa/main/queryobj.h
+++ b/src/mesa/main/queryobj.h
@@ -23,19 +23,24 @@
  */
 
 
-#ifndef OCCLUDE_H
-#define OCCLUDE_H
+#ifndef QUERYOBJ_H
+#define QUERYOBJ_H
 
 
-extern void
-_mesa_init_query(GLcontext *ctx);
+#include "main/mtypes.h"
 
-extern void
-_mesa_free_query_data(GLcontext *ctx);
 
-extern void
-_mesa_init_query_object_functions(struct dd_function_table *driver);
+#if FEATURE_queryobj
 
+#define _MESA_INIT_QUERYOBJ_FUNCTIONS(driver, impl)      \
+   do {                                                  \
+      (driver)->NewQueryObject = impl ## NewQueryObject; \
+      (driver)->DeleteQuery    = impl ## DeleteQuery;    \
+      (driver)->BeginQuery     = impl ## BeginQuery;     \
+      (driver)->EndQuery       = impl ## EndQuery;       \
+      (driver)->WaitQuery      = impl ## WaitQuery;      \
+      (driver)->CheckQuery     = impl ## CheckQuery;     \
+   } while (0)
 
 extern void GLAPIENTRY
 _mesa_GenQueriesARB(GLsizei n, GLuint *ids);
@@ -47,12 +52,6 @@ extern GLboolean GLAPIENTRY
 _mesa_IsQueryARB(GLuint id);
 
 extern void GLAPIENTRY
-_mesa_BeginQueryARB(GLenum target, GLuint id);
-
-extern void GLAPIENTRY
-_mesa_EndQueryARB(GLenum target);
-
-extern void GLAPIENTRY
 _mesa_GetQueryivARB(GLenum target, GLenum pname, GLint *params);
 
 extern void GLAPIENTRY
@@ -61,11 +60,33 @@ _mesa_GetQueryObjectivARB(GLuint id, GLenum pname, GLint *params);
 extern void GLAPIENTRY
 _mesa_GetQueryObjectuivARB(GLuint id, GLenum pname, GLuint *params);
 
-extern void GLAPIENTRY
-_mesa_GetQueryObjecti64vEXT(GLuint id, GLenum pname, GLint64EXT *params);
+extern void
+_mesa_init_query_object_functions(struct dd_function_table *driver);
 
-extern void GLAPIENTRY
-_mesa_GetQueryObjectui64vEXT(GLuint id, GLenum pname, GLuint64EXT *params);
+extern void
+_mesa_init_queryobj_dispatch(struct _glapi_table *disp);
+
+#else /* FEATURE_queryobj */
+
+#define _MESA_INIT_QUERYOBJ_FUNCTIONS(driver, impl) do { } while (0)
+
+static INLINE void
+_mesa_init_query_object_functions(struct dd_function_table *driver)
+{
+}
+
+static INLINE void
+_mesa_init_queryobj_dispatch(struct _glapi_table *disp)
+{
+}
+
+#endif /* FEATURE_queryobj */
+
+extern void
+_mesa_init_queryobj(GLcontext *ctx);
+
+extern void
+_mesa_free_queryobj_data(GLcontext *ctx);
 
 
-#endif /* OCCLUDE_H */
+#endif /* QUERYOBJ_H */
diff --git a/src/mesa/main/rastpos.c b/src/mesa/main/rastpos.c
index 9f309d6ab8..703b47ec74 100644
--- a/src/mesa/main/rastpos.c
+++ b/src/mesa/main/rastpos.c
@@ -34,6 +34,10 @@
 #include "macros.h"
 #include "rastpos.h"
 #include "state.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_rastpos
 
 
 /**
@@ -60,147 +64,147 @@ rasterpos(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2d(GLdouble x, GLdouble y)
 {
    rasterpos((GLfloat)x, (GLfloat)y, (GLfloat)0.0, (GLfloat)1.0);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2f(GLfloat x, GLfloat y)
 {
    rasterpos(x, y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2i(GLint x, GLint y)
 {
    rasterpos((GLfloat) x, (GLfloat) y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2s(GLshort x, GLshort y)
 {
    rasterpos(x, y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3d(GLdouble x, GLdouble y, GLdouble z)
 {
    rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3f(GLfloat x, GLfloat y, GLfloat z)
 {
    rasterpos(x, y, z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3i(GLint x, GLint y, GLint z)
 {
    rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3s(GLshort x, GLshort y, GLshort z)
 {
    rasterpos(x, y, z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4d(GLdouble x, GLdouble y, GLdouble z, GLdouble w)
 {
    rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
 {
    rasterpos(x, y, z, w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4i(GLint x, GLint y, GLint z, GLint w)
 {
    rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4s(GLshort x, GLshort y, GLshort z, GLshort w)
 {
    rasterpos(x, y, z, w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2dv(const GLdouble *v)
 {
    rasterpos((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2fv(const GLfloat *v)
 {
    rasterpos(v[0], v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2iv(const GLint *v)
 {
    rasterpos((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos2sv(const GLshort *v)
 {
    rasterpos(v[0], v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3dv(const GLdouble *v)
 {
    rasterpos((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3fv(const GLfloat *v)
 {
    rasterpos(v[0], v[1], v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3iv(const GLint *v)
 {
    rasterpos((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos3sv(const GLshort *v)
 {
    rasterpos(v[0], v[1], v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4dv(const GLdouble *v)
 {
    rasterpos((GLfloat) v[0], (GLfloat) v[1], 
 		     (GLfloat) v[2], (GLfloat) v[3]);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4fv(const GLfloat *v)
 {
    rasterpos(v[0], v[1], v[2], v[3]);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4iv(const GLint *v)
 {
    rasterpos((GLfloat) v[0], (GLfloat) v[1], 
 		     (GLfloat) v[2], (GLfloat) v[3]);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_RasterPos4sv(const GLshort *v)
 {
    rasterpos(v[0], v[1], v[2], v[3]);
@@ -211,7 +215,7 @@ _mesa_RasterPos4sv(const GLshort *v)
 /***           GL_ARB_window_pos / GL_MESA_window_pos               ***/
 /**********************************************************************/
 
-#if FEATURE_drawpix
+
 /**
  * All glWindowPosMESA and glWindowPosARB commands call this function to
  * update the current raster position.
@@ -290,153 +294,152 @@ window_pos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2dMESA(GLdouble x, GLdouble y)
 {
    window_pos4f((GLfloat) x, (GLfloat) y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2fMESA(GLfloat x, GLfloat y)
 {
    window_pos4f(x, y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2iMESA(GLint x, GLint y)
 {
    window_pos4f((GLfloat) x, (GLfloat) y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2sMESA(GLshort x, GLshort y)
 {
    window_pos4f(x, y, 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3dMESA(GLdouble x, GLdouble y, GLdouble z)
 {
    window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3fMESA(GLfloat x, GLfloat y, GLfloat z)
 {
    window_pos4f(x, y, z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3iMESA(GLint x, GLint y, GLint z)
 {
    window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3sMESA(GLshort x, GLshort y, GLshort z)
 {
    window_pos4f(x, y, z, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4dMESA(GLdouble x, GLdouble y, GLdouble z, GLdouble w)
 {
    window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4fMESA(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
 {
    window_pos4f(x, y, z, w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4iMESA(GLint x, GLint y, GLint z, GLint w)
 {
    window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4sMESA(GLshort x, GLshort y, GLshort z, GLshort w)
 {
    window_pos4f(x, y, z, w);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2dvMESA(const GLdouble *v)
 {
    window_pos4f((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2fvMESA(const GLfloat *v)
 {
    window_pos4f(v[0], v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2ivMESA(const GLint *v)
 {
    window_pos4f((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos2svMESA(const GLshort *v)
 {
    window_pos4f(v[0], v[1], 0.0F, 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3dvMESA(const GLdouble *v)
 {
    window_pos4f((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3fvMESA(const GLfloat *v)
 {
    window_pos4f(v[0], v[1], v[2], 1.0);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3ivMESA(const GLint *v)
 {
    window_pos4f((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos3svMESA(const GLshort *v)
 {
    window_pos4f(v[0], v[1], v[2], 1.0F);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4dvMESA(const GLdouble *v)
 {
    window_pos4f((GLfloat) v[0], (GLfloat) v[1], 
 			 (GLfloat) v[2], (GLfloat) v[3]);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4fvMESA(const GLfloat *v)
 {
    window_pos4f(v[0], v[1], v[2], v[3]);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4ivMESA(const GLint *v)
 {
    window_pos4f((GLfloat) v[0], (GLfloat) v[1], 
 			 (GLfloat) v[2], (GLfloat) v[3]);
 }
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_WindowPos4svMESA(const GLshort *v)
 {
    window_pos4f(v[0], v[1], v[2], v[3]);
 }
 
-#endif
 
 #if 0
 
@@ -477,6 +480,65 @@ void glWindowPos4fMESA( GLfloat x, GLfloat y, GLfloat z, GLfloat w )
 #endif
 
 
+void
+_mesa_init_rastpos_dispatch(struct _glapi_table *disp)
+{
+   SET_RasterPos2f(disp, _mesa_RasterPos2f);
+   SET_RasterPos2fv(disp, _mesa_RasterPos2fv);
+   SET_RasterPos2i(disp, _mesa_RasterPos2i);
+   SET_RasterPos2iv(disp, _mesa_RasterPos2iv);
+   SET_RasterPos2d(disp, _mesa_RasterPos2d);
+   SET_RasterPos2dv(disp, _mesa_RasterPos2dv);
+   SET_RasterPos2s(disp, _mesa_RasterPos2s);
+   SET_RasterPos2sv(disp, _mesa_RasterPos2sv);
+   SET_RasterPos3d(disp, _mesa_RasterPos3d);
+   SET_RasterPos3dv(disp, _mesa_RasterPos3dv);
+   SET_RasterPos3f(disp, _mesa_RasterPos3f);
+   SET_RasterPos3fv(disp, _mesa_RasterPos3fv);
+   SET_RasterPos3i(disp, _mesa_RasterPos3i);
+   SET_RasterPos3iv(disp, _mesa_RasterPos3iv);
+   SET_RasterPos3s(disp, _mesa_RasterPos3s);
+   SET_RasterPos3sv(disp, _mesa_RasterPos3sv);
+   SET_RasterPos4d(disp, _mesa_RasterPos4d);
+   SET_RasterPos4dv(disp, _mesa_RasterPos4dv);
+   SET_RasterPos4f(disp, _mesa_RasterPos4f);
+   SET_RasterPos4fv(disp, _mesa_RasterPos4fv);
+   SET_RasterPos4i(disp, _mesa_RasterPos4i);
+   SET_RasterPos4iv(disp, _mesa_RasterPos4iv);
+   SET_RasterPos4s(disp, _mesa_RasterPos4s);
+   SET_RasterPos4sv(disp, _mesa_RasterPos4sv);
+
+   /* 197. GL_MESA_window_pos */
+   SET_WindowPos2dMESA(disp, _mesa_WindowPos2dMESA);
+   SET_WindowPos2dvMESA(disp, _mesa_WindowPos2dvMESA);
+   SET_WindowPos2fMESA(disp, _mesa_WindowPos2fMESA);
+   SET_WindowPos2fvMESA(disp, _mesa_WindowPos2fvMESA);
+   SET_WindowPos2iMESA(disp, _mesa_WindowPos2iMESA);
+   SET_WindowPos2ivMESA(disp, _mesa_WindowPos2ivMESA);
+   SET_WindowPos2sMESA(disp, _mesa_WindowPos2sMESA);
+   SET_WindowPos2svMESA(disp, _mesa_WindowPos2svMESA);
+   SET_WindowPos3dMESA(disp, _mesa_WindowPos3dMESA);
+   SET_WindowPos3dvMESA(disp, _mesa_WindowPos3dvMESA);
+   SET_WindowPos3fMESA(disp, _mesa_WindowPos3fMESA);
+   SET_WindowPos3fvMESA(disp, _mesa_WindowPos3fvMESA);
+   SET_WindowPos3iMESA(disp, _mesa_WindowPos3iMESA);
+   SET_WindowPos3ivMESA(disp, _mesa_WindowPos3ivMESA);
+   SET_WindowPos3sMESA(disp, _mesa_WindowPos3sMESA);
+   SET_WindowPos3svMESA(disp, _mesa_WindowPos3svMESA);
+   SET_WindowPos4dMESA(disp, _mesa_WindowPos4dMESA);
+   SET_WindowPos4dvMESA(disp, _mesa_WindowPos4dvMESA);
+   SET_WindowPos4fMESA(disp, _mesa_WindowPos4fMESA);
+   SET_WindowPos4fvMESA(disp, _mesa_WindowPos4fvMESA);
+   SET_WindowPos4iMESA(disp, _mesa_WindowPos4iMESA);
+   SET_WindowPos4ivMESA(disp, _mesa_WindowPos4ivMESA);
+   SET_WindowPos4sMESA(disp, _mesa_WindowPos4sMESA);
+   SET_WindowPos4svMESA(disp, _mesa_WindowPos4svMESA);
+}
+
+
+#endif /* FEATURE_rastpos */
+
+
 /**********************************************************************/
 /** \name Initialization                                              */
 /**********************************************************************/
diff --git a/src/mesa/main/rastpos.h b/src/mesa/main/rastpos.h
index 363f86ad87..b2127225b6 100644
--- a/src/mesa/main/rastpos.h
+++ b/src/mesa/main/rastpos.h
@@ -32,162 +32,33 @@
 #define RASTPOS_H
 
 
-#include "glheader.h"
+#include "main/mtypes.h"
 
 
-extern void GLAPIENTRY
-_mesa_RasterPos2d(GLdouble x, GLdouble y);
+#if FEATURE_rastpos
 
-extern void GLAPIENTRY
-_mesa_RasterPos2f(GLfloat x, GLfloat y);
+#define _MESA_INIT_RASTPOS_FUNCTIONS(driver, impl) \
+   do {                                            \
+      (driver)->RasterPos = impl ## RasterPos;     \
+   } while (0)
 
-extern void GLAPIENTRY
-_mesa_RasterPos2i(GLint x, GLint y);
+extern void
+_mesa_init_rastpos_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY
-_mesa_RasterPos2s(GLshort x, GLshort y);
+#else /* FEATURE_rastpos */
 
-extern void GLAPIENTRY
-_mesa_RasterPos3d(GLdouble x, GLdouble y, GLdouble z);
+#define _MESA_INIT_RASTPOS_FUNCTIONS(driver, impl) do { } while (0)
 
-extern void GLAPIENTRY
-_mesa_RasterPos3f(GLfloat x, GLfloat y, GLfloat z);
+static INLINE void
+_mesa_init_rastpos_dispatch(struct _glapi_table *disp)
+{
+}
 
-extern void GLAPIENTRY
-_mesa_RasterPos3i(GLint x, GLint y, GLint z);
-
-extern void GLAPIENTRY
-_mesa_RasterPos3s(GLshort x, GLshort y, GLshort z);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4d(GLdouble x, GLdouble y, GLdouble z, GLdouble w);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4i(GLint x, GLint y, GLint z, GLint w);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4s(GLshort x, GLshort y, GLshort z, GLshort w);
-
-extern void GLAPIENTRY
-_mesa_RasterPos2dv(const GLdouble *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos2fv(const GLfloat *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos2iv(const GLint *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos2sv(const GLshort *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos3dv(const GLdouble *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos3fv(const GLfloat *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos3iv(const GLint *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos3sv(const GLshort *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4dv(const GLdouble *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4fv(const GLfloat *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4iv(const GLint *v);
-
-extern void GLAPIENTRY
-_mesa_RasterPos4sv(const GLshort *v);
-
-
-/**********************************************************************/
-/** \name GL_MESA_window_pos                                          */
-/**********************************************************************/
-/*@{*/
-
-extern void GLAPIENTRY
-_mesa_WindowPos2dMESA(GLdouble x, GLdouble y);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2fMESA(GLfloat x, GLfloat y);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2iMESA(GLint x, GLint y);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2sMESA(GLshort x, GLshort y);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3dMESA(GLdouble x, GLdouble y, GLdouble z);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3fMESA(GLfloat x, GLfloat y, GLfloat z);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3iMESA(GLint x, GLint y, GLint z);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3sMESA(GLshort x, GLshort y, GLshort z);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4dMESA(GLdouble x, GLdouble y, GLdouble z, GLdouble w);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4fMESA(GLfloat x, GLfloat y, GLfloat z, GLfloat w);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4iMESA(GLint x, GLint y, GLint z, GLint w);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4sMESA(GLshort x, GLshort y, GLshort z, GLshort w);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2dvMESA(const GLdouble *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2fvMESA(const GLfloat *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2ivMESA(const GLint *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos2svMESA(const GLshort *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3dvMESA(const GLdouble *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3fvMESA(const GLfloat *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3ivMESA(const GLint *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos3svMESA(const GLshort *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4dvMESA(const GLdouble *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4fvMESA(const GLfloat *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4ivMESA(const GLint *v);
-
-extern void GLAPIENTRY
-_mesa_WindowPos4svMESA(const GLshort *v);
+#endif /* FEATURE_rastpos */
 
 extern void 
-_mesa_init_rastpos( GLcontext * ctx );
+_mesa_init_rastpos(GLcontext *ctx);
 
 /*@}*/
 
-#endif
+#endif /* RASTPOS_H */
diff --git a/src/mesa/main/shaders.c b/src/mesa/main/shaders.c
index bc76b91291..96fd8695a5 100644
--- a/src/mesa/main/shaders.c
+++ b/src/mesa/main/shaders.c
@@ -388,7 +388,6 @@ _mesa_read_shader(const char *fname)
    int len;
 
    if (!f) {
-      _mesa_fprintf(stderr, "Unable to open shader file %s\n", fname);
       return NULL;
    }
 
@@ -401,11 +400,6 @@ _mesa_read_shader(const char *fname)
    shader = _mesa_strdup(buffer);
    free(buffer);
 
-   if (0) {
-      _mesa_fprintf(stderr, "Read shader %s:\n", fname);
-      _mesa_fprintf(stderr, "%s\n", shader);
-   }
-
    return shader;
 }
 
@@ -475,19 +469,25 @@ _mesa_ShaderSourceARB(GLhandleARB shaderObj, GLsizei count,
    source[totalLength - 1] = '\0';
    source[totalLength - 2] = '\0';
 
-#if 0
    if (0) {
+      /* Compute the shader's source code checksum then try to open a file
+       * named newshader_<CHECKSUM>.  If it exists, use it in place of the
+       * original shader source code.  For debugging.
+       */
+      const GLuint checksum = _mesa_str_checksum(source);
+      char filename[100];
       GLcharARB *newSource;
 
-      newSource = _mesa_read_shader("newshader.frag");
+      sprintf(filename, "newshader_%d", checksum);
+
+      newSource = _mesa_read_shader(filename);
       if (newSource) {
+         _mesa_fprintf(stderr, "Mesa: Replacing shader %u chksum=%d with %s\n",
+                       shaderObj, checksum, filename);
          _mesa_free(source);
          source = newSource;
       }
-   }
-#else
-   (void) _mesa_read_shader;
-#endif
+   }      
 
    ctx->Driver.ShaderSource(ctx, shaderObj, source);
 
diff --git a/src/mesa/main/shared.c b/src/mesa/main/shared.c
index 643ad3354e..4d01e8abc6 100644
--- a/src/mesa/main/shared.c
+++ b/src/mesa/main/shared.c
@@ -37,9 +37,7 @@
 #include "shared.h"
 #include "shader/program.h"
 #include "shader/shader_api.h"
-#if FEATURE_dlist
 #include "dlist.h"
-#endif
 #if FEATURE_ATI_fragment_shader
 #include "shader/atifragshader.h"
 #endif
@@ -143,11 +141,9 @@ _mesa_alloc_shared_state(GLcontext *ctx)
 static void
 delete_displaylist_cb(GLuint id, void *data, void *userData)
 {
-#if FEATURE_dlist
    struct gl_display_list *list = (struct gl_display_list *) data;
    GLcontext *ctx = (GLcontext *) userData;
    _mesa_delete_list(ctx, list);
-#endif
 }
 
 
diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c
index 140a998df2..f10e6b04b7 100644
--- a/src/mesa/main/state.c
+++ b/src/mesa/main/state.c
@@ -40,9 +40,7 @@
 #include "framebuffer.h"
 #include "light.h"
 #include "matrix.h"
-#if FEATURE_pixel_transfer
 #include "pixel.h"
-#endif
 #include "shader/program.h"
 #include "shader/prog_parameter.h"
 #include "state.h"
@@ -585,10 +583,8 @@ _mesa_update_state_locked( GLcontext *ctx )
    if (new_state & (_NEW_STENCIL | _NEW_BUFFERS))
       _mesa_update_stencil( ctx );
 
-#if FEATURE_pixel_transfer
    if (new_state & _MESA_NEW_TRANSFER_STATE)
       _mesa_update_pixel( ctx, new_state );
-#endif
 
    if (new_state & _DD_NEW_SEPARATE_SPECULAR)
       update_separate_specular( ctx );
diff --git a/src/mesa/main/syncobj.c b/src/mesa/main/syncobj.c
index 64f923ff91..ac3f9eb175 100644
--- a/src/mesa/main/syncobj.c
+++ b/src/mesa/main/syncobj.c
@@ -190,7 +190,7 @@ _mesa_unref_sync_object(GLcontext *ctx, struct gl_sync_object *syncObj)
 }
 
 
-GLboolean
+GLboolean GLAPIENTRY
 _mesa_IsSync(GLsync sync)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -201,7 +201,7 @@ _mesa_IsSync(GLsync sync)
 }
 
 
-void
+void GLAPIENTRY
 _mesa_DeleteSync(GLsync sync)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -231,7 +231,7 @@ _mesa_DeleteSync(GLsync sync)
 }
 
 
-GLsync
+GLsync GLAPIENTRY
 _mesa_FenceSync(GLenum condition, GLbitfield flags)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -278,7 +278,7 @@ _mesa_FenceSync(GLenum condition, GLbitfield flags)
 }
 
 
-GLenum
+GLenum GLAPIENTRY
 _mesa_ClientWaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -319,7 +319,7 @@ _mesa_ClientWaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout)
 }
 
 
-void
+void GLAPIENTRY
 _mesa_WaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -348,7 +348,7 @@ _mesa_WaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout)
 }
 
 
-void
+void GLAPIENTRY
 _mesa_GetSynciv(GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length,
 		GLint *values)
 {
diff --git a/src/mesa/main/syncobj.h b/src/mesa/main/syncobj.h
index fc160af289..f23fa281e2 100644
--- a/src/mesa/main/syncobj.h
+++ b/src/mesa/main/syncobj.h
@@ -48,22 +48,22 @@ _mesa_ref_sync_object(GLcontext *ctx, struct gl_sync_object *syncObj);
 extern void
 _mesa_unref_sync_object(GLcontext *ctx, struct gl_sync_object *syncObj);
 
-extern GLboolean
+extern GLboolean GLAPIENTRY
 _mesa_IsSync(GLsync sync);
 
-extern void
+extern void GLAPIENTRY
 _mesa_DeleteSync(GLsync sync);
 
-extern GLsync
+extern GLsync GLAPIENTRY
 _mesa_FenceSync(GLenum condition, GLbitfield flags);
 
-extern GLenum
+extern GLenum GLAPIENTRY
 _mesa_ClientWaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout);
 
-extern void
+extern void GLAPIENTRY
 _mesa_WaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout);
 
-extern void
+extern void GLAPIENTRY
 _mesa_GetSynciv(GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length,
 		GLint *values);
 
diff --git a/src/mesa/main/texgen.c b/src/mesa/main/texgen.c
index e3feb024c3..733e129fcf 100644
--- a/src/mesa/main/texgen.c
+++ b/src/mesa/main/texgen.c
@@ -35,7 +35,12 @@
 #include "main/enums.h"
 #include "main/macros.h"
 #include "main/texgen.h"
+#include "main/texstate.h"
 #include "math/m_matrix.h"
+#include "glapi/dispatch.h"
+
+
+#if FEATURE_texgen
 
 
 /**
@@ -79,7 +84,7 @@ _mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params )
       return;
    }
 
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   texUnit = _mesa_get_current_tex_unit(ctx);
 
    texgen = get_texgen(texUnit, coord);
    if (!texgen) {
@@ -161,7 +166,7 @@ _mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_TexGeniv(GLenum coord, GLenum pname, const GLint *params )
 {
    GLfloat p[4];
@@ -178,7 +183,7 @@ _mesa_TexGeniv(GLenum coord, GLenum pname, const GLint *params )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_TexGend(GLenum coord, GLenum pname, GLdouble param )
 {
    GLfloat p = (GLfloat) param;
@@ -186,7 +191,7 @@ _mesa_TexGend(GLenum coord, GLenum pname, GLdouble param )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_TexGendv(GLenum coord, GLenum pname, const GLdouble *params )
 {
    GLfloat p[4];
@@ -203,7 +208,7 @@ _mesa_TexGendv(GLenum coord, GLenum pname, const GLdouble *params )
 }
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_TexGenf( GLenum coord, GLenum pname, GLfloat param )
 {
    _mesa_TexGenfv(coord, pname, &param);
@@ -218,7 +223,7 @@ _mesa_TexGeni( GLenum coord, GLenum pname, GLint param )
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params )
 {
    struct gl_texture_unit *texUnit;
@@ -231,7 +236,7 @@ _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params )
       return;
    }
 
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   texUnit = _mesa_get_current_tex_unit(ctx);
 
    texgen = get_texgen(texUnit, coord);
    if (!texgen) {
@@ -256,7 +261,7 @@ _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params )
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params )
 {
    struct gl_texture_unit *texUnit;
@@ -269,7 +274,7 @@ _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params )
       return;
    }
 
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   texUnit = _mesa_get_current_tex_unit(ctx);
 
    texgen = get_texgen(texUnit, coord);
    if (!texgen) {
@@ -294,7 +299,7 @@ _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params )
 
 
 
-void GLAPIENTRY
+static void GLAPIENTRY
 _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params )
 {
    struct gl_texture_unit *texUnit;
@@ -307,7 +312,7 @@ _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params )
       return;
    }
 
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   texUnit = _mesa_get_current_tex_unit(ctx);
 
    texgen = get_texgen(texUnit, coord);
    if (!texgen) {
@@ -337,3 +342,19 @@ _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params )
 }
 
 
+void
+_mesa_init_texgen_dispatch(struct _glapi_table *disp)
+{
+   SET_GetTexGendv(disp, _mesa_GetTexGendv);
+   SET_GetTexGenfv(disp, _mesa_GetTexGenfv);
+   SET_GetTexGeniv(disp, _mesa_GetTexGeniv);
+   SET_TexGend(disp, _mesa_TexGend);
+   SET_TexGendv(disp, _mesa_TexGendv);
+   SET_TexGenf(disp, _mesa_TexGenf);
+   SET_TexGenfv(disp, _mesa_TexGenfv);
+   SET_TexGeni(disp, _mesa_TexGeni);
+   SET_TexGeniv(disp, _mesa_TexGeniv);
+}
+
+
+#endif /* FEATURE_texgen */
diff --git a/src/mesa/main/texgen.h b/src/mesa/main/texgen.h
index 073588efcd..f6924ef722 100644
--- a/src/mesa/main/texgen.h
+++ b/src/mesa/main/texgen.h
@@ -27,36 +27,45 @@
 #define TEXGEN_H
 
 
-#include "main/glheader.h"
+#include "main/mtypes.h"
 
 
-extern void GLAPIENTRY
-_mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params );
+#if FEATURE_texgen
 
-extern void GLAPIENTRY
-_mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params );
+#define _MESA_INIT_TEXGEN_FUNCTIONS(driver, impl) \
+   do {                                           \
+      (driver)->TexGen = impl ## TexGen;          \
+   } while (0)
 
 extern void GLAPIENTRY
-_mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params );
+_mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params );
 
 extern void GLAPIENTRY
-_mesa_TexGend( GLenum coord, GLenum pname, GLdouble param );
+_mesa_TexGeni( GLenum coord, GLenum pname, GLint param );
 
-extern void GLAPIENTRY
-_mesa_TexGendv( GLenum coord, GLenum pname, const GLdouble *params );
+extern void
+_mesa_init_texgen_dispatch(struct _glapi_table *disp);
 
-extern void GLAPIENTRY
-_mesa_TexGenf( GLenum coord, GLenum pname, GLfloat param );
+#else /* FEATURE_texgen */
 
-extern void GLAPIENTRY
-_mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params );
+#define _MESA_INIT_TEXGEN_FUNCTIONS(driver, impl) do { } while (0)
 
-extern void GLAPIENTRY
-_mesa_TexGeni( GLenum coord, GLenum pname, GLint param );
+static void
+_mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params )
+{
+}
 
-extern void GLAPIENTRY
-_mesa_TexGeniv( GLenum coord, GLenum pname, const GLint *params );
+static void INLINE
+_mesa_TexGeni( GLenum coord, GLenum pname, GLint param )
+{
+}
+
+static INLINE void
+_mesa_init_texgen_dispatch(struct _glapi_table *disp)
+{
+}
 
+#endif /* FEATURE_texgen */
 
 
 #endif /* TEXGEN_H */
diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index a8160c2735..465da6b046 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -33,9 +33,7 @@
 #include "glheader.h"
 #include "bufferobj.h"
 #include "context.h"
-#if FEATURE_convolve
 #include "convolve.h"
-#endif
 #include "enums.h"
 #include "fbobject.h"
 #include "framebuffer.h"
@@ -2076,6 +2074,23 @@ update_fbo_texture(GLcontext *ctx, struct gl_texture_object *texObj,
 }
 
 
+/**
+ * If the texture object's GenerateMipmap flag is set and we've
+ * changed the texture base level image, regenerate the rest of the
+ * mipmap levels now.
+ */
+static INLINE void
+check_gen_mipmap(GLcontext *ctx, GLenum target,
+                 struct gl_texture_object *texObj, GLint level)
+{
+   ASSERT(target != GL_TEXTURE_CUBE_MAP);
+   if (texObj->GenerateMipmap && level == texObj->BaseLevel) {
+      ASSERT(ctx->Driver.GenerateMipmap);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
+   }
+}
+
+
 /** Debug helper: override the user-requested internal format */
 static GLenum
 override_internal_format(GLenum internalFormat, GLint width, GLint height)
@@ -2169,36 +2184,36 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat,
 	 texImage = _mesa_get_tex_image(ctx, texObj, target, level);
 	 if (!texImage) {
 	    _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage1D");
-	    goto out;
-	 }
-      
-	 if (texImage->Data) {
-	    ctx->Driver.FreeTexImageData( ctx, texImage );
 	 }
+         else {
+            if (texImage->Data) {
+               ctx->Driver.FreeTexImageData( ctx, texImage );
+            }
+
+            ASSERT(texImage->Data == NULL);
+
+            clear_teximage_fields(texImage); /* not really needed, but helpful */
+            _mesa_init_teximage_fields(ctx, target, texImage,
+                                       postConvWidth, 1, 1,
+                                       border, internalFormat);
+
+            /* Give the texture to the driver.  <pixels> may be null. */
+            ASSERT(ctx->Driver.TexImage1D);
+            ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
+                                   width, border, format, type, pixels,
+                                   &ctx->Unpack, texObj, texImage);
 
-	 ASSERT(texImage->Data == NULL);
-
-	 clear_teximage_fields(texImage); /* not really needed, but helpful */
-	 _mesa_init_teximage_fields(ctx, target, texImage,
-				    postConvWidth, 1, 1,
-				    border, internalFormat);
-	 
-	 ASSERT(ctx->Driver.TexImage1D);
-
-	 /* Give the texture to the driver!  <pixels> may be null! */
-	 (*ctx->Driver.TexImage1D)(ctx, target, level, internalFormat,
-				   width, border, format, type, pixels,
-				   &ctx->Unpack, texObj, texImage);
-	 
-	 ASSERT(texImage->TexFormat);
-
-	 update_fbo_texture(ctx, texObj, face, level);
-	 
-	 /* state update */
-	 texObj->_Complete = GL_FALSE;
-	 ctx->NewState |= _NEW_TEXTURE;
-      }
-   out:
+            ASSERT(texImage->TexFormat);
+
+            check_gen_mipmap(ctx, target, texObj, level);
+
+            update_fbo_texture(ctx, texObj, face, level);
+
+            /* state update */
+            texObj->_Complete = GL_FALSE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
       _mesa_unlock_texture(ctx, texObj);
    }
    else if (target == GL_PROXY_TEXTURE_1D) {
@@ -2284,35 +2299,35 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat,
 	 texImage = _mesa_get_tex_image(ctx, texObj, target, level);
 	 if (!texImage) {
 	    _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage2D");
-	    goto out;
-	 }
-	 
-	 if (texImage->Data) {
-	    ctx->Driver.FreeTexImageData( ctx, texImage );
 	 }
-	 
-	 ASSERT(texImage->Data == NULL);
-	 clear_teximage_fields(texImage); /* not really needed, but helpful */
-	 _mesa_init_teximage_fields(ctx, target, texImage,
-				    postConvWidth, postConvHeight, 1,
-				    border, internalFormat);
-	 
-	 ASSERT(ctx->Driver.TexImage2D);
-
-	 /* Give the texture to the driver!  <pixels> may be null! */
-	 (*ctx->Driver.TexImage2D)(ctx, target, level, internalFormat,
-				   width, height, border, format, type, pixels,
-				   &ctx->Unpack, texObj, texImage);
-	 
-	 ASSERT(texImage->TexFormat);
-
-	 update_fbo_texture(ctx, texObj, face, level);
-
-	 /* state update */
-	 texObj->_Complete = GL_FALSE;
-	 ctx->NewState |= _NEW_TEXTURE;
-      }
-   out:
+         else {
+            if (texImage->Data) {
+               ctx->Driver.FreeTexImageData( ctx, texImage );
+            }
+
+            ASSERT(texImage->Data == NULL);
+            clear_teximage_fields(texImage); /* not really needed, but helpful */
+            _mesa_init_teximage_fields(ctx, target, texImage,
+                                       postConvWidth, postConvHeight, 1,
+                                       border, internalFormat);
+
+            /* Give the texture to the driver.  <pixels> may be null. */
+            ASSERT(ctx->Driver.TexImage2D);
+            ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
+                                   width, height, border, format, type,
+                                   pixels, &ctx->Unpack, texObj, texImage);
+
+            ASSERT(texImage->TexFormat);
+
+            check_gen_mipmap(ctx, target, texObj, level);
+
+            update_fbo_texture(ctx, texObj, face, level);
+
+            /* state update */
+            texObj->_Complete = GL_FALSE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
       _mesa_unlock_texture(ctx, texObj);
    }
    else if (target == GL_PROXY_TEXTURE_2D ||
@@ -2394,35 +2409,35 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat,
 	 texImage = _mesa_get_tex_image(ctx, texObj, target, level);
 	 if (!texImage) {
 	    _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage3D");
-	    goto out;
 	 }
-	 
-	 if (texImage->Data) {
-	    ctx->Driver.FreeTexImageData( ctx, texImage );
-	 }
-	 
-	 ASSERT(texImage->Data == NULL);
-	 clear_teximage_fields(texImage); /* not really needed, but helpful */
-	 _mesa_init_teximage_fields(ctx, target, texImage,
-				    width, height, depth,
-				    border, internalFormat);
+         else {
+            if (texImage->Data) {
+               ctx->Driver.FreeTexImageData( ctx, texImage );
+            }
 
-	 ASSERT(ctx->Driver.TexImage3D);
+            ASSERT(texImage->Data == NULL);
+            clear_teximage_fields(texImage); /* not really needed, but helpful */
+            _mesa_init_teximage_fields(ctx, target, texImage,
+                                       width, height, depth,
+                                       border, internalFormat);
 
-	 /* Give the texture to the driver!  <pixels> may be null! */
-	 (*ctx->Driver.TexImage3D)(ctx, target, level, internalFormat,
-				   width, height, depth, border, format, type,
-				   pixels, &ctx->Unpack, texObj, texImage);
+            /* Give the texture to the driver.  <pixels> may be null. */
+            ASSERT(ctx->Driver.TexImage3D);
+            ctx->Driver.TexImage3D(ctx, target, level, internalFormat,
+                                   width, height, depth, border, format, type,
+                                   pixels, &ctx->Unpack, texObj, texImage);
 
-	 ASSERT(texImage->TexFormat);
+            ASSERT(texImage->TexFormat);
 
-	 update_fbo_texture(ctx, texObj, face, level);
+            check_gen_mipmap(ctx, target, texObj, level);
 
-	 /* state update */
-	 texObj->_Complete = GL_FALSE;
-	 ctx->NewState |= _NEW_TEXTURE;
+            update_fbo_texture(ctx, texObj, face, level);
+
+            /* state update */
+            texObj->_Complete = GL_FALSE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
       }
-   out:
       _mesa_unlock_texture(ctx, texObj);
    }
    else if (target == GL_PROXY_TEXTURE_3D ||
@@ -2508,23 +2523,24 @@ _mesa_TexSubImage1D( GLenum target, GLint level,
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
 
       if (subtexture_error_check2(ctx, 1, target, level, xoffset, 0, 0,
-				  postConvWidth, 1, 1, format, type, texImage)) {
-	 goto out;   /* error was detected */
+				  postConvWidth, 1, 1,
+                                  format, type, texImage)) {
+         /* error was recorded */
       }
+      else if (width > 0) {
+         /* If we have a border, xoffset=-1 is legal.  Bias by border width */
+         xoffset += texImage->Border;
 
-      if (width == 0)
-	 goto out;  /* no-op, not an error */
+         ASSERT(ctx->Driver.TexSubImage1D);
+         ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width,
+                                   format, type, pixels, &ctx->Unpack,
+                                   texObj, texImage);
 
-      /* If we have a border, xoffset=-1 is legal.  Bias by border width */
-      xoffset += texImage->Border;
+         check_gen_mipmap(ctx, target, texObj, level);
 
-      ASSERT(ctx->Driver.TexSubImage1D);
-      (*ctx->Driver.TexSubImage1D)(ctx, target, level, xoffset, width,
-				   format, type, pixels, &ctx->Unpack,
-				   texObj, texImage);
-      ctx->NewState |= _NEW_TEXTURE;
+         ctx->NewState |= _NEW_TEXTURE;
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2568,30 +2584,31 @@ _mesa_TexSubImage2D( GLenum target, GLint level,
 
    texUnit = _mesa_get_current_tex_unit(ctx);
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
+
    _mesa_lock_texture(ctx, texObj);
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
 
       if (subtexture_error_check2(ctx, 2, target, level, xoffset, yoffset, 0,
-				  postConvWidth, postConvHeight, 1, format, type, 
-				  texImage)) {
-	 goto out;   /* error was detected */
+				  postConvWidth, postConvHeight, 1,
+                                  format, type, texImage)) {
+	 /* error was recorded */
       }
+      else if (width > 0 && height >= 0) {
+         /* If we have a border, xoffset=-1 is legal.  Bias by border width */
+         xoffset += texImage->Border;
+         yoffset += texImage->Border;
+
+         ASSERT(ctx->Driver.TexSubImage2D);
+         ctx->Driver.TexSubImage2D(ctx, target, level, xoffset, yoffset,
+                                   width, height, format, type, pixels,
+                                   &ctx->Unpack, texObj, texImage);
 
-      if (width == 0 || height == 0)
-	 goto out;  /* no-op, not an error */
+         check_gen_mipmap(ctx, target, texObj, level);
 
-      /* If we have a border, xoffset=-1 is legal.  Bias by border width */
-      xoffset += texImage->Border;
-      yoffset += texImage->Border;
-      
-      ASSERT(ctx->Driver.TexSubImage2D);
-      (*ctx->Driver.TexSubImage2D)(ctx, target, level, xoffset, yoffset,
-				   width, height, format, type, pixels,
-				   &ctx->Unpack, texObj, texImage);
-      ctx->NewState |= _NEW_TEXTURE;
+         ctx->NewState |= _NEW_TEXTURE;
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2632,28 +2649,30 @@ _mesa_TexSubImage3D( GLenum target, GLint level,
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
 
-      if (subtexture_error_check2(ctx, 3, target, level, xoffset, yoffset, zoffset,
-				  width, height, depth, format, type, texImage)) {
-	 goto out;   /* error was detected */
+      if (subtexture_error_check2(ctx, 3, target, level,
+                                  xoffset, yoffset, zoffset,
+				  width, height, depth,
+                                  format, type, texImage)) {
+         /* error was recorded */
       }
+      else if (width > 0 && height > 0 && height > 0) {
+         /* If we have a border, xoffset=-1 is legal.  Bias by border width */
+         xoffset += texImage->Border;
+         yoffset += texImage->Border;
+         zoffset += texImage->Border;
 
-      if (width == 0 || height == 0 || height == 0)
-	 goto out;  /* no-op, not an error */
+         ASSERT(ctx->Driver.TexSubImage3D);
+         ctx->Driver.TexSubImage3D(ctx, target, level,
+                                   xoffset, yoffset, zoffset,
+                                   width, height, depth,
+                                   format, type, pixels,
+                                   &ctx->Unpack, texObj, texImage );
 
-      /* If we have a border, xoffset=-1 is legal.  Bias by border width */
-      xoffset += texImage->Border;
-      yoffset += texImage->Border;
-      zoffset += texImage->Border;
+         check_gen_mipmap(ctx, target, texObj, level);
 
-      ASSERT(ctx->Driver.TexSubImage3D);
-      (*ctx->Driver.TexSubImage3D)(ctx, target, level,
-				   xoffset, yoffset, zoffset,
-				   width, height, depth,
-				   format, type, pixels,
-				   &ctx->Unpack, texObj, texImage );
-      ctx->NewState |= _NEW_TEXTURE;
+         ctx->NewState |= _NEW_TEXTURE;
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2694,38 +2713,39 @@ _mesa_CopyTexImage1D( GLenum target, GLint level,
 
    texUnit = _mesa_get_current_tex_unit(ctx);
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
+
    _mesa_lock_texture(ctx, texObj);
    {
       texImage = _mesa_get_tex_image(ctx, texObj, target, level);
       if (!texImage) {
 	 _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D");
-	 goto out;
       }
+      else {
+         if (texImage->Data) {
+            ctx->Driver.FreeTexImageData( ctx, texImage );
+         }
 
-      if (texImage->Data) {
-	 ctx->Driver.FreeTexImageData( ctx, texImage );
-      }
-      
-      ASSERT(texImage->Data == NULL);
+         ASSERT(texImage->Data == NULL);
 
-      clear_teximage_fields(texImage); /* not really needed, but helpful */
-      _mesa_init_teximage_fields(ctx, target, texImage, postConvWidth, 1, 1,
-				 border, internalFormat);
+         clear_teximage_fields(texImage); /* not really needed, but helpful */
+         _mesa_init_teximage_fields(ctx, target, texImage, postConvWidth, 1, 1,
+                                    border, internalFormat);
 
+         ASSERT(ctx->Driver.CopyTexImage1D);
+         ctx->Driver.CopyTexImage1D(ctx, target, level, internalFormat,
+                                    x, y, width, border);
 
-      ASSERT(ctx->Driver.CopyTexImage1D);
-      (*ctx->Driver.CopyTexImage1D)(ctx, target, level, internalFormat,
-				    x, y, width, border);
+         ASSERT(texImage->TexFormat);
 
-      ASSERT(texImage->TexFormat);
+         check_gen_mipmap(ctx, target, texObj, level);
 
-      update_fbo_texture(ctx, texObj, face, level);
+         update_fbo_texture(ctx, texObj, face, level);
 
-      /* state update */
-      texObj->_Complete = GL_FALSE;
-      ctx->NewState |= _NEW_TEXTURE;
+         /* state update */
+         texObj->_Complete = GL_FALSE;
+         ctx->NewState |= _NEW_TEXTURE;
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2773,33 +2793,34 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat,
 
       if (!texImage) {
 	 _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D");
-	 goto out;
-      }
-      
-      if (texImage->Data) {
-	 ctx->Driver.FreeTexImageData( ctx, texImage );
       }
-      
-      ASSERT(texImage->Data == NULL);
+      else {
+         if (texImage->Data) {
+            ctx->Driver.FreeTexImageData( ctx, texImage );
+         }
+
+         ASSERT(texImage->Data == NULL);
+
+         clear_teximage_fields(texImage); /* not really needed, but helpful */
+         _mesa_init_teximage_fields(ctx, target, texImage,
+                                    postConvWidth, postConvHeight, 1,
+                                    border, internalFormat);
+
+         ASSERT(ctx->Driver.CopyTexImage2D);
+         ctx->Driver.CopyTexImage2D(ctx, target, level, internalFormat,
+                                    x, y, width, height, border);
+
+         ASSERT(texImage->TexFormat);
 
-      clear_teximage_fields(texImage); /* not really needed, but helpful */
-      _mesa_init_teximage_fields(ctx, target, texImage,
-				 postConvWidth, postConvHeight, 1,
-				 border, internalFormat);
-      
-      ASSERT(ctx->Driver.CopyTexImage2D);
-      (*ctx->Driver.CopyTexImage2D)(ctx, target, level, internalFormat,
-				    x, y, width, height, border);
-      
-      ASSERT(texImage->TexFormat);
+         check_gen_mipmap(ctx, target, texObj, level);
 
-      update_fbo_texture(ctx, texObj, face, level);
+         update_fbo_texture(ctx, texObj, face, level);
 
-      /* state update */
-      texObj->_Complete = GL_FALSE;
-      ctx->NewState |= _NEW_TEXTURE;
+         /* state update */
+         texObj->_Complete = GL_FALSE;
+         ctx->NewState |= _NEW_TEXTURE;
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2844,23 +2865,25 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level,
 
       if (copytexsubimage_error_check2(ctx, 1, target, level,
 				       xoffset, 0, 0, postConvWidth, 1,
-				       texImage))
-	 goto out;
-      
+				       texImage)) {
+         /* error was recorded */
+      }
+      else {
+         /* If we have a border, xoffset=-1 is legal.  Bias by border width */
+         xoffset += texImage->Border;
 
-      /* If we have a border, xoffset=-1 is legal.  Bias by border width */
-      xoffset += texImage->Border;
+         if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y,
+                                        &width, &height)) {
+            ASSERT(ctx->Driver.CopyTexSubImage1D);
+            ctx->Driver.CopyTexSubImage1D(ctx, target, level,
+                                          xoffset, x, y, width);
 
-      if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y,
-                                     &width, &height)) {
-         ASSERT(ctx->Driver.CopyTexSubImage1D);
-         ctx->Driver.CopyTexSubImage1D(ctx, target, level,
-                                       xoffset, x, y, width);
-      }
+            check_gen_mipmap(ctx, target, texObj, level);
 
-      ctx->NewState |= _NEW_TEXTURE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2903,24 +2926,29 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level,
       }
 #endif
 
-      if (copytexsubimage_error_check2(ctx, 2, target, level, xoffset, yoffset, 0,
-				       postConvWidth, postConvHeight, texImage))
-	 goto out;
+      if (copytexsubimage_error_check2(ctx, 2, target, level,
+                                       xoffset, yoffset, 0,
+				       postConvWidth, postConvHeight,
+                                       texImage)) {
+         /* error was recorded */
+      }
+      else {
+         /* If we have a border, xoffset=-1 is legal.  Bias by border width */
+         xoffset += texImage->Border;
+         yoffset += texImage->Border;
+
+         if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y,
+                                        &width, &height)) {
+            ASSERT(ctx->Driver.CopyTexSubImage2D);
+            ctx->Driver.CopyTexSubImage2D(ctx, target, level, xoffset, yoffset,
+                                          x, y, width, height);
 
-      /* If we have a border, xoffset=-1 is legal.  Bias by border width */
-      xoffset += texImage->Border;
-      yoffset += texImage->Border;
+            check_gen_mipmap(ctx, target, texObj, level);
 
-      if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y,
-                                     &width, &height)) {
-         ASSERT(ctx->Driver.CopyTexSubImage2D);
-         ctx->Driver.CopyTexSubImage2D(ctx, target, level,
-				       xoffset, yoffset, x, y, width, height);
+            ctx->NewState |= _NEW_TEXTURE;
+         }
       }
-
-      ctx->NewState |= _NEW_TEXTURE;
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -2965,25 +2993,28 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level,
 
       if (copytexsubimage_error_check2(ctx, 3, target, level, xoffset, yoffset,
 				       zoffset, postConvWidth, postConvHeight,
-				       texImage))
-	 goto out;
-
-      /* If we have a border, xoffset=-1 is legal.  Bias by border width */
-      xoffset += texImage->Border;
-      yoffset += texImage->Border;
-      zoffset += texImage->Border;
-      
-      if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y,
-                                     &width, &height)) {
-         ASSERT(ctx->Driver.CopyTexSubImage3D);
-         ctx->Driver.CopyTexSubImage3D(ctx, target, level,
-				       xoffset, yoffset, zoffset,
-				       x, y, width, height);
+				       texImage)) {
+         /* error was recored */
       }
+      else {
+         /* If we have a border, xoffset=-1 is legal.  Bias by border width */
+         xoffset += texImage->Border;
+         yoffset += texImage->Border;
+         zoffset += texImage->Border;
+
+         if (_mesa_clip_copytexsubimage(ctx, &xoffset, &yoffset, &x, &y,
+                                        &width, &height)) {
+            ASSERT(ctx->Driver.CopyTexSubImage3D);
+            ctx->Driver.CopyTexSubImage3D(ctx, target, level,
+                                          xoffset, yoffset, zoffset,
+                                          x, y, width, height);
 
-      ctx->NewState |= _NEW_TEXTURE;
+            check_gen_mipmap(ctx, target, texObj, level);
+
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -3217,28 +3248,29 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level,
 	 texImage = _mesa_get_tex_image(ctx, texObj, target, level);
 	 if (!texImage) {
 	    _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage1D");
-	    goto out;
-	 }
-	 
-	 if (texImage->Data) {
-	    ctx->Driver.FreeTexImageData( ctx, texImage );
 	 }
-	 ASSERT(texImage->Data == NULL);
+         else {
+            if (texImage->Data) {
+               ctx->Driver.FreeTexImageData( ctx, texImage );
+            }
+            ASSERT(texImage->Data == NULL);
+
+            _mesa_init_teximage_fields(ctx, target, texImage, width, 1, 1,
+                                       border, internalFormat);
 
-	 _mesa_init_teximage_fields(ctx, target, texImage, width, 1, 1,
-				    border, internalFormat);
+            ASSERT(ctx->Driver.CompressedTexImage1D);
+            ctx->Driver.CompressedTexImage1D(ctx, target, level,
+                                             internalFormat, width, border,
+                                             imageSize, data,
+                                             texObj, texImage);
 
-	 ASSERT(ctx->Driver.CompressedTexImage1D);
-	 (*ctx->Driver.CompressedTexImage1D)(ctx, target, level,
-					     internalFormat, width, border,
-					     imageSize, data,
-					     texObj, texImage);
+            check_gen_mipmap(ctx, target, texObj, level);
 
-	 /* state update */
-	 texObj->_Complete = GL_FALSE;
-	 ctx->NewState |= _NEW_TEXTURE;
+            /* state update */
+            texObj->_Complete = GL_FALSE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
       }
-   out:
       _mesa_unlock_texture(ctx, texObj);
    }
    else if (target == GL_PROXY_TEXTURE_1D) {
@@ -3320,28 +3352,29 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level,
 	 texImage = _mesa_get_tex_image(ctx, texObj, target, level);
 	 if (!texImage) {
 	    _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage2D");
-	    goto out;
 	 }
-	 
-	 if (texImage->Data) {
-	    ctx->Driver.FreeTexImageData( ctx, texImage );
-	 }
-	 ASSERT(texImage->Data == NULL);
-
-	 _mesa_init_teximage_fields(ctx, target, texImage, width, height, 1,
-				    border, internalFormat);
-
-	 ASSERT(ctx->Driver.CompressedTexImage2D);
-	 (*ctx->Driver.CompressedTexImage2D)(ctx, target, level,
-					     internalFormat, width, height,
-					     border, imageSize, data,
-					     texObj, texImage);
-	 
-	 /* state update */
-	 texObj->_Complete = GL_FALSE;
-	 ctx->NewState |= _NEW_TEXTURE;
-      }
-   out:
+         else {
+            if (texImage->Data) {
+               ctx->Driver.FreeTexImageData( ctx, texImage );
+            }
+            ASSERT(texImage->Data == NULL);
+
+            _mesa_init_teximage_fields(ctx, target, texImage, width, height, 1,
+                                       border, internalFormat);
+
+            ASSERT(ctx->Driver.CompressedTexImage2D);
+            ctx->Driver.CompressedTexImage2D(ctx, target, level,
+                                             internalFormat, width, height,
+                                             border, imageSize, data,
+                                             texObj, texImage);
+
+            check_gen_mipmap(ctx, target, texObj, level);
+
+            /* state update */
+            texObj->_Complete = GL_FALSE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
       _mesa_unlock_texture(ctx, texObj);
    }
    else if (target == GL_PROXY_TEXTURE_2D ||
@@ -3421,29 +3454,31 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level,
 	 texImage = _mesa_get_tex_image(ctx, texObj, target, level);
 	 if (!texImage) {
 	    _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage3D");
-	    goto out;
 	 }
-	 
-	 if (texImage->Data) {
-	    ctx->Driver.FreeTexImageData( ctx, texImage );
-	 }
-	 ASSERT(texImage->Data == NULL);
-
-	 _mesa_init_teximage_fields(ctx, target, texImage, width, height, depth,
-				    border, internalFormat);
-
-	 ASSERT(ctx->Driver.CompressedTexImage3D);
-	 (*ctx->Driver.CompressedTexImage3D)(ctx, target, level,
-					     internalFormat,
-					     width, height, depth,
-					     border, imageSize, data,
-					     texObj, texImage);
-	 
-	 /* state update */
-	 texObj->_Complete = GL_FALSE;
-	 ctx->NewState |= _NEW_TEXTURE;
-      }
-   out:
+         else {
+            if (texImage->Data) {
+               ctx->Driver.FreeTexImageData( ctx, texImage );
+            }
+            ASSERT(texImage->Data == NULL);
+
+            _mesa_init_teximage_fields(ctx, target, texImage,
+                                       width, height, depth,
+                                       border, internalFormat);
+
+            ASSERT(ctx->Driver.CompressedTexImage3D);
+            ctx->Driver.CompressedTexImage3D(ctx, target, level,
+                                             internalFormat,
+                                             width, height, depth,
+                                             border, imageSize, data,
+                                             texObj, texImage);
+
+            check_gen_mipmap(ctx, target, texObj, level);
+
+            /* state update */
+            texObj->_Complete = GL_FALSE;
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
       _mesa_unlock_texture(ctx, texObj);
    }
    else if (target == GL_PROXY_TEXTURE_3D) {
@@ -3509,6 +3544,7 @@ _mesa_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset,
 
    texUnit = _mesa_get_current_tex_unit(ctx);
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
+
    _mesa_lock_texture(ctx, texObj);
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
@@ -3517,26 +3553,25 @@ _mesa_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset,
       if ((GLint) format != texImage->InternalFormat) {
 	 _mesa_error(ctx, GL_INVALID_OPERATION,
 		     "glCompressedTexSubImage1D(format)");
-	 goto out;
       }
+      else if ((width == 1 || width == 2) &&
+               (GLuint) width != texImage->Width) {
+	 _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glCompressedTexSubImage1D(width)");
+      }
+      else if (width > 0) {
+         if (ctx->Driver.CompressedTexSubImage1D) {
+            ctx->Driver.CompressedTexSubImage1D(ctx, target, level,
+                                                xoffset, width,
+                                                format, imageSize, data,
+                                                texObj, texImage);
+         }
 
-      if ((width == 1 || width == 2) && (GLuint) width != texImage->Width) {
-	 _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage1D(width)");
-	 goto out;
-      }
-      
-      if (width == 0)
-	 goto out;  /* no-op, not an error */
+         check_gen_mipmap(ctx, target, texObj, level);
 
-      if (ctx->Driver.CompressedTexSubImage1D) {
-	 (*ctx->Driver.CompressedTexSubImage1D)(ctx, target, level,
-						xoffset, width,
-						format, imageSize, data,
-						texObj, texImage);
+         ctx->NewState |= _NEW_TEXTURE;
       }
-      ctx->NewState |= _NEW_TEXTURE;
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -3566,6 +3601,7 @@ _mesa_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset,
 
    texUnit = _mesa_get_current_tex_unit(ctx);
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
+
    _mesa_lock_texture(ctx, texObj);
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
@@ -3574,27 +3610,26 @@ _mesa_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset,
       if ((GLint) format != texImage->InternalFormat) {
 	 _mesa_error(ctx, GL_INVALID_OPERATION,
 		     "glCompressedTexSubImage2D(format)");
-	 goto out;
       }
-
-      if (((width == 1 || width == 2) && (GLuint) width != texImage->Width) ||
-	  ((height == 1 || height == 2) && (GLuint) height != texImage->Height)) {
+      else if (((width == 1 || width == 2)
+                && (GLuint) width != texImage->Width) ||
+               ((height == 1 || height == 2)
+                && (GLuint) height != texImage->Height)) {
 	 _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage2D(size)");
-	 goto out;
       }
-      
-      if (width == 0 || height == 0)
-	 goto out;  /* no-op, not an error */
-
-      if (ctx->Driver.CompressedTexSubImage2D) {
-	 (*ctx->Driver.CompressedTexSubImage2D)(ctx, target, level,
+      else if (width > 0 && height > 0) {
+         if (ctx->Driver.CompressedTexSubImage2D) {
+            ctx->Driver.CompressedTexSubImage2D(ctx, target, level,
 						xoffset, yoffset, width, height,
 						format, imageSize, data,
 						texObj, texImage);
+         }
+
+         check_gen_mipmap(ctx, target, texObj, level);
+
+         ctx->NewState |= _NEW_TEXTURE;
       }
-      ctx->NewState |= _NEW_TEXTURE;
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
@@ -3623,6 +3658,7 @@ _mesa_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset,
 
    texUnit = _mesa_get_current_tex_unit(ctx);
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
+
    _mesa_lock_texture(ctx, texObj);
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
@@ -3631,29 +3667,29 @@ _mesa_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset,
       if ((GLint) format != texImage->InternalFormat) {
 	 _mesa_error(ctx, GL_INVALID_OPERATION,
 		     "glCompressedTexSubImage3D(format)");
-	 goto out;
       }
-
-      if (((width == 1 || width == 2) && (GLuint) width != texImage->Width) ||
-	  ((height == 1 || height == 2) && (GLuint) height != texImage->Height) ||
-	  ((depth == 1 || depth == 2) && (GLuint) depth != texImage->Depth)) {
+      else if (((width == 1 || width == 2)
+                && (GLuint) width != texImage->Width) ||
+               ((height == 1 || height == 2)
+                && (GLuint) height != texImage->Height) ||
+               ((depth == 1 || depth == 2)
+                && (GLuint) depth != texImage->Depth)) {
 	 _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage3D(size)");
-	 goto out;
       }
-      
-      if (width == 0 || height == 0 || depth == 0)
-	 goto out;  /* no-op, not an error */
-
-      if (ctx->Driver.CompressedTexSubImage3D) {
-	 (*ctx->Driver.CompressedTexSubImage3D)(ctx, target, level,
+      else if (width > 0 && height > 0 && depth > 0) {
+         if (ctx->Driver.CompressedTexSubImage3D) {
+            ctx->Driver.CompressedTexSubImage3D(ctx, target, level,
 						xoffset, yoffset, zoffset,
 						width, height, depth,
 						format, imageSize, data,
 						texObj, texImage);
+         }
+
+         check_gen_mipmap(ctx, target, texObj, level);
+
+         ctx->NewState |= _NEW_TEXTURE;
       }
-      ctx->NewState |= _NEW_TEXTURE;
    }
- out:
    _mesa_unlock_texture(ctx, texObj);
 }
 
diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c
index da55ac8697..31832184c0 100644
--- a/src/mesa/main/texobj.c
+++ b/src/mesa/main/texobj.c
@@ -29,9 +29,7 @@
 
 
 #include "mfeatures.h"
-#if FEATURE_colortable
 #include "colortab.h"
-#endif
 #include "context.h"
 #include "enums.h"
 #include "fbobject.h"
@@ -194,9 +192,7 @@ _mesa_delete_texture_object( GLcontext *ctx, struct gl_texture_object *texObj )
     */
    texObj->Target = 0x99;
 
-#if FEATURE_colortable
    _mesa_free_colortable_data(&texObj->Palette);
-#endif
 
    /* free the texture images */
    for (face = 0; face < 6; face++) {
@@ -495,7 +491,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
 	     t->Image[face][baseLevel]->Width2 != w ||
 	     t->Image[face][baseLevel]->Height2 != h) {
 	    t->_Complete = GL_FALSE;
-	    incomplete(t, "Non-quare cubemap image");
+	    incomplete(t, "Cube face missing or mismatched size");
 	    return;
 	 }
       }
@@ -1102,8 +1098,6 @@ _mesa_PrioritizeTextures( GLsizei n, const GLuint *texName,
          struct gl_texture_object *t = _mesa_lookup_texture(ctx, texName[i]);
          if (t) {
             t->Priority = CLAMP( priorities[i], 0.0F, 1.0F );
-	    if (ctx->Driver.PrioritizeTexture)
-	       ctx->Driver.PrioritizeTexture( ctx, t, t->Priority );
          }
       }
    }
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index 8292d43eb6..16492bd7bc 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.c
@@ -31,9 +31,7 @@
 #include "glheader.h"
 #include "mfeatures.h"
 #include "colormac.h"
-#if FEATURE_colortable
 #include "colortab.h"
-#endif
 #include "context.h"
 #include "enums.h"
 #include "macros.h"
@@ -309,10 +307,6 @@ _mesa_ActiveTextureARB(GLenum texture)
       /* update current stack pointer */
       ctx->CurrentStack = &ctx->TextureMatrixStack[texUnit];
    }
-
-   if (ctx->Driver.ActiveTexture) {
-      (*ctx->Driver.ActiveTexture)( ctx, (GLuint) texUnit );
-   }
 }
 
 
@@ -362,9 +356,6 @@ update_texture_matrices( GLcontext *ctx )
 	 if (ctx->Texture.Unit[u]._ReallyEnabled &&
 	     ctx->TextureMatrixStack[u].Top->type != MATRIX_IDENTITY)
 	    ctx->Texture._TexMatEnabled |= ENABLE_TEXMAT(u);
-
-	 if (ctx->Driver.TextureMatrix)
-	    ctx->Driver.TextureMatrix( ctx, u, ctx->TextureMatrixStack[u].Top);
       }
    }
 }
@@ -759,9 +750,7 @@ _mesa_init_texture(GLcontext *ctx)
    ctx->Texture.CurrentUnit = 0;      /* multitexture */
    ctx->Texture._EnabledUnits = 0x0;
    ctx->Texture.SharedPalette = GL_FALSE;
-#if FEATURE_colortable
    _mesa_init_colortable(&ctx->Texture.Palette);
-#endif
 
    for (u = 0; u < MAX_TEXTURE_UNITS; u++)
       init_texture_unit(ctx, u);
@@ -802,10 +791,8 @@ _mesa_free_texture_data(GLcontext *ctx)
    for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++)
       ctx->Driver.DeleteTexture(ctx, ctx->Texture.ProxyTex[tgt]);
 
-#if FEATURE_colortable
    for (u = 0; u < MAX_TEXTURE_IMAGE_UNITS; u++)
       _mesa_free_colortable_data(&ctx->Texture.Unit[u].ColorTable);
-#endif
 }
 
 
diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c
index a22db628d3..f553a898f9 100644
--- a/src/mesa/main/texstore.c
+++ b/src/mesa/main/texstore.c
@@ -56,9 +56,7 @@
 #include "bufferobj.h"
 #include "colormac.h"
 #include "context.h"
-#if FEATURE_convolve
 #include "convolve.h"
-#endif
 #include "image.h"
 #include "macros.h"
 #include "mipmap.h"
@@ -3302,32 +3300,9 @@ _mesa_set_fetch_functions(struct gl_texture_image *texImage, GLuint dims)
 }
 
 
-/**
- * Choose the actual storage format for a new texture image.
- * Mainly, this is a wrapper for the driver's ChooseTextureFormat() function.
- * Also set some other texImage fields related to texture compression, etc.
- * \param ctx  rendering context
- * \param texImage  the gl_texture_image
- * \param dims  texture dimensions (1, 2 or 3)
- * \param format  the user-specified format parameter
- * \param type  the user-specified type parameter
- * \param internalFormat  the user-specified internal format hint
- */
 static void
-choose_texture_format(GLcontext *ctx, struct gl_texture_image *texImage,
-                      GLuint dims,
-                      GLenum format, GLenum type, GLint internalFormat)
+compute_texture_size(GLcontext *ctx, struct gl_texture_image *texImage)
 {
-   ASSERT(dims == 1 || dims == 2 || dims == 3);
-   ASSERT(ctx->Driver.ChooseTextureFormat);
-
-   texImage->TexFormat
-      = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type);
-
-   ASSERT(texImage->TexFormat);
-
-   _mesa_set_fetch_functions(texImage, dims);
-
    if (texImage->TexFormat->TexelBytes == 0) {
       /* must be a compressed format */
       texImage->IsCompressed = GL_TRUE;
@@ -3365,7 +3340,12 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level,
    GLint sizeInBytes;
    (void) border;
 
-   choose_texture_format(ctx, texImage, 1, format, type, internalFormat);
+   texImage->TexFormat
+      = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type);
+   ASSERT(texImage->TexFormat);
+
+   _mesa_set_fetch_functions(texImage, 1);
+   compute_texture_size(ctx, texImage);
 
    /* allocate memory */
    if (texImage->IsCompressed)
@@ -3403,11 +3383,6 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
@@ -3434,7 +3409,12 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level,
    GLint texelBytes, sizeInBytes;
    (void) border;
 
-   choose_texture_format(ctx, texImage, 2, format, type, internalFormat);
+   texImage->TexFormat
+      = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type);
+   ASSERT(texImage->TexFormat);
+
+   _mesa_set_fetch_functions(texImage, 2);
+   compute_texture_size(ctx, texImage);
 
    texelBytes = texImage->TexFormat->TexelBytes;
 
@@ -3481,11 +3461,6 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
@@ -3508,7 +3483,12 @@ _mesa_store_teximage3d(GLcontext *ctx, GLenum target, GLint level,
    GLint texelBytes, sizeInBytes;
    (void) border;
 
-   choose_texture_format(ctx, texImage, 3, format, type, internalFormat);
+   texImage->TexFormat
+      = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, format, type);
+   ASSERT(texImage->TexFormat);
+
+   _mesa_set_fetch_functions(texImage, 3);
+   compute_texture_size(ctx, texImage);
 
    texelBytes = texImage->TexFormat->TexelBytes;
 
@@ -3555,11 +3535,6 @@ _mesa_store_teximage3d(GLcontext *ctx, GLenum target, GLint level,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
@@ -3601,11 +3576,6 @@ _mesa_store_texsubimage1d(GLcontext *ctx, GLenum target, GLint level,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
@@ -3654,11 +3624,6 @@ _mesa_store_texsubimage2d(GLcontext *ctx, GLenum target, GLint level,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
@@ -3707,11 +3672,6 @@ _mesa_store_texsubimage3d(GLcontext *ctx, GLenum target, GLint level,
       }
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, packing);
 }
 
@@ -3762,7 +3722,12 @@ _mesa_store_compressed_teximage2d(GLcontext *ctx, GLenum target, GLint level,
    ASSERT(texImage->Depth == 1);
    ASSERT(texImage->Data == NULL); /* was freed in glCompressedTexImage2DARB */
 
-   choose_texture_format(ctx, texImage, 2, 0, 0, internalFormat);
+   texImage->TexFormat
+      = ctx->Driver.ChooseTextureFormat(ctx, internalFormat, 0, 0);
+   ASSERT(texImage->TexFormat);
+
+   _mesa_set_fetch_functions(texImage, 2);
+   compute_texture_size(ctx, texImage);
 
    /* allocate storage */
    texImage->Data = _mesa_alloc_texmemory(imageSize);
@@ -3781,11 +3746,6 @@ _mesa_store_compressed_teximage2d(GLcontext *ctx, GLenum target, GLint level,
    ASSERT(texImage->CompressedSize == (GLuint) imageSize);
    MEMCPY(texImage->Data, data, imageSize);
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack);
 }
 
@@ -3891,11 +3851,6 @@ _mesa_store_compressed_texsubimage2d(GLcontext *ctx, GLenum target,
       src += srcRowStride;
    }
 
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-
    _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack);
 }
 
diff --git a/src/mesa/main/varray.h b/src/mesa/main/varray.h
index becc67c29d..ef790c504e 100644
--- a/src/mesa/main/varray.h
+++ b/src/mesa/main/varray.h
@@ -129,6 +129,11 @@ extern void GLAPIENTRY
 _mesa_MultiDrawElementsEXT( GLenum mode, const GLsizei *count, GLenum type,
                             const GLvoid **indices, GLsizei primcount );
 
+extern void GLAPIENTRY
+_mesa_MultiDrawElementsBaseVertex( GLenum mode,
+				   const GLsizei *count, GLenum type,
+				   const GLvoid **indices, GLsizei primcount,
+				   const GLint *basevertex);
 
 extern void GLAPIENTRY
 _mesa_MultiModeDrawArraysIBM( const GLenum * mode, const GLint * first,
@@ -159,6 +164,16 @@ extern void GLAPIENTRY
 _mesa_DrawRangeElements(GLenum mode, GLuint start, GLuint end, GLsizei count,
                         GLenum type, const GLvoid *indices);
 
+extern void GLAPIENTRY
+_mesa_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+			     const GLvoid *indices, GLint basevertex);
+
+extern void GLAPIENTRY
+_mesa_DrawRangeElementsBaseVertex(GLenum mode, GLuint start, GLuint end,
+				  GLsizei count, GLenum type,
+				  const GLvoid *indices,
+				  GLint basevertex);
+
 
 extern void
 _mesa_copy_client_array(GLcontext *ctx,
diff --git a/src/mesa/main/version.h b/src/mesa/main/version.h
index f24e11cc51..0cae1860a3 100644
--- a/src/mesa/main/version.h
+++ b/src/mesa/main/version.h
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.6.1
+ * Version:  7.7
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
@@ -30,9 +30,9 @@
 
 /* Mesa version */
 #define MESA_MAJOR 7
-#define MESA_MINOR 6
-#define MESA_PATCH 1
-#define MESA_VERSION_STRING "7.6.1-devel"
+#define MESA_MINOR 7
+#define MESA_PATCH 0
+#define MESA_VERSION_STRING "7.7-devel"
 
 /* To make version comparison easy */
 #define MESA_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))
diff --git a/src/mesa/main/vtxfmt.c b/src/mesa/main/vtxfmt.c
index 8d6f560a80..c9eea8ab83 100644
--- a/src/mesa/main/vtxfmt.c
+++ b/src/mesa/main/vtxfmt.c
@@ -27,12 +27,18 @@
  */
 
 #include "glheader.h"
+#include "api_arrayelt.h"
 #include "api_loopback.h"
 #include "context.h"
 #include "imports.h"
 #include "mtypes.h"
 #include "state.h"
 #include "vtxfmt.h"
+#include "eval.h"
+#include "dlist.h"
+
+
+#if FEATURE_beginend
 
 
 /* The neutral vertex format.  This wraps all tnl module functions,
@@ -82,18 +88,16 @@
 static void
 install_vtxfmt( struct _glapi_table *tab, const GLvertexformat *vfmt )
 {
-   SET_ArrayElement(tab, vfmt->ArrayElement);
+   _mesa_install_arrayelt_vtxfmt(tab, vfmt);
+
    SET_Color3f(tab, vfmt->Color3f);
    SET_Color3fv(tab, vfmt->Color3fv);
    SET_Color4f(tab, vfmt->Color4f);
    SET_Color4fv(tab, vfmt->Color4fv);
    SET_EdgeFlag(tab, vfmt->EdgeFlag);
-   SET_EvalCoord1f(tab, vfmt->EvalCoord1f);
-   SET_EvalCoord1fv(tab, vfmt->EvalCoord1fv);
-   SET_EvalCoord2f(tab, vfmt->EvalCoord2f);
-   SET_EvalCoord2fv(tab, vfmt->EvalCoord2fv);
-   SET_EvalPoint1(tab, vfmt->EvalPoint1);
-   SET_EvalPoint2(tab, vfmt->EvalPoint2);
+
+   _mesa_install_eval_vtxfmt(tab, vfmt);
+
    SET_FogCoordfEXT(tab, vfmt->FogCoordfEXT);
    SET_FogCoordfvEXT(tab, vfmt->FogCoordfvEXT);
    SET_Indexf(tab, vfmt->Indexf);
@@ -125,8 +129,9 @@ install_vtxfmt( struct _glapi_table *tab, const GLvertexformat *vfmt )
    SET_Vertex3fv(tab, vfmt->Vertex3fv);
    SET_Vertex4f(tab, vfmt->Vertex4f);
    SET_Vertex4fv(tab, vfmt->Vertex4fv);
-   SET_CallList(tab, vfmt->CallList);
-   SET_CallLists(tab, vfmt->CallLists);
+
+   _mesa_install_dlist_vtxfmt(tab, vfmt);
+
    SET_Begin(tab, vfmt->Begin);
    SET_End(tab, vfmt->End);
    SET_Rectf(tab, vfmt->Rectf);
@@ -134,9 +139,9 @@ install_vtxfmt( struct _glapi_table *tab, const GLvertexformat *vfmt )
    SET_DrawElements(tab, vfmt->DrawElements);
    SET_DrawRangeElements(tab, vfmt->DrawRangeElements);
    SET_MultiDrawElementsEXT(tab, vfmt->MultiDrawElementsEXT);
-   SET_EvalMesh1(tab, vfmt->EvalMesh1);
-   SET_EvalMesh2(tab, vfmt->EvalMesh2);
-   ASSERT(tab->EvalMesh2);
+   SET_DrawElementsBaseVertex(tab, vfmt->DrawElementsBaseVertex);
+   SET_DrawRangeElementsBaseVertex(tab, vfmt->DrawRangeElementsBaseVertex);
+   SET_MultiDrawElementsBaseVertex(tab, vfmt->MultiDrawElementsBaseVertex);
 
    /* GL_NV_vertex_program */
    SET_VertexAttrib1fNV(tab, vfmt->VertexAttrib1fNV);
@@ -193,3 +198,6 @@ void _mesa_restore_exec_vtxfmt( GLcontext *ctx )
 
    tnl->SwapCount = 0;
 }
+
+
+#endif /* FEATURE_beginend */
diff --git a/src/mesa/main/vtxfmt.h b/src/mesa/main/vtxfmt.h
index 76f108e023..fb6c23abe9 100644
--- a/src/mesa/main/vtxfmt.h
+++ b/src/mesa/main/vtxfmt.h
@@ -33,6 +33,8 @@
 #ifndef _VTXFMT_H_
 #define _VTXFMT_H_
 
+#if FEATURE_beginend
+
 extern void _mesa_init_exec_vtxfmt( GLcontext *ctx );
 
 extern void _mesa_install_exec_vtxfmt( GLcontext *ctx, const GLvertexformat *vfmt );
@@ -40,4 +42,28 @@ extern void _mesa_install_save_vtxfmt( GLcontext *ctx, const GLvertexformat *vfm
 
 extern void _mesa_restore_exec_vtxfmt( GLcontext *ctx );
 
-#endif
+#else /* FEATURE_beginend */
+
+static INLINE void
+_mesa_init_exec_vtxfmt( GLcontext *ctx )
+{
+}
+
+static INLINE void
+_mesa_install_exec_vtxfmt( GLcontext *ctx, const GLvertexformat *vfmt )
+{
+}
+
+static INLINE void
+_mesa_install_save_vtxfmt( GLcontext *ctx, const GLvertexformat *vfmt )
+{
+}
+
+static INLINE void
+_mesa_restore_exec_vtxfmt( GLcontext *ctx )
+{
+}
+
+#endif /* FEATURE_beginend */
+
+#endif /* _VTXFMT_H_ */
diff --git a/src/mesa/main/vtxfmt_tmp.h b/src/mesa/main/vtxfmt_tmp.h
index 1308d0aa46..d56a2bb95e 100644
--- a/src/mesa/main/vtxfmt_tmp.h
+++ b/src/mesa/main/vtxfmt_tmp.h
@@ -354,6 +354,44 @@ static void GLAPIENTRY TAG(DrawRangeElements)( GLenum mode, GLuint start,
    CALL_DrawRangeElements(GET_DISPATCH(), ( mode, start, end, count, type, indices ));
 }
 
+static void GLAPIENTRY TAG(DrawElementsBaseVertex)( GLenum mode,
+						    GLsizei count,
+						    GLenum type,
+						    const GLvoid *indices,
+						    GLint basevertex)
+{
+   PRE_LOOPBACK( DrawElementsBaseVertex );
+   CALL_DrawElementsBaseVertex(GET_DISPATCH(), ( mode, count, type,
+						 indices, basevertex ));
+}
+
+static void GLAPIENTRY TAG(DrawRangeElementsBaseVertex)( GLenum mode,
+							 GLuint start,
+							 GLuint end,
+							 GLsizei count,
+							 GLenum type,
+							 const GLvoid *indices,
+							 GLint basevertex)
+{
+   PRE_LOOPBACK( DrawRangeElementsBaseVertex );
+   CALL_DrawRangeElementsBaseVertex(GET_DISPATCH(), ( mode, start, end,
+						      count, type, indices,
+						      basevertex ));
+}
+
+static void GLAPIENTRY TAG(MultiDrawElementsBaseVertex)( GLenum mode,
+							 const GLsizei *count,
+							 GLenum type,
+							 const GLvoid **indices,
+							 GLsizei primcount,
+							 const GLint *basevertex)
+{
+   PRE_LOOPBACK( MultiDrawElementsBaseVertex );
+   CALL_MultiDrawElementsBaseVertex(GET_DISPATCH(), ( mode, count, type,
+						      indices,
+						      primcount, basevertex ));
+}
+
 static void GLAPIENTRY TAG(EvalMesh1)( GLenum mode, GLint i1, GLint i2 )
 {
    PRE_LOOPBACK( EvalMesh1 );
@@ -534,6 +572,9 @@ static GLvertexformat TAG(vtxfmt) = {
    TAG(DrawElements),
    TAG(DrawRangeElements),
    TAG(MultiDrawElementsEXT),
+   TAG(DrawElementsBaseVertex),
+   TAG(DrawRangeElementsBaseVertex),
+   TAG(MultiDrawElementsBaseVertex),
    TAG(EvalMesh1),
    TAG(EvalMesh2)
 };
diff --git a/src/mesa/math/m_clip_tmp.h b/src/mesa/math/m_clip_tmp.h
index f3a589be05..2e30964057 100644
--- a/src/mesa/math/m_clip_tmp.h
+++ b/src/mesa/math/m_clip_tmp.h
@@ -44,7 +44,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points4)( GLvector4f *clip_vec,
                                                      GLvector4f *proj_vec,
                                                      GLubyte clipMask[],
                                                      GLubyte *orMask,
-                                                     GLubyte *andMask )
+                                                     GLubyte *andMask,
+						     GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLfloat *from = (GLfloat *)clip_vec->start;
@@ -66,16 +67,20 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points4)( GLvector4f *clip_vec,
       mask |= (((cw < -cx) << CLIP_LEFT_SHIFT));
       mask |= (((cw < cy) << CLIP_TOP_SHIFT));
       mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT));
-      mask |= (((cw < cz) << CLIP_FAR_SHIFT));
-      mask |= (((cw < -cz) << CLIP_NEAR_SHIFT));
+      if (viewport_z_clip) {
+	 mask |= (((cw < cz) << CLIP_FAR_SHIFT));
+	 mask |= (((cw < -cz) << CLIP_NEAR_SHIFT));
+      }
 #else /* !defined(macintosh)) */
       GLubyte mask = 0;
       if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT;
       if ( cx + cw < 0) mask |= CLIP_LEFT_BIT;
       if (-cy + cw < 0) mask |= CLIP_TOP_BIT;
       if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT;
-      if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
-      if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
+      if (viewport_z_clip) {
+	 if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
+	 if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
+      }
 #endif /* defined(macintosh) */
 
       clipMask[i] = mask;
@@ -119,7 +124,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_np_points4)( GLvector4f *clip_vec,
 							GLvector4f *proj_vec,
 							GLubyte clipMask[],
 							GLubyte *orMask,
-							GLubyte *andMask )
+							GLubyte *andMask,
+							GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLuint count = clip_vec->count;
@@ -141,16 +147,20 @@ static GLvector4f * _XFORMAPI TAG(cliptest_np_points4)( GLvector4f *clip_vec,
       mask |= (((cw < -cx) << CLIP_LEFT_SHIFT));
       mask |= (((cw < cy) << CLIP_TOP_SHIFT));
       mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT));
-      mask |= (((cw < cz) << CLIP_FAR_SHIFT));
-      mask |= (((cw < -cz) << CLIP_NEAR_SHIFT));
+      if (viewport_z_clip) {
+	 mask |= (((cw < cz) << CLIP_FAR_SHIFT));
+	 mask |= (((cw < -cz) << CLIP_NEAR_SHIFT));
+      }
 #else /* !defined(macintosh)) */
       GLubyte mask = 0;
       if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT;
       if ( cx + cw < 0) mask |= CLIP_LEFT_BIT;
       if (-cy + cw < 0) mask |= CLIP_TOP_BIT;
       if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT;
-      if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
-      if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
+      if (viewport_z_clip) {
+	 if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
+	 if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
+      }
 #endif /* defined(macintosh) */
 
       clipMask[i] = mask;
@@ -171,7 +181,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points3)( GLvector4f *clip_vec,
                                                      GLvector4f *proj_vec,
                                                      GLubyte clipMask[],
                                                      GLubyte *orMask,
-                                                     GLubyte *andMask )
+                                                     GLubyte *andMask,
+						     GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLuint count = clip_vec->count;
@@ -187,8 +198,10 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points3)( GLvector4f *clip_vec,
       else if (cx < -1.0)  mask |= CLIP_LEFT_BIT;
       if (cy >  1.0)       mask |= CLIP_TOP_BIT;
       else if (cy < -1.0)  mask |= CLIP_BOTTOM_BIT;
-      if (cz >  1.0)       mask |= CLIP_FAR_BIT;
-      else if (cz < -1.0)  mask |= CLIP_NEAR_BIT;
+      if (viewport_z_clip) {
+	 if (cz >  1.0)       mask |= CLIP_FAR_BIT;
+	 else if (cz < -1.0)  mask |= CLIP_NEAR_BIT;
+      }
       clipMask[i] = mask;
       tmpOrMask |= mask;
       tmpAndMask &= mask;
@@ -204,7 +217,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points2)( GLvector4f *clip_vec,
                                                      GLvector4f *proj_vec,
                                                      GLubyte clipMask[],
                                                      GLubyte *orMask,
-                                                     GLubyte *andMask )
+                                                     GLubyte *andMask,
+						     GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLuint count = clip_vec->count;
@@ -231,7 +245,7 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points2)( GLvector4f *clip_vec,
 }
 
 
-static void TAG(init_c_cliptest)( void )
+void TAG(init_c_cliptest)( void )
 {
    _mesa_clip_tab[4] = TAG(cliptest_points4);
    _mesa_clip_tab[3] = TAG(cliptest_points3);
diff --git a/src/mesa/math/m_debug_clip.c b/src/mesa/math/m_debug_clip.c
index 460fed4a75..95ae5a347d 100644
--- a/src/mesa/math/m_debug_clip.c
+++ b/src/mesa/math/m_debug_clip.c
@@ -67,7 +67,8 @@ static GLvector4f *ref_cliptest_points4( GLvector4f *clip_vec,
 					 GLvector4f *proj_vec,
 					 GLubyte clipMask[],
 					 GLubyte *orMask,
-					 GLubyte *andMask )
+					 GLubyte *andMask,
+					 GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLuint count = clip_vec->count;
@@ -87,8 +88,10 @@ static GLvector4f *ref_cliptest_points4( GLvector4f *clip_vec,
       if (  cx + cw < 0 ) mask |= CLIP_LEFT_BIT;
       if ( -cy + cw < 0 ) mask |= CLIP_TOP_BIT;
       if (  cy + cw < 0 ) mask |= CLIP_BOTTOM_BIT;
-      if ( -cz + cw < 0 ) mask |= CLIP_FAR_BIT;
-      if (  cz + cw < 0 ) mask |= CLIP_NEAR_BIT;
+      if (viewport_z_clip) {
+	 if ( -cz + cw < 0 ) mask |= CLIP_FAR_BIT;
+	 if (  cz + cw < 0 ) mask |= CLIP_NEAR_BIT;
+      }
       clipMask[i] = mask;
       if ( mask ) {
 	 c++;
@@ -122,7 +125,8 @@ static GLvector4f *ref_cliptest_points3( GLvector4f *clip_vec,
 					 GLvector4f *proj_vec,
 					 GLubyte clipMask[],
 					 GLubyte *orMask,
-					 GLubyte *andMask )
+					 GLubyte *andMask,
+                                         GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLuint count = clip_vec->count;
@@ -138,8 +142,10 @@ static GLvector4f *ref_cliptest_points3( GLvector4f *clip_vec,
       else if ( cx < -1.0 )	mask |= CLIP_LEFT_BIT;
       if ( cy >  1.0 )		mask |= CLIP_TOP_BIT;
       else if ( cy < -1.0 )	mask |= CLIP_BOTTOM_BIT;
-      if ( cz >  1.0 )		mask |= CLIP_FAR_BIT;
-      else if ( cz < -1.0 )	mask |= CLIP_NEAR_BIT;
+      if (viewport_z_clip) {
+         if ( cz >  1.0 )		mask |= CLIP_FAR_BIT;
+         else if ( cz < -1.0 )	mask |= CLIP_NEAR_BIT;
+      }
       clipMask[i] = mask;
       tmpOrMask |= mask;
       tmpAndMask &= mask;
@@ -154,7 +160,8 @@ static GLvector4f * ref_cliptest_points2( GLvector4f *clip_vec,
 					  GLvector4f *proj_vec,
 					  GLubyte clipMask[],
 					  GLubyte *orMask,
-					  GLubyte *andMask )
+					  GLubyte *andMask,
+                                          GLboolean viewport_z_clip )
 {
    const GLuint stride = clip_vec->stride;
    const GLuint count = clip_vec->count;
@@ -163,6 +170,9 @@ static GLvector4f * ref_cliptest_points2( GLvector4f *clip_vec,
    GLubyte tmpOrMask = *orMask;
    GLubyte tmpAndMask = *andMask;
    GLuint i;
+
+   (void) viewport_z_clip;
+
    for ( i = 0 ; i < count ; i++, STRIDE_F(from, stride) ) {
       const GLfloat cx = from[0], cy = from[1];
       GLubyte mask = 0;
@@ -208,6 +218,7 @@ static int test_cliptest_function( clip_func func, int np,
 #ifdef  RUN_DEBUG_BENCHMARK
    int cycle_i;                /* the counter for the benchmarks we run */
 #endif
+   GLboolean viewport_z_clip = GL_TRUE;
 
    (void) cycles;
 
@@ -247,15 +258,15 @@ static int test_cliptest_function( clip_func func, int np,
    dco = rco = 0;
    dca = rca = CLIP_FRUSTUM_BITS;
 
-   ref_cliptest[psize]( source, ref, rm, &rco, &rca );
+   ref_cliptest[psize]( source, ref, rm, &rco, &rca, viewport_z_clip );
 
    if ( mesa_profile ) {
       BEGIN_RACE( *cycles );
-      func( source, dest, dm, &dco, &dca );
+      func( source, dest, dm, &dco, &dca, viewport_z_clip );
       END_RACE( *cycles );
    }
    else {
-      func( source, dest, dm, &dco, &dca );
+      func( source, dest, dm, &dco, &dca, viewport_z_clip );
    }
 
    if ( dco != rco ) {
diff --git a/src/mesa/math/m_xform.h b/src/mesa/math/m_xform.h
index 7ef76e0b92..33421ad1c0 100644
--- a/src/mesa/math/m_xform.h
+++ b/src/mesa/math/m_xform.h
@@ -43,7 +43,8 @@
 
 extern void
 _math_init_transformation(void);
-
+extern void
+init_c_cliptest(void);
 
 /* KW: Clip functions now do projective divide as well.  The projected
  * coordinates are very useful to us because they let us cull
@@ -102,7 +103,8 @@ typedef GLvector4f * (_XFORMAPIP clip_func)( GLvector4f *vClip,
 					     GLvector4f *vProj,
 					     GLubyte clipMask[],
 					     GLubyte *orMask,
-					     GLubyte *andMask );
+					     GLubyte *andMask,
+					     GLboolean viewport_z_clip );
 
 typedef void (*dotprod_func)( GLfloat *out,
 			      GLuint out_stride,
diff --git a/src/mesa/shader/lex.yy.c b/src/mesa/shader/lex.yy.c
index fefef573ee..17843c2121 100644
--- a/src/mesa/shader/lex.yy.c
+++ b/src/mesa/shader/lex.yy.c
@@ -53,7 +53,6 @@ typedef int flex_int32_t;
 typedef unsigned char flex_uint8_t; 
 typedef unsigned short int flex_uint16_t;
 typedef unsigned int flex_uint32_t;
-#endif /* ! C99 */
 
 /* Limits of integral types. */
 #ifndef INT8_MIN
@@ -84,6 +83,8 @@ typedef unsigned int flex_uint32_t;
 #define UINT32_MAX             (4294967295U)
 #endif
 
+#endif /* ! C99 */
+
 #endif /* ! FLEXINT_H */
 
 #ifdef __cplusplus
@@ -348,8 +349,8 @@ static void yy_fatal_error (yyconst char msg[] ,yyscan_t yyscanner );
 	*yy_cp = '\0'; \
 	yyg->yy_c_buf_p = yy_cp;
 
-#define YY_NUM_RULES 183
-#define YY_END_OF_BUFFER 184
+#define YY_NUM_RULES 170
+#define YY_END_OF_BUFFER 171
 /* This struct is not used in this scanner,
    but its presence is necessary. */
 struct yy_trans_info
@@ -357,82 +358,101 @@ struct yy_trans_info
 	flex_int32_t yy_verify;
 	flex_int32_t yy_nxt;
 	};
-static yyconst flex_int16_t yy_accept[675] =
+static yyconst flex_int16_t yy_accept[850] =
     {   0,
-        0,    0,  184,  182,  180,  179,  182,  182,  152,  178,
-      154,  154,  154,  154,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  180,    0,    0,  181,  152,    0,
-      153,  155,  175,  175,    0,    0,    0,    0,  175,    0,
-        0,    0,    0,    0,    0,    0,  132,  176,  133,  134,
-      166,  166,  166,  166,    0,  154,    0,  140,  141,  142,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,    0,    0,    0,    0,
-        0,    0,    0,    0,    0,  174,    0,    0,    0,    0,
-        0,    0,    0,    0,    0,  173,  173,    0,    0,    0,
+        0,    0,  171,  169,  167,  166,  169,  169,  139,  165,
+      141,  141,  141,  141,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  167,    0,    0,  168,  139,
+        0,  140,  142,  162,  162,    0,    0,    0,    0,  162,
+        0,    0,    0,    0,    0,    0,    0,  119,  163,  120,
+      121,  153,  153,  153,  153,    0,  141,    0,  127,  128,
+      129,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,    0,    0,    0,    0,
+        0,    0,    0,    0,    0,  161,    0,    0,    0,    0,
+        0,    0,    0,    0,    0,  160,  160,    0,    0,    0,
         0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
-      172,  172,  172,    0,    0,    0,    0,    0,    0,    0,
-        0,    0,    0,  163,  163,  163,  164,  164,  165,  156,
-      155,  156,    0,  157,   11,   13,  152,   15,  152,  152,
-       16,   18,  152,   20,   22,   24,   26,    6,   28,   30,
-       31,   33,   35,   38,   36,   40,   41,   43,   45,   47,
-
-       49,   51,  152,  152,  152,   53,   55,  152,   57,   59,
-       61,  152,   63,   65,   67,   69,  152,   71,   73,   75,
-       77,  152,  152,  152,  152,  152,  152,    0,    0,    0,
-        0,  155,    0,    0,    0,    0,    0,    0,    0,    0,
-        0,    0,   93,   94,   96,    0,  171,    0,    0,    0,
-        0,    0,    0,  110,    0,    0,    0,    0,    0,    0,
-        0,    0,    0,    0,  170,  169,  169,  122,    0,    0,
-        0,    0,    0,    0,    0,    0,    0,    0,  160,  160,
-      161,  162,    0,  158,  152,  152,  152,  152,  152,  152,
-      152,  152,  143,  152,  152,  152,  152,  152,  152,  152,
-
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  144,  152,  152,  152,  152,  152,  152,
-      152,  152,   10,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,    0,  177,    0,    0,    0,   86,   87,
-        0,    0,    0,    0,    0,    0,    0,   98,    0,    0,
+      159,  159,  159,    0,    0,    0,    0,    0,    0,    0,
+        0,    0,    0,  150,  150,  150,  151,  151,  152,  143,
+      142,  143,    0,  144,   11,   12,  139,   13,  139,  139,
+       14,   15,  139,   16,   17,   18,   19,   20,   21,    6,
+
+       22,   23,   24,   25,   26,   28,   27,   29,   30,   31,
+       32,   33,   34,   35,  139,  139,  139,  139,  139,   40,
+       41,  139,   42,   43,   44,   45,   46,   47,   48,  139,
+       49,   50,   51,   52,   53,   54,   55,  139,   56,   57,
+       58,   59,  139,  139,   64,   65,  139,  139,  139,  139,
+      139,  139,    0,    0,    0,    0,  142,    0,    0,    0,
+        0,    0,    0,    0,    0,    0,    0,   80,   81,   83,
+        0,  158,    0,    0,    0,    0,    0,    0,   97,    0,
+        0,    0,    0,    0,    0,    0,    0,    0,    0,  157,
+      156,  156,  109,    0,    0,    0,    0,    0,    0,    0,
+
+        0,    0,    0,  147,  147,  148,  149,    0,  145,   11,
+       11,  139,   12,   12,   12,  139,  139,  139,  139,  139,
+       15,   15,  139,  130,   16,   16,  139,   17,   17,  139,
+       18,   18,  139,   19,   19,  139,   20,   20,  139,   21,
+       21,  139,   22,   22,  139,   24,   24,  139,   25,   25,
+      139,   28,   28,  139,   27,   27,  139,   30,   30,  139,
+       31,   31,  139,   32,   32,  139,   33,   33,  139,   34,
+       34,  139,   35,   35,  139,  139,  139,  139,   36,  139,
+       38,  139,   40,   40,  139,   41,   41,  139,  131,   42,
+       42,  139,   43,   43,  139,  139,   45,   45,  139,   46,
+
+       46,  139,   47,   47,  139,   48,   48,  139,  139,   49,
+       49,  139,   50,   50,  139,   51,   51,  139,   52,   52,
+      139,   53,   53,  139,   54,   54,  139,  139,   10,   56,
+      139,   57,  139,   58,  139,   59,  139,   60,  139,   62,
+      139,   64,   64,  139,  139,  139,  139,  139,  139,  139,
+      139,    0,  164,    0,    0,    0,   73,   74,    0,    0,
+        0,    0,    0,    0,    0,   85,    0,    0,    0,    0,
         0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
-        0,    0,    0,    0,    0,    0,  168,    0,    0,    0,
-      126,    0,  128,    0,    0,    0,    0,    0,    0,  167,
-      159,  152,  152,  152,    4,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-
-      152,  152,  152,  152,  152,  152,    9,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,   82,  152,  152,    0,    0,    0,
-        0,    0,   88,   89,    0,    0,    0,    0,   97,    0,
-        0,  101,  104,    0,    0,    0,    0,    0,    0,    0,
-      115,  116,    0,    0,    0,    0,  121,    0,    0,    0,
-        0,    0,    0,    0,    0,    0,    0,  152,  152,  152,
-      152,  152,  152,    5,  152,  152,  152,  152,  152,  152,
-      152,  152,  152,  152,  152,  152,  152,  152,  152,  152,
-      152,    7,    8,  152,  152,  152,  152,  152,  152,  152,
-
-      152,  152,  152,  152,  152,  152,  152,  152,  152,   83,
-      152,   79,    0,    0,    0,    0,  137,    0,    0,    0,
-        0,    0,    0,    0,    0,    0,  107,    0,  111,  112,
-        0,  114,    0,    0,    0,    0,    0,    0,    0,    0,
-        0,    0,  130,  131,    0,    0,  138,   12,    3,   14,
-      148,  149,  152,   17,   19,   21,   23,   25,   27,   29,
-       32,   34,   39,   37,   42,   44,   46,   48,   50,   52,
-       54,   56,   58,   60,   62,  152,  152,  152,   64,   66,
-       68,   70,   72,   74,   76,   78,  152,   81,  139,    0,
-        0,   84,    0,   90,    0,    0,    0,   99,    0,    0,
-
-        0,    0,    0,    0,  113,    0,    0,  119,  106,    0,
-        0,    0,    0,    0,    0,  135,    0,  152,  145,  146,
-      152,   80,    0,    0,    0,    0,   92,   95,  100,    0,
-        0,  105,    0,    0,    0,  118,    0,    0,    0,    0,
-      127,  129,    0,  152,  152,    2,    1,    0,   91,    0,
-      103,    0,  109,  117,    0,    0,  124,  125,  136,  152,
-      147,    0,  102,    0,  120,  123,  152,   85,  108,  152,
-      152,  150,  151,    0
+        0,    0,    0,    0,  155,    0,    0,    0,  113,    0,
+      115,    0,    0,    0,    0,    0,    0,  154,  146,  139,
+
+      139,  139,    4,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,    9,   37,   39,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+       60,  139,   61,   62,  139,   63,  139,  139,  139,  139,
+      139,   69,  139,  139,    0,    0,    0,    0,    0,   75,
+       76,    0,    0,    0,    0,   84,    0,    0,   88,   91,
+        0,    0,    0,    0,    0,    0,    0,  102,  103,    0,
+        0,    0,    0,  108,    0,    0,    0,    0,    0,    0,
+
+        0,    0,    0,    0,  139,  139,  139,  139,  139,  139,
+        5,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+        7,    8,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,  139,  139,  139,  139,  139,  139,
+      139,  139,  139,  139,   61,  139,  139,   63,  139,  139,
+      139,  139,  139,   70,  139,   66,    0,    0,    0,    0,
+      124,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+       94,    0,   98,   99,    0,  101,    0,    0,    0,    0,
+        0,    0,    0,    0,    0,    0,  117,  118,    0,    0,
+
+      125,   11,    3,   12,  135,  136,  139,   14,   15,   16,
+       17,   18,   19,   20,   21,   22,   24,   25,   28,   27,
+       30,   31,   32,   33,   34,   35,   40,   41,   42,   43,
+       44,   45,   46,   47,   48,  139,  139,  139,   49,   50,
+       51,   52,   53,   54,   55,   56,   57,   58,   59,  139,
+      139,  139,  139,   64,   65,  139,   68,  126,    0,    0,
+       71,    0,   77,    0,    0,    0,   86,    0,    0,    0,
+        0,    0,    0,  100,    0,    0,  106,   93,    0,    0,
+        0,    0,    0,    0,  122,    0,  139,  132,  133,  139,
+       60,  139,   62,  139,   67,    0,    0,    0,    0,   79,
+
+       82,   87,    0,    0,   92,    0,    0,    0,  105,    0,
+        0,    0,    0,  114,  116,    0,  139,  139,   61,   63,
+        2,    1,    0,   78,    0,   90,    0,   96,  104,    0,
+        0,  111,  112,  123,  139,  134,    0,   89,    0,  107,
+      110,  139,   72,   95,  139,  139,  137,  138,    0
     } ;
 
 static yyconst flex_int32_t yy_ec[256] =
@@ -478,280 +498,357 @@ static yyconst flex_int32_t yy_meta[68] =
         2,    2,    2,    2,    2,    2,    2
     } ;
 
-static yyconst flex_int16_t yy_base[678] =
+static yyconst flex_int16_t yy_base[853] =
     {   0,
-        0,    0,  954,  955,   66,  955,  948,  949,    0,   69,
-       85,  128,  140,  152,  151,   58,   39,   48,   75,  927,
-      158,  160,   73,   59,   71,  170,   54,  920,  890,  889,
-      901,  885,  899,  898,  142,  927,  939,  955,    0,  206,
-      955,  189,  168,  171,   53,   27,   66,  119,  175,  899,
-      885,  123,  170,  883,  895,  183,  955,  198,  225,   99,
-      212,  219,  223,  227,  285,  297,  308,  955,  955,  955,
-      904,  917,  911,  165,  900,  903,  899,  914,  224,  896,
-      910,  194,  896,  909,  900,  913,  890,  901,  892,  294,
-      893,  884,  893,  884,  883,  884,  878,  884,  895,  881,
-
-      878,  890,  893,  880,  873,  889,  865,  193,  139,  885,
-      861,  846,  841,  858,  834,  839,  865,  167,  854,  259,
-      849,  325,  282,  851,  832,  302,  842,  838,  833,   43,
-      839,  825,  841,  838,  829,  305,  309,  831,  820,  834,
-      837,  819,  834,  821,  818,  825,  275,  833,  254,  299,
-      317,  327,  331,  810,  827,  828,  821,  803,  310,  804,
-      826,  817,  316,  327,  331,  335,  339,  343,  347,  955,
-      405,  416,  422,  428,  825,  240,  849,    0,  848,  831,
-      821,  820,  840,  818,  817,  816,  815,    0,  814,    0,
-      813,  812,    0,  811,  810,    0,  809,  808,  807,  806,
-
-      805,  804,  820,  813,  826,  800,  799,  805,  797,  796,
-      795,  816,  793,  792,  791,  790,  800,  788,  787,  786,
-      785,  777,  776,  761,  761,  760,  759,  802,  774,  762,
-      434,  442,  416,  766,  186,  763,  757,  757,  751,  764,
-      764,  749,  955,  955,  764,  752,  418,  759,  281,  756,
-      762,  308,  757,  955,  748,  755,  754,  757,  743,  742,
-      746,  741,  278,  746,  420,  428,  430,  955,  738,  736,
-      736,  744,  745,  727,  421,  732,  738,  419,  426,  430,
-      434,  438,  496,  502,  752,  764,  750,  749,  742,  756,
-      746,  745,    0,  744,  743,  742,  741,  740,  739,  738,
-
-      737,  736,  735,  734,  733,  732,  731,  730,  733,  726,
-      733,  726,  725,    0,  724,  723,  722,  725,  720,  719,
-      718,  717,    0,  716,  715,  714,  713,  691,  685,  690,
-      696,  679,  694,  315,  955,  693,  683,  687,  955,  955,
-      677,  686,  672,  689,  672,  675,  669,  955,  670,  669,
-      666,  673,  666,  674,  670,  680,  677,  659,  665,  672,
-      656,  655,  673,  655,  667,  666,  955,  665,  655,  659,
-      955,  646,  955,  651,  651,  659,  642,  643,  653,  955,
-      955,  685,  667,  683,    0,  507,  681,  681,  680,  679,
-      678,  677,  676,  675,  674,  673,  672,  671,  670,  669,
-
-      668,  667,  666,  665,  652,  645,    0,  662,  661,  660,
-      659,  658,  636,  656,  655,  654,  653,  652,  651,  650,
-      649,  618,  621,  601,    0,  602,  595,  602,  601,  602,
-      594,  612,  955,  955,  594,  592,  602,  595,  955,  590,
-      607,  330,  955,  598,  582,  583,  592,  583,  582,  582,
-      955,  581,  590,  580,  596,  593,  955,  592,  590,  579,
-      580,  576,  568,  575,  570,  571,  566,  592,  592,  590,
-      604,  603,  598,    0,  586,  585,  584,  583,  582,  581,
-      580,  579,  578,  577,  576,  575,  574,  573,  572,  571,
-      570,    0,    0,  569,  568,  567,  566,  565,  509,  564,
-
-      563,  562,  561,  560,  559,  558,  557,  535,  535,    0,
-      542,    0,  576,  575,  524,  542,  955,  537,  532,  525,
-      521,  533,  523,  521,  517,  533,  524,  523,  955,  955,
-      526,  955,  521,  514,  503,  514,  506,  510,  523,  518,
-      521,  503,  955,  955,  515,  504,  955,    0,    0,    0,
-        0,    0,  543,    0,    0,    0,    0,    0,    0,    0,
+        0,    0, 1299, 1300,   66, 1300, 1293, 1294,    0,   69,
+       85,  128,  140,  152,  151,   58,   56,   63,   76, 1272,
+      158,  160,   39,  163,  173,  189,   52, 1265,   76, 1235,
+     1234, 1246, 1230, 1244, 1243,  105, 1272, 1284, 1300,    0,
+      225, 1300,  218,  160,  157,   20,  123,   66,  119,  192,
+     1244, 1230,   54,  162, 1228, 1240,  194, 1300,  200,  195,
+       98,  227,  196,  231,  235,  293,  305,  316, 1300, 1300,
+     1300, 1249, 1262, 1256,  223, 1245, 1248, 1244, 1259,  107,
+      298, 1241, 1255,  246, 1241, 1254, 1245, 1258, 1235, 1246,
+     1237,  182, 1238, 1229, 1238, 1229, 1228, 1229,  144, 1223,
+
+     1229, 1240, 1231, 1225, 1222, 1223, 1227,  289, 1236, 1223,
+      302, 1230, 1217, 1231, 1207,   65,  315,  276, 1227, 1226,
+     1202, 1187, 1182, 1199, 1175, 1180, 1206,  279, 1195,  293,
+     1190,  342,  299, 1192, 1173,  317, 1183, 1179, 1174,  207,
+     1180, 1166, 1182, 1179, 1170,  320,  324, 1172, 1161, 1175,
+     1178, 1160, 1175, 1162, 1159, 1166,  284, 1174,  227,  288,
+      327,  342,  345, 1151, 1168, 1169, 1162, 1144,  318, 1145,
+     1167, 1158,  330,  341,  345,  349,  353,  357,  361, 1300,
+      419,  430,  436,  442,  440,  441, 1191,    0, 1190, 1173,
+     1163,  443, 1183,  444,  451,  468,  470,  472,  471,    0,
+
+      496,    0,  497,  498,    0,  499,  500,    0,  524,  525,
+      526,  536,  537,  553, 1178, 1171, 1184,  354,  356,  561,
+      563, 1165,  564,  565, 1157,  580,  590,  591,  592, 1178,
+      593,  617,  618,  619,  629,  630, 1155, 1165,  330,  362,
+      419,  483,  445,  364,  646, 1153, 1145, 1144, 1129, 1129,
+     1128, 1127, 1170, 1142, 1130,  662,  669,  643, 1134,  487,
+     1131, 1125, 1125, 1119, 1132, 1132, 1117, 1300, 1300, 1132,
+     1120,  646, 1127,  135, 1124, 1130,  561, 1125, 1300, 1116,
+     1123, 1122, 1125, 1111, 1110, 1114, 1109,  448, 1114,  650,
+      653,  665, 1300, 1106, 1104, 1104, 1112, 1113, 1095,  670,
+
+     1100, 1106,  486,  579,  655,  661,  668,  726,  732, 1112,
+      682, 1119, 1110,  688,  730, 1117, 1116, 1109, 1123, 1113,
+     1104,  712, 1111,    0, 1102,  731, 1109, 1100,  733, 1107,
+     1098,  734, 1105, 1096,  736, 1103, 1094,  737, 1101, 1092,
+      738, 1099, 1090,  739, 1097, 1088,  740, 1095, 1086,  741,
+     1093, 1084,  742, 1091, 1082,  743, 1089, 1080,  744, 1087,
+     1078,  745, 1085, 1076,  746, 1083, 1074,  747, 1081, 1072,
+      748, 1079, 1070,  749, 1077, 1080, 1073, 1080,    0, 1073,
+        0, 1088, 1063,  750, 1070, 1061,  751, 1068,    0, 1059,
+      752, 1066, 1057,  755, 1064, 1063, 1054,  758, 1061, 1052,
+
+      776, 1059, 1050,  777, 1057, 1048,  779, 1055, 1058, 1045,
+      780, 1052, 1043,  782, 1050, 1041,  783, 1048, 1039,  784,
+     1046, 1037,  785, 1044, 1035,  786, 1042, 1041,    0, 1032,
+     1039, 1030, 1037, 1028, 1035, 1026, 1033,  787, 1032,  788,
+     1047, 1022,  789, 1029, 1028, 1006, 1000, 1005, 1011,  994,
+     1009,  424, 1300, 1008,  998, 1002, 1300, 1300,  992, 1001,
+      987, 1004,  987,  990,  984, 1300,  985,  984,  981,  988,
+      981,  989,  985,  995,  992,  974,  980,  987,  971,  970,
+      988,  970,  982,  981, 1300,  980,  970,  974, 1300,  961,
+     1300,  966,  966,  974,  957,  958,  968, 1300, 1300, 1000,
+
+      982,  998,    0,  798,  996,  996,  995,  994,  993,  992,
+      991,  990,  989,  988,  987,  986,  985,  984,  983,  982,
+      981,  980,  979,  978,  965,  958,    0,    0,    0,  975,
+      974,  973,  972,  971,  970,  969,  968,  967,  945,  965,
+      964,  963,  962,  961,  960,  959,  958,  957,  956,  955,
+      929,  936,  793,  927,  934,  794,  950,  949,  918,  921,
+      901,    0,  902,  895,  902,  901,  902,  894,  912, 1300,
+     1300,  894,  892,  902,  895, 1300,  890,  907,  516, 1300,
+      898,  882,  883,  892,  883,  882,  882, 1300,  881,  890,
+      880,  896,  893, 1300,  892,  890,  879,  880,  876,  868,
+
+      875,  870,  871,  866,  892,  892,  890,  904,  903,  898,
+        0,  886,  885,  884,  883,  882,  881,  880,  879,  878,
+      877,  876,  875,  874,  873,  872,  871,  870,  869,  868,
+        0,    0,  867,  866,  865,  864,  863,  862,  861,  860,
+      859,  804,  858,  857,  856,  855,  854,  853,  852,  851,
+      850,  849,  848,  865,  839,  846,  862,  836,  843,  841,
+      840,  818,  818,    0,  825,    0,  859,  858,  807,  825,
+     1300,  820,  815,  808,  804,  816,  806,  804,  800,  816,
+      807,  806, 1300, 1300,  809, 1300,  804,  797,  786,  797,
+      789,  793,  806,  801,  804,  786, 1300, 1300,  798,  787,
+
+     1300,    0,    0,    0,    0,    0,  826,    0,    0,    0,
         0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
-        0,    0,    0,    0,    0,  539,  538,  536,    0,    0,
-        0,    0,    0,    0,    0,    0,  494,    0,    0,  545,
-      544,  955,  491,  955,  495,  495,  504,  955,  488,  502,
-
-      483,  485,  482,  490,  955,  468,  479,  955,  955,  483,
-      479,  472,  470,  470,  483,  955,  467,  507,    0,    0,
-      507,    0,  514,  513,  472,  433,  955,  955,  955,  435,
-      435,  955,  429,  386,  377,  955,  366,  365,  323,  328,
-      955,  955,  339,  348,  337,  955,  955,  307,  955,  305,
-      955,  257,  955,  955,  247,  221,  955,  955,  955,  236,
-        0,  213,  955,  150,  955,  955,  232,  955,  955,  162,
-      138,    0,    0,  955,  541,  108,  544
+        0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+        0,    0,    0,    0,    0,  814,  813,  802,    0,    0,
+        0,    0,    0,    0,    0,    0,    0,    0,    0,  785,
+      798,  779,  792,    0,    0,  656,    0,    0,  706,  702,
+     1300,  649, 1300,  648,  648,  654, 1300,  637,  645,  610,
+      612,  608,  608, 1300,  572,  583, 1300, 1300,  577,  573,
+      560,  557,  542,  555, 1300,  539,  573,    0,    0,  572,
+        0,  555,    0,  546,    0,  562,  551,  495,  479, 1300,
+
+     1300, 1300,  481,  481, 1300,  480,  443,   31, 1300,  141,
+      166,  171,  186, 1300, 1300,  211,  236,  276,    0,    0,
+     1300, 1300,  290, 1300,  325, 1300,  346, 1300, 1300,  343,
+      341, 1300, 1300, 1300,  365,    0,  380, 1300,  371, 1300,
+     1300,  486, 1300, 1300,  451,  458,    0,    0, 1300,  836,
+      503,  839
     } ;
 
-static yyconst flex_int16_t yy_def[678] =
+static yyconst flex_int16_t yy_def[853] =
     {   0,
-      674,    1,  674,  674,  674,  674,  674,  675,  676,  674,
-      674,  674,  674,  674,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  674,  674,  675,  674,  676,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  677,  674,  674,  674,  674,  674,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  676,
-      676,  676,  676,  676,  676,  676,  676,  676,  676,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  676,  676,  676,
-      676,  676,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  676,  676,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  676,
-      676,  674,  674,  674,  674,  674,  676,  674,  674,  676,
-      676,  676,  676,    0,  674,  674,  674
+      849,    1,  849,  849,  849,  849,  849,  850,  851,  849,
+      849,  849,  849,  849,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  849,  849,  850,  849,  851,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  852,  849,  849,  849,  849,
+      849,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  851,
+
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+
+      849,  849,  849,  849,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+
+      849,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  851,  851,  851,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  851,  851,  851,  851,
+      851,  851,  851,  851,  851,  849,  849,  849,  849,  849,
+
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  851,  851,  851,  851,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  851,  851,  849,  849,  849,  849,
+      849,  851,  849,  849,  851,  851,  851,  851,    0,  849,
+      849,  849
     } ;
 
-static yyconst flex_int16_t yy_nxt[1023] =
+static yyconst flex_int16_t yy_nxt[1368] =
     {   0,
         4,    5,    6,    5,    7,    8,    9,    4,   10,   11,
        12,   13,   14,   11,   11,   15,    9,   16,   17,   18,
        19,    9,    9,    9,   20,   21,   22,    9,   23,   24,
-        9,   25,   26,   27,    9,    9,    9,   28,    9,    9,
-        9,    9,    9,    9,    9,    9,   29,    9,    9,    9,
-        9,    9,    9,    9,    9,    9,   30,    9,   31,   32,
-       33,    9,   34,    9,    9,    9,    9,   35,   79,   35,
-       40,   80,  129,  108,   96,   81,  130,   41,   42,   42,
-       42,   42,   42,   42,   76,   82,   77,   97,   98,  240,
-       99,  109,   78,   65,   66,   66,   66,   66,   66,   66,
-
-       83,  241,   94,  100,   67,  127,   84,   95,  128,   39,
-       43,   44,   45,   46,   47,   48,   49,   50,   51,  131,
-      132,   52,   53,   54,   55,   56,   57,   58,   59,   60,
-       67,  133,   61,   62,   63,   64,   65,   66,   66,   66,
-       66,   66,   66,   35,  160,   35,   68,   67,   65,   66,
-       66,   66,   66,   66,   66,  219,  673,  161,   69,   67,
-       65,   66,   66,   66,   66,   66,   66,   71,  220,   72,
-       70,   67,  140,   67,  134,   90,   73,  135,  141,   86,
-      672,   87,   74,   91,   75,   67,   88,  101,   92,   89,
-      178,  102,  103,  104,   93,  105,  179,   67,   42,   42,
-
-       42,   42,   42,   42,  106,  189,  107,   40,  122,  123,
-      123,  142,  126,  123,  669,  123,  136,  137,  123,  217,
-      124,  124,  123,  190,  147,  143,  123,  125,  125,  123,
-      218,  337,  144,  123,  122,  148,  184,  185,  149,  151,
-      152,  150,  670,  671,  338,  153,  186,  118,  119,   45,
-       46,   47,   48,  154,   50,   51,  123,  162,   52,   53,
-       54,   55,   56,   57,  120,   59,   60,  668,  155,  121,
-      156,  286,  667,  157,  158,  163,  163,  163,  163,  666,
-      287,  159,  164,  163,  165,  166,  167,  163,  163,  168,
-      169,  163,  163,  163,  171,  171,  171,  171,  171,  171,
-
-      230,  665,  664,  260,  172,   65,   66,   66,   66,   66,
-       66,   66,  198,  261,  154,  173,   67,  174,  174,  174,
-      174,  174,  174,  233,  233,  364,  349,  257,  365,  233,
-      172,  199,  231,  258,  232,  232,  232,  232,  232,  232,
-      233,  350,   67,  233,  233,  236,  233,  233,  262,  233,
-      247,  233,  233,  353,  263,  273,  233,  663,  233,  233,
-      233,  428,  662,  233,  233,  274,  354,  233,  265,  233,
-      661,  264,  266,  267,  233,  233,  660,  429,  233,  278,
-      278,  278,  278,  524,  659,  233,  525,  658,  657,  233,
-      278,  278,  278,  278,  279,  278,  278,  280,  281,  278,
-
-      278,  278,  278,  278,  278,  278,  282,  278,  278,  278,
-      278,  278,  278,  278,   42,   42,   42,   42,   42,   42,
-      656,  655,  654,  283,  122,  284,  284,  284,  284,  284,
-      284,  174,  174,  174,  174,  174,  174,  174,  174,  174,
-      174,  174,  174,  232,  232,  232,  232,  232,  232,  653,
-      122,  232,  232,  232,  232,  232,  232,  335,  335,  335,
-      335,  335,  335,  335,  374,  335,  375,  335,  376,  335,
-      335,  367,  335,  652,  335,  335,  335,  335,  335,  651,
-      650,  377,  380,  380,  380,  380,  335,  649,  335,  380,
-      380,  380,  380,  381,  380,  380,  380,  380,  380,  380,
-
-      380,  380,  380,  380,  380,  284,  284,  284,  284,  284,
-      284,  284,  284,  284,  284,  284,  284,  471,  472,  576,
-      577,  648,  647,  646,  645,  644,  643,  642,  641,  640,
-      639,  638,  637,  636,  635,  634,  633,  632,  631,  473,
-      578,   37,   37,   37,  170,  170,  630,  629,  628,  627,
-      626,  625,  624,  623,  622,  621,  620,  619,  618,  617,
-      616,  615,  614,  613,  612,  611,  610,  609,  608,  607,
-      606,  605,  604,  603,  602,  601,  600,  599,  598,  597,
-      596,  595,  594,  593,  592,  591,  590,  589,  588,  587,
-      586,  585,  584,  583,  582,  581,  580,  579,  575,  574,
-
-      573,  572,  571,  570,  569,  568,  567,  566,  565,  564,
-      563,  562,  561,  560,  559,  558,  557,  556,  555,  554,
-      553,  552,  551,  550,  549,  548,  547,  546,  545,  544,
-      543,  542,  541,  540,  539,  538,  537,  536,  535,  534,
-      533,  532,  531,  530,  529,  528,  527,  526,  523,  522,
-      521,  520,  519,  518,  517,  516,  515,  514,  513,  512,
-      511,  510,  509,  508,  507,  506,  505,  504,  503,  502,
-      501,  500,  499,  498,  497,  496,  495,  494,  493,  492,
-      491,  490,  489,  488,  487,  486,  485,  484,  483,  482,
-      481,  480,  479,  478,  477,  476,  475,  474,  470,  469,
-
-      468,  467,  466,  465,  464,  463,  462,  461,  460,  459,
-      458,  457,  456,  455,  454,  453,  452,  451,  450,  449,
-      448,  447,  446,  445,  444,  443,  442,  441,  440,  439,
-      438,  437,  436,  435,  434,  433,  432,  431,  430,  427,
-      426,  425,  424,  423,  422,  421,  420,  419,  418,  417,
-      416,  415,  414,  413,  412,  411,  410,  409,  408,  407,
-      406,  405,  404,  403,  402,  401,  400,  399,  398,  397,
-      396,  395,  394,  393,  392,  391,  390,  389,  388,  387,
-      386,  385,  384,  383,  382,  379,  378,  373,  372,  371,
-      370,  369,  368,  366,  363,  362,  361,  360,  359,  358,
-
-      357,  356,  355,  352,  351,  348,  347,  346,  345,  344,
-      343,  342,  341,  340,  339,  336,  264,  236,  334,  333,
-      332,  331,  330,  329,  328,  327,  326,  325,  324,  323,
-      322,  321,  320,  319,  318,  317,  316,  315,  314,  313,
-      312,  311,  310,  309,  308,  307,  306,  305,  304,  303,
-      302,  301,  300,  299,  298,  297,  296,  295,  294,  293,
-      292,  291,  290,  289,  288,  285,  277,  276,  275,  272,
-      271,  270,  269,  268,  259,  256,  255,  254,  253,  252,
-      251,  250,  249,  248,  246,  245,  244,  243,  242,  239,
-      238,  237,  235,  234,  162,  229,  228,  227,  226,  225,
-
-      224,  223,  222,  221,  216,  215,  214,  213,  212,  211,
-      210,  209,  208,  207,  206,  205,  204,  203,  202,  201,
-      200,  197,  196,  195,  194,  193,  192,  191,  188,  187,
-      183,  182,  181,  180,  177,  176,  175,  146,  145,  139,
-      138,   38,  117,  116,  115,  114,  113,  112,  111,  110,
-       85,   38,   36,  674,    3,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674
+        9,   25,   26,   27,   28,    9,    9,   29,    9,    9,
+        9,    9,    9,    9,    9,    9,   30,    9,    9,    9,
+        9,    9,    9,    9,    9,    9,   31,    9,   32,   33,
+       34,    9,   35,    9,    9,    9,    9,   36,   96,   36,
+       41,  116,  137,   97,   80,  138,  829,   42,   43,   43,
+       43,   43,   43,   43,   77,   81,   78,  119,   82,  117,
+       83,  238,   79,   66,   67,   67,   67,   67,   67,   67,
+
+       84,   85,  239,  150,   68,  120,   36,   86,   36,  151,
+       44,   45,   46,   47,   48,   49,   50,   51,   52,  141,
+      142,   53,   54,   55,   56,   57,   58,   59,   60,   61,
+       68,  143,   62,   63,   64,   65,   66,   67,   67,   67,
+       67,   67,   67,  170,  194,  195,   69,   68,   66,   67,
+       67,   67,   67,   67,   67,  218,  171,  219,   70,   68,
+       66,   67,   67,   67,   67,   67,   67,   72,  139,   73,
+       71,   68,  140,   68,  144,   92,   74,  145,   98,   88,
+      467,   89,   75,   93,   76,   68,   90,   99,   94,   91,
+      101,  100,  102,  103,   95,  468,  830,   68,  136,  133,
+
+      210,  133,  133,  152,  133,  104,  105,  133,  106,  107,
+      108,  109,  110,  134,  111,  133,  112,  153,  133,  211,
+      135,  831,  113,  114,  154,  115,   41,   43,   43,   43,
+       43,   43,   43,  146,  147,  157,  832,  132,  165,  133,
+      166,  161,  162,  167,  168,  833,  158,  163,  188,  159,
+      133,  169,  160,  265,  189,  164,  834,  201,  133,  174,
+      173,  175,  176,  132,  835,  266,  128,  129,   46,   47,
+       48,   49,  172,   51,   52,  202,  285,   53,   54,   55,
+       56,   57,   58,  130,   60,   61,  286,  243,  131,  244,
+      173,  173,  173,  173,  177,  173,  173,  178,  179,  173,
+
+      173,  173,  181,  181,  181,  181,  181,  181,  228,  836,
+      196,  197,  182,   66,   67,   67,   67,   67,   67,   67,
+      198,  232,  229,  183,   68,  184,  184,  184,  184,  184,
+      184,  240,  134,  241,  255,  233,  282,  287,  182,  135,
+      258,  258,  283,  288,  242,  837,  258,  430,  164,  256,
+       68,  257,  257,  257,  257,  257,  257,  258,  258,  258,
+      261,  258,  258,  298,  258,  272,  258,  258,  258,  258,
+      431,  258,  381,  299,  258,  258,  379,  838,  258,  432,
+      440,  289,  258,  290,  258,  258,  291,  292,  380,  258,
+      382,  839,  258,  303,  303,  303,  303,  840,  441,  841,
+
+      258,  842,  433,  258,  303,  303,  303,  303,  304,  303,
+      303,  305,  306,  303,  303,  303,  303,  303,  303,  303,
+      307,  303,  303,  303,  303,  303,  303,  303,   43,   43,
+       43,   43,   43,   43,  843,  844,  434,  308,  132,  309,
+      309,  309,  309,  309,  309,  184,  184,  184,  184,  184,
+      184,  184,  184,  184,  184,  184,  184,  310,  313,  435,
+      321,  325,  311,  314,  132,  322,  326,  438,  328,  847,
+      565,  311,  315,  329,  322,  326,  848,  311,  314,  439,
+      312,  316,  329,  323,  327,  331,  566,  334,  340,  337,
+      332,  330,  335,  341,  338,  482,  845,  846,  483,  332,
+
+      436,  335,  341,  338,   40,  332,  828,  335,  333,  338,
+      336,  342,  339,  343,  346,  349,  352,  355,  344,  347,
+      350,  353,  356,  437,  827,  826,  825,  344,  347,  350,
+      353,  356,  455,  824,  347,  350,  345,  348,  351,  354,
+      357,  358,  361,  364,  823,  456,  359,  362,  365,  498,
+      498,  498,  498,  367,  370,  359,  362,  365,  368,  371,
+      822,  359,  362,  365,  360,  363,  366,  368,  371,  678,
+      373,  821,  679,  368,  371,  374,  369,  372,  383,  820,
+      386,  390,  393,  384,  374,  387,  391,  394,  819,  818,
+      374,  817,  384,  375,  387,  391,  394,  397,  816,  815,
+
+      814,  385,  398,  388,  392,  395,  471,  400,  403,  406,
+      410,  398,  401,  404,  407,  411,  813,  398,  812,  472,
+      399,  401,  404,  407,  411,  811,  810,  401,  404,  407,
+      402,  405,  408,  412,  413,  416,  419,  809,  808,  414,
+      417,  420,  498,  498,  498,  498,  422,  425,  414,  417,
+      420,  423,  426,  807,  414,  417,  420,  415,  418,  421,
+      423,  426,  806,  442,  805,  804,  423,  426,  443,  424,
+      427,  257,  257,  257,  257,  257,  257,  443,  257,  257,
+      257,  257,  257,  257,  453,  453,  444,  453,  453,  803,
+      453,  453,  453,  453,  453,  453,  802,  453,  801,  310,
+
+      453,  453,  800,  799,  453,  313,  485,  453,  453,  798,
+      797,  453,  453,  492,  796,  493,  795,  494,  499,  498,
+      498,  498,  312,  453,  498,  498,  498,  498,  316,  321,
+      495,  498,  498,  498,  498,  309,  309,  309,  309,  309,
+      309,  309,  309,  309,  309,  309,  309,  313,  325,  501,
+      328,  331,  323,  334,  337,  340,  343,  346,  349,  352,
+      355,  358,  361,  364,  367,  370,  373,  383,  386,  390,
+      316,  327,  393,  330,  333,  397,  336,  339,  342,  345,
+      348,  351,  354,  357,  360,  363,  366,  369,  372,  375,
+      385,  388,  392,  400,  403,  395,  406,  410,  399,  413,
+
+      416,  419,  422,  425,  551,  554,  442,  794,  608,  609,
+      655,  658,  793,  792,  736,  737,  402,  405,  791,  408,
+      412,  790,  415,  418,  421,  424,  427,  552,  555,  444,
+      610,  789,  788,  656,  659,  738,   38,   38,   38,  180,
+      180,  787,  786,  785,  784,  783,  782,  781,  780,  779,
+      778,  777,  776,  775,  774,  773,  772,  771,  770,  769,
+      768,  767,  766,  765,  764,  763,  762,  761,  760,  759,
+      758,  757,  756,  755,  754,  753,  659,  752,  751,  656,
+      750,  749,  748,  747,  746,  745,  744,  743,  742,  741,
+      740,  739,  735,  734,  733,  732,  731,  730,  729,  728,
+
+      727,  726,  725,  724,  723,  722,  721,  720,  719,  718,
+      717,  716,  715,  714,  713,  712,  711,  710,  709,  708,
+      707,  706,  705,  704,  703,  702,  701,  700,  699,  698,
+      697,  696,  695,  694,  693,  692,  691,  690,  689,  688,
+      687,  686,  685,  684,  683,  682,  681,  680,  677,  676,
+      675,  674,  673,  672,  671,  670,  669,  668,  667,  666,
+      665,  664,  663,  662,  661,  660,  657,  555,  654,  552,
+      653,  652,  651,  650,  649,  648,  647,  646,  645,  644,
+      643,  642,  641,  640,  639,  638,  637,  636,  635,  634,
+      633,  632,  631,  630,  629,  628,  627,  626,  625,  624,
+
+      623,  622,  621,  620,  619,  618,  617,  616,  615,  614,
+      613,  612,  611,  607,  606,  605,  604,  603,  602,  601,
+      600,  599,  598,  597,  596,  595,  594,  593,  592,  591,
+      590,  589,  588,  587,  586,  585,  584,  583,  582,  581,
+      580,  579,  578,  577,  576,  575,  574,  573,  572,  571,
+      570,  569,  568,  567,  564,  563,  562,  561,  560,  559,
+      558,  557,  444,  556,  553,  550,  437,  549,  435,  548,
+      433,  547,  431,  546,  545,  427,  544,  424,  543,  421,
+      542,  418,  541,  415,  540,  412,  539,  538,  408,  537,
+      405,  536,  402,  535,  399,  534,  533,  395,  532,  392,
+
+      531,  388,  530,  385,  529,  528,  527,  526,  525,  524,
+      375,  523,  372,  522,  369,  521,  366,  520,  363,  519,
+      360,  518,  357,  517,  354,  516,  351,  515,  348,  514,
+      345,  513,  342,  512,  339,  511,  336,  510,  333,  509,
+      330,  508,  327,  507,  323,  506,  505,  504,  503,  502,
+      316,  500,  312,  497,  496,  491,  490,  489,  488,  487,
+      486,  484,  481,  480,  479,  478,  477,  476,  475,  474,
+      473,  470,  469,  466,  465,  464,  463,  462,  461,  460,
+      459,  458,  457,  454,  289,  261,  452,  451,  450,  449,
+      448,  447,  446,  445,  429,  428,  409,  396,  389,  378,
+
+      377,  376,  324,  320,  319,  318,  317,  302,  301,  300,
+      297,  296,  295,  294,  293,  284,  281,  280,  279,  278,
+      277,  276,  275,  274,  273,  271,  270,  269,  268,  267,
+      264,  263,  262,  260,  259,  172,  254,  253,  252,  251,
+      250,  249,  248,  247,  246,  245,  237,  236,  235,  234,
+      231,  230,  227,  226,  225,  224,  223,  222,  221,  220,
+      217,  216,  215,  214,  213,  212,  209,  208,  207,  206,
+      205,  204,  203,  200,  199,  193,  192,  191,  190,  187,
+      186,  185,  156,  155,  149,  148,   39,  127,  126,  125,
+      124,  123,  122,  121,  118,   87,   39,   37,  849,    3,
+
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849
     } ;
 
-static yyconst flex_int16_t yy_chk[1023] =
+static yyconst flex_int16_t yy_chk[1368] =
     {   0,
         1,    1,    1,    1,    1,    1,    1,    1,    1,    1,
         1,    1,    1,    1,    1,    1,    1,    1,    1,    1,
@@ -759,113 +856,150 @@ static yyconst flex_int16_t yy_chk[1023] =
         1,    1,    1,    1,    1,    1,    1,    1,    1,    1,
         1,    1,    1,    1,    1,    1,    1,    1,    1,    1,
         1,    1,    1,    1,    1,    1,    1,    1,    1,    1,
-        1,    1,    1,    1,    1,    1,    1,    5,   17,    5,
-       10,   17,   46,   27,   24,   18,   46,   10,   10,   10,
-       10,   10,   10,   10,   16,   18,   16,   24,   25,  130,
-       25,   27,   16,   11,   11,   11,   11,   11,   11,   11,
-
-       19,  130,   23,   25,   11,   45,   19,   23,   45,  676,
-       10,   10,   10,   10,   10,   10,   10,   10,   10,   47,
-       47,   10,   10,   10,   10,   10,   10,   10,   10,   10,
-       11,   47,   10,   10,   10,   10,   12,   12,   12,   12,
-       12,   12,   12,   35,   60,   35,   12,   12,   13,   13,
-       13,   13,   13,   13,   13,  109,  671,   60,   13,   13,
-       14,   14,   14,   14,   14,   14,   14,   15,  109,   15,
-       14,   14,   52,   12,   48,   22,   15,   48,   52,   21,
-      670,   21,   15,   22,   15,   13,   21,   26,   22,   21,
-       74,   26,   26,   26,   22,   26,   74,   14,   42,   42,
-
-       42,   42,   42,   42,   26,   82,   26,   40,   42,   43,
-       43,   53,   44,   44,  664,   43,   49,   49,   44,  108,
-      118,   43,   49,   82,   56,   53,   43,  118,   43,   44,
-      108,  235,   53,   49,   42,   56,   79,   79,   56,   58,
-       58,   56,  667,  667,  235,   58,   79,   40,   40,   40,
-       40,   40,   40,   58,   40,   40,   58,   61,   40,   40,
-       40,   40,   40,   40,   40,   40,   40,  662,   59,   40,
-       59,  176,  660,   59,   59,   61,   61,   61,   61,  656,
-      176,   59,   62,   62,   62,   62,   63,   63,   63,   63,
-       64,   64,   64,   64,   65,   65,   65,   65,   65,   65,
-
-      120,  655,  652,  149,   65,   66,   66,   66,   66,   66,
-       66,   66,   90,  149,  120,   67,   66,   67,   67,   67,
-       67,   67,   67,  123,  123,  263,  249,  147,  263,  123,
-       65,   90,  122,  147,  122,  122,  122,  122,  122,  122,
-      123,  249,   66,  126,  126,  126,  136,  136,  150,  126,
-      137,  137,  136,  252,  150,  159,  137,  650,  151,  151,
-      126,  334,  648,  136,  151,  159,  252,  137,  152,  152,
-      645,  151,  153,  153,  152,  151,  644,  334,  153,  163,
-      163,  163,  163,  442,  643,  152,  442,  640,  639,  153,
-      164,  164,  164,  164,  165,  165,  165,  165,  166,  166,
-
-      166,  166,  167,  167,  167,  167,  168,  168,  168,  168,
-      169,  169,  169,  169,  171,  171,  171,  171,  171,  171,
-      638,  637,  635,  172,  171,  172,  172,  172,  172,  172,
-      172,  173,  173,  173,  173,  173,  173,  174,  174,  174,
-      174,  174,  174,  231,  231,  231,  231,  231,  231,  634,
-      171,  232,  232,  232,  232,  232,  232,  233,  233,  247,
-      247,  265,  265,  233,  275,  247,  275,  265,  275,  266,
-      266,  267,  267,  633,  233,  266,  247,  267,  265,  631,
-      630,  275,  278,  278,  278,  278,  266,  626,  267,  279,
-      279,  279,  279,  280,  280,  280,  280,  281,  281,  281,
-
-      281,  282,  282,  282,  282,  283,  283,  283,  283,  283,
-      283,  284,  284,  284,  284,  284,  284,  386,  386,  499,
-      499,  625,  624,  623,  621,  618,  617,  615,  614,  613,
-      612,  611,  610,  607,  606,  604,  603,  602,  601,  386,
-      499,  675,  675,  675,  677,  677,  600,  599,  597,  596,
-      595,  593,  591,  590,  587,  578,  577,  576,  553,  546,
-      545,  542,  541,  540,  539,  538,  537,  536,  535,  534,
-      533,  531,  528,  527,  526,  525,  524,  523,  522,  521,
-      520,  519,  518,  516,  515,  514,  513,  511,  509,  508,
-      507,  506,  505,  504,  503,  502,  501,  500,  498,  497,
-
-      496,  495,  494,  491,  490,  489,  488,  487,  486,  485,
-      484,  483,  482,  481,  480,  479,  478,  477,  476,  475,
-      473,  472,  471,  470,  469,  468,  467,  466,  465,  464,
-      463,  462,  461,  460,  459,  458,  456,  455,  454,  453,
-      452,  450,  449,  448,  447,  446,  445,  444,  441,  440,
-      438,  437,  436,  435,  432,  431,  430,  429,  428,  427,
-      426,  424,  423,  422,  421,  420,  419,  418,  417,  416,
-      415,  414,  413,  412,  411,  410,  409,  408,  406,  405,
-      404,  403,  402,  401,  400,  399,  398,  397,  396,  395,
-      394,  393,  392,  391,  390,  389,  388,  387,  384,  383,
-
-      382,  379,  378,  377,  376,  375,  374,  372,  370,  369,
-      368,  366,  365,  364,  363,  362,  361,  360,  359,  358,
-      357,  356,  355,  354,  353,  352,  351,  350,  349,  347,
-      346,  345,  344,  343,  342,  341,  338,  337,  336,  333,
-      332,  331,  330,  329,  328,  327,  326,  325,  324,  322,
-      321,  320,  319,  318,  317,  316,  315,  313,  312,  311,
-      310,  309,  308,  307,  306,  305,  304,  303,  302,  301,
-      300,  299,  298,  297,  296,  295,  294,  292,  291,  290,
-      289,  288,  287,  286,  285,  277,  276,  274,  273,  272,
-      271,  270,  269,  264,  262,  261,  260,  259,  258,  257,
-
-      256,  255,  253,  251,  250,  248,  246,  245,  242,  241,
-      240,  239,  238,  237,  236,  234,  230,  229,  228,  227,
-      226,  225,  224,  223,  222,  221,  220,  219,  218,  217,
-      216,  215,  214,  213,  212,  211,  210,  209,  208,  207,
-      206,  205,  204,  203,  202,  201,  200,  199,  198,  197,
-      195,  194,  192,  191,  189,  187,  186,  185,  184,  183,
-      182,  181,  180,  179,  177,  175,  162,  161,  160,  158,
-      157,  156,  155,  154,  148,  146,  145,  144,  143,  142,
-      141,  140,  139,  138,  135,  134,  133,  132,  131,  129,
-      128,  127,  125,  124,  121,  119,  117,  116,  115,  114,
-
-      113,  112,  111,  110,  107,  106,  105,  104,  103,  102,
-      101,  100,   99,   98,   97,   96,   95,   94,   93,   92,
-       91,   89,   88,   87,   86,   85,   84,   83,   81,   80,
-       78,   77,   76,   75,   73,   72,   71,   55,   54,   51,
-       50,   37,   36,   34,   33,   32,   31,   30,   29,   28,
-       20,    8,    7,    3,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674,  674,  674,  674,  674,  674,  674,  674,  674,
-      674,  674
+        1,    1,    1,    1,    1,    1,    1,    5,   23,    5,
+       10,   27,   46,   23,   17,   46,  808,   10,   10,   10,
+       10,   10,   10,   10,   16,   17,   16,   29,   17,   27,
+       18,  116,   16,   11,   11,   11,   11,   11,   11,   11,
+
+       18,   19,  116,   53,   11,   29,   36,   19,   36,   53,
+       10,   10,   10,   10,   10,   10,   10,   10,   10,   48,
+       48,   10,   10,   10,   10,   10,   10,   10,   10,   10,
+       11,   48,   10,   10,   10,   10,   12,   12,   12,   12,
+       12,   12,   12,   61,   80,   80,   12,   12,   13,   13,
+       13,   13,   13,   13,   13,   99,   61,   99,   13,   13,
+       14,   14,   14,   14,   14,   14,   14,   15,   47,   15,
+       14,   14,   47,   12,   49,   22,   15,   49,   24,   21,
+      274,   21,   15,   22,   15,   13,   21,   24,   22,   21,
+       25,   24,   25,   25,   22,  274,  810,   14,   45,   45,
+
+       92,   44,   44,   54,   45,   25,   26,   44,   26,   26,
+       26,   26,   26,   44,   26,   45,   26,   54,   44,   92,
+       44,  811,   26,   26,   54,   26,   41,   43,   43,   43,
+       43,   43,   43,   50,   50,   57,  812,   43,   60,   50,
+       60,   59,   59,   60,   60,  813,   57,   59,   75,   57,
+       50,   60,   57,  140,   75,   59,  816,   84,   59,   63,
+       63,   63,   63,   43,  817,  140,   41,   41,   41,   41,
+       41,   41,   62,   41,   41,   84,  159,   41,   41,   41,
+       41,   41,   41,   41,   41,   41,  159,  118,   41,  118,
+       62,   62,   62,   62,   64,   64,   64,   64,   65,   65,
+
+       65,   65,   66,   66,   66,   66,   66,   66,  108,  818,
+       81,   81,   66,   67,   67,   67,   67,   67,   67,   67,
+       81,  111,  108,   68,   67,   68,   68,   68,   68,   68,
+       68,  117,  128,  117,  130,  111,  157,  160,   66,  128,
+      133,  133,  157,  160,  117,  823,  133,  239,  130,  132,
+       67,  132,  132,  132,  132,  132,  132,  133,  136,  136,
+      136,  146,  146,  169,  136,  147,  147,  146,  161,  161,
+      239,  147,  219,  169,  161,  136,  218,  825,  146,  240,
+      244,  161,  147,  162,  162,  161,  163,  163,  218,  162,
+      219,  827,  163,  173,  173,  173,  173,  830,  244,  831,
+
+      162,  835,  240,  163,  174,  174,  174,  174,  175,  175,
+      175,  175,  176,  176,  176,  176,  177,  177,  177,  177,
+      178,  178,  178,  178,  179,  179,  179,  179,  181,  181,
+      181,  181,  181,  181,  837,  839,  241,  182,  181,  182,
+      182,  182,  182,  182,  182,  183,  183,  183,  183,  183,
+      183,  184,  184,  184,  184,  184,  184,  185,  186,  241,
+      192,  194,  185,  186,  181,  192,  194,  243,  195,  845,
+      452,  185,  186,  195,  192,  194,  846,  185,  186,  243,
+      185,  186,  195,  192,  194,  196,  452,  197,  199,  198,
+      196,  195,  197,  199,  198,  288,  842,  842,  288,  196,
+
+      242,  197,  199,  198,  851,  196,  807,  197,  196,  198,
+      197,  199,  198,  201,  203,  204,  206,  207,  201,  203,
+      204,  206,  207,  242,  806,  804,  803,  201,  203,  204,
+      206,  207,  260,  799,  203,  204,  201,  203,  204,  206,
+      207,  209,  210,  211,  798,  260,  209,  210,  211,  303,
+      303,  303,  303,  212,  213,  209,  210,  211,  212,  213,
+      797,  209,  210,  211,  209,  210,  211,  212,  213,  579,
+      214,  796,  579,  212,  213,  214,  212,  213,  220,  794,
+      221,  223,  224,  220,  214,  221,  223,  224,  792,  790,
+      214,  787,  220,  214,  221,  223,  224,  226,  786,  784,
+
+      783,  220,  226,  221,  223,  224,  277,  227,  228,  229,
+      231,  226,  227,  228,  229,  231,  782,  226,  781,  277,
+      226,  227,  228,  229,  231,  780,  779,  227,  228,  229,
+      227,  228,  229,  231,  232,  233,  234,  776,  775,  232,
+      233,  234,  304,  304,  304,  304,  235,  236,  232,  233,
+      234,  235,  236,  773,  232,  233,  234,  232,  233,  234,
+      235,  236,  772,  245,  771,  770,  235,  236,  245,  235,
+      236,  256,  256,  256,  256,  256,  256,  245,  257,  257,
+      257,  257,  257,  257,  258,  258,  245,  272,  272,  769,
+      258,  290,  290,  272,  291,  291,  768,  290,  766,  311,
+
+      291,  258,  765,  764,  272,  314,  292,  292,  290,  762,
+      760,  291,  292,  300,  759,  300,  756,  300,  305,  305,
+      305,  305,  311,  292,  306,  306,  306,  306,  314,  322,
+      300,  307,  307,  307,  307,  308,  308,  308,  308,  308,
+      308,  309,  309,  309,  309,  309,  309,  315,  326,  315,
+      329,  332,  322,  335,  338,  341,  344,  347,  350,  353,
+      356,  359,  362,  365,  368,  371,  374,  384,  387,  391,
+      315,  326,  394,  329,  332,  398,  335,  338,  341,  344,
+      347,  350,  353,  356,  359,  362,  365,  368,  371,  374,
+      384,  387,  391,  401,  404,  394,  407,  411,  398,  414,
+
+      417,  420,  423,  426,  438,  440,  443,  753,  504,  504,
+      553,  556,  752,  751,  642,  642,  401,  404,  750,  407,
+      411,  738,  414,  417,  420,  423,  426,  438,  440,  443,
+      504,  737,  736,  553,  556,  642,  850,  850,  850,  852,
+      852,  707,  700,  699,  696,  695,  694,  693,  692,  691,
+      690,  689,  688,  687,  685,  682,  681,  680,  679,  678,
+      677,  676,  675,  674,  673,  672,  670,  669,  668,  667,
+      665,  663,  662,  661,  660,  659,  658,  657,  656,  655,
+      654,  653,  652,  651,  650,  649,  648,  647,  646,  645,
+      644,  643,  641,  640,  639,  638,  637,  636,  635,  634,
+
+      633,  630,  629,  628,  627,  626,  625,  624,  623,  622,
+      621,  620,  619,  618,  617,  616,  615,  614,  613,  612,
+      610,  609,  608,  607,  606,  605,  604,  603,  602,  601,
+      600,  599,  598,  597,  596,  595,  593,  592,  591,  590,
+      589,  587,  586,  585,  584,  583,  582,  581,  578,  577,
+      575,  574,  573,  572,  569,  568,  567,  566,  565,  564,
+      563,  561,  560,  559,  558,  557,  555,  554,  552,  551,
+      550,  549,  548,  547,  546,  545,  544,  543,  542,  541,
+      540,  539,  538,  537,  536,  535,  534,  533,  532,  531,
+      530,  526,  525,  524,  523,  522,  521,  520,  519,  518,
+
+      517,  516,  515,  514,  513,  512,  511,  510,  509,  508,
+      507,  506,  505,  502,  501,  500,  497,  496,  495,  494,
+      493,  492,  490,  488,  487,  486,  484,  483,  482,  481,
+      480,  479,  478,  477,  476,  475,  474,  473,  472,  471,
+      470,  469,  468,  467,  465,  464,  463,  462,  461,  460,
+      459,  456,  455,  454,  451,  450,  449,  448,  447,  446,
+      445,  444,  442,  441,  439,  437,  436,  435,  434,  433,
+      432,  431,  430,  428,  427,  425,  424,  422,  421,  419,
+      418,  416,  415,  413,  412,  410,  409,  408,  406,  405,
+      403,  402,  400,  399,  397,  396,  395,  393,  392,  390,
+
+      388,  386,  385,  383,  382,  380,  378,  377,  376,  375,
+      373,  372,  370,  369,  367,  366,  364,  363,  361,  360,
+      358,  357,  355,  354,  352,  351,  349,  348,  346,  345,
+      343,  342,  340,  339,  337,  336,  334,  333,  331,  330,
+      328,  327,  325,  323,  321,  320,  319,  318,  317,  316,
+      313,  312,  310,  302,  301,  299,  298,  297,  296,  295,
+      294,  289,  287,  286,  285,  284,  283,  282,  281,  280,
+      278,  276,  275,  273,  271,  270,  267,  266,  265,  264,
+      263,  262,  261,  259,  255,  254,  253,  252,  251,  250,
+      249,  248,  247,  246,  238,  237,  230,  225,  222,  217,
+
+      216,  215,  193,  191,  190,  189,  187,  172,  171,  170,
+      168,  167,  166,  165,  164,  158,  156,  155,  154,  153,
+      152,  151,  150,  149,  148,  145,  144,  143,  142,  141,
+      139,  138,  137,  135,  134,  131,  129,  127,  126,  125,
+      124,  123,  122,  121,  120,  119,  115,  114,  113,  112,
+      110,  109,  107,  106,  105,  104,  103,  102,  101,  100,
+       98,   97,   96,   95,   94,   93,   91,   90,   89,   88,
+       87,   86,   85,   83,   82,   79,   78,   77,   76,   74,
+       73,   72,   56,   55,   52,   51,   38,   37,   35,   34,
+       33,   32,   31,   30,   28,   20,    8,    7,    3,  849,
+
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849,  849,  849,  849,
+      849,  849,  849,  849,  849,  849,  849
     } ;
 
 /* The intent behind this definition is that it'll catch
@@ -904,22 +1038,27 @@ static yyconst flex_int16_t yy_chk[1023] =
 #include "prog_instruction.h"
 #include "prog_statevars.h"
 
+#include "symbol_table.h"
 #include "program_parser.h"
 #include "program_parse.tab.h"
 
 #define require_ARB_vp (yyextra->mode == ARB_vertex)
 #define require_ARB_fp (yyextra->mode == ARB_fragment)
+#define require_NV_fp  (yyextra->option.NV_fragment)
 #define require_shadow (yyextra->option.Shadow)
 #define require_rect   (yyextra->option.TexRect)
 #define require_texarray        (yyextra->option.TexArray)
 
+#ifndef HAVE_UNISTD_H
+#define YY_NO_UNISTD_H
+#endif
+
 #define return_token_or_IDENTIFIER(condition, token)	\
    do {							\
       if (condition) {					\
 	 return token;					\
       } else {						\
-	 yylval->string = strdup(yytext);		\
-	 return IDENTIFIER;				\
+	 return handle_ident(yyextra, yytext, yylval);	\
       }							\
    } while (0)
 
@@ -934,15 +1073,16 @@ static yyconst flex_int16_t yy_chk[1023] =
    } while (0)
 
 
-#define return_opcode(condition, token, opcode, sat)	\
+#define return_opcode(condition, token, opcode, len)	\
    do {							\
-      if (condition) {					\
+      if (condition &&					\
+	  _mesa_parse_instruction_suffix(yyextra,	\
+					 yytext + len,	\
+					 & yylval->temp_inst)) {	\
 	 yylval->temp_inst.Opcode = OPCODE_ ## opcode;	\
-	 yylval->temp_inst.SaturateMode = SATURATE_ ## sat; \
 	 return token;					\
       } else {						\
-	 yylval->string = strdup(yytext);		\
-	 return IDENTIFIER;				\
+	 return handle_ident(yyextra, yytext, yylval);	\
       }							\
    } while (0)
 
@@ -991,6 +1131,15 @@ swiz_from_char(char c)
    return 0;
 }
 
+static int
+handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval)
+{
+   lval->string = strdup(text);
+
+   return (_mesa_symbol_table_find_symbol(state->st, 0, text) == NULL)
+      ? IDENTIFIER : USED_IDENTIFIER;
+}
+
 #define YY_USER_ACTION							\
    do {									\
       yylloc->first_column = yylloc->last_column;			\
@@ -1004,7 +1153,7 @@ swiz_from_char(char c)
    } while(0);
 
 #define YY_EXTRA_TYPE struct asm_parser_state *
-#line 1008 "lex.yy.c"
+#line 1157 "lex.yy.c"
 
 #define INITIAL 0
 
@@ -1149,7 +1298,7 @@ static int input (yyscan_t yyscanner );
 /* This used to be an fputs(), but since the string might contain NUL's,
  * we now use fwrite().
  */
-#define ECHO fwrite( yytext, yyleng, 1, yyout )
+#define ECHO do { if (fwrite( yytext, yyleng, 1, yyout )) {} } while (0)
 #endif
 
 /* Gets input and stuffs it into "buf".  number of characters read, or YY_NULL,
@@ -1160,7 +1309,7 @@ static int input (yyscan_t yyscanner );
 	if ( YY_CURRENT_BUFFER_LVALUE->yy_is_interactive ) \
 		{ \
 		int c = '*'; \
-		unsigned n; \
+		size_t n; \
 		for ( n = 0; n < max_size && \
 			     (c = getc( yyin )) != EOF && c != '\n'; ++n ) \
 			buf[n] = (char) c; \
@@ -1245,10 +1394,10 @@ YY_DECL
 	register int yy_act;
     struct yyguts_t * yyg = (struct yyguts_t*)yyscanner;
 
-#line 137 "program_lexer.l"
+#line 157 "program_lexer.l"
 
 
-#line 1252 "lex.yy.c"
+#line 1401 "lex.yy.c"
 
     yylval = yylval_param;
 
@@ -1305,13 +1454,13 @@ yy_match:
 			while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state )
 				{
 				yy_current_state = (int) yy_def[yy_current_state];
-				if ( yy_current_state >= 675 )
+				if ( yy_current_state >= 850 )
 					yy_c = yy_meta[(unsigned int) yy_c];
 				}
 			yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c];
 			++yy_cp;
 			}
-		while ( yy_base[yy_current_state] != 955 );
+		while ( yy_base[yy_current_state] != 1300 );
 
 yy_find_action:
 		yy_act = yy_accept[yy_current_state];
@@ -1337,17 +1486,17 @@ do_action:	/* This label is used only to access EOF actions. */
 
 case 1:
 YY_RULE_SETUP
-#line 139 "program_lexer.l"
+#line 159 "program_lexer.l"
 { return ARBvp_10; }
 	YY_BREAK
 case 2:
 YY_RULE_SETUP
-#line 140 "program_lexer.l"
+#line 160 "program_lexer.l"
 { return ARBfp_10; }
 	YY_BREAK
 case 3:
 YY_RULE_SETUP
-#line 141 "program_lexer.l"
+#line 161 "program_lexer.l"
 {
    yylval->integer = at_address;
    return_token_or_IDENTIFIER(require_ARB_vp, ADDRESS);
@@ -1355,813 +1504,745 @@ YY_RULE_SETUP
 	YY_BREAK
 case 4:
 YY_RULE_SETUP
-#line 145 "program_lexer.l"
+#line 165 "program_lexer.l"
 { return ALIAS; }
 	YY_BREAK
 case 5:
 YY_RULE_SETUP
-#line 146 "program_lexer.l"
+#line 166 "program_lexer.l"
 { return ATTRIB; }
 	YY_BREAK
 case 6:
 YY_RULE_SETUP
-#line 147 "program_lexer.l"
+#line 167 "program_lexer.l"
 { return END; }
 	YY_BREAK
 case 7:
 YY_RULE_SETUP
-#line 148 "program_lexer.l"
+#line 168 "program_lexer.l"
 { return OPTION; }
 	YY_BREAK
 case 8:
 YY_RULE_SETUP
-#line 149 "program_lexer.l"
+#line 169 "program_lexer.l"
 { return OUTPUT; }
 	YY_BREAK
 case 9:
 YY_RULE_SETUP
-#line 150 "program_lexer.l"
+#line 170 "program_lexer.l"
 { return PARAM; }
 	YY_BREAK
 case 10:
 YY_RULE_SETUP
-#line 151 "program_lexer.l"
+#line 171 "program_lexer.l"
 { yylval->integer = at_temp; return TEMP; }
 	YY_BREAK
 case 11:
 YY_RULE_SETUP
-#line 153 "program_lexer.l"
-{ return_opcode(             1, VECTOR_OP, ABS, OFF); }
+#line 173 "program_lexer.l"
+{ return_opcode(             1, VECTOR_OP, ABS, 3); }
 	YY_BREAK
 case 12:
 YY_RULE_SETUP
-#line 154 "program_lexer.l"
-{ return_opcode(require_ARB_fp, VECTOR_OP, ABS, ZERO_ONE); }
+#line 174 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, ADD, 3); }
 	YY_BREAK
 case 13:
 YY_RULE_SETUP
-#line 155 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, ADD, OFF); }
+#line 175 "program_lexer.l"
+{ return_opcode(require_ARB_vp, ARL, ARL, 3); }
 	YY_BREAK
 case 14:
 YY_RULE_SETUP
-#line 156 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, ADD, ZERO_ONE); }
+#line 177 "program_lexer.l"
+{ return_opcode(require_ARB_fp, TRI_OP, CMP, 3); }
 	YY_BREAK
 case 15:
 YY_RULE_SETUP
-#line 157 "program_lexer.l"
-{ return_opcode(require_ARB_vp, ARL, ARL, OFF); }
+#line 178 "program_lexer.l"
+{ return_opcode(require_ARB_fp, SCALAR_OP, COS, 3); }
 	YY_BREAK
 case 16:
 YY_RULE_SETUP
-#line 159 "program_lexer.l"
-{ return_opcode(require_ARB_fp, TRI_OP, CMP, OFF); }
+#line 180 "program_lexer.l"
+{ return_opcode(require_NV_fp,  VECTOR_OP, DDX, 3); }
 	YY_BREAK
 case 17:
 YY_RULE_SETUP
-#line 160 "program_lexer.l"
-{ return_opcode(require_ARB_fp, TRI_OP, CMP, ZERO_ONE); }
+#line 181 "program_lexer.l"
+{ return_opcode(require_NV_fp,  VECTOR_OP, DDY, 3); }
 	YY_BREAK
 case 18:
 YY_RULE_SETUP
-#line 161 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, COS, OFF); }
+#line 182 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, DP3, 3); }
 	YY_BREAK
 case 19:
 YY_RULE_SETUP
-#line 162 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, COS, ZERO_ONE); }
+#line 183 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, DP4, 3); }
 	YY_BREAK
 case 20:
 YY_RULE_SETUP
-#line 164 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, DP3, OFF); }
+#line 184 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, DPH, 3); }
 	YY_BREAK
 case 21:
 YY_RULE_SETUP
-#line 165 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, DP3, ZERO_ONE); }
+#line 185 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, DST, 3); }
 	YY_BREAK
 case 22:
 YY_RULE_SETUP
-#line 166 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, DP4, OFF); }
+#line 187 "program_lexer.l"
+{ return_opcode(             1, SCALAR_OP, EX2, 3); }
 	YY_BREAK
 case 23:
 YY_RULE_SETUP
-#line 167 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, DP4, ZERO_ONE); }
+#line 188 "program_lexer.l"
+{ return_opcode(require_ARB_vp, SCALAR_OP, EXP, 3); }
 	YY_BREAK
 case 24:
 YY_RULE_SETUP
-#line 168 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, DPH, OFF); }
+#line 190 "program_lexer.l"
+{ return_opcode(             1, VECTOR_OP, FLR, 3); }
 	YY_BREAK
 case 25:
 YY_RULE_SETUP
-#line 169 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, DPH, ZERO_ONE); }
+#line 191 "program_lexer.l"
+{ return_opcode(             1, VECTOR_OP, FRC, 3); }
 	YY_BREAK
 case 26:
 YY_RULE_SETUP
-#line 170 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, DST, OFF); }
+#line 193 "program_lexer.l"
+{ return_opcode(require_ARB_fp, KIL, KIL, 3); }
 	YY_BREAK
 case 27:
 YY_RULE_SETUP
-#line 171 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, DST, ZERO_ONE); }
+#line 195 "program_lexer.l"
+{ return_opcode(             1, VECTOR_OP, LIT, 3); }
 	YY_BREAK
 case 28:
 YY_RULE_SETUP
-#line 173 "program_lexer.l"
-{ return_opcode(             1, SCALAR_OP, EX2, OFF); }
+#line 196 "program_lexer.l"
+{ return_opcode(             1, SCALAR_OP, LG2, 3); }
 	YY_BREAK
 case 29:
 YY_RULE_SETUP
-#line 174 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, EX2, ZERO_ONE); }
+#line 197 "program_lexer.l"
+{ return_opcode(require_ARB_vp, SCALAR_OP, LOG, 3); }
 	YY_BREAK
 case 30:
 YY_RULE_SETUP
-#line 175 "program_lexer.l"
-{ return_opcode(require_ARB_vp, SCALAR_OP, EXP, OFF); }
+#line 198 "program_lexer.l"
+{ return_opcode(require_ARB_fp, TRI_OP, LRP, 3); }
 	YY_BREAK
 case 31:
 YY_RULE_SETUP
-#line 177 "program_lexer.l"
-{ return_opcode(             1, VECTOR_OP, FLR, OFF); }
+#line 200 "program_lexer.l"
+{ return_opcode(             1, TRI_OP, MAD, 3); }
 	YY_BREAK
 case 32:
 YY_RULE_SETUP
-#line 178 "program_lexer.l"
-{ return_opcode(require_ARB_fp, VECTOR_OP, FLR, ZERO_ONE); }
+#line 201 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, MAX, 3); }
 	YY_BREAK
 case 33:
 YY_RULE_SETUP
-#line 179 "program_lexer.l"
-{ return_opcode(             1, VECTOR_OP, FRC, OFF); }
+#line 202 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, MIN, 3); }
 	YY_BREAK
 case 34:
 YY_RULE_SETUP
-#line 180 "program_lexer.l"
-{ return_opcode(require_ARB_fp, VECTOR_OP, FRC, ZERO_ONE); }
+#line 203 "program_lexer.l"
+{ return_opcode(             1, VECTOR_OP, MOV, 3); }
 	YY_BREAK
 case 35:
 YY_RULE_SETUP
-#line 182 "program_lexer.l"
-{ return_opcode(require_ARB_fp, KIL, KIL, OFF); }
+#line 204 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, MUL, 3); }
 	YY_BREAK
 case 36:
 YY_RULE_SETUP
-#line 184 "program_lexer.l"
-{ return_opcode(             1, VECTOR_OP, LIT, OFF); }
+#line 206 "program_lexer.l"
+{ return_opcode(require_NV_fp,  VECTOR_OP, PK2H, 4); }
 	YY_BREAK
 case 37:
 YY_RULE_SETUP
-#line 185 "program_lexer.l"
-{ return_opcode(require_ARB_fp, VECTOR_OP, LIT, ZERO_ONE); }
+#line 207 "program_lexer.l"
+{ return_opcode(require_NV_fp,  VECTOR_OP, PK2US, 5); }
 	YY_BREAK
 case 38:
 YY_RULE_SETUP
-#line 186 "program_lexer.l"
-{ return_opcode(             1, SCALAR_OP, LG2, OFF); }
+#line 208 "program_lexer.l"
+{ return_opcode(require_NV_fp,  VECTOR_OP, PK4B, 4); }
 	YY_BREAK
 case 39:
 YY_RULE_SETUP
-#line 187 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, LG2, ZERO_ONE); }
+#line 209 "program_lexer.l"
+{ return_opcode(require_NV_fp,  VECTOR_OP, PK4UB, 5); }
 	YY_BREAK
 case 40:
 YY_RULE_SETUP
-#line 188 "program_lexer.l"
-{ return_opcode(require_ARB_vp, SCALAR_OP, LOG, OFF); }
+#line 210 "program_lexer.l"
+{ return_opcode(             1, BINSC_OP, POW, 3); }
 	YY_BREAK
 case 41:
 YY_RULE_SETUP
-#line 189 "program_lexer.l"
-{ return_opcode(require_ARB_fp, TRI_OP, LRP, OFF); }
+#line 212 "program_lexer.l"
+{ return_opcode(             1, SCALAR_OP, RCP, 3); }
 	YY_BREAK
 case 42:
 YY_RULE_SETUP
-#line 190 "program_lexer.l"
-{ return_opcode(require_ARB_fp, TRI_OP, LRP, ZERO_ONE); }
+#line 213 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP,    RFL, 3); }
 	YY_BREAK
 case 43:
 YY_RULE_SETUP
-#line 192 "program_lexer.l"
-{ return_opcode(             1, TRI_OP, MAD, OFF); }
+#line 214 "program_lexer.l"
+{ return_opcode(             1, SCALAR_OP, RSQ, 3); }
 	YY_BREAK
 case 44:
 YY_RULE_SETUP
-#line 193 "program_lexer.l"
-{ return_opcode(require_ARB_fp, TRI_OP, MAD, ZERO_ONE); }
+#line 216 "program_lexer.l"
+{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, 3); }
 	YY_BREAK
 case 45:
 YY_RULE_SETUP
-#line 194 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, MAX, OFF); }
+#line 217 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP, SEQ, 3); }
 	YY_BREAK
 case 46:
 YY_RULE_SETUP
-#line 195 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, MAX, ZERO_ONE); }
+#line 218 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP, SFL, 3); }
 	YY_BREAK
 case 47:
 YY_RULE_SETUP
-#line 196 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, MIN, OFF); }
+#line 219 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, SGE, 3); }
 	YY_BREAK
 case 48:
 YY_RULE_SETUP
-#line 197 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, MIN, ZERO_ONE); }
+#line 220 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP, SGT, 3); }
 	YY_BREAK
 case 49:
 YY_RULE_SETUP
-#line 198 "program_lexer.l"
-{ return_opcode(             1, VECTOR_OP, MOV, OFF); }
+#line 221 "program_lexer.l"
+{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, 3); }
 	YY_BREAK
 case 50:
 YY_RULE_SETUP
-#line 199 "program_lexer.l"
-{ return_opcode(require_ARB_fp, VECTOR_OP, MOV, ZERO_ONE); }
+#line 222 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP, SLE, 3); }
 	YY_BREAK
 case 51:
 YY_RULE_SETUP
-#line 200 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, MUL, OFF); }
+#line 223 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, SLT, 3); }
 	YY_BREAK
 case 52:
 YY_RULE_SETUP
-#line 201 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, MUL, ZERO_ONE); }
+#line 224 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP, SNE, 3); }
 	YY_BREAK
 case 53:
 YY_RULE_SETUP
-#line 203 "program_lexer.l"
-{ return_opcode(             1, BINSC_OP, POW, OFF); }
+#line 225 "program_lexer.l"
+{ return_opcode(require_NV_fp,  BIN_OP, STR, 3); }
 	YY_BREAK
 case 54:
 YY_RULE_SETUP
-#line 204 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BINSC_OP, POW, ZERO_ONE); }
+#line 226 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, SUB, 3); }
 	YY_BREAK
 case 55:
 YY_RULE_SETUP
-#line 206 "program_lexer.l"
-{ return_opcode(             1, SCALAR_OP, RCP, OFF); }
+#line 227 "program_lexer.l"
+{ return_opcode(             1, SWZ, SWZ, 3); }
 	YY_BREAK
 case 56:
 YY_RULE_SETUP
-#line 207 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, RCP, ZERO_ONE); }
+#line 229 "program_lexer.l"
+{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, 3); }
 	YY_BREAK
 case 57:
 YY_RULE_SETUP
-#line 208 "program_lexer.l"
-{ return_opcode(             1, SCALAR_OP, RSQ, OFF); }
+#line 230 "program_lexer.l"
+{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, 3); }
 	YY_BREAK
 case 58:
 YY_RULE_SETUP
-#line 209 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, RSQ, ZERO_ONE); }
+#line 231 "program_lexer.l"
+{ return_opcode(require_NV_fp,  TXD_OP, TXD, 3); }
 	YY_BREAK
 case 59:
 YY_RULE_SETUP
-#line 211 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, OFF); }
+#line 232 "program_lexer.l"
+{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, 3); }
 	YY_BREAK
 case 60:
 YY_RULE_SETUP
-#line 212 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, ZERO_ONE); }
+#line 234 "program_lexer.l"
+{ return_opcode(require_NV_fp,  SCALAR_OP, UP2H, 4); }
 	YY_BREAK
 case 61:
 YY_RULE_SETUP
-#line 213 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, SGE, OFF); }
+#line 235 "program_lexer.l"
+{ return_opcode(require_NV_fp,  SCALAR_OP, UP2US, 5); }
 	YY_BREAK
 case 62:
 YY_RULE_SETUP
-#line 214 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, SGE, ZERO_ONE); }
+#line 236 "program_lexer.l"
+{ return_opcode(require_NV_fp,  SCALAR_OP, UP4B, 4); }
 	YY_BREAK
 case 63:
 YY_RULE_SETUP
-#line 215 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, OFF); }
+#line 237 "program_lexer.l"
+{ return_opcode(require_NV_fp,  SCALAR_OP, UP4UB, 5); }
 	YY_BREAK
 case 64:
 YY_RULE_SETUP
-#line 216 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, ZERO_ONE); }
+#line 239 "program_lexer.l"
+{ return_opcode(require_NV_fp,  TRI_OP, X2D, 3); }
 	YY_BREAK
 case 65:
 YY_RULE_SETUP
-#line 217 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, SLT, OFF); }
+#line 240 "program_lexer.l"
+{ return_opcode(             1, BIN_OP, XPD, 3); }
 	YY_BREAK
 case 66:
 YY_RULE_SETUP
-#line 218 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, SLT, ZERO_ONE); }
+#line 242 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); }
 	YY_BREAK
 case 67:
 YY_RULE_SETUP
-#line 219 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, SUB, OFF); }
+#line 243 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); }
 	YY_BREAK
 case 68:
 YY_RULE_SETUP
-#line 220 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, SUB, ZERO_ONE); }
+#line 244 "program_lexer.l"
+{ return PROGRAM; }
 	YY_BREAK
 case 69:
 YY_RULE_SETUP
-#line 221 "program_lexer.l"
-{ return_opcode(             1, SWZ, SWZ, OFF); }
+#line 245 "program_lexer.l"
+{ return STATE; }
 	YY_BREAK
 case 70:
 YY_RULE_SETUP
-#line 222 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SWZ, SWZ, ZERO_ONE); }
+#line 246 "program_lexer.l"
+{ return RESULT; }
 	YY_BREAK
 case 71:
 YY_RULE_SETUP
-#line 224 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, OFF); }
+#line 248 "program_lexer.l"
+{ return AMBIENT; }
 	YY_BREAK
 case 72:
 YY_RULE_SETUP
-#line 225 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, ZERO_ONE); }
+#line 249 "program_lexer.l"
+{ return ATTENUATION; }
 	YY_BREAK
 case 73:
 YY_RULE_SETUP
-#line 226 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, OFF); }
+#line 250 "program_lexer.l"
+{ return BACK; }
 	YY_BREAK
 case 74:
 YY_RULE_SETUP
-#line 227 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, ZERO_ONE); }
+#line 251 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, CLIP); }
 	YY_BREAK
 case 75:
 YY_RULE_SETUP
-#line 228 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, OFF); }
+#line 252 "program_lexer.l"
+{ return COLOR; }
 	YY_BREAK
 case 76:
 YY_RULE_SETUP
-#line 229 "program_lexer.l"
-{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, ZERO_ONE); }
+#line 253 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_fp, DEPTH); }
 	YY_BREAK
 case 77:
 YY_RULE_SETUP
-#line 231 "program_lexer.l"
-{ return_opcode(             1, BIN_OP, XPD, OFF); }
+#line 254 "program_lexer.l"
+{ return DIFFUSE; }
 	YY_BREAK
 case 78:
 YY_RULE_SETUP
-#line 232 "program_lexer.l"
-{ return_opcode(require_ARB_fp, BIN_OP, XPD, ZERO_ONE); }
+#line 255 "program_lexer.l"
+{ return DIRECTION; }
 	YY_BREAK
 case 79:
 YY_RULE_SETUP
-#line 234 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); }
+#line 256 "program_lexer.l"
+{ return EMISSION; }
 	YY_BREAK
 case 80:
 YY_RULE_SETUP
-#line 235 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); }
+#line 257 "program_lexer.l"
+{ return ENV; }
 	YY_BREAK
 case 81:
 YY_RULE_SETUP
-#line 236 "program_lexer.l"
-{ return PROGRAM; }
+#line 258 "program_lexer.l"
+{ return EYE; }
 	YY_BREAK
 case 82:
 YY_RULE_SETUP
-#line 237 "program_lexer.l"
-{ return STATE; }
+#line 259 "program_lexer.l"
+{ return FOGCOORD; }
 	YY_BREAK
 case 83:
 YY_RULE_SETUP
-#line 238 "program_lexer.l"
-{ return RESULT; }
+#line 260 "program_lexer.l"
+{ return FOG; }
 	YY_BREAK
 case 84:
 YY_RULE_SETUP
-#line 240 "program_lexer.l"
-{ return AMBIENT; }
+#line 261 "program_lexer.l"
+{ return FRONT; }
 	YY_BREAK
 case 85:
 YY_RULE_SETUP
-#line 241 "program_lexer.l"
-{ return ATTENUATION; }
+#line 262 "program_lexer.l"
+{ return HALF; }
 	YY_BREAK
 case 86:
 YY_RULE_SETUP
-#line 242 "program_lexer.l"
-{ return BACK; }
+#line 263 "program_lexer.l"
+{ return INVERSE; }
 	YY_BREAK
 case 87:
 YY_RULE_SETUP
-#line 243 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, CLIP); }
+#line 264 "program_lexer.l"
+{ return INVTRANS; }
 	YY_BREAK
 case 88:
 YY_RULE_SETUP
-#line 244 "program_lexer.l"
-{ return COLOR; }
+#line 265 "program_lexer.l"
+{ return LIGHT; }
 	YY_BREAK
 case 89:
 YY_RULE_SETUP
-#line 245 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_fp, DEPTH); }
+#line 266 "program_lexer.l"
+{ return LIGHTMODEL; }
 	YY_BREAK
 case 90:
 YY_RULE_SETUP
-#line 246 "program_lexer.l"
-{ return DIFFUSE; }
+#line 267 "program_lexer.l"
+{ return LIGHTPROD; }
 	YY_BREAK
 case 91:
 YY_RULE_SETUP
-#line 247 "program_lexer.l"
-{ return DIRECTION; }
+#line 268 "program_lexer.l"
+{ return LOCAL; }
 	YY_BREAK
 case 92:
 YY_RULE_SETUP
-#line 248 "program_lexer.l"
-{ return EMISSION; }
+#line 269 "program_lexer.l"
+{ return MATERIAL; }
 	YY_BREAK
 case 93:
 YY_RULE_SETUP
-#line 249 "program_lexer.l"
-{ return ENV; }
+#line 270 "program_lexer.l"
+{ return MAT_PROGRAM; }
 	YY_BREAK
 case 94:
 YY_RULE_SETUP
-#line 250 "program_lexer.l"
-{ return EYE; }
+#line 271 "program_lexer.l"
+{ return MATRIX; }
 	YY_BREAK
 case 95:
 YY_RULE_SETUP
-#line 251 "program_lexer.l"
-{ return FOGCOORD; }
+#line 272 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, MATRIXINDEX); }
 	YY_BREAK
 case 96:
 YY_RULE_SETUP
-#line 252 "program_lexer.l"
-{ return FOG; }
+#line 273 "program_lexer.l"
+{ return MODELVIEW; }
 	YY_BREAK
 case 97:
 YY_RULE_SETUP
-#line 253 "program_lexer.l"
-{ return FRONT; }
+#line 274 "program_lexer.l"
+{ return MVP; }
 	YY_BREAK
 case 98:
 YY_RULE_SETUP
-#line 254 "program_lexer.l"
-{ return HALF; }
+#line 275 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, NORMAL); }
 	YY_BREAK
 case 99:
 YY_RULE_SETUP
-#line 255 "program_lexer.l"
-{ return INVERSE; }
+#line 276 "program_lexer.l"
+{ return OBJECT; }
 	YY_BREAK
 case 100:
 YY_RULE_SETUP
-#line 256 "program_lexer.l"
-{ return INVTRANS; }
+#line 277 "program_lexer.l"
+{ return PALETTE; }
 	YY_BREAK
 case 101:
 YY_RULE_SETUP
-#line 257 "program_lexer.l"
-{ return LIGHT; }
+#line 278 "program_lexer.l"
+{ return PARAMS; }
 	YY_BREAK
 case 102:
 YY_RULE_SETUP
-#line 258 "program_lexer.l"
-{ return LIGHTMODEL; }
+#line 279 "program_lexer.l"
+{ return PLANE; }
 	YY_BREAK
 case 103:
 YY_RULE_SETUP
-#line 259 "program_lexer.l"
-{ return LIGHTPROD; }
+#line 280 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, POINT_TOK); }
 	YY_BREAK
 case 104:
 YY_RULE_SETUP
-#line 260 "program_lexer.l"
-{ return LOCAL; }
+#line 281 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, POINTSIZE); }
 	YY_BREAK
 case 105:
 YY_RULE_SETUP
-#line 261 "program_lexer.l"
-{ return MATERIAL; }
+#line 282 "program_lexer.l"
+{ return POSITION; }
 	YY_BREAK
 case 106:
 YY_RULE_SETUP
-#line 262 "program_lexer.l"
-{ return MAT_PROGRAM; }
+#line 283 "program_lexer.l"
+{ return PRIMARY; }
 	YY_BREAK
 case 107:
 YY_RULE_SETUP
-#line 263 "program_lexer.l"
-{ return MATRIX; }
+#line 284 "program_lexer.l"
+{ return PROJECTION; }
 	YY_BREAK
 case 108:
 YY_RULE_SETUP
-#line 264 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, MATRIXINDEX); }
+#line 285 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_fp, RANGE); }
 	YY_BREAK
 case 109:
 YY_RULE_SETUP
-#line 265 "program_lexer.l"
-{ return MODELVIEW; }
+#line 286 "program_lexer.l"
+{ return ROW; }
 	YY_BREAK
 case 110:
 YY_RULE_SETUP
-#line 266 "program_lexer.l"
-{ return MVP; }
+#line 287 "program_lexer.l"
+{ return SCENECOLOR; }
 	YY_BREAK
 case 111:
 YY_RULE_SETUP
-#line 267 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, NORMAL); }
+#line 288 "program_lexer.l"
+{ return SECONDARY; }
 	YY_BREAK
 case 112:
 YY_RULE_SETUP
-#line 268 "program_lexer.l"
-{ return OBJECT; }
+#line 289 "program_lexer.l"
+{ return SHININESS; }
 	YY_BREAK
 case 113:
 YY_RULE_SETUP
-#line 269 "program_lexer.l"
-{ return PALETTE; }
+#line 290 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, SIZE_TOK); }
 	YY_BREAK
 case 114:
 YY_RULE_SETUP
-#line 270 "program_lexer.l"
-{ return PARAMS; }
+#line 291 "program_lexer.l"
+{ return SPECULAR; }
 	YY_BREAK
 case 115:
 YY_RULE_SETUP
-#line 271 "program_lexer.l"
-{ return PLANE; }
+#line 292 "program_lexer.l"
+{ return SPOT; }
 	YY_BREAK
 case 116:
 YY_RULE_SETUP
-#line 272 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, POINT_TOK); }
+#line 293 "program_lexer.l"
+{ return TEXCOORD; }
 	YY_BREAK
 case 117:
 YY_RULE_SETUP
-#line 273 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, POINTSIZE); }
+#line 294 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_fp, TEXENV); }
 	YY_BREAK
 case 118:
 YY_RULE_SETUP
-#line 274 "program_lexer.l"
-{ return POSITION; }
+#line 295 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, TEXGEN); }
 	YY_BREAK
 case 119:
 YY_RULE_SETUP
-#line 275 "program_lexer.l"
-{ return PRIMARY; }
+#line 296 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, TEXGEN_Q); }
 	YY_BREAK
 case 120:
 YY_RULE_SETUP
-#line 276 "program_lexer.l"
-{ return PROJECTION; }
+#line 297 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, TEXGEN_S); }
 	YY_BREAK
 case 121:
 YY_RULE_SETUP
-#line 277 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_fp, RANGE); }
+#line 298 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, TEXGEN_T); }
 	YY_BREAK
 case 122:
 YY_RULE_SETUP
-#line 278 "program_lexer.l"
-{ return ROW; }
+#line 299 "program_lexer.l"
+{ return TEXTURE; }
 	YY_BREAK
 case 123:
 YY_RULE_SETUP
-#line 279 "program_lexer.l"
-{ return SCENECOLOR; }
+#line 300 "program_lexer.l"
+{ return TRANSPOSE; }
 	YY_BREAK
 case 124:
 YY_RULE_SETUP
-#line 280 "program_lexer.l"
-{ return SECONDARY; }
+#line 301 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, VTXATTRIB); }
 	YY_BREAK
 case 125:
 YY_RULE_SETUP
-#line 281 "program_lexer.l"
-{ return SHININESS; }
+#line 302 "program_lexer.l"
+{ return_token_or_DOT(require_ARB_vp, WEIGHT); }
 	YY_BREAK
 case 126:
 YY_RULE_SETUP
-#line 282 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, SIZE_TOK); }
+#line 304 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp, TEXTURE_UNIT); }
 	YY_BREAK
 case 127:
 YY_RULE_SETUP
-#line 283 "program_lexer.l"
-{ return SPECULAR; }
+#line 305 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_1D); }
 	YY_BREAK
 case 128:
 YY_RULE_SETUP
-#line 284 "program_lexer.l"
-{ return SPOT; }
+#line 306 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_2D); }
 	YY_BREAK
 case 129:
 YY_RULE_SETUP
-#line 285 "program_lexer.l"
-{ return TEXCOORD; }
+#line 307 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_3D); }
 	YY_BREAK
 case 130:
 YY_RULE_SETUP
-#line 286 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_fp, TEXENV); }
+#line 308 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_CUBE); }
 	YY_BREAK
 case 131:
 YY_RULE_SETUP
-#line 287 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, TEXGEN); }
+#line 309 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_rect, TEX_RECT); }
 	YY_BREAK
 case 132:
 YY_RULE_SETUP
-#line 288 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, TEXGEN_Q); }
+#line 310 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW1D); }
 	YY_BREAK
 case 133:
 YY_RULE_SETUP
-#line 289 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, TEXGEN_S); }
+#line 311 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW2D); }
 	YY_BREAK
 case 134:
 YY_RULE_SETUP
-#line 290 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, TEXGEN_T); }
+#line 312 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_rect, TEX_SHADOWRECT); }
 	YY_BREAK
 case 135:
 YY_RULE_SETUP
-#line 291 "program_lexer.l"
-{ return TEXTURE; }
+#line 313 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY1D); }
 	YY_BREAK
 case 136:
 YY_RULE_SETUP
-#line 292 "program_lexer.l"
-{ return TRANSPOSE; }
+#line 314 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY2D); }
 	YY_BREAK
 case 137:
 YY_RULE_SETUP
-#line 293 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, VTXATTRIB); }
+#line 315 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); }
 	YY_BREAK
 case 138:
 YY_RULE_SETUP
-#line 294 "program_lexer.l"
-{ return_token_or_DOT(require_ARB_vp, WEIGHT); }
+#line 316 "program_lexer.l"
+{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); }
 	YY_BREAK
 case 139:
 YY_RULE_SETUP
-#line 296 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp, TEXTURE_UNIT); }
+#line 318 "program_lexer.l"
+{ return handle_ident(yyextra, yytext, yylval); }
 	YY_BREAK
 case 140:
 YY_RULE_SETUP
-#line 297 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_1D); }
-	YY_BREAK
-case 141:
-YY_RULE_SETUP
-#line 298 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_2D); }
-	YY_BREAK
-case 142:
-YY_RULE_SETUP
-#line 299 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_3D); }
-	YY_BREAK
-case 143:
-YY_RULE_SETUP
-#line 300 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_CUBE); }
-	YY_BREAK
-case 144:
-YY_RULE_SETUP
-#line 301 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_rect, TEX_RECT); }
-	YY_BREAK
-case 145:
-YY_RULE_SETUP
-#line 302 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW1D); }
-	YY_BREAK
-case 146:
-YY_RULE_SETUP
-#line 303 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW2D); }
-	YY_BREAK
-case 147:
-YY_RULE_SETUP
-#line 304 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_rect, TEX_SHADOWRECT); }
-	YY_BREAK
-case 148:
-YY_RULE_SETUP
-#line 305 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY1D); }
-	YY_BREAK
-case 149:
-YY_RULE_SETUP
-#line 306 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY2D); }
-	YY_BREAK
-case 150:
-YY_RULE_SETUP
-#line 307 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); }
-	YY_BREAK
-case 151:
-YY_RULE_SETUP
-#line 308 "program_lexer.l"
-{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); }
-	YY_BREAK
-case 152:
-YY_RULE_SETUP
-#line 310 "program_lexer.l"
-{
-   yylval->string = strdup(yytext);
-   return IDENTIFIER;
-}
-	YY_BREAK
-case 153:
-YY_RULE_SETUP
-#line 315 "program_lexer.l"
+#line 320 "program_lexer.l"
 { return DOT_DOT; }
 	YY_BREAK
-case 154:
+case 141:
 YY_RULE_SETUP
-#line 317 "program_lexer.l"
+#line 322 "program_lexer.l"
 {
    yylval->integer = strtol(yytext, NULL, 10);
    return INTEGER;
 }
 	YY_BREAK
-case 155:
+case 142:
 YY_RULE_SETUP
-#line 321 "program_lexer.l"
+#line 326 "program_lexer.l"
 {
    yylval->real = _mesa_strtod(yytext, NULL);
    return REAL;
 }
 	YY_BREAK
-case 156:
-/* rule 156 can match eol */
+case 143:
+/* rule 143 can match eol */
 *yy_cp = yyg->yy_hold_char; /* undo effects of setting up yytext */
 yyg->yy_c_buf_p = yy_cp -= 1;
 YY_DO_BEFORE_ACTION; /* set up yytext again */
 YY_RULE_SETUP
-#line 325 "program_lexer.l"
+#line 330 "program_lexer.l"
 {
    yylval->real = _mesa_strtod(yytext, NULL);
    return REAL;
 }
 	YY_BREAK
-case 157:
+case 144:
 YY_RULE_SETUP
-#line 329 "program_lexer.l"
+#line 334 "program_lexer.l"
 {
    yylval->real = _mesa_strtod(yytext, NULL);
    return REAL;
 }
 	YY_BREAK
-case 158:
+case 145:
 YY_RULE_SETUP
-#line 333 "program_lexer.l"
+#line 338 "program_lexer.l"
 {
    yylval->real = _mesa_strtod(yytext, NULL);
    return REAL;
 }
 	YY_BREAK
-case 159:
+case 146:
 YY_RULE_SETUP
-#line 338 "program_lexer.l"
+#line 343 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_NOOP;
    yylval->swiz_mask.mask = WRITEMASK_XYZW;
    return MASK4;
 }
 	YY_BREAK
-case 160:
+case 147:
 YY_RULE_SETUP
-#line 344 "program_lexer.l"
+#line 349 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_XY
@@ -2169,27 +2250,27 @@ YY_RULE_SETUP
    return MASK3;
 }
 	YY_BREAK
-case 161:
+case 148:
 YY_RULE_SETUP
-#line 350 "program_lexer.l"
+#line 355 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_XZW;
    return MASK3;
 }
 	YY_BREAK
-case 162:
+case 149:
 YY_RULE_SETUP
-#line 355 "program_lexer.l"
+#line 360 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_YZW;
    return MASK3;
 }
 	YY_BREAK
-case 163:
+case 150:
 YY_RULE_SETUP
-#line 361 "program_lexer.l"
+#line 366 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_X
@@ -2197,9 +2278,9 @@ YY_RULE_SETUP
    return MASK2;
 }
 	YY_BREAK
-case 164:
+case 151:
 YY_RULE_SETUP
-#line 367 "program_lexer.l"
+#line 372 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_Y
@@ -2207,18 +2288,18 @@ YY_RULE_SETUP
    return MASK2;
 }
 	YY_BREAK
-case 165:
+case 152:
 YY_RULE_SETUP
-#line 373 "program_lexer.l"
+#line 378 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_ZW;
    return MASK2;
 }
 	YY_BREAK
-case 166:
+case 153:
 YY_RULE_SETUP
-#line 379 "program_lexer.l"
+#line 384 "program_lexer.l"
 {
    const unsigned s = swiz_from_char(yytext[1]);
    yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s);
@@ -2226,9 +2307,9 @@ YY_RULE_SETUP
    return MASK1; 
 }
 	YY_BREAK
-case 167:
+case 154:
 YY_RULE_SETUP
-#line 386 "program_lexer.l"
+#line 391 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]),
 					    swiz_from_char(yytext[2]),
@@ -2238,18 +2319,18 @@ YY_RULE_SETUP
    return SWIZZLE;
 }
 	YY_BREAK
-case 168:
+case 155:
 YY_RULE_SETUP
-#line 395 "program_lexer.l"
+#line 400 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_NOOP;
    yylval->swiz_mask.mask = WRITEMASK_XYZW;
    return_token_or_DOT(require_ARB_fp, MASK4);
 }
 	YY_BREAK
-case 169:
+case 156:
 YY_RULE_SETUP
-#line 401 "program_lexer.l"
+#line 406 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_XY
@@ -2257,27 +2338,27 @@ YY_RULE_SETUP
    return_token_or_DOT(require_ARB_fp, MASK3);
 }
 	YY_BREAK
-case 170:
+case 157:
 YY_RULE_SETUP
-#line 407 "program_lexer.l"
+#line 412 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_XZW;
    return_token_or_DOT(require_ARB_fp, MASK3);
 }
 	YY_BREAK
-case 171:
+case 158:
 YY_RULE_SETUP
-#line 412 "program_lexer.l"
+#line 417 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_YZW;
    return_token_or_DOT(require_ARB_fp, MASK3);
 }
 	YY_BREAK
-case 172:
+case 159:
 YY_RULE_SETUP
-#line 418 "program_lexer.l"
+#line 423 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_X
@@ -2285,9 +2366,9 @@ YY_RULE_SETUP
    return_token_or_DOT(require_ARB_fp, MASK2);
 }
 	YY_BREAK
-case 173:
+case 160:
 YY_RULE_SETUP
-#line 424 "program_lexer.l"
+#line 429 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_Y
@@ -2295,18 +2376,18 @@ YY_RULE_SETUP
    return_token_or_DOT(require_ARB_fp, MASK2);
 }
 	YY_BREAK
-case 174:
+case 161:
 YY_RULE_SETUP
-#line 430 "program_lexer.l"
+#line 435 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = SWIZZLE_INVAL;
    yylval->swiz_mask.mask = WRITEMASK_ZW;
    return_token_or_DOT(require_ARB_fp, MASK2);
 }
 	YY_BREAK
-case 175:
+case 162:
 YY_RULE_SETUP
-#line 436 "program_lexer.l"
+#line 441 "program_lexer.l"
 {
    const unsigned s = swiz_from_char(yytext[1]);
    yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s);
@@ -2314,9 +2395,9 @@ YY_RULE_SETUP
    return_token_or_DOT(require_ARB_fp, MASK1);
 }
 	YY_BREAK
-case 176:
+case 163:
 YY_RULE_SETUP
-#line 444 "program_lexer.l"
+#line 449 "program_lexer.l"
 {
    if (require_ARB_vp) {
       return TEXGEN_R;
@@ -2328,9 +2409,9 @@ YY_RULE_SETUP
    }
 }
 	YY_BREAK
-case 177:
+case 164:
 YY_RULE_SETUP
-#line 455 "program_lexer.l"
+#line 460 "program_lexer.l"
 {
    yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]),
 					    swiz_from_char(yytext[2]),
@@ -2340,15 +2421,15 @@ YY_RULE_SETUP
    return_token_or_DOT(require_ARB_fp, SWIZZLE);
 }
 	YY_BREAK
-case 178:
+case 165:
 YY_RULE_SETUP
-#line 464 "program_lexer.l"
+#line 469 "program_lexer.l"
 { return DOT; }
 	YY_BREAK
-case 179:
-/* rule 179 can match eol */
+case 166:
+/* rule 166 can match eol */
 YY_RULE_SETUP
-#line 466 "program_lexer.l"
+#line 471 "program_lexer.l"
 {
    yylloc->first_line++;
    yylloc->first_column = 1;
@@ -2357,30 +2438,30 @@ YY_RULE_SETUP
    yylloc->position++;
 }
 	YY_BREAK
-case 180:
+case 167:
 YY_RULE_SETUP
-#line 473 "program_lexer.l"
+#line 478 "program_lexer.l"
 /* eat whitespace */ ;
 	YY_BREAK
-case 181:
+case 168:
 *yy_cp = yyg->yy_hold_char; /* undo effects of setting up yytext */
 yyg->yy_c_buf_p = yy_cp -= 1;
 YY_DO_BEFORE_ACTION; /* set up yytext again */
 YY_RULE_SETUP
-#line 474 "program_lexer.l"
+#line 479 "program_lexer.l"
 /* eat comments */ ;
 	YY_BREAK
-case 182:
+case 169:
 YY_RULE_SETUP
-#line 475 "program_lexer.l"
+#line 480 "program_lexer.l"
 { return yytext[0]; }
 	YY_BREAK
-case 183:
+case 170:
 YY_RULE_SETUP
-#line 476 "program_lexer.l"
+#line 481 "program_lexer.l"
 ECHO;
 	YY_BREAK
-#line 2384 "lex.yy.c"
+#line 2465 "lex.yy.c"
 case YY_STATE_EOF(INITIAL):
 	yyterminate();
 
@@ -2674,7 +2755,7 @@ static int yy_get_next_buffer (yyscan_t yyscanner)
 		while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state )
 			{
 			yy_current_state = (int) yy_def[yy_current_state];
-			if ( yy_current_state >= 675 )
+			if ( yy_current_state >= 850 )
 				yy_c = yy_meta[(unsigned int) yy_c];
 			}
 		yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c];
@@ -2703,11 +2784,11 @@ static int yy_get_next_buffer (yyscan_t yyscanner)
 	while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state )
 		{
 		yy_current_state = (int) yy_def[yy_current_state];
-		if ( yy_current_state >= 675 )
+		if ( yy_current_state >= 850 )
 			yy_c = yy_meta[(unsigned int) yy_c];
 		}
 	yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c];
-	yy_is_jam = (yy_current_state == 674);
+	yy_is_jam = (yy_current_state == 849);
 
 	return yy_is_jam ? 0 : yy_current_state;
 }
@@ -3555,7 +3636,7 @@ void yyfree (void * ptr , yyscan_t yyscanner)
 
 #define YYTABLES_NAME "yytables"
 
-#line 476 "program_lexer.l"
+#line 481 "program_lexer.l"
 
 
 
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index 39a221eeab..1c687bc16c 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -121,7 +121,6 @@
 /*@{*/
 #define SATURATE_OFF            0
 #define SATURATE_ZERO_ONE       1
-#define SATURATE_PLUS_MINUS_ONE 2
 /*@}*/
 
 
diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c
index b71735aa80..ba4d39452f 100644
--- a/src/mesa/shader/prog_print.c
+++ b/src/mesa/shader/prog_print.c
@@ -788,7 +788,7 @@ _mesa_fprint_program_opt(FILE *f,
       else if (mode == PROG_PRINT_NV)
          _mesa_fprintf(f, "!!VP1.0\n");
       else
-         _mesa_fprintf(f, "# Vertex Program/Shader\n");
+         _mesa_fprintf(f, "# Vertex Program/Shader %u\n", prog->Id);
       break;
    case GL_FRAGMENT_PROGRAM_ARB:
    case GL_FRAGMENT_PROGRAM_NV:
@@ -797,7 +797,7 @@ _mesa_fprint_program_opt(FILE *f,
       else if (mode == PROG_PRINT_NV)
          _mesa_fprintf(f, "!!FP1.0\n");
       else
-         _mesa_fprintf(f, "# Fragment Program/Shader\n");
+         _mesa_fprintf(f, "# Fragment Program/Shader %u\n", prog->Id);
       break;
    }
 
@@ -906,7 +906,8 @@ _mesa_fprint_parameter_list(FILE *f,
    if (!list)
       return;
 
-   _mesa_fprintf(f, "param list %p\n", (void *) list);
+   if (0)
+      _mesa_fprintf(f, "param list %p\n", (void *) list);
    _mesa_fprintf(f, "dirty state flags: 0x%x\n", list->StateFlags);
    for (i = 0; i < list->NumParameters; i++){
       struct gl_program_parameter *param = list->Parameters + i;
diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c
index 2cd6eb8a38..532adf4d36 100644
--- a/src/mesa/shader/program.c
+++ b/src/mesa/shader/program.c
@@ -351,13 +351,6 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog)
       _mesa_free_parameter_list(prog->Attributes);
    }
 
-   /* XXX this is a little ugly */
-   if (prog->Target == GL_VERTEX_PROGRAM_ARB) {
-      struct gl_vertex_program *vprog = (struct gl_vertex_program *) prog;
-      if (vprog->TnlData)
-         _mesa_free(vprog->TnlData);
-   }
-
    _mesa_free(prog);
 }
 
diff --git a/src/mesa/shader/program_lexer.l b/src/mesa/shader/program_lexer.l
index c2803ff707..e2acb3c0c9 100644
--- a/src/mesa/shader/program_lexer.l
+++ b/src/mesa/shader/program_lexer.l
@@ -26,22 +26,27 @@
 #include "prog_instruction.h"
 #include "prog_statevars.h"
 
+#include "symbol_table.h"
 #include "program_parser.h"
 #include "program_parse.tab.h"
 
 #define require_ARB_vp (yyextra->mode == ARB_vertex)
 #define require_ARB_fp (yyextra->mode == ARB_fragment)
+#define require_NV_fp  (yyextra->option.NV_fragment)
 #define require_shadow (yyextra->option.Shadow)
 #define require_rect   (yyextra->option.TexRect)
 #define require_texarray        (yyextra->option.TexArray)
 
+#ifndef HAVE_UNISTD_H
+#define YY_NO_UNISTD_H
+#endif
+
 #define return_token_or_IDENTIFIER(condition, token)	\
    do {							\
       if (condition) {					\
 	 return token;					\
       } else {						\
-	 yylval->string = strdup(yytext);		\
-	 return IDENTIFIER;				\
+	 return handle_ident(yyextra, yytext, yylval);	\
       }							\
    } while (0)
 
@@ -56,15 +61,16 @@
    } while (0)
 
 
-#define return_opcode(condition, token, opcode, sat)	\
+#define return_opcode(condition, token, opcode, len)	\
    do {							\
-      if (condition) {					\
+      if (condition &&					\
+	  _mesa_parse_instruction_suffix(yyextra,	\
+					 yytext + len,	\
+					 & yylval->temp_inst)) {	\
 	 yylval->temp_inst.Opcode = OPCODE_ ## opcode;	\
-	 yylval->temp_inst.SaturateMode = SATURATE_ ## sat; \
 	 return token;					\
       } else {						\
-	 yylval->string = strdup(yytext);		\
-	 return IDENTIFIER;				\
+	 return handle_ident(yyextra, yytext, yylval);	\
       }							\
    } while (0)
 
@@ -113,6 +119,15 @@ swiz_from_char(char c)
    return 0;
 }
 
+static int
+handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval)
+{
+   lval->string = strdup(text);
+
+   return (_mesa_symbol_table_find_symbol(state->st, 0, text) == NULL)
+      ? IDENTIFIER : USED_IDENTIFIER;
+}
+
 #define YY_USER_ACTION							\
    do {									\
       yylloc->first_column = yylloc->last_column;			\
@@ -133,6 +148,11 @@ exp    [Ee][-+]?[0-9]+
 frac   "."[0-9]+
 dot    "."[ \t]*
 
+sz     [HRX]?
+szf    [HR]?
+cc     C?
+sat    (_SAT)?
+
 %option bison-bridge bison-locations reentrant noyywrap
 %%
 
@@ -150,86 +170,74 @@ OUTPUT                    { return OUTPUT; }
 PARAM                     { return PARAM; }
 TEMP                      { yylval->integer = at_temp; return TEMP; }
 
-ABS       { return_opcode(             1, VECTOR_OP, ABS, OFF); }
-ABS_SAT   { return_opcode(require_ARB_fp, VECTOR_OP, ABS, ZERO_ONE); }
-ADD       { return_opcode(             1, BIN_OP, ADD, OFF); }
-ADD_SAT   { return_opcode(require_ARB_fp, BIN_OP, ADD, ZERO_ONE); }
-ARL       { return_opcode(require_ARB_vp, ARL, ARL, OFF); }
-
-CMP       { return_opcode(require_ARB_fp, TRI_OP, CMP, OFF); }
-CMP_SAT   { return_opcode(require_ARB_fp, TRI_OP, CMP, ZERO_ONE); }
-COS       { return_opcode(require_ARB_fp, SCALAR_OP, COS, OFF); }
-COS_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, COS, ZERO_ONE); }
-
-DP3       { return_opcode(             1, BIN_OP, DP3, OFF); }
-DP3_SAT   { return_opcode(require_ARB_fp, BIN_OP, DP3, ZERO_ONE); }
-DP4       { return_opcode(             1, BIN_OP, DP4, OFF); }
-DP4_SAT   { return_opcode(require_ARB_fp, BIN_OP, DP4, ZERO_ONE); }
-DPH       { return_opcode(             1, BIN_OP, DPH, OFF); }
-DPH_SAT   { return_opcode(require_ARB_fp, BIN_OP, DPH, ZERO_ONE); }
-DST       { return_opcode(             1, BIN_OP, DST, OFF); }
-DST_SAT   { return_opcode(require_ARB_fp, BIN_OP, DST, ZERO_ONE); }
-
-EX2       { return_opcode(             1, SCALAR_OP, EX2, OFF); }
-EX2_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, EX2, ZERO_ONE); }
-EXP       { return_opcode(require_ARB_vp, SCALAR_OP, EXP, OFF); }
-
-FLR       { return_opcode(             1, VECTOR_OP, FLR, OFF); }
-FLR_SAT   { return_opcode(require_ARB_fp, VECTOR_OP, FLR, ZERO_ONE); }
-FRC       { return_opcode(             1, VECTOR_OP, FRC, OFF); }
-FRC_SAT   { return_opcode(require_ARB_fp, VECTOR_OP, FRC, ZERO_ONE); }
-
-KIL       { return_opcode(require_ARB_fp, KIL, KIL, OFF); }
-
-LIT       { return_opcode(             1, VECTOR_OP, LIT, OFF); }
-LIT_SAT   { return_opcode(require_ARB_fp, VECTOR_OP, LIT, ZERO_ONE); }
-LG2       { return_opcode(             1, SCALAR_OP, LG2, OFF); }
-LG2_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, LG2, ZERO_ONE); }
-LOG       { return_opcode(require_ARB_vp, SCALAR_OP, LOG, OFF); }
-LRP       { return_opcode(require_ARB_fp, TRI_OP, LRP, OFF); }
-LRP_SAT   { return_opcode(require_ARB_fp, TRI_OP, LRP, ZERO_ONE); }
-
-MAD       { return_opcode(             1, TRI_OP, MAD, OFF); }
-MAD_SAT   { return_opcode(require_ARB_fp, TRI_OP, MAD, ZERO_ONE); }
-MAX       { return_opcode(             1, BIN_OP, MAX, OFF); }
-MAX_SAT   { return_opcode(require_ARB_fp, BIN_OP, MAX, ZERO_ONE); }
-MIN       { return_opcode(             1, BIN_OP, MIN, OFF); }
-MIN_SAT   { return_opcode(require_ARB_fp, BIN_OP, MIN, ZERO_ONE); }
-MOV       { return_opcode(             1, VECTOR_OP, MOV, OFF); }
-MOV_SAT   { return_opcode(require_ARB_fp, VECTOR_OP, MOV, ZERO_ONE); }
-MUL       { return_opcode(             1, BIN_OP, MUL, OFF); }
-MUL_SAT   { return_opcode(require_ARB_fp, BIN_OP, MUL, ZERO_ONE); }
-
-POW       { return_opcode(             1, BINSC_OP, POW, OFF); }
-POW_SAT   { return_opcode(require_ARB_fp, BINSC_OP, POW, ZERO_ONE); }
-
-RCP       { return_opcode(             1, SCALAR_OP, RCP, OFF); }
-RCP_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, RCP, ZERO_ONE); }
-RSQ       { return_opcode(             1, SCALAR_OP, RSQ, OFF); }
-RSQ_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, RSQ, ZERO_ONE); }
-
-SCS       { return_opcode(require_ARB_fp, SCALAR_OP, SCS, OFF); }
-SCS_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, SCS, ZERO_ONE); }
-SGE       { return_opcode(             1, BIN_OP, SGE, OFF); }
-SGE_SAT   { return_opcode(require_ARB_fp, BIN_OP, SGE, ZERO_ONE); }
-SIN       { return_opcode(require_ARB_fp, SCALAR_OP, SIN, OFF); }
-SIN_SAT   { return_opcode(require_ARB_fp, SCALAR_OP, SIN, ZERO_ONE); }
-SLT       { return_opcode(             1, BIN_OP, SLT, OFF); }
-SLT_SAT   { return_opcode(require_ARB_fp, BIN_OP, SLT, ZERO_ONE); }
-SUB       { return_opcode(             1, BIN_OP, SUB, OFF); }
-SUB_SAT   { return_opcode(require_ARB_fp, BIN_OP, SUB, ZERO_ONE); }
-SWZ       { return_opcode(             1, SWZ, SWZ, OFF); }
-SWZ_SAT   { return_opcode(require_ARB_fp, SWZ, SWZ, ZERO_ONE); }
-
-TEX       { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, OFF); }
-TEX_SAT   { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, ZERO_ONE); }
-TXB       { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, OFF); }
-TXB_SAT   { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, ZERO_ONE); }
-TXP       { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, OFF); }
-TXP_SAT   { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, ZERO_ONE); }
-
-XPD       { return_opcode(             1, BIN_OP, XPD, OFF); }
-XPD_SAT   { return_opcode(require_ARB_fp, BIN_OP, XPD, ZERO_ONE); }
+ABS{sz}{cc}{sat}   { return_opcode(             1, VECTOR_OP, ABS, 3); }
+ADD{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, ADD, 3); }
+ARL                { return_opcode(require_ARB_vp, ARL, ARL, 3); }
+
+CMP{sat}           { return_opcode(require_ARB_fp, TRI_OP, CMP, 3); }
+COS{szf}{cc}{sat}  { return_opcode(require_ARB_fp, SCALAR_OP, COS, 3); }
+
+DDX{szf}{cc}{sat}  { return_opcode(require_NV_fp,  VECTOR_OP, DDX, 3); }
+DDY{szf}{cc}{sat}  { return_opcode(require_NV_fp,  VECTOR_OP, DDY, 3); }
+DP3{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, DP3, 3); }
+DP4{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, DP4, 3); }
+DPH{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, DPH, 3); }
+DST{szf}{cc}{sat}  { return_opcode(             1, BIN_OP, DST, 3); }
+
+EX2{szf}{cc}{sat}  { return_opcode(             1, SCALAR_OP, EX2, 3); }
+EXP                { return_opcode(require_ARB_vp, SCALAR_OP, EXP, 3); }
+
+FLR{sz}{cc}{sat}   { return_opcode(             1, VECTOR_OP, FLR, 3); }
+FRC{sz}{cc}{sat}   { return_opcode(             1, VECTOR_OP, FRC, 3); }
+
+KIL                { return_opcode(require_ARB_fp, KIL, KIL, 3); }
+
+LIT{szf}{cc}{sat}  { return_opcode(             1, VECTOR_OP, LIT, 3); }
+LG2{szf}{cc}{sat}  { return_opcode(             1, SCALAR_OP, LG2, 3); }
+LOG                { return_opcode(require_ARB_vp, SCALAR_OP, LOG, 3); }
+LRP{sz}{cc}{sat}   { return_opcode(require_ARB_fp, TRI_OP, LRP, 3); }
+
+MAD{sz}{cc}{sat}   { return_opcode(             1, TRI_OP, MAD, 3); }
+MAX{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, MAX, 3); }
+MIN{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, MIN, 3); }
+MOV{sz}{cc}{sat}   { return_opcode(             1, VECTOR_OP, MOV, 3); }
+MUL{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, MUL, 3); }
+
+PK2H               { return_opcode(require_NV_fp,  VECTOR_OP, PK2H, 4); }
+PK2US              { return_opcode(require_NV_fp,  VECTOR_OP, PK2US, 5); }
+PK4B               { return_opcode(require_NV_fp,  VECTOR_OP, PK4B, 4); }
+PK4UB              { return_opcode(require_NV_fp,  VECTOR_OP, PK4UB, 5); }
+POW{szf}{cc}{sat}  { return_opcode(             1, BINSC_OP, POW, 3); }
+
+RCP{szf}{cc}{sat}  { return_opcode(             1, SCALAR_OP, RCP, 3); }
+RFL{szf}{cc}{sat}  { return_opcode(require_NV_fp,  BIN_OP,    RFL, 3); }
+RSQ{szf}{cc}{sat}  { return_opcode(             1, SCALAR_OP, RSQ, 3); }
+
+SCS{sat}           { return_opcode(require_ARB_fp, SCALAR_OP, SCS, 3); }
+SEQ{sz}{cc}{sat}   { return_opcode(require_NV_fp,  BIN_OP, SEQ, 3); }
+SFL{sz}{cc}{sat}   { return_opcode(require_NV_fp,  BIN_OP, SFL, 3); }
+SGE{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, SGE, 3); }
+SGT{sz}{cc}{sat}   { return_opcode(require_NV_fp,  BIN_OP, SGT, 3); }
+SIN{szf}{cc}{sat}  { return_opcode(require_ARB_fp, SCALAR_OP, SIN, 3); }
+SLE{sz}{cc}{sat}   { return_opcode(require_NV_fp,  BIN_OP, SLE, 3); }
+SLT{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, SLT, 3); }
+SNE{sz}{cc}{sat}   { return_opcode(require_NV_fp,  BIN_OP, SNE, 3); }
+STR{sz}{cc}{sat}   { return_opcode(require_NV_fp,  BIN_OP, STR, 3); }
+SUB{sz}{cc}{sat}   { return_opcode(             1, BIN_OP, SUB, 3); }
+SWZ{sat}           { return_opcode(             1, SWZ, SWZ, 3); }
+
+TEX{cc}{sat}       { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, 3); }
+TXB{cc}{sat}       { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, 3); }
+TXD{cc}{sat}       { return_opcode(require_NV_fp,  TXD_OP, TXD, 3); }
+TXP{cc}{sat}       { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, 3); }
+
+UP2H{cc}{sat}      { return_opcode(require_NV_fp,  SCALAR_OP, UP2H, 4); }
+UP2US{cc}{sat}     { return_opcode(require_NV_fp,  SCALAR_OP, UP2US, 5); }
+UP4B{cc}{sat}      { return_opcode(require_NV_fp,  SCALAR_OP, UP4B, 4); }
+UP4UB{cc}{sat}     { return_opcode(require_NV_fp,  SCALAR_OP, UP4UB, 5); }
+
+X2D{szf}{cc}{sat}  { return_opcode(require_NV_fp,  TRI_OP, X2D, 3); }
+XPD{sat}           { return_opcode(             1, BIN_OP, XPD, 3); }
 
 vertex                    { return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); }
 fragment                  { return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); }
@@ -307,10 +315,7 @@ ARRAY2D                   { return_token_or_IDENTIFIER(require_ARB_fp && require
 ARRAYSHADOW1D             { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); }
 ARRAYSHADOW2D             { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); }
 
-[_a-zA-Z$][_a-zA-Z0-9$]*  {
-   yylval->string = strdup(yytext);
-   return IDENTIFIER;
-}
+[_a-zA-Z$][_a-zA-Z0-9$]*  { return handle_ident(yyextra, yytext, yylval); }
 
 ".."                      { return DOT_DOT; }
 
diff --git a/src/mesa/shader/program_parse.tab.c b/src/mesa/shader/program_parse.tab.c
index 9f2d4de90f..c0f50147bd 100644
--- a/src/mesa/shader/program_parse.tab.c
+++ b/src/mesa/shader/program_parse.tab.c
@@ -139,10 +139,19 @@ static void init_dst_reg(struct prog_dst_register *r);
 
 static void init_src_reg(struct asm_src_register *r);
 
+static void asm_instruction_set_operands(struct asm_instruction *inst,
+    const struct prog_dst_register *dst, const struct asm_src_register *src0,
+    const struct asm_src_register *src1, const struct asm_src_register *src2);
+
 static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
     const struct prog_dst_register *dst, const struct asm_src_register *src0,
     const struct asm_src_register *src1, const struct asm_src_register *src2);
 
+static struct asm_instruction *asm_instruction_copy_ctor(
+    const struct prog_instruction *base, const struct prog_dst_register *dst,
+    const struct asm_src_register *src0, const struct asm_src_register *src1,
+    const struct asm_src_register *src2);
+
 #ifndef FALSE
 #define FALSE 0
 #define TRUE (!FALSE)
@@ -170,7 +179,7 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
 
 
 /* Line 189 of yacc.c  */
-#line 174 "program_parse.tab.c"
+#line 183 "program_parse.tab.c"
 
 /* Enabling traces.  */
 #ifndef YYDEBUG
@@ -216,90 +225,92 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
      ARL = 274,
      KIL = 275,
      SWZ = 276,
-     INTEGER = 277,
-     REAL = 278,
-     AMBIENT = 279,
-     ATTENUATION = 280,
-     BACK = 281,
-     CLIP = 282,
-     COLOR = 283,
-     DEPTH = 284,
-     DIFFUSE = 285,
-     DIRECTION = 286,
-     EMISSION = 287,
-     ENV = 288,
-     EYE = 289,
-     FOG = 290,
-     FOGCOORD = 291,
-     FRAGMENT = 292,
-     FRONT = 293,
-     HALF = 294,
-     INVERSE = 295,
-     INVTRANS = 296,
-     LIGHT = 297,
-     LIGHTMODEL = 298,
-     LIGHTPROD = 299,
-     LOCAL = 300,
-     MATERIAL = 301,
-     MAT_PROGRAM = 302,
-     MATRIX = 303,
-     MATRIXINDEX = 304,
-     MODELVIEW = 305,
-     MVP = 306,
-     NORMAL = 307,
-     OBJECT = 308,
-     PALETTE = 309,
-     PARAMS = 310,
-     PLANE = 311,
-     POINT_TOK = 312,
-     POINTSIZE = 313,
-     POSITION = 314,
-     PRIMARY = 315,
-     PROGRAM = 316,
-     PROJECTION = 317,
-     RANGE = 318,
-     RESULT = 319,
-     ROW = 320,
-     SCENECOLOR = 321,
-     SECONDARY = 322,
-     SHININESS = 323,
-     SIZE_TOK = 324,
-     SPECULAR = 325,
-     SPOT = 326,
-     STATE = 327,
-     TEXCOORD = 328,
-     TEXENV = 329,
-     TEXGEN = 330,
-     TEXGEN_Q = 331,
-     TEXGEN_R = 332,
-     TEXGEN_S = 333,
-     TEXGEN_T = 334,
-     TEXTURE = 335,
-     TRANSPOSE = 336,
-     TEXTURE_UNIT = 337,
-     TEX_1D = 338,
-     TEX_2D = 339,
-     TEX_3D = 340,
-     TEX_CUBE = 341,
-     TEX_RECT = 342,
-     TEX_SHADOW1D = 343,
-     TEX_SHADOW2D = 344,
-     TEX_SHADOWRECT = 345,
-     TEX_ARRAY1D = 346,
-     TEX_ARRAY2D = 347,
-     TEX_ARRAYSHADOW1D = 348,
-     TEX_ARRAYSHADOW2D = 349,
-     VERTEX = 350,
-     VTXATTRIB = 351,
-     WEIGHT = 352,
-     IDENTIFIER = 353,
-     MASK4 = 354,
-     MASK3 = 355,
-     MASK2 = 356,
-     MASK1 = 357,
-     SWIZZLE = 358,
-     DOT_DOT = 359,
-     DOT = 360
+     TXD_OP = 277,
+     INTEGER = 278,
+     REAL = 279,
+     AMBIENT = 280,
+     ATTENUATION = 281,
+     BACK = 282,
+     CLIP = 283,
+     COLOR = 284,
+     DEPTH = 285,
+     DIFFUSE = 286,
+     DIRECTION = 287,
+     EMISSION = 288,
+     ENV = 289,
+     EYE = 290,
+     FOG = 291,
+     FOGCOORD = 292,
+     FRAGMENT = 293,
+     FRONT = 294,
+     HALF = 295,
+     INVERSE = 296,
+     INVTRANS = 297,
+     LIGHT = 298,
+     LIGHTMODEL = 299,
+     LIGHTPROD = 300,
+     LOCAL = 301,
+     MATERIAL = 302,
+     MAT_PROGRAM = 303,
+     MATRIX = 304,
+     MATRIXINDEX = 305,
+     MODELVIEW = 306,
+     MVP = 307,
+     NORMAL = 308,
+     OBJECT = 309,
+     PALETTE = 310,
+     PARAMS = 311,
+     PLANE = 312,
+     POINT_TOK = 313,
+     POINTSIZE = 314,
+     POSITION = 315,
+     PRIMARY = 316,
+     PROGRAM = 317,
+     PROJECTION = 318,
+     RANGE = 319,
+     RESULT = 320,
+     ROW = 321,
+     SCENECOLOR = 322,
+     SECONDARY = 323,
+     SHININESS = 324,
+     SIZE_TOK = 325,
+     SPECULAR = 326,
+     SPOT = 327,
+     STATE = 328,
+     TEXCOORD = 329,
+     TEXENV = 330,
+     TEXGEN = 331,
+     TEXGEN_Q = 332,
+     TEXGEN_R = 333,
+     TEXGEN_S = 334,
+     TEXGEN_T = 335,
+     TEXTURE = 336,
+     TRANSPOSE = 337,
+     TEXTURE_UNIT = 338,
+     TEX_1D = 339,
+     TEX_2D = 340,
+     TEX_3D = 341,
+     TEX_CUBE = 342,
+     TEX_RECT = 343,
+     TEX_SHADOW1D = 344,
+     TEX_SHADOW2D = 345,
+     TEX_SHADOWRECT = 346,
+     TEX_ARRAY1D = 347,
+     TEX_ARRAY2D = 348,
+     TEX_ARRAYSHADOW1D = 349,
+     TEX_ARRAYSHADOW2D = 350,
+     VERTEX = 351,
+     VTXATTRIB = 352,
+     WEIGHT = 353,
+     IDENTIFIER = 354,
+     USED_IDENTIFIER = 355,
+     MASK4 = 356,
+     MASK3 = 357,
+     MASK2 = 358,
+     MASK1 = 359,
+     SWIZZLE = 360,
+     DOT_DOT = 361,
+     DOT = 362
    };
 #endif
 
@@ -310,7 +321,7 @@ typedef union YYSTYPE
 {
 
 /* Line 214 of yacc.c  */
-#line 107 "program_parse.y"
+#line 116 "program_parse.y"
 
    struct asm_instruction *inst;
    struct asm_symbol *sym;
@@ -339,7 +350,7 @@ typedef union YYSTYPE
 
 
 /* Line 214 of yacc.c  */
-#line 343 "program_parse.tab.c"
+#line 354 "program_parse.tab.c"
 } YYSTYPE;
 # define YYSTYPE_IS_TRIVIAL 1
 # define yystype YYSTYPE /* obsolescent; will be withdrawn */
@@ -363,14 +374,14 @@ typedef struct YYLTYPE
 /* Copy the second part of user declarations.  */
 
 /* Line 264 of yacc.c  */
-#line 249 "program_parse.y"
+#line 261 "program_parse.y"
 
 extern int yylex(YYSTYPE *yylval_param, YYLTYPE *yylloc_param,
     void *yyscanner);
 
 
 /* Line 264 of yacc.c  */
-#line 374 "program_parse.tab.c"
+#line 385 "program_parse.tab.c"
 
 #ifdef short
 # undef short
@@ -587,20 +598,20 @@ union yyalloc
 /* YYFINAL -- State number of the termination state.  */
 #define YYFINAL  5
 /* YYLAST -- Last index in YYTABLE.  */
-#define YYLAST   342
+#define YYLAST   396
 
 /* YYNTOKENS -- Number of terminals.  */
-#define YYNTOKENS  115
+#define YYNTOKENS  120
 /* YYNNTS -- Number of nonterminals.  */
-#define YYNNTS  134
+#define YYNNTS  143
 /* YYNRULES -- Number of rules.  */
-#define YYNRULES  264
+#define YYNRULES  282
 /* YYNRULES -- Number of states.  */
-#define YYNSTATES  436
+#define YYNSTATES  475
 
 /* YYTRANSLATE(YYLEX) -- Bison symbol number corresponding to YYLEX.  */
 #define YYUNDEFTOK  2
-#define YYMAXUTOK   360
+#define YYMAXUTOK   362
 
 #define YYTRANSLATE(YYX)						\
   ((unsigned int) (YYX) <= YYMAXUTOK ? yytranslate[YYX] : YYUNDEFTOK)
@@ -612,15 +623,15 @@ static const yytype_uint8 yytranslate[] =
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
-       2,     2,     2,   110,   107,   111,     2,     2,     2,     2,
-       2,     2,     2,     2,     2,     2,     2,     2,     2,   106,
-       2,   112,     2,     2,     2,     2,     2,     2,     2,     2,
+     115,   116,     2,   113,   109,   114,     2,     2,     2,     2,
+       2,     2,     2,     2,     2,     2,     2,     2,     2,   108,
+       2,   117,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
-       2,   108,     2,   109,     2,     2,     2,     2,     2,     2,
+       2,   111,     2,   112,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
-       2,     2,     2,   113,     2,   114,     2,     2,     2,     2,
+       2,     2,     2,   118,   110,   119,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
@@ -644,7 +655,7 @@ static const yytype_uint8 yytranslate[] =
       75,    76,    77,    78,    79,    80,    81,    82,    83,    84,
       85,    86,    87,    88,    89,    90,    91,    92,    93,    94,
       95,    96,    97,    98,    99,   100,   101,   102,   103,   104,
-     105
+     105,   106,   107
 };
 
 #if YYDEBUG
@@ -654,142 +665,152 @@ static const yytype_uint16 yyprhs[] =
 {
        0,     0,     3,     8,    10,    12,    15,    16,    20,    23,
       24,    27,    30,    32,    34,    36,    38,    40,    42,    44,
-      46,    48,    50,    52,    57,    62,    67,    74,    81,    90,
-      99,   102,   105,   107,   109,   111,   113,   115,   117,   119,
-     121,   123,   125,   127,   129,   136,   140,   144,   147,   150,
-     158,   161,   163,   165,   167,   169,   174,   176,   178,   180,
-     182,   184,   186,   188,   192,   193,   196,   199,   201,   203,
-     205,   207,   209,   211,   213,   215,   217,   218,   220,   222,
-     224,   226,   227,   229,   231,   233,   235,   237,   239,   244,
-     247,   250,   252,   255,   257,   260,   262,   265,   270,   275,
-     277,   278,   282,   284,   286,   289,   291,   294,   296,   298,
-     302,   309,   310,   312,   315,   320,   322,   326,   328,   330,
-     332,   334,   336,   338,   340,   342,   344,   346,   349,   352,
-     355,   358,   361,   364,   367,   370,   373,   376,   379,   382,
-     386,   388,   390,   392,   398,   400,   402,   404,   407,   409,
-     411,   414,   416,   419,   426,   428,   432,   434,   436,   438,
-     440,   442,   447,   449,   451,   453,   455,   457,   459,   462,
-     464,   466,   472,   474,   477,   479,   481,   487,   490,   491,
-     498,   502,   503,   505,   507,   509,   511,   513,   516,   518,
-     520,   523,   528,   533,   534,   538,   540,   542,   544,   547,
-     549,   551,   553,   555,   561,   563,   567,   573,   579,   581,
-     585,   591,   593,   595,   597,   599,   601,   603,   605,   607,
-     609,   613,   619,   627,   637,   640,   643,   645,   647,   648,
-     649,   653,   654,   658,   662,   664,   669,   672,   675,   678,
-     681,   685,   688,   692,   693,   695,   697,   698,   700,   702,
-     703,   705,   707,   708,   710,   712,   713,   717,   718,   722,
-     723,   727,   729,   731,   733
+      46,    48,    50,    52,    54,    59,    64,    69,    76,    83,
+      92,   101,   104,   107,   120,   123,   125,   127,   129,   131,
+     133,   135,   137,   139,   141,   143,   145,   147,   154,   157,
+     162,   165,   167,   171,   177,   181,   184,   192,   195,   197,
+     199,   201,   203,   208,   210,   212,   214,   216,   218,   220,
+     222,   226,   227,   230,   233,   235,   237,   239,   241,   243,
+     245,   247,   249,   251,   252,   254,   256,   258,   260,   261,
+     265,   269,   270,   273,   276,   278,   280,   282,   284,   286,
+     288,   290,   292,   297,   300,   303,   305,   308,   310,   313,
+     315,   318,   323,   328,   330,   331,   335,   337,   339,   342,
+     344,   347,   349,   351,   355,   362,   363,   365,   368,   373,
+     375,   379,   381,   383,   385,   387,   389,   391,   393,   395,
+     397,   399,   402,   405,   408,   411,   414,   417,   420,   423,
+     426,   429,   432,   435,   439,   441,   443,   445,   451,   453,
+     455,   457,   460,   462,   464,   467,   469,   472,   479,   481,
+     485,   487,   489,   491,   493,   495,   500,   502,   504,   506,
+     508,   510,   512,   515,   517,   519,   525,   527,   530,   532,
+     534,   540,   543,   544,   551,   555,   556,   558,   560,   562,
+     564,   566,   569,   571,   573,   576,   581,   586,   587,   591,
+     593,   595,   597,   600,   602,   604,   606,   608,   614,   616,
+     620,   626,   632,   634,   638,   644,   646,   648,   650,   652,
+     654,   656,   658,   660,   662,   666,   672,   680,   690,   693,
+     696,   698,   700,   701,   702,   707,   709,   710,   711,   715,
+     719,   721,   727,   730,   733,   736,   739,   743,   746,   750,
+     751,   753,   755,   756,   758,   760,   761,   763,   765,   766,
+     768,   770,   771,   775,   776,   780,   781,   785,   787,   789,
+     791,   796,   798
 };
 
 /* YYRHS -- A `-1'-separated list of the rules' RHS.  */
 static const yytype_int16 yyrhs[] =
 {
-     116,     0,    -1,   117,   118,   120,    12,    -1,     3,    -1,
-       4,    -1,   118,   119,    -1,    -1,     8,    98,   106,    -1,
-     120,   121,    -1,    -1,   122,   106,    -1,   158,   106,    -1,
-     123,    -1,   124,    -1,   125,    -1,   126,    -1,   127,    -1,
-     128,    -1,   129,    -1,   130,    -1,   135,    -1,   131,    -1,
-     132,    -1,    19,   139,   107,   136,    -1,    18,   138,   107,
-     137,    -1,    16,   138,   107,   136,    -1,    14,   138,   107,
-     136,   107,   136,    -1,    13,   138,   107,   137,   107,   137,
-      -1,    17,   138,   107,   137,   107,   137,   107,   137,    -1,
-      15,   138,   107,   137,   107,   133,   107,   134,    -1,    20,
-     137,    -1,    82,   243,    -1,    83,    -1,    84,    -1,    85,
-      -1,    86,    -1,    87,    -1,    88,    -1,    89,    -1,    90,
-      -1,    91,    -1,    92,    -1,    93,    -1,    94,    -1,    21,
-     138,   107,   143,   107,   140,    -1,   229,   143,   155,    -1,
-     229,   143,   156,    -1,   144,   157,    -1,   152,   154,    -1,
-     141,   107,   141,   107,   141,   107,   141,    -1,   229,   142,
-      -1,    22,    -1,    98,    -1,    98,    -1,   160,    -1,   145,
-     108,   146,   109,    -1,   174,    -1,   236,    -1,    98,    -1,
-      98,    -1,   147,    -1,   148,    -1,    22,    -1,   152,   153,
-     149,    -1,    -1,   110,   150,    -1,   111,   151,    -1,    22,
-      -1,    22,    -1,    98,    -1,   102,    -1,   102,    -1,   102,
-      -1,   102,    -1,    99,    -1,   103,    -1,    -1,    99,    -1,
-     100,    -1,   101,    -1,   102,    -1,    -1,   159,    -1,   166,
-      -1,   230,    -1,   232,    -1,   235,    -1,   248,    -1,     7,
-      98,   112,   160,    -1,    95,   161,    -1,    37,   165,    -1,
-      59,    -1,    97,   163,    -1,    52,    -1,    28,   241,    -1,
-      36,    -1,    73,   242,    -1,    49,   108,   164,   109,    -1,
-      96,   108,   162,   109,    -1,    22,    -1,    -1,   108,   164,
-     109,    -1,    22,    -1,    59,    -1,    28,   241,    -1,    36,
-      -1,    73,   242,    -1,   167,    -1,   168,    -1,    10,    98,
-     170,    -1,    10,    98,   108,   169,   109,   171,    -1,    -1,
-      22,    -1,   112,   173,    -1,   112,   113,   172,   114,    -1,
-     175,    -1,   172,   107,   175,    -1,   177,    -1,   213,    -1,
-     223,    -1,   177,    -1,   213,    -1,   224,    -1,   176,    -1,
-     214,    -1,   223,    -1,   177,    -1,    72,   201,    -1,    72,
-     178,    -1,    72,   180,    -1,    72,   183,    -1,    72,   185,
-      -1,    72,   191,    -1,    72,   187,    -1,    72,   194,    -1,
-      72,   196,    -1,    72,   198,    -1,    72,   200,    -1,    72,
-     212,    -1,    46,   240,   179,    -1,   189,    -1,    32,    -1,
-      68,    -1,    42,   108,   190,   109,   181,    -1,   189,    -1,
-      59,    -1,    25,    -1,    71,   182,    -1,    39,    -1,    31,
-      -1,    43,   184,    -1,    24,    -1,   240,    66,    -1,    44,
-     108,   190,   109,   240,   186,    -1,   189,    -1,    74,   244,
-     188,    -1,    28,    -1,    24,    -1,    30,    -1,    70,    -1,
-      22,    -1,    75,   242,   192,   193,    -1,    34,    -1,    53,
-      -1,    78,    -1,    79,    -1,    77,    -1,    76,    -1,    35,
-     195,    -1,    28,    -1,    55,    -1,    27,   108,   197,   109,
-      56,    -1,    22,    -1,    57,   199,    -1,    69,    -1,    25,
-      -1,   203,    65,   108,   206,   109,    -1,   203,   202,    -1,
-      -1,    65,   108,   206,   104,   206,   109,    -1,    48,   207,
-     204,    -1,    -1,   205,    -1,    40,    -1,    81,    -1,    41,
-      -1,    22,    -1,    50,   208,    -1,    62,    -1,    51,    -1,
-      80,   242,    -1,    54,   108,   210,   109,    -1,    47,   108,
-     211,   109,    -1,    -1,   108,   209,   109,    -1,    22,    -1,
-      22,    -1,    22,    -1,    29,    63,    -1,   217,    -1,   220,
-      -1,   215,    -1,   218,    -1,    61,    33,   108,   216,   109,
-      -1,   221,    -1,   221,   104,   221,    -1,    61,    33,   108,
-     221,   109,    -1,    61,    45,   108,   219,   109,    -1,   222,
-      -1,   222,   104,   222,    -1,    61,    45,   108,   222,   109,
-      -1,    22,    -1,    22,    -1,   225,    -1,   227,    -1,   226,
-      -1,   227,    -1,   228,    -1,    23,    -1,    22,    -1,   113,
-     228,   114,    -1,   113,   228,   107,   228,   114,    -1,   113,
-     228,   107,   228,   107,   228,   114,    -1,   113,   228,   107,
-     228,   107,   228,   107,   228,   114,    -1,   229,    23,    -1,
-     229,    22,    -1,   110,    -1,   111,    -1,    -1,    -1,    11,
-     231,   234,    -1,    -1,     5,   233,   234,    -1,   234,   107,
-      98,    -1,    98,    -1,     9,    98,   112,   236,    -1,    64,
-      59,    -1,    64,    36,    -1,    64,   237,    -1,    64,    58,
-      -1,    64,    73,   242,    -1,    64,    29,    -1,    28,   238,
-     239,    -1,    -1,    38,    -1,    26,    -1,    -1,    60,    -1,
-      67,    -1,    -1,    38,    -1,    26,    -1,    -1,    60,    -1,
-      67,    -1,    -1,   108,   245,   109,    -1,    -1,   108,   246,
-     109,    -1,    -1,   108,   247,   109,    -1,    22,    -1,    22,
-      -1,    22,    -1,     6,    98,   112,    98,    -1
+     121,     0,    -1,   122,   123,   125,    12,    -1,     3,    -1,
+       4,    -1,   123,   124,    -1,    -1,     8,   262,   108,    -1,
+     125,   126,    -1,    -1,   127,   108,    -1,   170,   108,    -1,
+     128,    -1,   129,    -1,   130,    -1,   131,    -1,   132,    -1,
+     133,    -1,   134,    -1,   135,    -1,   141,    -1,   136,    -1,
+     137,    -1,   138,    -1,    19,   146,   109,   142,    -1,    18,
+     145,   109,   144,    -1,    16,   145,   109,   142,    -1,    14,
+     145,   109,   142,   109,   142,    -1,    13,   145,   109,   144,
+     109,   144,    -1,    17,   145,   109,   144,   109,   144,   109,
+     144,    -1,    15,   145,   109,   144,   109,   139,   109,   140,
+      -1,    20,   144,    -1,    20,   166,    -1,    22,   145,   109,
+     144,   109,   144,   109,   144,   109,   139,   109,   140,    -1,
+      83,   256,    -1,    84,    -1,    85,    -1,    86,    -1,    87,
+      -1,    88,    -1,    89,    -1,    90,    -1,    91,    -1,    92,
+      -1,    93,    -1,    94,    -1,    95,    -1,    21,   145,   109,
+     150,   109,   147,    -1,   241,   143,    -1,   241,   110,   143,
+     110,    -1,   150,   162,    -1,   238,    -1,   241,   150,   163,
+      -1,   241,   110,   150,   163,   110,    -1,   151,   164,   165,
+      -1,   159,   161,    -1,   148,   109,   148,   109,   148,   109,
+     148,    -1,   241,   149,    -1,    23,    -1,   262,    -1,   100,
+      -1,   172,    -1,   152,   111,   153,   112,    -1,   186,    -1,
+     249,    -1,   100,    -1,   100,    -1,   154,    -1,   155,    -1,
+      23,    -1,   159,   160,   156,    -1,    -1,   113,   157,    -1,
+     114,   158,    -1,    23,    -1,    23,    -1,   100,    -1,   104,
+      -1,   104,    -1,   104,    -1,   104,    -1,   101,    -1,   105,
+      -1,    -1,   101,    -1,   102,    -1,   103,    -1,   104,    -1,
+      -1,   115,   166,   116,    -1,   115,   167,   116,    -1,    -1,
+     168,   163,    -1,   169,   163,    -1,    99,    -1,   100,    -1,
+     171,    -1,   178,    -1,   242,    -1,   245,    -1,   248,    -1,
+     261,    -1,     7,    99,   117,   172,    -1,    96,   173,    -1,
+      38,   177,    -1,    60,    -1,    98,   175,    -1,    53,    -1,
+      29,   254,    -1,    37,    -1,    74,   255,    -1,    50,   111,
+     176,   112,    -1,    97,   111,   174,   112,    -1,    23,    -1,
+      -1,   111,   176,   112,    -1,    23,    -1,    60,    -1,    29,
+     254,    -1,    37,    -1,    74,   255,    -1,   179,    -1,   180,
+      -1,    10,    99,   182,    -1,    10,    99,   111,   181,   112,
+     183,    -1,    -1,    23,    -1,   117,   185,    -1,   117,   118,
+     184,   119,    -1,   187,    -1,   184,   109,   187,    -1,   189,
+      -1,   225,    -1,   235,    -1,   189,    -1,   225,    -1,   236,
+      -1,   188,    -1,   226,    -1,   235,    -1,   189,    -1,    73,
+     213,    -1,    73,   190,    -1,    73,   192,    -1,    73,   195,
+      -1,    73,   197,    -1,    73,   203,    -1,    73,   199,    -1,
+      73,   206,    -1,    73,   208,    -1,    73,   210,    -1,    73,
+     212,    -1,    73,   224,    -1,    47,   253,   191,    -1,   201,
+      -1,    33,    -1,    69,    -1,    43,   111,   202,   112,   193,
+      -1,   201,    -1,    60,    -1,    26,    -1,    72,   194,    -1,
+      40,    -1,    32,    -1,    44,   196,    -1,    25,    -1,   253,
+      67,    -1,    45,   111,   202,   112,   253,   198,    -1,   201,
+      -1,    75,   257,   200,    -1,    29,    -1,    25,    -1,    31,
+      -1,    71,    -1,    23,    -1,    76,   255,   204,   205,    -1,
+      35,    -1,    54,    -1,    79,    -1,    80,    -1,    78,    -1,
+      77,    -1,    36,   207,    -1,    29,    -1,    56,    -1,    28,
+     111,   209,   112,    57,    -1,    23,    -1,    58,   211,    -1,
+      70,    -1,    26,    -1,   215,    66,   111,   218,   112,    -1,
+     215,   214,    -1,    -1,    66,   111,   218,   106,   218,   112,
+      -1,    49,   219,   216,    -1,    -1,   217,    -1,    41,    -1,
+      82,    -1,    42,    -1,    23,    -1,    51,   220,    -1,    63,
+      -1,    52,    -1,    81,   255,    -1,    55,   111,   222,   112,
+      -1,    48,   111,   223,   112,    -1,    -1,   111,   221,   112,
+      -1,    23,    -1,    23,    -1,    23,    -1,    30,    64,    -1,
+     229,    -1,   232,    -1,   227,    -1,   230,    -1,    62,    34,
+     111,   228,   112,    -1,   233,    -1,   233,   106,   233,    -1,
+      62,    34,   111,   233,   112,    -1,    62,    46,   111,   231,
+     112,    -1,   234,    -1,   234,   106,   234,    -1,    62,    46,
+     111,   234,   112,    -1,    23,    -1,    23,    -1,   237,    -1,
+     239,    -1,   238,    -1,   239,    -1,   240,    -1,    24,    -1,
+      23,    -1,   118,   240,   119,    -1,   118,   240,   109,   240,
+     119,    -1,   118,   240,   109,   240,   109,   240,   119,    -1,
+     118,   240,   109,   240,   109,   240,   109,   240,   119,    -1,
+     241,    24,    -1,   241,    23,    -1,   113,    -1,   114,    -1,
+      -1,    -1,   244,    11,   243,   247,    -1,   262,    -1,    -1,
+      -1,     5,   246,   247,    -1,   247,   109,    99,    -1,    99,
+      -1,   244,     9,    99,   117,   249,    -1,    65,    60,    -1,
+      65,    37,    -1,    65,   250,    -1,    65,    59,    -1,    65,
+      74,   255,    -1,    65,    30,    -1,    29,   251,   252,    -1,
+      -1,    39,    -1,    27,    -1,    -1,    61,    -1,    68,    -1,
+      -1,    39,    -1,    27,    -1,    -1,    61,    -1,    68,    -1,
+      -1,   111,   258,   112,    -1,    -1,   111,   259,   112,    -1,
+      -1,   111,   260,   112,    -1,    23,    -1,    23,    -1,    23,
+      -1,     6,    99,   117,   100,    -1,    99,    -1,   100,    -1
 };
 
 /* YYRLINE[YYN] -- source line where rule number YYN was defined.  */
 static const yytype_uint16 yyrline[] =
 {
-       0,   256,   256,   259,   267,   279,   280,   283,   305,   306,
-     309,   324,   327,   332,   339,   340,   341,   342,   343,   344,
-     345,   348,   349,   352,   358,   365,   372,   380,   387,   395,
-     440,   447,   453,   454,   455,   456,   457,   458,   459,   460,
-     461,   462,   463,   464,   467,   480,   493,   506,   528,   537,
-     570,   577,   592,   642,   684,   695,   716,   726,   732,   763,
-     780,   780,   782,   789,   801,   802,   803,   806,   818,   830,
-     848,   859,   871,   873,   874,   875,   876,   879,   879,   879,
-     879,   880,   883,   884,   885,   886,   887,   888,   891,   909,
-     913,   919,   923,   927,   931,   940,   949,   953,   958,   964,
-     975,   975,   976,   978,   982,   986,   990,   996,   996,   998,
-    1014,  1037,  1040,  1051,  1057,  1063,  1064,  1071,  1077,  1083,
-    1091,  1097,  1103,  1111,  1117,  1123,  1131,  1132,  1135,  1136,
-    1137,  1138,  1139,  1140,  1141,  1142,  1143,  1144,  1145,  1148,
-    1157,  1161,  1165,  1171,  1180,  1184,  1188,  1197,  1201,  1207,
-    1213,  1220,  1225,  1233,  1243,  1245,  1253,  1259,  1263,  1267,
-    1273,  1284,  1293,  1297,  1302,  1306,  1310,  1314,  1320,  1327,
-    1331,  1337,  1345,  1356,  1363,  1367,  1373,  1383,  1394,  1398,
-    1416,  1425,  1428,  1434,  1438,  1442,  1448,  1459,  1464,  1469,
-    1474,  1479,  1484,  1492,  1495,  1500,  1513,  1521,  1532,  1540,
-    1540,  1542,  1542,  1544,  1554,  1559,  1566,  1576,  1585,  1590,
-    1597,  1607,  1617,  1629,  1629,  1630,  1630,  1632,  1642,  1650,
-    1660,  1668,  1676,  1685,  1696,  1700,  1706,  1707,  1708,  1711,
-    1711,  1714,  1714,  1717,  1723,  1731,  1744,  1753,  1762,  1766,
-    1775,  1784,  1795,  1802,  1807,  1816,  1828,  1831,  1840,  1851,
-    1852,  1853,  1856,  1857,  1858,  1861,  1862,  1865,  1866,  1869,
-    1870,  1873,  1884,  1895,  1906
+       0,   268,   268,   271,   279,   291,   292,   295,   317,   318,
+     321,   336,   339,   344,   351,   352,   353,   354,   355,   356,
+     357,   360,   361,   362,   365,   371,   377,   383,   390,   396,
+     403,   447,   452,   462,   506,   512,   513,   514,   515,   516,
+     517,   518,   519,   520,   521,   522,   523,   526,   538,   546,
+     563,   570,   589,   600,   620,   645,   654,   687,   694,   709,
+     759,   801,   812,   833,   843,   849,   880,   897,   897,   899,
+     906,   918,   919,   920,   923,   937,   951,   969,   980,   992,
+     994,   995,   996,   997,  1000,  1000,  1000,  1000,  1001,  1004,
+    1008,  1013,  1020,  1027,  1034,  1057,  1080,  1081,  1082,  1083,
+    1084,  1085,  1088,  1106,  1110,  1116,  1120,  1124,  1128,  1137,
+    1146,  1150,  1155,  1161,  1172,  1172,  1173,  1175,  1179,  1183,
+    1187,  1193,  1193,  1195,  1211,  1234,  1237,  1248,  1254,  1260,
+    1261,  1268,  1274,  1280,  1288,  1294,  1300,  1308,  1314,  1320,
+    1328,  1329,  1332,  1333,  1334,  1335,  1336,  1337,  1338,  1339,
+    1340,  1341,  1342,  1345,  1354,  1358,  1362,  1368,  1377,  1381,
+    1385,  1394,  1398,  1404,  1410,  1417,  1422,  1430,  1440,  1442,
+    1450,  1456,  1460,  1464,  1470,  1481,  1490,  1494,  1499,  1503,
+    1507,  1511,  1517,  1524,  1528,  1534,  1542,  1553,  1560,  1564,
+    1570,  1580,  1591,  1595,  1613,  1622,  1625,  1631,  1635,  1639,
+    1645,  1656,  1661,  1666,  1671,  1676,  1681,  1689,  1692,  1697,
+    1710,  1718,  1729,  1737,  1737,  1739,  1739,  1741,  1751,  1756,
+    1763,  1773,  1782,  1787,  1794,  1804,  1814,  1826,  1826,  1827,
+    1827,  1829,  1839,  1847,  1857,  1865,  1873,  1882,  1893,  1897,
+    1903,  1904,  1905,  1908,  1908,  1911,  1946,  1950,  1950,  1953,
+    1959,  1967,  1980,  1989,  1998,  2002,  2011,  2020,  2031,  2038,
+    2043,  2052,  2064,  2067,  2076,  2087,  2088,  2089,  2092,  2093,
+    2094,  2097,  2098,  2101,  2102,  2105,  2106,  2109,  2120,  2131,
+    2142,  2163,  2164
 };
 #endif
 
@@ -801,9 +822,9 @@ static const char *const yytname[] =
   "$end", "error", "$undefined", "ARBvp_10", "ARBfp_10", "ADDRESS",
   "ALIAS", "ATTRIB", "OPTION", "OUTPUT", "PARAM", "TEMP", "END", "BIN_OP",
   "BINSC_OP", "SAMPLE_OP", "SCALAR_OP", "TRI_OP", "VECTOR_OP", "ARL",
-  "KIL", "SWZ", "INTEGER", "REAL", "AMBIENT", "ATTENUATION", "BACK",
-  "CLIP", "COLOR", "DEPTH", "DIFFUSE", "DIRECTION", "EMISSION", "ENV",
-  "EYE", "FOG", "FOGCOORD", "FRAGMENT", "FRONT", "HALF", "INVERSE",
+  "KIL", "SWZ", "TXD_OP", "INTEGER", "REAL", "AMBIENT", "ATTENUATION",
+  "BACK", "CLIP", "COLOR", "DEPTH", "DIFFUSE", "DIRECTION", "EMISSION",
+  "ENV", "EYE", "FOG", "FOGCOORD", "FRAGMENT", "FRONT", "HALF", "INVERSE",
   "INVTRANS", "LIGHT", "LIGHTMODEL", "LIGHTPROD", "LOCAL", "MATERIAL",
   "MAT_PROGRAM", "MATRIX", "MATRIXINDEX", "MODELVIEW", "MVP", "NORMAL",
   "OBJECT", "PALETTE", "PARAMS", "PLANE", "POINT_TOK", "POINTSIZE",
@@ -814,49 +835,51 @@ static const char *const yytname[] =
   "TEX_2D", "TEX_3D", "TEX_CUBE", "TEX_RECT", "TEX_SHADOW1D",
   "TEX_SHADOW2D", "TEX_SHADOWRECT", "TEX_ARRAY1D", "TEX_ARRAY2D",
   "TEX_ARRAYSHADOW1D", "TEX_ARRAYSHADOW2D", "VERTEX", "VTXATTRIB",
-  "WEIGHT", "IDENTIFIER", "MASK4", "MASK3", "MASK2", "MASK1", "SWIZZLE",
-  "DOT_DOT", "DOT", "';'", "','", "'['", "']'", "'+'", "'-'", "'='", "'{'",
-  "'}'", "$accept", "program", "language", "optionSequence", "option",
-  "statementSequence", "statement", "instruction", "ALU_instruction",
-  "TexInstruction", "ARL_instruction", "VECTORop_instruction",
-  "SCALARop_instruction", "BINSCop_instruction", "BINop_instruction",
-  "TRIop_instruction", "SAMPLE_instruction", "KIL_instruction",
-  "texImageUnit", "texTarget", "SWZ_instruction", "scalarSrcReg",
-  "swizzleSrcReg", "maskedDstReg", "maskedAddrReg", "extendedSwizzle",
-  "extSwizComp", "extSwizSel", "srcReg", "dstReg", "progParamArray",
-  "progParamArrayMem", "progParamArrayAbs", "progParamArrayRel",
-  "addrRegRelOffset", "addrRegPosOffset", "addrRegNegOffset", "addrReg",
-  "addrComponent", "addrWriteMask", "scalarSuffix", "swizzleSuffix",
-  "optionalMask", "namingStatement", "ATTRIB_statement", "attribBinding",
-  "vtxAttribItem", "vtxAttribNum", "vtxOptWeightNum", "vtxWeightNum",
-  "fragAttribItem", "PARAM_statement", "PARAM_singleStmt",
-  "PARAM_multipleStmt", "optArraySize", "paramSingleInit",
-  "paramMultipleInit", "paramMultInitList", "paramSingleItemDecl",
-  "paramSingleItemUse", "paramMultipleItem", "stateMultipleItem",
-  "stateSingleItem", "stateMaterialItem", "stateMatProperty",
-  "stateLightItem", "stateLightProperty", "stateSpotProperty",
-  "stateLightModelItem", "stateLModProperty", "stateLightProdItem",
-  "stateLProdProperty", "stateTexEnvItem", "stateTexEnvProperty",
-  "ambDiffSpecProperty", "stateLightNumber", "stateTexGenItem",
-  "stateTexGenType", "stateTexGenCoord", "stateFogItem",
-  "stateFogProperty", "stateClipPlaneItem", "stateClipPlaneNum",
-  "statePointItem", "statePointProperty", "stateMatrixRow",
-  "stateMatrixRows", "optMatrixRows", "stateMatrixItem",
-  "stateOptMatModifier", "stateMatModifier", "stateMatrixRowNum",
-  "stateMatrixName", "stateOptModMatNum", "stateModMatNum",
-  "statePaletteMatNum", "stateProgramMatNum", "stateDepthItem",
-  "programSingleItem", "programMultipleItem", "progEnvParams",
-  "progEnvParamNums", "progEnvParam", "progLocalParams",
-  "progLocalParamNums", "progLocalParam", "progEnvParamNum",
-  "progLocalParamNum", "paramConstDecl", "paramConstUse",
-  "paramConstScalarDecl", "paramConstScalarUse", "paramConstVector",
-  "signedFloatConstant", "optionalSign", "TEMP_statement", "@1",
+  "WEIGHT", "IDENTIFIER", "USED_IDENTIFIER", "MASK4", "MASK3", "MASK2",
+  "MASK1", "SWIZZLE", "DOT_DOT", "DOT", "';'", "','", "'|'", "'['", "']'",
+  "'+'", "'-'", "'('", "')'", "'='", "'{'", "'}'", "$accept", "program",
+  "language", "optionSequence", "option", "statementSequence", "statement",
+  "instruction", "ALU_instruction", "TexInstruction", "ARL_instruction",
+  "VECTORop_instruction", "SCALARop_instruction", "BINSCop_instruction",
+  "BINop_instruction", "TRIop_instruction", "SAMPLE_instruction",
+  "KIL_instruction", "TXD_instruction", "texImageUnit", "texTarget",
+  "SWZ_instruction", "scalarSrcReg", "scalarUse", "swizzleSrcReg",
+  "maskedDstReg", "maskedAddrReg", "extendedSwizzle", "extSwizComp",
+  "extSwizSel", "srcReg", "dstReg", "progParamArray", "progParamArrayMem",
+  "progParamArrayAbs", "progParamArrayRel", "addrRegRelOffset",
+  "addrRegPosOffset", "addrRegNegOffset", "addrReg", "addrComponent",
+  "addrWriteMask", "scalarSuffix", "swizzleSuffix", "optionalMask",
+  "optionalCcMask", "ccTest", "ccTest2", "ccMaskRule", "ccMaskRule2",
+  "namingStatement", "ATTRIB_statement", "attribBinding", "vtxAttribItem",
+  "vtxAttribNum", "vtxOptWeightNum", "vtxWeightNum", "fragAttribItem",
+  "PARAM_statement", "PARAM_singleStmt", "PARAM_multipleStmt",
+  "optArraySize", "paramSingleInit", "paramMultipleInit",
+  "paramMultInitList", "paramSingleItemDecl", "paramSingleItemUse",
+  "paramMultipleItem", "stateMultipleItem", "stateSingleItem",
+  "stateMaterialItem", "stateMatProperty", "stateLightItem",
+  "stateLightProperty", "stateSpotProperty", "stateLightModelItem",
+  "stateLModProperty", "stateLightProdItem", "stateLProdProperty",
+  "stateTexEnvItem", "stateTexEnvProperty", "ambDiffSpecProperty",
+  "stateLightNumber", "stateTexGenItem", "stateTexGenType",
+  "stateTexGenCoord", "stateFogItem", "stateFogProperty",
+  "stateClipPlaneItem", "stateClipPlaneNum", "statePointItem",
+  "statePointProperty", "stateMatrixRow", "stateMatrixRows",
+  "optMatrixRows", "stateMatrixItem", "stateOptMatModifier",
+  "stateMatModifier", "stateMatrixRowNum", "stateMatrixName",
+  "stateOptModMatNum", "stateModMatNum", "statePaletteMatNum",
+  "stateProgramMatNum", "stateDepthItem", "programSingleItem",
+  "programMultipleItem", "progEnvParams", "progEnvParamNums",
+  "progEnvParam", "progLocalParams", "progLocalParamNums",
+  "progLocalParam", "progEnvParamNum", "progLocalParamNum",
+  "paramConstDecl", "paramConstUse", "paramConstScalarDecl",
+  "paramConstScalarUse", "paramConstVector", "signedFloatConstant",
+  "optionalSign", "TEMP_statement", "@1", "optVarSize",
   "ADDRESS_statement", "@2", "varNameList", "OUTPUT_statement",
   "resultBinding", "resultColBinding", "optResultFaceType",
   "optResultColorType", "optFaceType", "optColorType",
   "optTexCoordUnitNum", "optTexImageUnitNum", "optLegacyTexUnitNum",
   "texCoordUnitNum", "texImageUnitNum", "legacyTexUnitNum",
-  "ALIAS_statement", 0
+  "ALIAS_statement", "string", 0
 };
 #endif
 
@@ -875,41 +898,43 @@ static const yytype_uint16 yytoknum[] =
      325,   326,   327,   328,   329,   330,   331,   332,   333,   334,
      335,   336,   337,   338,   339,   340,   341,   342,   343,   344,
      345,   346,   347,   348,   349,   350,   351,   352,   353,   354,
-     355,   356,   357,   358,   359,   360,    59,    44,    91,    93,
-      43,    45,    61,   123,   125
+     355,   356,   357,   358,   359,   360,   361,   362,    59,    44,
+     124,    91,    93,    43,    45,    40,    41,    61,   123,   125
 };
 # endif
 
 /* YYR1[YYN] -- Symbol number of symbol that rule YYN derives.  */
-static const yytype_uint8 yyr1[] =
+static const yytype_uint16 yyr1[] =
 {
-       0,   115,   116,   117,   117,   118,   118,   119,   120,   120,
-     121,   121,   122,   122,   123,   123,   123,   123,   123,   123,
-     123,   124,   124,   125,   126,   127,   128,   129,   130,   131,
-     132,   133,   134,   134,   134,   134,   134,   134,   134,   134,
-     134,   134,   134,   134,   135,   136,   137,   138,   139,   140,
-     141,   142,   142,   143,   143,   143,   143,   144,   144,   145,
-     146,   146,   147,   148,   149,   149,   149,   150,   151,   152,
-     153,   154,   155,   156,   156,   156,   156,   157,   157,   157,
-     157,   157,   158,   158,   158,   158,   158,   158,   159,   160,
-     160,   161,   161,   161,   161,   161,   161,   161,   161,   162,
-     163,   163,   164,   165,   165,   165,   165,   166,   166,   167,
-     168,   169,   169,   170,   171,   172,   172,   173,   173,   173,
-     174,   174,   174,   175,   175,   175,   176,   176,   177,   177,
-     177,   177,   177,   177,   177,   177,   177,   177,   177,   178,
-     179,   179,   179,   180,   181,   181,   181,   181,   181,   182,
-     183,   184,   184,   185,   186,   187,   188,   189,   189,   189,
-     190,   191,   192,   192,   193,   193,   193,   193,   194,   195,
-     195,   196,   197,   198,   199,   199,   200,   201,   202,   202,
-     203,   204,   204,   205,   205,   205,   206,   207,   207,   207,
-     207,   207,   207,   208,   208,   209,   210,   211,   212,   213,
-     213,   214,   214,   215,   216,   216,   217,   218,   219,   219,
-     220,   221,   222,   223,   223,   224,   224,   225,   226,   226,
-     227,   227,   227,   227,   228,   228,   229,   229,   229,   231,
-     230,   233,   232,   234,   234,   235,   236,   236,   236,   236,
-     236,   236,   237,   238,   238,   238,   239,   239,   239,   240,
-     240,   240,   241,   241,   241,   242,   242,   243,   243,   244,
-     244,   245,   246,   247,   248
+       0,   120,   121,   122,   122,   123,   123,   124,   125,   125,
+     126,   126,   127,   127,   128,   128,   128,   128,   128,   128,
+     128,   129,   129,   129,   130,   131,   132,   133,   134,   135,
+     136,   137,   137,   138,   139,   140,   140,   140,   140,   140,
+     140,   140,   140,   140,   140,   140,   140,   141,   142,   142,
+     143,   143,   144,   144,   145,   146,   147,   148,   149,   149,
+     150,   150,   150,   150,   151,   151,   152,   153,   153,   154,
+     155,   156,   156,   156,   157,   158,   159,   160,   161,   162,
+     163,   163,   163,   163,   164,   164,   164,   164,   164,   165,
+     165,   165,   166,   167,   168,   169,   170,   170,   170,   170,
+     170,   170,   171,   172,   172,   173,   173,   173,   173,   173,
+     173,   173,   173,   174,   175,   175,   176,   177,   177,   177,
+     177,   178,   178,   179,   180,   181,   181,   182,   183,   184,
+     184,   185,   185,   185,   186,   186,   186,   187,   187,   187,
+     188,   188,   189,   189,   189,   189,   189,   189,   189,   189,
+     189,   189,   189,   190,   191,   191,   191,   192,   193,   193,
+     193,   193,   193,   194,   195,   196,   196,   197,   198,   199,
+     200,   201,   201,   201,   202,   203,   204,   204,   205,   205,
+     205,   205,   206,   207,   207,   208,   209,   210,   211,   211,
+     212,   213,   214,   214,   215,   216,   216,   217,   217,   217,
+     218,   219,   219,   219,   219,   219,   219,   220,   220,   221,
+     222,   223,   224,   225,   225,   226,   226,   227,   228,   228,
+     229,   230,   231,   231,   232,   233,   234,   235,   235,   236,
+     236,   237,   238,   238,   239,   239,   239,   239,   240,   240,
+     241,   241,   241,   243,   242,   244,   244,   246,   245,   247,
+     247,   248,   249,   249,   249,   249,   249,   249,   250,   251,
+     251,   251,   252,   252,   252,   253,   253,   253,   254,   254,
+     254,   255,   255,   256,   256,   257,   257,   258,   259,   260,
+     261,   262,   262
 };
 
 /* YYR2[YYN] -- Number of symbols composing right hand side of rule YYN.  */
@@ -917,31 +942,33 @@ static const yytype_uint8 yyr2[] =
 {
        0,     2,     4,     1,     1,     2,     0,     3,     2,     0,
        2,     2,     1,     1,     1,     1,     1,     1,     1,     1,
-       1,     1,     1,     4,     4,     4,     6,     6,     8,     8,
-       2,     2,     1,     1,     1,     1,     1,     1,     1,     1,
-       1,     1,     1,     1,     6,     3,     3,     2,     2,     7,
-       2,     1,     1,     1,     1,     4,     1,     1,     1,     1,
-       1,     1,     1,     3,     0,     2,     2,     1,     1,     1,
-       1,     1,     1,     1,     1,     1,     0,     1,     1,     1,
-       1,     0,     1,     1,     1,     1,     1,     1,     4,     2,
-       2,     1,     2,     1,     2,     1,     2,     4,     4,     1,
-       0,     3,     1,     1,     2,     1,     2,     1,     1,     3,
-       6,     0,     1,     2,     4,     1,     3,     1,     1,     1,
-       1,     1,     1,     1,     1,     1,     1,     2,     2,     2,
-       2,     2,     2,     2,     2,     2,     2,     2,     2,     3,
-       1,     1,     1,     5,     1,     1,     1,     2,     1,     1,
-       2,     1,     2,     6,     1,     3,     1,     1,     1,     1,
-       1,     4,     1,     1,     1,     1,     1,     1,     2,     1,
-       1,     5,     1,     2,     1,     1,     5,     2,     0,     6,
-       3,     0,     1,     1,     1,     1,     1,     2,     1,     1,
-       2,     4,     4,     0,     3,     1,     1,     1,     2,     1,
-       1,     1,     1,     5,     1,     3,     5,     5,     1,     3,
-       5,     1,     1,     1,     1,     1,     1,     1,     1,     1,
-       3,     5,     7,     9,     2,     2,     1,     1,     0,     0,
-       3,     0,     3,     3,     1,     4,     2,     2,     2,     2,
-       3,     2,     3,     0,     1,     1,     0,     1,     1,     0,
-       1,     1,     0,     1,     1,     0,     3,     0,     3,     0,
-       3,     1,     1,     1,     4
+       1,     1,     1,     1,     4,     4,     4,     6,     6,     8,
+       8,     2,     2,    12,     2,     1,     1,     1,     1,     1,
+       1,     1,     1,     1,     1,     1,     1,     6,     2,     4,
+       2,     1,     3,     5,     3,     2,     7,     2,     1,     1,
+       1,     1,     4,     1,     1,     1,     1,     1,     1,     1,
+       3,     0,     2,     2,     1,     1,     1,     1,     1,     1,
+       1,     1,     1,     0,     1,     1,     1,     1,     0,     3,
+       3,     0,     2,     2,     1,     1,     1,     1,     1,     1,
+       1,     1,     4,     2,     2,     1,     2,     1,     2,     1,
+       2,     4,     4,     1,     0,     3,     1,     1,     2,     1,
+       2,     1,     1,     3,     6,     0,     1,     2,     4,     1,
+       3,     1,     1,     1,     1,     1,     1,     1,     1,     1,
+       1,     2,     2,     2,     2,     2,     2,     2,     2,     2,
+       2,     2,     2,     3,     1,     1,     1,     5,     1,     1,
+       1,     2,     1,     1,     2,     1,     2,     6,     1,     3,
+       1,     1,     1,     1,     1,     4,     1,     1,     1,     1,
+       1,     1,     2,     1,     1,     5,     1,     2,     1,     1,
+       5,     2,     0,     6,     3,     0,     1,     1,     1,     1,
+       1,     2,     1,     1,     2,     4,     4,     0,     3,     1,
+       1,     1,     2,     1,     1,     1,     1,     5,     1,     3,
+       5,     5,     1,     3,     5,     1,     1,     1,     1,     1,
+       1,     1,     1,     1,     3,     5,     7,     9,     2,     2,
+       1,     1,     0,     0,     4,     1,     0,     0,     3,     3,
+       1,     5,     2,     2,     2,     2,     3,     2,     3,     0,
+       1,     1,     0,     1,     1,     0,     1,     1,     0,     1,
+       1,     0,     3,     0,     3,     0,     3,     1,     1,     1,
+       4,     1,     1
 };
 
 /* YYDEFACT[STATE-NAME] -- Default rule to reduce with in state
@@ -949,272 +976,296 @@ static const yytype_uint8 yyr2[] =
    means the default is an error.  */
 static const yytype_uint16 yydefact[] =
 {
-       0,     3,     4,     0,     6,     1,     9,     0,     5,     0,
-       0,   231,     0,     0,     0,     0,   229,     2,     0,     0,
-       0,     0,     0,     0,     0,   228,     0,     8,     0,    12,
-      13,    14,    15,    16,    17,    18,    19,    21,    22,    20,
-       0,    82,    83,   107,   108,    84,    85,    86,    87,     7,
-       0,     0,     0,     0,     0,     0,     0,    58,     0,    81,
-      57,     0,     0,     0,     0,     0,    69,     0,     0,   226,
-     227,    30,     0,     0,    10,    11,   234,   232,     0,     0,
-       0,   111,   228,   109,   230,   243,   241,   237,   239,   236,
-     255,   238,   228,    77,    78,    79,    80,    47,   228,   228,
-     228,   228,   228,   228,    71,    48,   219,   218,     0,     0,
-       0,     0,    53,   228,    76,     0,    54,    56,   120,   121,
-     199,   200,   122,   215,   216,     0,     0,   264,    88,   235,
-     112,     0,   113,   117,   118,   119,   213,   214,   217,     0,
-     245,   244,   246,     0,   240,     0,     0,     0,     0,    25,
-       0,    24,    23,   252,   105,   103,   255,    90,     0,     0,
-       0,     0,     0,     0,   249,     0,   249,     0,     0,   259,
-     255,   128,   129,   130,   131,   133,   132,   134,   135,   136,
-     137,     0,   138,   252,    95,     0,    93,    91,   255,     0,
-     100,    89,     0,    74,    73,    75,    46,     0,     0,   233,
-       0,   225,   224,   247,   248,   242,   261,     0,   228,   228,
-       0,     0,   228,   253,   254,   104,   106,     0,     0,     0,
-     198,   169,   170,   168,     0,   151,   251,   250,   150,     0,
-       0,     0,     0,   193,   189,     0,   188,   255,   181,   175,
-     174,   173,     0,     0,     0,     0,    94,     0,    96,     0,
-       0,    92,   228,   220,    62,     0,    60,    61,     0,   228,
-       0,   110,   256,    27,    26,    72,    45,   257,     0,     0,
-     211,     0,   212,     0,   172,     0,   160,     0,   152,     0,
-     157,   158,   141,   142,   159,   139,   140,     0,     0,   187,
-       0,   190,   183,   185,   184,   180,   182,   263,     0,   156,
-     155,   162,   163,     0,     0,   102,     0,    99,     0,     0,
-       0,    55,    70,    64,    44,     0,     0,   228,     0,    31,
-       0,   228,   206,   210,     0,     0,   249,   197,     0,   195,
-       0,   196,     0,   260,   167,   166,   164,   165,   161,   186,
-       0,    97,    98,   101,   228,   221,     0,     0,    63,   228,
-      51,    52,    50,     0,     0,     0,   115,   123,   126,   124,
-     201,   202,   125,   262,     0,    32,    33,    34,    35,    36,
-      37,    38,    39,    40,    41,    42,    43,    29,    28,   171,
-     146,   148,   145,     0,   143,   144,     0,   192,   194,   191,
-     176,     0,    67,    65,    68,    66,     0,     0,     0,   127,
-     178,   228,   114,   258,   149,   147,   153,   154,   228,   222,
-     228,     0,     0,     0,   177,   116,     0,     0,     0,   204,
-       0,   208,     0,   223,   228,   203,     0,   207,     0,     0,
-      49,   205,   209,     0,     0,   179
+       0,     3,     4,     0,     6,     1,     9,     0,     5,   246,
+     281,   282,     0,   247,     0,     0,     0,     2,     0,     0,
+       0,     0,     0,     0,     0,   242,     0,     0,     8,     0,
+      12,    13,    14,    15,    16,    17,    18,    19,    21,    22,
+      23,    20,     0,    96,    97,   121,   122,    98,     0,    99,
+     100,   101,   245,     7,     0,     0,     0,     0,     0,    65,
+       0,    88,    64,     0,     0,     0,     0,     0,    76,     0,
+       0,    94,   240,   241,    31,    32,    83,     0,     0,     0,
+      10,    11,     0,   243,   250,   248,     0,     0,   125,   242,
+     123,   259,   257,   253,   255,   252,   271,   254,   242,    84,
+      85,    86,    87,    91,   242,   242,   242,   242,   242,   242,
+      78,    55,    81,    80,    82,    92,   233,   232,     0,     0,
+       0,     0,    60,     0,   242,    83,     0,    61,    63,   134,
+     135,   213,   214,   136,   229,   230,     0,   242,     0,     0,
+       0,   280,   102,   126,     0,   127,   131,   132,   133,   227,
+     228,   231,     0,   261,   260,   262,     0,   256,     0,     0,
+      54,     0,     0,     0,    26,     0,    25,    24,   268,   119,
+     117,   271,   104,     0,     0,     0,     0,     0,     0,   265,
+       0,   265,     0,     0,   275,   271,   142,   143,   144,   145,
+     147,   146,   148,   149,   150,   151,     0,   152,   268,   109,
+       0,   107,   105,   271,     0,   114,   103,    83,     0,    52,
+       0,     0,     0,     0,   244,   249,     0,   239,   238,   263,
+     264,   258,   277,     0,   242,    95,     0,     0,    83,   242,
+       0,    48,     0,    51,     0,   242,   269,   270,   118,   120,
+       0,     0,     0,   212,   183,   184,   182,     0,   165,   267,
+     266,   164,     0,     0,     0,     0,   207,   203,     0,   202,
+     271,   195,   189,   188,   187,     0,     0,     0,     0,   108,
+       0,   110,     0,     0,   106,     0,   242,   234,    69,     0,
+      67,    68,     0,   242,   242,   251,     0,   124,   272,    28,
+      89,    90,    93,    27,     0,    79,    50,   273,     0,     0,
+     225,     0,   226,     0,   186,     0,   174,     0,   166,     0,
+     171,   172,   155,   156,   173,   153,   154,     0,     0,   201,
+       0,   204,   197,   199,   198,   194,   196,   279,     0,   170,
+     169,   176,   177,     0,     0,   116,     0,   113,     0,     0,
+      53,     0,    62,    77,    71,    47,     0,     0,     0,   242,
+      49,     0,    34,     0,   242,   220,   224,     0,     0,   265,
+     211,     0,   209,     0,   210,     0,   276,   181,   180,   178,
+     179,   175,   200,     0,   111,   112,   115,   242,   235,     0,
+       0,    70,   242,    58,    57,    59,   242,     0,     0,     0,
+     129,   137,   140,   138,   215,   216,   139,   278,     0,    35,
+      36,    37,    38,    39,    40,    41,    42,    43,    44,    45,
+      46,    30,    29,   185,   160,   162,   159,     0,   157,   158,
+       0,   206,   208,   205,   190,     0,    74,    72,    75,    73,
+       0,     0,     0,     0,   141,   192,   242,   128,   274,   163,
+     161,   167,   168,   242,   236,   242,     0,     0,     0,     0,
+     191,   130,     0,     0,     0,     0,   218,     0,   222,     0,
+     237,   242,     0,   217,     0,   221,     0,     0,    56,    33,
+     219,   223,     0,     0,   193
 };
 
 /* YYDEFGOTO[NTERM-NUM].  */
 static const yytype_int16 yydefgoto[] =
 {
-      -1,     3,     4,     6,     8,     9,    27,    28,    29,    30,
-      31,    32,    33,    34,    35,    36,    37,    38,   268,   377,
-      39,   146,    71,    58,    67,   314,   315,   352,   114,    59,
-     115,   255,   256,   257,   348,   393,   395,    68,   313,   105,
-     266,   196,    97,    40,    41,   116,   191,   308,   251,   306,
-     157,    42,    43,    44,   131,    83,   261,   355,   132,   117,
-     356,   357,   118,   171,   285,   172,   384,   405,   173,   228,
-     174,   406,   175,   300,   286,   277,   176,   303,   338,   177,
-     223,   178,   275,   179,   241,   180,   399,   414,   181,   295,
-     296,   340,   238,   289,   330,   332,   328,   182,   119,   359,
-     360,   418,   120,   361,   420,   121,   271,   273,   362,   122,
-     136,   123,   124,   138,    72,    45,    55,    46,    50,    77,
-      47,    60,    91,   142,   205,   229,   215,   144,   319,   243,
-     207,   364,   298,    48
+      -1,     3,     4,     6,     8,     9,    28,    29,    30,    31,
+      32,    33,    34,    35,    36,    37,    38,    39,    40,   298,
+     411,    41,   161,   231,    74,    60,    69,   345,   346,   384,
+     232,    61,   126,   279,   280,   281,   381,   427,   429,    70,
+     344,   111,   296,   115,   103,   160,    75,   227,    76,   228,
+      42,    43,   127,   206,   338,   274,   336,   172,    44,    45,
+      46,   144,    90,   287,   389,   145,   128,   390,   391,   129,
+     186,   315,   187,   418,   440,   188,   251,   189,   441,   190,
+     330,   316,   307,   191,   333,   371,   192,   246,   193,   305,
+     194,   264,   195,   434,   450,   196,   325,   326,   373,   261,
+     319,   363,   365,   361,   197,   130,   393,   394,   455,   131,
+     395,   457,   132,   301,   303,   396,   133,   149,   134,   135,
+     151,    77,    47,   139,    48,    49,    54,    85,    50,    62,
+      97,   155,   221,   252,   238,   157,   352,   266,   223,   398,
+     328,    51,    12
 };
 
 /* YYPACT[STATE-NUM] -- Index in YYTABLE of the portion describing
    STATE-NUM.  */
-#define YYPACT_NINF -369
+#define YYPACT_NINF -401
 static const yytype_int16 yypact[] =
 {
-     143,  -369,  -369,    36,  -369,  -369,    45,   -39,  -369,   169,
-     -33,  -369,   -19,    -6,    -4,    12,  -369,  -369,   -34,   -34,
-     -34,   -34,   -34,   -34,    15,    62,   -34,  -369,    26,  -369,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-      60,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-      20,    56,   107,   110,    37,    20,    -3,  -369,   111,   109,
-    -369,   113,   114,   116,   117,   118,  -369,   119,   125,  -369,
-    -369,  -369,   -15,   121,  -369,  -369,  -369,   122,   132,   -18,
-     167,   210,   -11,  -369,   122,    63,  -369,  -369,  -369,  -369,
-     130,  -369,    62,  -369,  -369,  -369,  -369,  -369,    62,    62,
-      62,    62,    62,    62,  -369,  -369,  -369,  -369,     9,    72,
-      87,    -1,   131,    62,   104,   134,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,  -369,  -369,   -15,   142,  -369,  -369,  -369,
-    -369,   135,  -369,  -369,  -369,  -369,  -369,  -369,  -369,   182,
-    -369,  -369,    52,   219,  -369,   138,   139,   -15,   140,  -369,
-     141,  -369,  -369,    61,  -369,  -369,   130,  -369,   144,   145,
-     146,   180,    11,   147,    85,   148,    99,    89,    -2,   149,
-     130,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-    -369,   184,  -369,    61,  -369,   150,  -369,  -369,   130,   151,
-     152,  -369,    27,  -369,  -369,  -369,  -369,   -10,   154,  -369,
-     153,  -369,  -369,  -369,  -369,  -369,  -369,   155,    62,    62,
-     161,   168,    62,  -369,  -369,  -369,  -369,   229,   244,   246,
-    -369,  -369,  -369,  -369,   247,  -369,  -369,  -369,  -369,   204,
-     247,    17,   163,   164,  -369,   165,  -369,   130,    67,  -369,
-    -369,  -369,   252,   248,    18,   170,  -369,   253,  -369,   255,
-     253,  -369,    62,  -369,  -369,   171,  -369,  -369,   177,    62,
-     172,  -369,  -369,  -369,  -369,  -369,  -369,   173,   175,   176,
-    -369,   178,  -369,   179,  -369,   181,  -369,   183,  -369,   185,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,   262,   264,  -369,
-     267,  -369,  -369,  -369,  -369,  -369,  -369,  -369,   186,  -369,
-    -369,  -369,  -369,   136,   269,  -369,   187,  -369,   188,   190,
-      43,  -369,  -369,   106,  -369,   193,    -5,    -7,   271,  -369,
-     108,    62,  -369,  -369,   245,     4,    99,  -369,   194,  -369,
-     195,  -369,   196,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-     197,  -369,  -369,  -369,    62,  -369,   280,   285,  -369,    62,
-    -369,  -369,  -369,    93,    87,    53,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,  -369,   199,  -369,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,   278,  -369,  -369,     8,  -369,  -369,  -369,
-    -369,    57,  -369,  -369,  -369,  -369,   203,   205,   206,  -369,
-     250,    -7,  -369,  -369,  -369,  -369,  -369,  -369,    62,  -369,
-      62,   229,   244,   208,  -369,  -369,   198,   211,   202,   213,
-     214,   218,   269,  -369,    62,  -369,   229,  -369,   244,    54,
-    -369,  -369,  -369,   269,   215,  -369
+     193,  -401,  -401,    27,  -401,  -401,    62,   143,  -401,    24,
+    -401,  -401,   -30,  -401,   -18,    12,    83,  -401,    15,    15,
+      15,    15,    15,    15,    67,    61,    15,    15,  -401,   127,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,
+    -401,  -401,   144,  -401,  -401,  -401,  -401,  -401,   204,  -401,
+    -401,  -401,  -401,  -401,   155,   136,   138,    34,   140,  -401,
+     147,   108,  -401,   150,   156,   157,   158,   160,  -401,   162,
+     159,  -401,  -401,  -401,  -401,  -401,   102,   -13,   163,   164,
+    -401,  -401,   165,  -401,  -401,   166,   170,    10,   235,     0,
+    -401,   141,  -401,  -401,  -401,  -401,   167,  -401,   131,  -401,
+    -401,  -401,  -401,   168,   131,   131,   131,   131,   131,   131,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,   104,    97,
+     114,    38,   169,    30,   131,   102,   171,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,  -401,  -401,    30,   131,   172,   155,
+     175,  -401,  -401,  -401,   173,  -401,  -401,  -401,  -401,  -401,
+    -401,  -401,   223,  -401,  -401,   123,   253,  -401,   177,   149,
+    -401,   178,   -10,   181,  -401,   182,  -401,  -401,   134,  -401,
+    -401,   167,  -401,   183,   184,   185,   213,    99,   186,   154,
+     187,   146,   153,     7,   188,   167,  -401,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,  -401,  -401,   215,  -401,   134,  -401,
+     190,  -401,  -401,   167,   191,   192,  -401,   102,   -48,  -401,
+       1,   195,   196,   214,   166,  -401,   189,  -401,  -401,  -401,
+    -401,  -401,  -401,   180,   131,  -401,   194,   197,   102,   131,
+      30,  -401,   203,   205,   201,   131,  -401,  -401,  -401,  -401,
+     285,   288,   289,  -401,  -401,  -401,  -401,   291,  -401,  -401,
+    -401,  -401,   248,   291,    33,   206,   207,  -401,   208,  -401,
+     167,    14,  -401,  -401,  -401,   293,   292,    92,   209,  -401,
+     299,  -401,   301,   299,  -401,   216,   131,  -401,  -401,   217,
+    -401,  -401,   221,   131,   131,  -401,   212,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,   218,  -401,  -401,   220,   224,   225,
+    -401,   226,  -401,   227,  -401,   228,  -401,   230,  -401,   231,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,   304,   309,  -401,
+     312,  -401,  -401,  -401,  -401,  -401,  -401,  -401,   232,  -401,
+    -401,  -401,  -401,   161,   313,  -401,   233,  -401,   234,   238,
+    -401,    13,  -401,  -401,   137,  -401,   242,   -15,   243,     3,
+    -401,   314,  -401,   133,   131,  -401,  -401,   296,    94,   146,
+    -401,   245,  -401,   246,  -401,   247,  -401,  -401,  -401,  -401,
+    -401,  -401,  -401,   249,  -401,  -401,  -401,   131,  -401,   332,
+     337,  -401,   131,  -401,  -401,  -401,   131,   142,   114,    28,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,   250,  -401,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,   331,  -401,  -401,
+      68,  -401,  -401,  -401,  -401,    43,  -401,  -401,  -401,  -401,
+     255,   256,   257,   258,  -401,   300,     3,  -401,  -401,  -401,
+    -401,  -401,  -401,   131,  -401,   131,   201,   285,   288,   259,
+    -401,  -401,   252,   264,   265,   263,   261,   266,   270,   313,
+    -401,   131,   133,  -401,   285,  -401,   288,    80,  -401,  -401,
+    -401,  -401,   313,   267,  -401
 };
 
 /* YYPGOTO[NTERM-NUM].  */
 static const yytype_int16 yypgoto[] =
 {
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-    -369,   -94,   -88,   133,  -369,  -369,  -334,  -369,   -85,  -369,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,   128,  -369,  -369,
-    -369,  -369,  -369,  -369,  -369,   251,  -369,  -369,  -369,    77,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-     -72,  -369,   -81,  -369,  -369,  -369,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,  -369,  -305,   101,  -369,  -369,  -369,  -369,
-    -369,  -369,  -369,  -369,  -369,  -369,  -369,  -369,   -22,  -369,
-    -369,  -336,  -369,  -369,  -369,  -369,  -369,  -369,   254,  -369,
-    -369,  -369,  -369,  -369,  -369,  -369,  -342,  -368,   256,  -369,
-    -369,  -369,   -80,  -110,   -82,  -369,  -369,  -369,  -369,   279,
-    -369,   257,  -369,  -369,  -369,  -161,   156,  -146,  -369,  -369,
-    -369,  -369,  -369,  -369
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,   -69,
+     -82,  -401,  -100,   151,   -86,   210,  -401,  -401,  -366,  -401,
+     -54,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,   174,
+    -401,  -401,  -401,  -118,  -401,  -401,   229,  -401,  -401,  -401,
+    -401,  -401,   295,  -401,  -401,  -401,   110,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,   -51,  -401,   -88,
+    -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,  -401,
+    -401,  -311,   139,  -401,  -401,  -401,  -401,  -401,  -401,  -401,
+    -401,  -401,  -401,  -401,  -401,    -2,  -401,  -401,  -400,  -401,
+    -401,  -401,  -401,  -401,  -401,   298,  -401,  -401,  -401,  -401,
+    -401,  -401,  -401,  -390,  -295,   302,  -401,  -401,  -136,   -87,
+    -120,   -89,  -401,  -401,  -401,  -401,  -401,   251,  -401,   176,
+    -401,  -401,  -401,  -176,   198,  -153,  -401,  -401,  -401,  -401,
+    -401,  -401,    -6
 };
 
 /* YYTABLE[YYPACT[STATE-NUM]].  What to do in state STATE-NUM.  If
    positive, shift that token.  If negative, reduce the rule which
    number is the opposite.  If zero, do what YYDEFACT says.
    If YYTABLE_NINF, syntax error.  */
-#define YYTABLE_NINF -60
+#define YYTABLE_NINF -230
 static const yytype_int16 yytable[] =
 {
-     139,   133,   137,   192,   145,   231,   149,   106,   107,   152,
-     216,   148,   254,   150,   151,   396,   147,   350,   147,   108,
-     385,   147,   108,   239,   244,    85,    86,   183,   280,   380,
-      56,   139,   280,    87,   281,   184,     5,   153,   281,   221,
-     198,   280,   248,   381,   421,   154,   109,   281,   185,   282,
-     109,   186,   301,     7,   353,    88,    89,   110,   187,    10,
-     432,   110,   210,   382,    57,   354,   222,   240,   155,   419,
-      90,   302,   188,    49,   284,   383,   417,   111,   284,    51,
-     111,   407,   156,   112,   431,   283,   429,   284,    66,   140,
-     430,   291,    52,   351,    53,   189,   190,   434,   113,    69,
-      70,   141,   113,    69,    70,   158,   113,   292,   293,   225,
-      54,   226,   203,    66,   160,   264,   161,   159,    76,   204,
-     263,   213,   162,   227,   269,   226,   397,   147,   214,   163,
-     164,   165,    74,   166,   252,   167,   232,   227,   398,   233,
-     234,   253,   310,   235,   168,    81,     1,     2,   294,    82,
-     344,   236,    61,    62,    63,    64,    65,   345,   433,    73,
-     401,   169,   170,   390,   408,   386,    75,   402,    78,   237,
-     139,   409,    69,    70,    11,    12,    13,   316,    14,    15,
-      16,    17,    18,    19,    20,    21,    22,    23,    24,    25,
-      26,   365,   366,   367,   368,   369,   370,   371,   372,   373,
-     374,   375,   376,   193,   201,   202,   194,   195,    93,    94,
-      95,    96,   334,   335,   336,   337,   346,   347,    92,    79,
-      98,    99,    80,   100,   101,   102,   103,   104,   125,   126,
-     127,    56,   130,   378,   391,   139,   358,   137,   143,   -59,
-     199,   206,   197,   220,   200,   208,   209,   211,   212,   245,
-     267,   270,   217,   218,   219,   224,   230,   242,   247,   249,
-     250,   259,   139,   265,   262,   260,   272,   316,   274,   276,
-     278,   287,   288,   290,   297,   305,   299,   307,   304,   312,
-     311,   318,   320,   321,   327,   317,   329,   322,   323,   331,
-     324,   339,   325,   363,   326,   333,   341,   342,   416,   343,
-     349,   379,   392,   387,   388,   389,   390,   394,   403,   404,
-     410,   425,   423,   411,   412,   413,   422,   426,   424,   139,
-     358,   137,   428,   427,   435,   258,   139,   309,   316,   415,
-     128,   279,   400,     0,    84,     0,   134,   129,   135,   246,
-       0,     0,   316
+     152,   146,   150,    52,   208,   254,   164,   209,   383,   167,
+     116,   117,   158,   116,   117,   162,   430,   162,   239,   163,
+     162,   165,   166,   125,   278,   118,   233,     5,   118,    13,
+      14,    15,   267,   262,    16,   152,    17,    18,    19,    20,
+      21,    22,    23,    24,    25,    26,    27,   419,   118,   119,
+     271,   212,   119,   116,   117,   322,   323,   456,   310,   467,
+     120,   276,   119,   120,   311,   387,   312,   198,   118,   207,
+       7,   277,   473,   120,   470,   199,   388,   263,    53,   453,
+      58,    55,   211,   121,    10,    11,   121,   122,   200,   275,
+     122,   201,   119,   310,   233,   468,   324,   123,   202,   311,
+     230,    68,   313,   120,   314,   124,   121,   321,   124,   442,
+     292,    56,   203,    72,    73,    59,    72,    73,   124,   310,
+     414,   124,   377,    10,    11,   311,   121,   331,   244,   293,
+     122,   173,   378,   168,   415,   204,   205,   436,   289,   314,
+     162,   169,   175,   174,   176,    88,   332,   437,   124,   299,
+     177,    89,   443,   458,   416,   245,   341,   178,   179,   180,
+      71,   181,   444,   182,   170,   314,   417,    68,   153,    91,
+      92,   471,   183,   249,    72,    73,   432,    93,   171,   248,
+     154,   249,    57,   420,   219,   250,   472,   152,   433,   184,
+     185,   220,   424,   250,   347,   236,     1,     2,   348,    94,
+      95,   255,   237,   112,   256,   257,   113,   114,   258,    99,
+     100,   101,   102,    82,    96,    83,   259,   399,   400,   401,
+     402,   403,   404,   405,   406,   407,   408,   409,   410,    63,
+      64,    65,    66,    67,   260,    80,    78,    79,   367,   368,
+     369,   370,    10,    11,    72,    73,   217,   218,    71,   225,
+     379,   380,    81,    86,    84,    87,    98,   425,   143,   104,
+     152,   392,   150,   110,   138,   105,   106,   107,   412,   108,
+     141,   109,   136,   137,   215,   140,   222,   243,   156,    58,
+     -66,   268,   210,   159,   297,   216,   224,   229,   152,   213,
+     234,   235,   288,   347,   240,   241,   242,   247,   253,   265,
+     431,   270,   272,   273,   283,   284,   286,   295,   300,  -229,
+     290,   302,   304,   291,   306,   308,   327,   317,   318,   320,
+     334,   329,   335,   452,   337,   343,   340,   360,   350,   342,
+     349,   351,   362,   353,   354,   364,   372,   397,   355,   356,
+     357,   385,   358,   359,   366,   374,   375,   152,   392,   150,
+     376,   382,   386,   413,   152,   426,   347,   421,   422,   423,
+     428,   424,   438,   439,   445,   446,   449,   464,   447,   448,
+     459,   460,   347,   461,   462,   463,   466,   454,   465,   474,
+     469,   294,   142,   339,   282,   451,   435,   147,   226,   285,
+     214,   148,   309,     0,     0,     0,   269
 };
 
 static const yytype_int16 yycheck[] =
 {
-      82,    82,    82,   113,    92,   166,   100,    22,    23,   103,
-     156,    99,    22,   101,   102,   349,    98,    22,   100,    37,
-     325,   103,    37,    25,   170,    28,    29,    28,    24,    25,
-      64,   113,    24,    36,    30,    36,     0,    28,    30,    28,
-     125,    24,   188,    39,   412,    36,    61,    30,    49,    32,
-      61,    52,    34,     8,    61,    58,    59,    72,    59,    98,
-     428,    72,   147,    59,    98,    72,    55,    69,    59,   411,
-      73,    53,    73,   106,    70,    71,   410,    95,    70,    98,
-      95,   386,    73,    98,   426,    68,   422,    70,    98,    26,
-     424,   237,    98,    98,    98,    96,    97,   433,   113,   110,
-     111,    38,   113,   110,   111,    33,   113,    40,    41,    24,
-      98,    26,    60,    98,    27,   209,    29,    45,    98,    67,
-     208,    60,    35,    38,   212,    26,    33,   209,    67,    42,
-      43,    44,   106,    46,   107,    48,    47,    38,    45,    50,
-      51,   114,   252,    54,    57,   108,     3,     4,    81,   112,
-     107,    62,    19,    20,    21,    22,    23,   114,   104,    26,
-     107,    74,    75,   109,   107,   326,   106,   114,   112,    80,
-     252,   114,   110,   111,     5,     6,     7,   259,     9,    10,
-      11,    12,    13,    14,    15,    16,    17,    18,    19,    20,
-      21,    83,    84,    85,    86,    87,    88,    89,    90,    91,
-      92,    93,    94,    99,    22,    23,   102,   103,    99,   100,
-     101,   102,    76,    77,    78,    79,   110,   111,   107,   112,
-     107,   107,   112,   107,   107,   107,   107,   102,   107,   107,
-      98,    64,    22,   321,   344,   317,   317,   317,   108,   108,
-      98,    22,   108,    63,   109,   107,   107,   107,   107,    65,
-      82,    22,   108,   108,   108,   108,   108,   108,   108,   108,
-     108,   107,   344,   102,   109,   112,    22,   349,    22,    22,
-      66,   108,   108,   108,    22,    22,    28,    22,   108,   102,
-     109,   108,   107,   107,    22,   113,    22,   109,   109,    22,
-     109,    22,   109,    22,   109,   109,   109,   109,   408,   109,
-     107,    56,    22,   109,   109,   109,   109,    22,   109,    31,
-     107,   109,   114,   108,   108,    65,   108,   104,   107,   401,
-     401,   401,   104,   109,   109,   197,   408,   250,   410,   401,
-      79,   230,   354,    -1,    55,    -1,    82,    80,    82,   183,
-      -1,    -1,   424
+      89,    89,    89,     9,   124,   181,   106,   125,    23,   109,
+      23,    24,    98,    23,    24,   104,   382,   106,   171,   105,
+     109,   107,   108,    77,    23,    38,   162,     0,    38,     5,
+       6,     7,   185,    26,    10,   124,    12,    13,    14,    15,
+      16,    17,    18,    19,    20,    21,    22,   358,    38,    62,
+     203,   137,    62,    23,    24,    41,    42,   447,    25,   459,
+      73,   109,    62,    73,    31,    62,    33,    29,    38,   123,
+       8,   119,   472,    73,   464,    37,    73,    70,   108,   445,
+      65,    99,   136,    96,    99,   100,    96,   100,    50,   207,
+     100,    53,    62,    25,   230,   461,    82,   110,    60,    31,
+     110,   100,    69,    73,    71,   118,    96,   260,   118,   420,
+     228,    99,    74,   113,   114,   100,   113,   114,   118,    25,
+      26,   118,   109,    99,   100,    31,    96,    35,    29,   229,
+     100,    34,   119,    29,    40,    97,    98,   109,   224,    71,
+     229,    37,    28,    46,    30,   111,    54,   119,   118,   235,
+      36,   117,   109,   448,    60,    56,   276,    43,    44,    45,
+      99,    47,   119,    49,    60,    71,    72,   100,    27,    29,
+      30,   466,    58,    27,   113,   114,    34,    37,    74,    25,
+      39,    27,    99,   359,    61,    39,   106,   276,    46,    75,
+      76,    68,   112,    39,   283,    61,     3,     4,   284,    59,
+      60,    48,    68,   101,    51,    52,   104,   105,    55,   101,
+     102,   103,   104,     9,    74,    11,    63,    84,    85,    86,
+      87,    88,    89,    90,    91,    92,    93,    94,    95,    19,
+      20,    21,    22,    23,    81,   108,    26,    27,    77,    78,
+      79,    80,    99,   100,   113,   114,    23,    24,    99,   100,
+     113,   114,   108,   117,    99,   117,   109,   377,    23,   109,
+     349,   349,   349,   104,    99,   109,   109,   109,   354,   109,
+     100,   109,   109,   109,    99,   109,    23,    64,   111,    65,
+     111,    66,   111,   115,    83,   112,   109,   109,   377,   117,
+     109,   109,   112,   382,   111,   111,   111,   111,   111,   111,
+     386,   111,   111,   111,   109,   109,   117,   104,    23,   104,
+     116,    23,    23,   116,    23,    67,    23,   111,   111,   111,
+     111,    29,    23,   443,    23,   104,   110,    23,   110,   112,
+     118,   111,    23,   109,   109,    23,    23,    23,   112,   112,
+     112,   347,   112,   112,   112,   112,   112,   436,   436,   436,
+     112,   109,   109,    57,   443,    23,   445,   112,   112,   112,
+      23,   112,   112,    32,   109,   109,    66,   106,   111,   111,
+     111,   119,   461,   109,   109,   112,   106,   446,   112,   112,
+     462,   230,    87,   273,   210,   436,   388,    89,   159,   213,
+     139,    89,   253,    -1,    -1,    -1,   198
 };
 
 /* YYSTOS[STATE-NUM] -- The (internal number of the) accessing
    symbol of state STATE-NUM.  */
-static const yytype_uint8 yystos[] =
+static const yytype_uint16 yystos[] =
 {
-       0,     3,     4,   116,   117,     0,   118,     8,   119,   120,
-      98,     5,     6,     7,     9,    10,    11,    12,    13,    14,
-      15,    16,    17,    18,    19,    20,    21,   121,   122,   123,
-     124,   125,   126,   127,   128,   129,   130,   131,   132,   135,
-     158,   159,   166,   167,   168,   230,   232,   235,   248,   106,
-     233,    98,    98,    98,    98,   231,    64,    98,   138,   144,
-     236,   138,   138,   138,   138,   138,    98,   139,   152,   110,
-     111,   137,   229,   138,   106,   106,    98,   234,   112,   112,
-     112,   108,   112,   170,   234,    28,    29,    36,    58,    59,
-      73,   237,   107,    99,   100,   101,   102,   157,   107,   107,
-     107,   107,   107,   107,   102,   154,    22,    23,    37,    61,
-      72,    95,    98,   113,   143,   145,   160,   174,   177,   213,
-     217,   220,   224,   226,   227,   107,   107,    98,   160,   236,
-      22,   169,   173,   177,   213,   223,   225,   227,   228,   229,
-      26,    38,   238,   108,   242,   137,   136,   229,   137,   136,
-     137,   137,   136,    28,    36,    59,    73,   165,    33,    45,
-      27,    29,    35,    42,    43,    44,    46,    48,    57,    74,
-      75,   178,   180,   183,   185,   187,   191,   194,   196,   198,
-     200,   203,   212,    28,    36,    49,    52,    59,    73,    96,
-      97,   161,   228,    99,   102,   103,   156,   108,   143,    98,
-     109,    22,    23,    60,    67,   239,    22,   245,   107,   107,
-     143,   107,   107,    60,    67,   241,   242,   108,   108,   108,
-      63,    28,    55,   195,   108,    24,    26,    38,   184,   240,
-     108,   240,    47,    50,    51,    54,    62,    80,   207,    25,
-      69,   199,   108,   244,   242,    65,   241,   108,   242,   108,
-     108,   163,   107,   114,    22,   146,   147,   148,   152,   107,
-     112,   171,   109,   137,   136,   102,   155,    82,   133,   137,
-      22,   221,    22,   222,    22,   197,    22,   190,    66,   190,
-      24,    30,    32,    68,    70,   179,   189,   108,   108,   208,
-     108,   242,    40,    41,    81,   204,   205,    22,   247,    28,
-     188,    34,    53,   192,   108,    22,   164,    22,   162,   164,
-     228,   109,   102,   153,   140,   141,   229,   113,   108,   243,
-     107,   107,   109,   109,   109,   109,   109,    22,   211,    22,
-     209,    22,   210,   109,    76,    77,    78,    79,   193,    22,
-     206,   109,   109,   109,   107,   114,   110,   111,   149,   107,
-      22,    98,   142,    61,    72,   172,   175,   176,   177,   214,
-     215,   218,   223,    22,   246,    83,    84,    85,    86,    87,
-      88,    89,    90,    91,    92,    93,    94,   134,   137,    56,
-      25,    39,    59,    71,   181,   189,   240,   109,   109,   109,
-     109,   228,    22,   150,    22,   151,   141,    33,    45,   201,
-     203,   107,   114,   109,    31,   182,   186,   189,   107,   114,
-     107,   108,   108,    65,   202,   175,   228,   141,   216,   221,
-     219,   222,   108,   114,   107,   109,   104,   109,   104,   206,
-     141,   221,   222,   104,   206,   109
+       0,     3,     4,   121,   122,     0,   123,     8,   124,   125,
+      99,   100,   262,     5,     6,     7,    10,    12,    13,    14,
+      15,    16,    17,    18,    19,    20,    21,    22,   126,   127,
+     128,   129,   130,   131,   132,   133,   134,   135,   136,   137,
+     138,   141,   170,   171,   178,   179,   180,   242,   244,   245,
+     248,   261,   262,   108,   246,    99,    99,    99,    65,   100,
+     145,   151,   249,   145,   145,   145,   145,   145,   100,   146,
+     159,    99,   113,   114,   144,   166,   168,   241,   145,   145,
+     108,   108,     9,    11,    99,   247,   117,   117,   111,   117,
+     182,    29,    30,    37,    59,    60,    74,   250,   109,   101,
+     102,   103,   104,   164,   109,   109,   109,   109,   109,   109,
+     104,   161,   101,   104,   105,   163,    23,    24,    38,    62,
+      73,    96,   100,   110,   118,   150,   152,   172,   186,   189,
+     225,   229,   232,   236,   238,   239,   109,   109,    99,   243,
+     109,   100,   172,    23,   181,   185,   189,   225,   235,   237,
+     239,   240,   241,    27,    39,   251,   111,   255,   144,   115,
+     165,   142,   241,   144,   142,   144,   144,   142,    29,    37,
+      60,    74,   177,    34,    46,    28,    30,    36,    43,    44,
+      45,    47,    49,    58,    75,    76,   190,   192,   195,   197,
+     199,   203,   206,   208,   210,   212,   215,   224,    29,    37,
+      50,    53,    60,    74,    97,    98,   173,   150,   240,   163,
+     111,   150,   144,   117,   247,    99,   112,    23,    24,    61,
+      68,   252,    23,   258,   109,   100,   166,   167,   169,   109,
+     110,   143,   150,   238,   109,   109,    61,    68,   254,   255,
+     111,   111,   111,    64,    29,    56,   207,   111,    25,    27,
+      39,   196,   253,   111,   253,    48,    51,    52,    55,    63,
+      81,   219,    26,    70,   211,   111,   257,   255,    66,   254,
+     111,   255,   111,   111,   175,   163,   109,   119,    23,   153,
+     154,   155,   159,   109,   109,   249,   117,   183,   112,   144,
+     116,   116,   163,   142,   143,   104,   162,    83,   139,   144,
+      23,   233,    23,   234,    23,   209,    23,   202,    67,   202,
+      25,    31,    33,    69,    71,   191,   201,   111,   111,   220,
+     111,   255,    41,    42,    82,   216,   217,    23,   260,    29,
+     200,    35,    54,   204,   111,    23,   176,    23,   174,   176,
+     110,   240,   112,   104,   160,   147,   148,   241,   144,   118,
+     110,   111,   256,   109,   109,   112,   112,   112,   112,   112,
+      23,   223,    23,   221,    23,   222,   112,    77,    78,    79,
+      80,   205,    23,   218,   112,   112,   112,   109,   119,   113,
+     114,   156,   109,    23,   149,   262,   109,    62,    73,   184,
+     187,   188,   189,   226,   227,   230,   235,    23,   259,    84,
+      85,    86,    87,    88,    89,    90,    91,    92,    93,    94,
+      95,   140,   144,    57,    26,    40,    60,    72,   193,   201,
+     253,   112,   112,   112,   112,   240,    23,   157,    23,   158,
+     148,   144,    34,    46,   213,   215,   109,   119,   112,    32,
+     194,   198,   201,   109,   119,   109,   109,   111,   111,    66,
+     214,   187,   240,   148,   139,   228,   233,   231,   234,   111,
+     119,   109,   109,   112,   106,   112,   106,   218,   148,   140,
+     233,   234,   106,   218,   112
 };
 
 #define yyerrok		(yyerrstatus = 0)
@@ -2068,7 +2119,7 @@ yyreduce:
         case 3:
 
 /* Line 1455 of yacc.c  */
-#line 260 "program_parse.y"
+#line 272 "program_parse.y"
     {
 	   if (state->prog->Target != GL_VERTEX_PROGRAM_ARB) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid fragment program header");
@@ -2081,7 +2132,7 @@ yyreduce:
   case 4:
 
 /* Line 1455 of yacc.c  */
-#line 268 "program_parse.y"
+#line 280 "program_parse.y"
     {
 	   if (state->prog->Target != GL_FRAGMENT_PROGRAM_ARB) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex program header");
@@ -2096,7 +2147,7 @@ yyreduce:
   case 7:
 
 /* Line 1455 of yacc.c  */
-#line 284 "program_parse.y"
+#line 296 "program_parse.y"
     {
 	   int valid = 0;
 
@@ -2121,7 +2172,7 @@ yyreduce:
   case 10:
 
 /* Line 1455 of yacc.c  */
-#line 310 "program_parse.y"
+#line 322 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (2)].inst) != NULL) {
 	      if (state->inst_tail == NULL) {
@@ -2141,7 +2192,7 @@ yyreduce:
   case 12:
 
 /* Line 1455 of yacc.c  */
-#line 328 "program_parse.y"
+#line 340 "program_parse.y"
     {
 	   (yyval.inst) = (yyvsp[(1) - (1)].inst);
 	   state->prog->NumAluInstructions++;
@@ -2151,85 +2202,79 @@ yyreduce:
   case 13:
 
 /* Line 1455 of yacc.c  */
-#line 333 "program_parse.y"
+#line 345 "program_parse.y"
     {
 	   (yyval.inst) = (yyvsp[(1) - (1)].inst);
 	   state->prog->NumTexInstructions++;
 	;}
     break;
 
-  case 23:
+  case 24:
 
 /* Line 1455 of yacc.c  */
-#line 353 "program_parse.y"
+#line 366 "program_parse.y"
     {
 	   (yyval.inst) = asm_instruction_ctor(OPCODE_ARL, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL);
 	;}
     break;
 
-  case 24:
+  case 25:
 
 /* Line 1455 of yacc.c  */
-#line 359 "program_parse.y"
+#line 372 "program_parse.y"
     {
-	   (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (4)].temp_inst).Opcode, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL);
-	   (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (4)].temp_inst).SaturateMode;
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (4)].temp_inst), & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL);
 	;}
     break;
 
-  case 25:
+  case 26:
 
 /* Line 1455 of yacc.c  */
-#line 366 "program_parse.y"
+#line 378 "program_parse.y"
     {
-	   (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (4)].temp_inst).Opcode, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL);
-	   (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (4)].temp_inst).SaturateMode;
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (4)].temp_inst), & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL);
 	;}
     break;
 
-  case 26:
+  case 27:
 
 /* Line 1455 of yacc.c  */
-#line 373 "program_parse.y"
+#line 384 "program_parse.y"
     {
-	   (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (6)].temp_inst).Opcode, & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL);
-	   (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (6)].temp_inst).SaturateMode;
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL);
 	;}
     break;
 
-  case 27:
+  case 28:
 
 /* Line 1455 of yacc.c  */
-#line 381 "program_parse.y"
+#line 391 "program_parse.y"
     {
-	   (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (6)].temp_inst).Opcode, & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL);
-	   (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (6)].temp_inst).SaturateMode;
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL);
 	;}
     break;
 
-  case 28:
+  case 29:
 
 /* Line 1455 of yacc.c  */
-#line 389 "program_parse.y"
+#line 398 "program_parse.y"
     {
-	   (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (8)].temp_inst).Opcode, & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), & (yyvsp[(6) - (8)].src_reg), & (yyvsp[(8) - (8)].src_reg));
-	   (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (8)].temp_inst).SaturateMode;
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (8)].temp_inst), & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), & (yyvsp[(6) - (8)].src_reg), & (yyvsp[(8) - (8)].src_reg));
 	;}
     break;
 
-  case 29:
+  case 30:
 
 /* Line 1455 of yacc.c  */
-#line 396 "program_parse.y"
+#line 404 "program_parse.y"
     {
-	   (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (8)].temp_inst).Opcode, & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), NULL, NULL);
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (8)].temp_inst), & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), NULL, NULL);
 	   if ((yyval.inst) != NULL) {
 	      const GLbitfield tex_mask = (1U << (yyvsp[(6) - (8)].integer));
 	      GLbitfield shadow_tex = 0;
 	      GLbitfield target_mask = 0;
 
 
-	      (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (8)].temp_inst).SaturateMode;
 	      (yyval.inst)->Base.TexSrcUnit = (yyvsp[(6) - (8)].integer);
 
 	      if ((yyvsp[(8) - (8)].integer) < 0) {
@@ -2265,113 +2310,173 @@ yyreduce:
 	;}
     break;
 
-  case 30:
+  case 31:
 
 /* Line 1455 of yacc.c  */
-#line 441 "program_parse.y"
+#line 448 "program_parse.y"
     {
 	   (yyval.inst) = asm_instruction_ctor(OPCODE_KIL, NULL, & (yyvsp[(2) - (2)].src_reg), NULL, NULL);
 	   state->fragment.UsesKill = 1;
 	;}
     break;
 
-  case 31:
+  case 32:
 
 /* Line 1455 of yacc.c  */
-#line 448 "program_parse.y"
+#line 453 "program_parse.y"
+    {
+	   (yyval.inst) = asm_instruction_ctor(OPCODE_KIL_NV, NULL, NULL, NULL, NULL);
+	   (yyval.inst)->Base.DstReg.CondMask = (yyvsp[(2) - (2)].dst_reg).CondMask;
+	   (yyval.inst)->Base.DstReg.CondSwizzle = (yyvsp[(2) - (2)].dst_reg).CondSwizzle;
+	   (yyval.inst)->Base.DstReg.CondSrc = (yyvsp[(2) - (2)].dst_reg).CondSrc;
+	   state->fragment.UsesKill = 1;
+	;}
+    break;
+
+  case 33:
+
+/* Line 1455 of yacc.c  */
+#line 463 "program_parse.y"
+    {
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (12)].temp_inst), & (yyvsp[(2) - (12)].dst_reg), & (yyvsp[(4) - (12)].src_reg), & (yyvsp[(6) - (12)].src_reg), & (yyvsp[(8) - (12)].src_reg));
+	   if ((yyval.inst) != NULL) {
+	      const GLbitfield tex_mask = (1U << (yyvsp[(10) - (12)].integer));
+	      GLbitfield shadow_tex = 0;
+	      GLbitfield target_mask = 0;
+
+
+	      (yyval.inst)->Base.TexSrcUnit = (yyvsp[(10) - (12)].integer);
+
+	      if ((yyvsp[(12) - (12)].integer) < 0) {
+		 shadow_tex = tex_mask;
+
+		 (yyval.inst)->Base.TexSrcTarget = -(yyvsp[(12) - (12)].integer);
+		 (yyval.inst)->Base.TexShadow = 1;
+	      } else {
+		 (yyval.inst)->Base.TexSrcTarget = (yyvsp[(12) - (12)].integer);
+	      }
+
+	      target_mask = (1U << (yyval.inst)->Base.TexSrcTarget);
+
+	      /* If this texture unit was previously accessed and that access
+	       * had a different texture target, generate an error.
+	       *
+	       * If this texture unit was previously accessed and that access
+	       * had a different shadow mode, generate an error.
+	       */
+	      if ((state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] != 0)
+		  && ((state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] != target_mask)
+		      || ((state->prog->ShadowSamplers & tex_mask)
+			  != shadow_tex))) {
+		 yyerror(& (yylsp[(12) - (12)]), state,
+			 "multiple targets used on one texture image unit");
+		 YYERROR;
+	      }
+
+
+	      state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] |= target_mask;
+	      state->prog->ShadowSamplers |= shadow_tex;
+	   }
+	;}
+    break;
+
+  case 34:
+
+/* Line 1455 of yacc.c  */
+#line 507 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 32:
+  case 35:
 
 /* Line 1455 of yacc.c  */
-#line 453 "program_parse.y"
+#line 512 "program_parse.y"
     { (yyval.integer) = TEXTURE_1D_INDEX; ;}
     break;
 
-  case 33:
+  case 36:
 
 /* Line 1455 of yacc.c  */
-#line 454 "program_parse.y"
+#line 513 "program_parse.y"
     { (yyval.integer) = TEXTURE_2D_INDEX; ;}
     break;
 
-  case 34:
+  case 37:
 
 /* Line 1455 of yacc.c  */
-#line 455 "program_parse.y"
+#line 514 "program_parse.y"
     { (yyval.integer) = TEXTURE_3D_INDEX; ;}
     break;
 
-  case 35:
+  case 38:
 
 /* Line 1455 of yacc.c  */
-#line 456 "program_parse.y"
+#line 515 "program_parse.y"
     { (yyval.integer) = TEXTURE_CUBE_INDEX; ;}
     break;
 
-  case 36:
+  case 39:
 
 /* Line 1455 of yacc.c  */
-#line 457 "program_parse.y"
+#line 516 "program_parse.y"
     { (yyval.integer) = TEXTURE_RECT_INDEX; ;}
     break;
 
-  case 37:
+  case 40:
 
 /* Line 1455 of yacc.c  */
-#line 458 "program_parse.y"
+#line 517 "program_parse.y"
     { (yyval.integer) = -TEXTURE_1D_INDEX; ;}
     break;
 
-  case 38:
+  case 41:
 
 /* Line 1455 of yacc.c  */
-#line 459 "program_parse.y"
+#line 518 "program_parse.y"
     { (yyval.integer) = -TEXTURE_2D_INDEX; ;}
     break;
 
-  case 39:
+  case 42:
 
 /* Line 1455 of yacc.c  */
-#line 460 "program_parse.y"
+#line 519 "program_parse.y"
     { (yyval.integer) = -TEXTURE_RECT_INDEX; ;}
     break;
 
-  case 40:
+  case 43:
 
 /* Line 1455 of yacc.c  */
-#line 461 "program_parse.y"
+#line 520 "program_parse.y"
     { (yyval.integer) = TEXTURE_1D_ARRAY_INDEX; ;}
     break;
 
-  case 41:
+  case 44:
 
 /* Line 1455 of yacc.c  */
-#line 462 "program_parse.y"
+#line 521 "program_parse.y"
     { (yyval.integer) = TEXTURE_2D_ARRAY_INDEX; ;}
     break;
 
-  case 42:
+  case 45:
 
 /* Line 1455 of yacc.c  */
-#line 463 "program_parse.y"
+#line 522 "program_parse.y"
     { (yyval.integer) = -TEXTURE_1D_ARRAY_INDEX; ;}
     break;
 
-  case 43:
+  case 46:
 
 /* Line 1455 of yacc.c  */
-#line 464 "program_parse.y"
+#line 523 "program_parse.y"
     { (yyval.integer) = -TEXTURE_2D_ARRAY_INDEX; ;}
     break;
 
-  case 44:
+  case 47:
 
 /* Line 1455 of yacc.c  */
-#line 468 "program_parse.y"
+#line 527 "program_parse.y"
     {
 	   /* FIXME: Is this correct?  Should the extenedSwizzle be applied
 	    * FIXME: to the existing swizzle?
@@ -2379,31 +2484,81 @@ yyreduce:
 	   (yyvsp[(4) - (6)].src_reg).Base.Swizzle = (yyvsp[(6) - (6)].swiz_mask).swizzle;
 	   (yyvsp[(4) - (6)].src_reg).Base.Negate = (yyvsp[(6) - (6)].swiz_mask).mask;
 
-	   (yyval.inst) = asm_instruction_ctor(OPCODE_SWZ, & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), NULL, NULL);
-	   (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (6)].temp_inst).SaturateMode;
+	   (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), NULL, NULL);
 	;}
     break;
 
-  case 45:
+  case 48:
 
 /* Line 1455 of yacc.c  */
-#line 481 "program_parse.y"
+#line 539 "program_parse.y"
     {
-	   (yyval.src_reg) = (yyvsp[(2) - (3)].src_reg);
+	   (yyval.src_reg) = (yyvsp[(2) - (2)].src_reg);
 
-	   if ((yyvsp[(1) - (3)].negate)) {
+	   if ((yyvsp[(1) - (2)].negate)) {
+	      (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate;
+	   }
+	;}
+    break;
+
+  case 49:
+
+/* Line 1455 of yacc.c  */
+#line 547 "program_parse.y"
+    {
+	   (yyval.src_reg) = (yyvsp[(3) - (4)].src_reg);
+
+	   if (!state->option.NV_fragment) {
+	      yyerror(& (yylsp[(2) - (4)]), state, "unexpected character '|'");
+	      YYERROR;
+	   }
+
+	   if ((yyvsp[(1) - (4)].negate)) {
 	      (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate;
 	   }
 
+	   (yyval.src_reg).Base.Abs = 1;
+	;}
+    break;
+
+  case 50:
+
+/* Line 1455 of yacc.c  */
+#line 564 "program_parse.y"
+    {
+	   (yyval.src_reg) = (yyvsp[(1) - (2)].src_reg);
+
 	   (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle,
-						    (yyvsp[(3) - (3)].swiz_mask).swizzle);
+						    (yyvsp[(2) - (2)].swiz_mask).swizzle);
 	;}
     break;
 
-  case 46:
+  case 51:
+
+/* Line 1455 of yacc.c  */
+#line 571 "program_parse.y"
+    {
+	   struct asm_symbol temp_sym;
+
+	   if (!state->option.NV_fragment) {
+	      yyerror(& (yylsp[(1) - (1)]), state, "expected scalar suffix");
+	      YYERROR;
+	   }
+
+	   memset(& temp_sym, 0, sizeof(temp_sym));
+	   temp_sym.param_binding_begin = ~0;
+	   initialize_symbol_from_const(state->prog, & temp_sym, & (yyvsp[(1) - (1)].vector));
+
+	   init_src_reg(& (yyval.src_reg));
+	   (yyval.src_reg).Base.File = PROGRAM_CONSTANT;
+	   (yyval.src_reg).Base.Index = temp_sym.param_binding_begin;
+	;}
+    break;
+
+  case 52:
 
 /* Line 1455 of yacc.c  */
-#line 494 "program_parse.y"
+#line 590 "program_parse.y"
     {
 	   (yyval.src_reg) = (yyvsp[(2) - (3)].src_reg);
 
@@ -2416,13 +2571,38 @@ yyreduce:
 	;}
     break;
 
-  case 47:
+  case 53:
 
 /* Line 1455 of yacc.c  */
-#line 507 "program_parse.y"
+#line 601 "program_parse.y"
     {
-	   (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg);
-	   (yyval.dst_reg).WriteMask = (yyvsp[(2) - (2)].swiz_mask).mask;
+	   (yyval.src_reg) = (yyvsp[(3) - (5)].src_reg);
+
+	   if (!state->option.NV_fragment) {
+	      yyerror(& (yylsp[(2) - (5)]), state, "unexpected character '|'");
+	      YYERROR;
+	   }
+
+	   if ((yyvsp[(1) - (5)].negate)) {
+	      (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate;
+	   }
+
+	   (yyval.src_reg).Base.Abs = 1;
+	   (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle,
+						    (yyvsp[(4) - (5)].swiz_mask).swizzle);
+	;}
+    break;
+
+  case 54:
+
+/* Line 1455 of yacc.c  */
+#line 621 "program_parse.y"
+    {
+	   (yyval.dst_reg) = (yyvsp[(1) - (3)].dst_reg);
+	   (yyval.dst_reg).WriteMask = (yyvsp[(2) - (3)].swiz_mask).mask;
+	   (yyval.dst_reg).CondMask = (yyvsp[(3) - (3)].dst_reg).CondMask;
+	   (yyval.dst_reg).CondSwizzle = (yyvsp[(3) - (3)].dst_reg).CondSwizzle;
+	   (yyval.dst_reg).CondSrc = (yyvsp[(3) - (3)].dst_reg).CondSrc;
 
 	   if ((yyval.dst_reg).File == PROGRAM_OUTPUT) {
 	      /* Technically speaking, this should check that it is in
@@ -2431,7 +2611,7 @@ yyreduce:
 	       */
 	      if (state->option.PositionInvariant
 	       && ((yyval.dst_reg).Index == VERT_RESULT_HPOS)) {
-		 yyerror(& (yylsp[(1) - (2)]), state, "position-invariant programs cannot "
+		 yyerror(& (yylsp[(1) - (3)]), state, "position-invariant programs cannot "
 			 "write position");
 		 YYERROR;
 	      }
@@ -2441,10 +2621,10 @@ yyreduce:
 	;}
     break;
 
-  case 48:
+  case 55:
 
 /* Line 1455 of yacc.c  */
-#line 529 "program_parse.y"
+#line 646 "program_parse.y"
     {
 	   init_dst_reg(& (yyval.dst_reg));
 	   (yyval.dst_reg).File = PROGRAM_ADDRESS;
@@ -2453,10 +2633,10 @@ yyreduce:
 	;}
     break;
 
-  case 49:
+  case 56:
 
 /* Line 1455 of yacc.c  */
-#line 538 "program_parse.y"
+#line 655 "program_parse.y"
     {
 	   const unsigned xyzw_valid =
 	      ((yyvsp[(1) - (7)].ext_swizzle).xyzw_valid << 0)
@@ -2489,20 +2669,20 @@ yyreduce:
 	;}
     break;
 
-  case 50:
+  case 57:
 
 /* Line 1455 of yacc.c  */
-#line 571 "program_parse.y"
+#line 688 "program_parse.y"
     {
 	   (yyval.ext_swizzle) = (yyvsp[(2) - (2)].ext_swizzle);
 	   (yyval.ext_swizzle).negate = ((yyvsp[(1) - (2)].negate)) ? 1 : 0;
 	;}
     break;
 
-  case 51:
+  case 58:
 
 /* Line 1455 of yacc.c  */
-#line 578 "program_parse.y"
+#line 695 "program_parse.y"
     {
 	   if (((yyvsp[(1) - (1)].integer) != 0) && ((yyvsp[(1) - (1)].integer) != 1)) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector");
@@ -2519,10 +2699,10 @@ yyreduce:
 	;}
     break;
 
-  case 52:
+  case 59:
 
 /* Line 1455 of yacc.c  */
-#line 593 "program_parse.y"
+#line 710 "program_parse.y"
     {
 	   if (strlen((yyvsp[(1) - (1)].string)) > 1) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector");
@@ -2572,10 +2752,10 @@ yyreduce:
 	;}
     break;
 
-  case 53:
+  case 60:
 
 /* Line 1455 of yacc.c  */
-#line 643 "program_parse.y"
+#line 760 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -2619,10 +2799,10 @@ yyreduce:
 	;}
     break;
 
-  case 54:
+  case 61:
 
 /* Line 1455 of yacc.c  */
-#line 685 "program_parse.y"
+#line 802 "program_parse.y"
     {
 	   init_src_reg(& (yyval.src_reg));
 	   (yyval.src_reg).Base.File = PROGRAM_INPUT;
@@ -2635,10 +2815,10 @@ yyreduce:
 	;}
     break;
 
-  case 55:
+  case 62:
 
 /* Line 1455 of yacc.c  */
-#line 696 "program_parse.y"
+#line 813 "program_parse.y"
     {
 	   if (! (yyvsp[(3) - (4)].src_reg).Base.RelAddr
 	       && ((unsigned) (yyvsp[(3) - (4)].src_reg).Base.Index >= (yyvsp[(1) - (4)].sym)->param_binding_length)) {
@@ -2661,10 +2841,10 @@ yyreduce:
 	;}
     break;
 
-  case 56:
+  case 63:
 
 /* Line 1455 of yacc.c  */
-#line 717 "program_parse.y"
+#line 834 "program_parse.y"
     {
 	   init_src_reg(& (yyval.src_reg));
 	   (yyval.src_reg).Base.File = ((yyvsp[(1) - (1)].temp_sym).name != NULL) 
@@ -2674,10 +2854,10 @@ yyreduce:
 	;}
     break;
 
-  case 57:
+  case 64:
 
 /* Line 1455 of yacc.c  */
-#line 727 "program_parse.y"
+#line 844 "program_parse.y"
     {
 	   init_dst_reg(& (yyval.dst_reg));
 	   (yyval.dst_reg).File = PROGRAM_OUTPUT;
@@ -2685,10 +2865,10 @@ yyreduce:
 	;}
     break;
 
-  case 58:
+  case 65:
 
 /* Line 1455 of yacc.c  */
-#line 733 "program_parse.y"
+#line 850 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -2719,10 +2899,10 @@ yyreduce:
 	;}
     break;
 
-  case 59:
+  case 66:
 
 /* Line 1455 of yacc.c  */
-#line 764 "program_parse.y"
+#line 881 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -2739,20 +2919,20 @@ yyreduce:
 	;}
     break;
 
-  case 62:
+  case 69:
 
 /* Line 1455 of yacc.c  */
-#line 783 "program_parse.y"
+#line 900 "program_parse.y"
     {
 	   init_src_reg(& (yyval.src_reg));
 	   (yyval.src_reg).Base.Index = (yyvsp[(1) - (1)].integer);
 	;}
     break;
 
-  case 63:
+  case 70:
 
 /* Line 1455 of yacc.c  */
-#line 790 "program_parse.y"
+#line 907 "program_parse.y"
     {
 	   /* FINISHME: Add support for multiple address registers.
 	    */
@@ -2764,35 +2944,37 @@ yyreduce:
 	;}
     break;
 
-  case 64:
+  case 71:
 
 /* Line 1455 of yacc.c  */
-#line 801 "program_parse.y"
+#line 918 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 65:
+  case 72:
 
 /* Line 1455 of yacc.c  */
-#line 802 "program_parse.y"
+#line 919 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;}
     break;
 
-  case 66:
+  case 73:
 
 /* Line 1455 of yacc.c  */
-#line 803 "program_parse.y"
+#line 920 "program_parse.y"
     { (yyval.integer) = -(yyvsp[(2) - (2)].integer); ;}
     break;
 
-  case 67:
+  case 74:
 
 /* Line 1455 of yacc.c  */
-#line 807 "program_parse.y"
+#line 924 "program_parse.y"
     {
 	   if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 63)) {
-	      yyerror(& (yylsp[(1) - (1)]), state,
-		      "relative address offset too large (positive)");
+              char s[100];
+              _mesa_snprintf(s, sizeof(s),
+                             "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer));
+	      yyerror(& (yylsp[(1) - (1)]), state, s);
 	      YYERROR;
 	   } else {
 	      (yyval.integer) = (yyvsp[(1) - (1)].integer);
@@ -2800,14 +2982,16 @@ yyreduce:
 	;}
     break;
 
-  case 68:
+  case 75:
 
 /* Line 1455 of yacc.c  */
-#line 819 "program_parse.y"
+#line 938 "program_parse.y"
     {
 	   if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 64)) {
-	      yyerror(& (yylsp[(1) - (1)]), state,
-		      "relative address offset too large (negative)");
+              char s[100];
+              _mesa_snprintf(s, sizeof(s),
+                             "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer));
+	      yyerror(& (yylsp[(1) - (1)]), state, s);
 	      YYERROR;
 	   } else {
 	      (yyval.integer) = (yyvsp[(1) - (1)].integer);
@@ -2815,10 +2999,10 @@ yyreduce:
 	;}
     break;
 
-  case 69:
+  case 76:
 
 /* Line 1455 of yacc.c  */
-#line 831 "program_parse.y"
+#line 952 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -2836,10 +3020,10 @@ yyreduce:
 	;}
     break;
 
-  case 70:
+  case 77:
 
 /* Line 1455 of yacc.c  */
-#line 849 "program_parse.y"
+#line 970 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid address component selector");
@@ -2850,10 +3034,10 @@ yyreduce:
 	;}
     break;
 
-  case 71:
+  case 78:
 
 /* Line 1455 of yacc.c  */
-#line 860 "program_parse.y"
+#line 981 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) {
 	      yyerror(& (yylsp[(1) - (1)]), state,
@@ -2865,24 +3049,125 @@ yyreduce:
 	;}
     break;
 
-  case 76:
+  case 83:
 
 /* Line 1455 of yacc.c  */
-#line 876 "program_parse.y"
+#line 997 "program_parse.y"
     { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;}
     break;
 
-  case 81:
+  case 88:
 
 /* Line 1455 of yacc.c  */
-#line 880 "program_parse.y"
+#line 1001 "program_parse.y"
     { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;}
     break;
 
-  case 88:
+  case 89:
+
+/* Line 1455 of yacc.c  */
+#line 1005 "program_parse.y"
+    {
+	   (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg);
+	;}
+    break;
+
+  case 90:
+
+/* Line 1455 of yacc.c  */
+#line 1009 "program_parse.y"
+    {
+	   (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg);
+	;}
+    break;
+
+  case 91:
+
+/* Line 1455 of yacc.c  */
+#line 1013 "program_parse.y"
+    {
+	   (yyval.dst_reg).CondMask = COND_TR;
+	   (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP;
+	   (yyval.dst_reg).CondSrc = 0;
+	;}
+    break;
+
+  case 92:
+
+/* Line 1455 of yacc.c  */
+#line 1021 "program_parse.y"
+    {
+	   (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg);
+	   (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle;
+	;}
+    break;
+
+  case 93:
+
+/* Line 1455 of yacc.c  */
+#line 1028 "program_parse.y"
+    {
+	   (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg);
+	   (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle;
+	;}
+    break;
+
+  case 94:
 
 /* Line 1455 of yacc.c  */
-#line 892 "program_parse.y"
+#line 1035 "program_parse.y"
+    {
+	   const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string));
+	   if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) {
+	      char *const err_str =
+		 make_error_string("invalid condition code \"%s\"", (yyvsp[(1) - (1)].string));
+
+	      yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL)
+		      ? err_str : "invalid condition code");
+
+	      if (err_str != NULL) {
+		 _mesa_free(err_str);
+	      }
+
+	      YYERROR;
+	   }
+
+	   (yyval.dst_reg).CondMask = cond;
+	   (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP;
+	   (yyval.dst_reg).CondSrc = 0;
+	;}
+    break;
+
+  case 95:
+
+/* Line 1455 of yacc.c  */
+#line 1058 "program_parse.y"
+    {
+	   const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string));
+	   if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) {
+	      char *const err_str =
+		 make_error_string("invalid condition code \"%s\"", (yyvsp[(1) - (1)].string));
+
+	      yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL)
+		      ? err_str : "invalid condition code");
+
+	      if (err_str != NULL) {
+		 _mesa_free(err_str);
+	      }
+
+	      YYERROR;
+	   }
+
+	   (yyval.dst_reg).CondMask = cond;
+	   (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP;
+	   (yyval.dst_reg).CondSrc = 0;
+	;}
+    break;
+
+  case 102:
+
+/* Line 1455 of yacc.c  */
+#line 1089 "program_parse.y"
     {
 	   struct asm_symbol *const s =
 	      declare_variable(state, (yyvsp[(2) - (4)].string), at_attrib, & (yylsp[(2) - (4)]));
@@ -2900,55 +3185,55 @@ yyreduce:
 	;}
     break;
 
-  case 89:
+  case 103:
 
 /* Line 1455 of yacc.c  */
-#line 910 "program_parse.y"
+#line 1107 "program_parse.y"
     {
 	   (yyval.attrib) = (yyvsp[(2) - (2)].attrib);
 	;}
     break;
 
-  case 90:
+  case 104:
 
 /* Line 1455 of yacc.c  */
-#line 914 "program_parse.y"
+#line 1111 "program_parse.y"
     {
 	   (yyval.attrib) = (yyvsp[(2) - (2)].attrib);
 	;}
     break;
 
-  case 91:
+  case 105:
 
 /* Line 1455 of yacc.c  */
-#line 920 "program_parse.y"
+#line 1117 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_POS;
 	;}
     break;
 
-  case 92:
+  case 106:
 
 /* Line 1455 of yacc.c  */
-#line 924 "program_parse.y"
+#line 1121 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_WEIGHT;
 	;}
     break;
 
-  case 93:
+  case 107:
 
 /* Line 1455 of yacc.c  */
-#line 928 "program_parse.y"
+#line 1125 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_NORMAL;
 	;}
     break;
 
-  case 94:
+  case 108:
 
 /* Line 1455 of yacc.c  */
-#line 932 "program_parse.y"
+#line 1129 "program_parse.y"
     {
 	   if (!state->ctx->Extensions.EXT_secondary_color) {
 	      yyerror(& (yylsp[(2) - (2)]), state, "GL_EXT_secondary_color not supported");
@@ -2959,10 +3244,10 @@ yyreduce:
 	;}
     break;
 
-  case 95:
+  case 109:
 
 /* Line 1455 of yacc.c  */
-#line 941 "program_parse.y"
+#line 1138 "program_parse.y"
     {
 	   if (!state->ctx->Extensions.EXT_fog_coord) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "GL_EXT_fog_coord not supported");
@@ -2973,38 +3258,38 @@ yyreduce:
 	;}
     break;
 
-  case 96:
+  case 110:
 
 /* Line 1455 of yacc.c  */
-#line 950 "program_parse.y"
+#line 1147 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 97:
+  case 111:
 
 /* Line 1455 of yacc.c  */
-#line 954 "program_parse.y"
+#line 1151 "program_parse.y"
     {
 	   yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported");
 	   YYERROR;
 	;}
     break;
 
-  case 98:
+  case 112:
 
 /* Line 1455 of yacc.c  */
-#line 959 "program_parse.y"
+#line 1156 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_GENERIC0 + (yyvsp[(3) - (4)].integer);
 	;}
     break;
 
-  case 99:
+  case 113:
 
 /* Line 1455 of yacc.c  */
-#line 965 "program_parse.y"
+#line 1162 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxAttribs) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex attribute reference");
@@ -3015,46 +3300,46 @@ yyreduce:
 	;}
     break;
 
-  case 103:
+  case 117:
 
 /* Line 1455 of yacc.c  */
-#line 979 "program_parse.y"
+#line 1176 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_WPOS;
 	;}
     break;
 
-  case 104:
+  case 118:
 
 /* Line 1455 of yacc.c  */
-#line 983 "program_parse.y"
+#line 1180 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_COL0 + (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 105:
+  case 119:
 
 /* Line 1455 of yacc.c  */
-#line 987 "program_parse.y"
+#line 1184 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_FOGC;
 	;}
     break;
 
-  case 106:
+  case 120:
 
 /* Line 1455 of yacc.c  */
-#line 991 "program_parse.y"
+#line 1188 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 109:
+  case 123:
 
 /* Line 1455 of yacc.c  */
-#line 999 "program_parse.y"
+#line 1196 "program_parse.y"
     {
 	   struct asm_symbol *const s =
 	      declare_variable(state, (yyvsp[(2) - (3)].string), at_param, & (yylsp[(2) - (3)]));
@@ -3070,10 +3355,10 @@ yyreduce:
 	;}
     break;
 
-  case 110:
+  case 124:
 
 /* Line 1455 of yacc.c  */
-#line 1015 "program_parse.y"
+#line 1212 "program_parse.y"
     {
 	   if (((yyvsp[(4) - (6)].integer) != 0) && ((unsigned) (yyvsp[(4) - (6)].integer) != (yyvsp[(6) - (6)].temp_sym).param_binding_length)) {
 	      yyerror(& (yylsp[(4) - (6)]), state, 
@@ -3095,19 +3380,19 @@ yyreduce:
 	;}
     break;
 
-  case 111:
+  case 125:
 
 /* Line 1455 of yacc.c  */
-#line 1037 "program_parse.y"
+#line 1234 "program_parse.y"
     {
 	   (yyval.integer) = 0;
 	;}
     break;
 
-  case 112:
+  case 126:
 
 /* Line 1455 of yacc.c  */
-#line 1041 "program_parse.y"
+#line 1238 "program_parse.y"
     {
 	   if (((yyvsp[(1) - (1)].integer) < 1) || ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxParameters)) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid parameter array size");
@@ -3118,38 +3403,38 @@ yyreduce:
 	;}
     break;
 
-  case 113:
+  case 127:
 
 /* Line 1455 of yacc.c  */
-#line 1052 "program_parse.y"
+#line 1249 "program_parse.y"
     {
 	   (yyval.temp_sym) = (yyvsp[(2) - (2)].temp_sym);
 	;}
     break;
 
-  case 114:
+  case 128:
 
 /* Line 1455 of yacc.c  */
-#line 1058 "program_parse.y"
+#line 1255 "program_parse.y"
     {
 	   (yyval.temp_sym) = (yyvsp[(3) - (4)].temp_sym);
 	;}
     break;
 
-  case 116:
+  case 130:
 
 /* Line 1455 of yacc.c  */
-#line 1065 "program_parse.y"
+#line 1262 "program_parse.y"
     {
 	   (yyvsp[(1) - (3)].temp_sym).param_binding_length += (yyvsp[(3) - (3)].temp_sym).param_binding_length;
 	   (yyval.temp_sym) = (yyvsp[(1) - (3)].temp_sym);
 	;}
     break;
 
-  case 117:
+  case 131:
 
 /* Line 1455 of yacc.c  */
-#line 1072 "program_parse.y"
+#line 1269 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3157,10 +3442,10 @@ yyreduce:
 	;}
     break;
 
-  case 118:
+  case 132:
 
 /* Line 1455 of yacc.c  */
-#line 1078 "program_parse.y"
+#line 1275 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3168,10 +3453,10 @@ yyreduce:
 	;}
     break;
 
-  case 119:
+  case 133:
 
 /* Line 1455 of yacc.c  */
-#line 1084 "program_parse.y"
+#line 1281 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3179,10 +3464,10 @@ yyreduce:
 	;}
     break;
 
-  case 120:
+  case 134:
 
 /* Line 1455 of yacc.c  */
-#line 1092 "program_parse.y"
+#line 1289 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3190,10 +3475,10 @@ yyreduce:
 	;}
     break;
 
-  case 121:
+  case 135:
 
 /* Line 1455 of yacc.c  */
-#line 1098 "program_parse.y"
+#line 1295 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3201,10 +3486,10 @@ yyreduce:
 	;}
     break;
 
-  case 122:
+  case 136:
 
 /* Line 1455 of yacc.c  */
-#line 1104 "program_parse.y"
+#line 1301 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3212,10 +3497,10 @@ yyreduce:
 	;}
     break;
 
-  case 123:
+  case 137:
 
 /* Line 1455 of yacc.c  */
-#line 1112 "program_parse.y"
+#line 1309 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3223,10 +3508,10 @@ yyreduce:
 	;}
     break;
 
-  case 124:
+  case 138:
 
 /* Line 1455 of yacc.c  */
-#line 1118 "program_parse.y"
+#line 1315 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3234,10 +3519,10 @@ yyreduce:
 	;}
     break;
 
-  case 125:
+  case 139:
 
 /* Line 1455 of yacc.c  */
-#line 1124 "program_parse.y"
+#line 1321 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3245,101 +3530,101 @@ yyreduce:
 	;}
     break;
 
-  case 126:
+  case 140:
 
 /* Line 1455 of yacc.c  */
-#line 1131 "program_parse.y"
+#line 1328 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(1) - (1)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 127:
+  case 141:
 
 /* Line 1455 of yacc.c  */
-#line 1132 "program_parse.y"
+#line 1329 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 128:
+  case 142:
 
 /* Line 1455 of yacc.c  */
-#line 1135 "program_parse.y"
+#line 1332 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 129:
+  case 143:
 
 /* Line 1455 of yacc.c  */
-#line 1136 "program_parse.y"
+#line 1333 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 130:
+  case 144:
 
 /* Line 1455 of yacc.c  */
-#line 1137 "program_parse.y"
+#line 1334 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 131:
+  case 145:
 
 /* Line 1455 of yacc.c  */
-#line 1138 "program_parse.y"
+#line 1335 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 132:
+  case 146:
 
 /* Line 1455 of yacc.c  */
-#line 1139 "program_parse.y"
+#line 1336 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 133:
+  case 147:
 
 /* Line 1455 of yacc.c  */
-#line 1140 "program_parse.y"
+#line 1337 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 134:
+  case 148:
 
 /* Line 1455 of yacc.c  */
-#line 1141 "program_parse.y"
+#line 1338 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 135:
+  case 149:
 
 /* Line 1455 of yacc.c  */
-#line 1142 "program_parse.y"
+#line 1339 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 136:
+  case 150:
 
 /* Line 1455 of yacc.c  */
-#line 1143 "program_parse.y"
+#line 1340 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 137:
+  case 151:
 
 /* Line 1455 of yacc.c  */
-#line 1144 "program_parse.y"
+#line 1341 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 138:
+  case 152:
 
 /* Line 1455 of yacc.c  */
-#line 1145 "program_parse.y"
+#line 1342 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
-  case 139:
+  case 153:
 
 /* Line 1455 of yacc.c  */
-#line 1149 "program_parse.y"
+#line 1346 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_MATERIAL;
@@ -3348,37 +3633,37 @@ yyreduce:
 	;}
     break;
 
-  case 140:
+  case 154:
 
 /* Line 1455 of yacc.c  */
-#line 1158 "program_parse.y"
+#line 1355 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(1) - (1)].integer);
 	;}
     break;
 
-  case 141:
+  case 155:
 
 /* Line 1455 of yacc.c  */
-#line 1162 "program_parse.y"
+#line 1359 "program_parse.y"
     {
 	   (yyval.integer) = STATE_EMISSION;
 	;}
     break;
 
-  case 142:
+  case 156:
 
 /* Line 1455 of yacc.c  */
-#line 1166 "program_parse.y"
+#line 1363 "program_parse.y"
     {
 	   (yyval.integer) = STATE_SHININESS;
 	;}
     break;
 
-  case 143:
+  case 157:
 
 /* Line 1455 of yacc.c  */
-#line 1172 "program_parse.y"
+#line 1369 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHT;
@@ -3387,28 +3672,28 @@ yyreduce:
 	;}
     break;
 
-  case 144:
+  case 158:
 
 /* Line 1455 of yacc.c  */
-#line 1181 "program_parse.y"
+#line 1378 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(1) - (1)].integer);
 	;}
     break;
 
-  case 145:
+  case 159:
 
 /* Line 1455 of yacc.c  */
-#line 1185 "program_parse.y"
+#line 1382 "program_parse.y"
     {
 	   (yyval.integer) = STATE_POSITION;
 	;}
     break;
 
-  case 146:
+  case 160:
 
 /* Line 1455 of yacc.c  */
-#line 1189 "program_parse.y"
+#line 1386 "program_parse.y"
     {
 	   if (!state->ctx->Extensions.EXT_point_parameters) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "GL_ARB_point_parameters not supported");
@@ -3419,57 +3704,57 @@ yyreduce:
 	;}
     break;
 
-  case 147:
+  case 161:
 
 /* Line 1455 of yacc.c  */
-#line 1198 "program_parse.y"
+#line 1395 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 148:
+  case 162:
 
 /* Line 1455 of yacc.c  */
-#line 1202 "program_parse.y"
+#line 1399 "program_parse.y"
     {
 	   (yyval.integer) = STATE_HALF_VECTOR;
 	;}
     break;
 
-  case 149:
+  case 163:
 
 /* Line 1455 of yacc.c  */
-#line 1208 "program_parse.y"
+#line 1405 "program_parse.y"
     {
 	   (yyval.integer) = STATE_SPOT_DIRECTION;
 	;}
     break;
 
-  case 150:
+  case 164:
 
 /* Line 1455 of yacc.c  */
-#line 1214 "program_parse.y"
+#line 1411 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(2) - (2)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(2) - (2)].state)[1];
 	;}
     break;
 
-  case 151:
+  case 165:
 
 /* Line 1455 of yacc.c  */
-#line 1221 "program_parse.y"
+#line 1418 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHTMODEL_AMBIENT;
 	;}
     break;
 
-  case 152:
+  case 166:
 
 /* Line 1455 of yacc.c  */
-#line 1226 "program_parse.y"
+#line 1423 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHTMODEL_SCENECOLOR;
@@ -3477,10 +3762,10 @@ yyreduce:
 	;}
     break;
 
-  case 153:
+  case 167:
 
 /* Line 1455 of yacc.c  */
-#line 1234 "program_parse.y"
+#line 1431 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHTPROD;
@@ -3490,10 +3775,10 @@ yyreduce:
 	;}
     break;
 
-  case 155:
+  case 169:
 
 /* Line 1455 of yacc.c  */
-#line 1246 "program_parse.y"
+#line 1443 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = (yyvsp[(3) - (3)].integer);
@@ -3501,46 +3786,46 @@ yyreduce:
 	;}
     break;
 
-  case 156:
+  case 170:
 
 /* Line 1455 of yacc.c  */
-#line 1254 "program_parse.y"
+#line 1451 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXENV_COLOR;
 	;}
     break;
 
-  case 157:
+  case 171:
 
 /* Line 1455 of yacc.c  */
-#line 1260 "program_parse.y"
+#line 1457 "program_parse.y"
     {
 	   (yyval.integer) = STATE_AMBIENT;
 	;}
     break;
 
-  case 158:
+  case 172:
 
 /* Line 1455 of yacc.c  */
-#line 1264 "program_parse.y"
+#line 1461 "program_parse.y"
     {
 	   (yyval.integer) = STATE_DIFFUSE;
 	;}
     break;
 
-  case 159:
+  case 173:
 
 /* Line 1455 of yacc.c  */
-#line 1268 "program_parse.y"
+#line 1465 "program_parse.y"
     {
 	   (yyval.integer) = STATE_SPECULAR;
 	;}
     break;
 
-  case 160:
+  case 174:
 
 /* Line 1455 of yacc.c  */
-#line 1274 "program_parse.y"
+#line 1471 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxLights) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid light selector");
@@ -3551,10 +3836,10 @@ yyreduce:
 	;}
     break;
 
-  case 161:
+  case 175:
 
 /* Line 1455 of yacc.c  */
-#line 1285 "program_parse.y"
+#line 1482 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_TEXGEN;
@@ -3563,92 +3848,92 @@ yyreduce:
 	;}
     break;
 
-  case 162:
+  case 176:
 
 /* Line 1455 of yacc.c  */
-#line 1294 "program_parse.y"
+#line 1491 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_S;
 	;}
     break;
 
-  case 163:
+  case 177:
 
 /* Line 1455 of yacc.c  */
-#line 1298 "program_parse.y"
+#line 1495 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_OBJECT_S;
 	;}
     break;
 
-  case 164:
+  case 178:
 
 /* Line 1455 of yacc.c  */
-#line 1303 "program_parse.y"
+#line 1500 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_S - STATE_TEXGEN_EYE_S;
 	;}
     break;
 
-  case 165:
+  case 179:
 
 /* Line 1455 of yacc.c  */
-#line 1307 "program_parse.y"
+#line 1504 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_T - STATE_TEXGEN_EYE_S;
 	;}
     break;
 
-  case 166:
+  case 180:
 
 /* Line 1455 of yacc.c  */
-#line 1311 "program_parse.y"
+#line 1508 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_R - STATE_TEXGEN_EYE_S;
 	;}
     break;
 
-  case 167:
+  case 181:
 
 /* Line 1455 of yacc.c  */
-#line 1315 "program_parse.y"
+#line 1512 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_Q - STATE_TEXGEN_EYE_S;
 	;}
     break;
 
-  case 168:
+  case 182:
 
 /* Line 1455 of yacc.c  */
-#line 1321 "program_parse.y"
+#line 1518 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 169:
+  case 183:
 
 /* Line 1455 of yacc.c  */
-#line 1328 "program_parse.y"
+#line 1525 "program_parse.y"
     {
 	   (yyval.integer) = STATE_FOG_COLOR;
 	;}
     break;
 
-  case 170:
+  case 184:
 
 /* Line 1455 of yacc.c  */
-#line 1332 "program_parse.y"
+#line 1529 "program_parse.y"
     {
 	   (yyval.integer) = STATE_FOG_PARAMS;
 	;}
     break;
 
-  case 171:
+  case 185:
 
 /* Line 1455 of yacc.c  */
-#line 1338 "program_parse.y"
+#line 1535 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_CLIPPLANE;
@@ -3656,10 +3941,10 @@ yyreduce:
 	;}
     break;
 
-  case 172:
+  case 186:
 
 /* Line 1455 of yacc.c  */
-#line 1346 "program_parse.y"
+#line 1543 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxClipPlanes) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid clip plane selector");
@@ -3670,38 +3955,38 @@ yyreduce:
 	;}
     break;
 
-  case 173:
+  case 187:
 
 /* Line 1455 of yacc.c  */
-#line 1357 "program_parse.y"
+#line 1554 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 174:
+  case 188:
 
 /* Line 1455 of yacc.c  */
-#line 1364 "program_parse.y"
+#line 1561 "program_parse.y"
     {
 	   (yyval.integer) = STATE_POINT_SIZE;
 	;}
     break;
 
-  case 175:
+  case 189:
 
 /* Line 1455 of yacc.c  */
-#line 1368 "program_parse.y"
+#line 1565 "program_parse.y"
     {
 	   (yyval.integer) = STATE_POINT_ATTENUATION;
 	;}
     break;
 
-  case 176:
+  case 190:
 
 /* Line 1455 of yacc.c  */
-#line 1374 "program_parse.y"
+#line 1571 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (5)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(1) - (5)].state)[1];
@@ -3711,10 +3996,10 @@ yyreduce:
 	;}
     break;
 
-  case 177:
+  case 191:
 
 /* Line 1455 of yacc.c  */
-#line 1384 "program_parse.y"
+#line 1581 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (2)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(1) - (2)].state)[1];
@@ -3724,20 +4009,20 @@ yyreduce:
 	;}
     break;
 
-  case 178:
+  case 192:
 
 /* Line 1455 of yacc.c  */
-#line 1394 "program_parse.y"
+#line 1591 "program_parse.y"
     {
 	   (yyval.state)[2] = 0;
 	   (yyval.state)[3] = 3;
 	;}
     break;
 
-  case 179:
+  case 193:
 
 /* Line 1455 of yacc.c  */
-#line 1399 "program_parse.y"
+#line 1596 "program_parse.y"
     {
 	   /* It seems logical that the matrix row range specifier would have
 	    * to specify a range or more than one row (i.e., $5 > $3).
@@ -3755,10 +4040,10 @@ yyreduce:
 	;}
     break;
 
-  case 180:
+  case 194:
 
 /* Line 1455 of yacc.c  */
-#line 1417 "program_parse.y"
+#line 1614 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(2) - (3)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(2) - (3)].state)[1];
@@ -3766,55 +4051,55 @@ yyreduce:
 	;}
     break;
 
-  case 181:
+  case 195:
 
 /* Line 1455 of yacc.c  */
-#line 1425 "program_parse.y"
+#line 1622 "program_parse.y"
     {
 	   (yyval.integer) = 0;
 	;}
     break;
 
-  case 182:
+  case 196:
 
 /* Line 1455 of yacc.c  */
-#line 1429 "program_parse.y"
+#line 1626 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(1) - (1)].integer);
 	;}
     break;
 
-  case 183:
+  case 197:
 
 /* Line 1455 of yacc.c  */
-#line 1435 "program_parse.y"
+#line 1632 "program_parse.y"
     {
 	   (yyval.integer) = STATE_MATRIX_INVERSE;
 	;}
     break;
 
-  case 184:
+  case 198:
 
 /* Line 1455 of yacc.c  */
-#line 1439 "program_parse.y"
+#line 1636 "program_parse.y"
     {
 	   (yyval.integer) = STATE_MATRIX_TRANSPOSE;
 	;}
     break;
 
-  case 185:
+  case 199:
 
 /* Line 1455 of yacc.c  */
-#line 1443 "program_parse.y"
+#line 1640 "program_parse.y"
     {
 	   (yyval.integer) = STATE_MATRIX_INVTRANS;
 	;}
     break;
 
-  case 186:
+  case 200:
 
 /* Line 1455 of yacc.c  */
-#line 1449 "program_parse.y"
+#line 1646 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (1)].integer) > 3) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid matrix row reference");
@@ -3825,88 +4110,88 @@ yyreduce:
 	;}
     break;
 
-  case 187:
+  case 201:
 
 /* Line 1455 of yacc.c  */
-#line 1460 "program_parse.y"
+#line 1657 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_MODELVIEW_MATRIX;
 	   (yyval.state)[1] = (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 188:
+  case 202:
 
 /* Line 1455 of yacc.c  */
-#line 1465 "program_parse.y"
+#line 1662 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_PROJECTION_MATRIX;
 	   (yyval.state)[1] = 0;
 	;}
     break;
 
-  case 189:
+  case 203:
 
 /* Line 1455 of yacc.c  */
-#line 1470 "program_parse.y"
+#line 1667 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_MVP_MATRIX;
 	   (yyval.state)[1] = 0;
 	;}
     break;
 
-  case 190:
+  case 204:
 
 /* Line 1455 of yacc.c  */
-#line 1475 "program_parse.y"
+#line 1672 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_TEXTURE_MATRIX;
 	   (yyval.state)[1] = (yyvsp[(2) - (2)].integer);
 	;}
     break;
 
-  case 191:
+  case 205:
 
 /* Line 1455 of yacc.c  */
-#line 1480 "program_parse.y"
+#line 1677 "program_parse.y"
     {
 	   yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported");
 	   YYERROR;
 	;}
     break;
 
-  case 192:
+  case 206:
 
 /* Line 1455 of yacc.c  */
-#line 1485 "program_parse.y"
+#line 1682 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_PROGRAM_MATRIX;
 	   (yyval.state)[1] = (yyvsp[(3) - (4)].integer);
 	;}
     break;
 
-  case 193:
+  case 207:
 
 /* Line 1455 of yacc.c  */
-#line 1492 "program_parse.y"
+#line 1689 "program_parse.y"
     {
 	   (yyval.integer) = 0;
 	;}
     break;
 
-  case 194:
+  case 208:
 
 /* Line 1455 of yacc.c  */
-#line 1496 "program_parse.y"
+#line 1693 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(2) - (3)].integer);
 	;}
     break;
 
-  case 195:
+  case 209:
 
 /* Line 1455 of yacc.c  */
-#line 1501 "program_parse.y"
+#line 1698 "program_parse.y"
     {
 	   /* Since GL_ARB_vertex_blend isn't supported, only modelview matrix
 	    * zero is valid.
@@ -3920,10 +4205,10 @@ yyreduce:
 	;}
     break;
 
-  case 196:
+  case 210:
 
 /* Line 1455 of yacc.c  */
-#line 1514 "program_parse.y"
+#line 1711 "program_parse.y"
     {
 	   /* Since GL_ARB_matrix_palette isn't supported, just let any value
 	    * through here.  The error will be generated later.
@@ -3932,10 +4217,10 @@ yyreduce:
 	;}
     break;
 
-  case 197:
+  case 211:
 
 /* Line 1455 of yacc.c  */
-#line 1522 "program_parse.y"
+#line 1719 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxProgramMatrices) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid program matrix selector");
@@ -3946,20 +4231,20 @@ yyreduce:
 	;}
     break;
 
-  case 198:
+  case 212:
 
 /* Line 1455 of yacc.c  */
-#line 1533 "program_parse.y"
+#line 1730 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_DEPTH_RANGE;
 	;}
     break;
 
-  case 203:
+  case 217:
 
 /* Line 1455 of yacc.c  */
-#line 1545 "program_parse.y"
+#line 1742 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -3969,30 +4254,30 @@ yyreduce:
 	;}
     break;
 
-  case 204:
+  case 218:
 
 /* Line 1455 of yacc.c  */
-#line 1555 "program_parse.y"
+#line 1752 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (1)].integer);
 	   (yyval.state)[1] = (yyvsp[(1) - (1)].integer);
 	;}
     break;
 
-  case 205:
+  case 219:
 
 /* Line 1455 of yacc.c  */
-#line 1560 "program_parse.y"
+#line 1757 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (3)].integer);
 	   (yyval.state)[1] = (yyvsp[(3) - (3)].integer);
 	;}
     break;
 
-  case 206:
+  case 220:
 
 /* Line 1455 of yacc.c  */
-#line 1567 "program_parse.y"
+#line 1764 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4002,10 +4287,10 @@ yyreduce:
 	;}
     break;
 
-  case 207:
+  case 221:
 
 /* Line 1455 of yacc.c  */
-#line 1577 "program_parse.y"
+#line 1774 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4015,30 +4300,30 @@ yyreduce:
 	;}
     break;
 
-  case 208:
+  case 222:
 
 /* Line 1455 of yacc.c  */
-#line 1586 "program_parse.y"
+#line 1783 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (1)].integer);
 	   (yyval.state)[1] = (yyvsp[(1) - (1)].integer);
 	;}
     break;
 
-  case 209:
+  case 223:
 
 /* Line 1455 of yacc.c  */
-#line 1591 "program_parse.y"
+#line 1788 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (3)].integer);
 	   (yyval.state)[1] = (yyvsp[(3) - (3)].integer);
 	;}
     break;
 
-  case 210:
+  case 224:
 
 /* Line 1455 of yacc.c  */
-#line 1598 "program_parse.y"
+#line 1795 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4048,10 +4333,10 @@ yyreduce:
 	;}
     break;
 
-  case 211:
+  case 225:
 
 /* Line 1455 of yacc.c  */
-#line 1608 "program_parse.y"
+#line 1805 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxEnvParams) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid environment parameter reference");
@@ -4061,10 +4346,10 @@ yyreduce:
 	;}
     break;
 
-  case 212:
+  case 226:
 
 /* Line 1455 of yacc.c  */
-#line 1618 "program_parse.y"
+#line 1815 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxLocalParams) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid local parameter reference");
@@ -4074,10 +4359,10 @@ yyreduce:
 	;}
     break;
 
-  case 217:
+  case 231:
 
 /* Line 1455 of yacc.c  */
-#line 1633 "program_parse.y"
+#line 1830 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(1) - (1)].real);
@@ -4087,10 +4372,10 @@ yyreduce:
 	;}
     break;
 
-  case 218:
+  case 232:
 
 /* Line 1455 of yacc.c  */
-#line 1643 "program_parse.y"
+#line 1840 "program_parse.y"
     {
 	   (yyval.vector).count = 1;
 	   (yyval.vector).data[0] = (yyvsp[(1) - (1)].real);
@@ -4100,10 +4385,10 @@ yyreduce:
 	;}
     break;
 
-  case 219:
+  case 233:
 
 /* Line 1455 of yacc.c  */
-#line 1651 "program_parse.y"
+#line 1848 "program_parse.y"
     {
 	   (yyval.vector).count = 1;
 	   (yyval.vector).data[0] = (float) (yyvsp[(1) - (1)].integer);
@@ -4113,10 +4398,10 @@ yyreduce:
 	;}
     break;
 
-  case 220:
+  case 234:
 
 /* Line 1455 of yacc.c  */
-#line 1661 "program_parse.y"
+#line 1858 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (3)].real);
@@ -4126,10 +4411,10 @@ yyreduce:
 	;}
     break;
 
-  case 221:
+  case 235:
 
 /* Line 1455 of yacc.c  */
-#line 1669 "program_parse.y"
+#line 1866 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (5)].real);
@@ -4139,10 +4424,10 @@ yyreduce:
 	;}
     break;
 
-  case 222:
+  case 236:
 
 /* Line 1455 of yacc.c  */
-#line 1678 "program_parse.y"
+#line 1875 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (7)].real);
@@ -4152,10 +4437,10 @@ yyreduce:
 	;}
     break;
 
-  case 223:
+  case 237:
 
 /* Line 1455 of yacc.c  */
-#line 1687 "program_parse.y"
+#line 1884 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (9)].real);
@@ -4165,63 +4450,110 @@ yyreduce:
 	;}
     break;
 
-  case 224:
+  case 238:
 
 /* Line 1455 of yacc.c  */
-#line 1697 "program_parse.y"
+#line 1894 "program_parse.y"
     {
 	   (yyval.real) = ((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].real) : (yyvsp[(2) - (2)].real);
 	;}
     break;
 
-  case 225:
+  case 239:
 
 /* Line 1455 of yacc.c  */
-#line 1701 "program_parse.y"
+#line 1898 "program_parse.y"
     {
 	   (yyval.real) = (float)(((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].integer) : (yyvsp[(2) - (2)].integer));
 	;}
     break;
 
-  case 226:
+  case 240:
 
 /* Line 1455 of yacc.c  */
-#line 1706 "program_parse.y"
+#line 1903 "program_parse.y"
     { (yyval.negate) = FALSE; ;}
     break;
 
-  case 227:
+  case 241:
 
 /* Line 1455 of yacc.c  */
-#line 1707 "program_parse.y"
+#line 1904 "program_parse.y"
     { (yyval.negate) = TRUE;  ;}
     break;
 
-  case 228:
+  case 242:
 
 /* Line 1455 of yacc.c  */
-#line 1708 "program_parse.y"
+#line 1905 "program_parse.y"
     { (yyval.negate) = FALSE; ;}
     break;
 
-  case 229:
+  case 243:
 
 /* Line 1455 of yacc.c  */
-#line 1711 "program_parse.y"
-    { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;}
+#line 1908 "program_parse.y"
+    { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;}
     break;
 
-  case 231:
+  case 245:
+
+/* Line 1455 of yacc.c  */
+#line 1912 "program_parse.y"
+    {
+	   /* NV_fragment_program_option defines the size qualifiers in a
+	    * fairly broken way.  "SHORT" or "LONG" can optionally be used
+	    * before TEMP or OUTPUT.  However, neither is a reserved word!
+	    * This means that we have to parse it as an identifier, then check
+	    * to make sure it's one of the valid values.  *sigh*
+	    *
+	    * In addition, the grammar in the extension spec does *not* allow
+	    * the size specifier to be optional, but all known implementations
+	    * do.
+	    */
+	   if (!state->option.NV_fragment) {
+	      yyerror(& (yylsp[(1) - (1)]), state, "unexpected IDENTIFIER");
+	      YYERROR;
+	   }
+
+	   if (strcmp("SHORT", (yyvsp[(1) - (1)].string)) == 0) {
+	   } else if (strcmp("LONG", (yyvsp[(1) - (1)].string)) == 0) {
+	   } else {
+	      char *const err_str =
+		 make_error_string("invalid storage size specifier \"%s\"",
+				   (yyvsp[(1) - (1)].string));
+
+	      yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL)
+		      ? err_str : "invalid storage size specifier");
+
+	      if (err_str != NULL) {
+		 _mesa_free(err_str);
+	      }
+
+	      YYERROR;
+	   }
+	;}
+    break;
+
+  case 246:
+
+/* Line 1455 of yacc.c  */
+#line 1946 "program_parse.y"
+    {
+	;}
+    break;
+
+  case 247:
 
 /* Line 1455 of yacc.c  */
-#line 1714 "program_parse.y"
+#line 1950 "program_parse.y"
     { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;}
     break;
 
-  case 233:
+  case 249:
 
 /* Line 1455 of yacc.c  */
-#line 1718 "program_parse.y"
+#line 1954 "program_parse.y"
     {
 	   if (!declare_variable(state, (yyvsp[(3) - (3)].string), (yyvsp[(0) - (3)].integer), & (yylsp[(3) - (3)]))) {
 	      YYERROR;
@@ -4229,10 +4561,10 @@ yyreduce:
 	;}
     break;
 
-  case 234:
+  case 250:
 
 /* Line 1455 of yacc.c  */
-#line 1724 "program_parse.y"
+#line 1960 "program_parse.y"
     {
 	   if (!declare_variable(state, (yyvsp[(1) - (1)].string), (yyvsp[(0) - (1)].integer), & (yylsp[(1) - (1)]))) {
 	      YYERROR;
@@ -4240,26 +4572,26 @@ yyreduce:
 	;}
     break;
 
-  case 235:
+  case 251:
 
 /* Line 1455 of yacc.c  */
-#line 1732 "program_parse.y"
+#line 1968 "program_parse.y"
     {
 	   struct asm_symbol *const s =
-	      declare_variable(state, (yyvsp[(2) - (4)].string), at_output, & (yylsp[(2) - (4)]));
+	      declare_variable(state, (yyvsp[(3) - (5)].string), at_output, & (yylsp[(3) - (5)]));
 
 	   if (s == NULL) {
 	      YYERROR;
 	   } else {
-	      s->output_binding = (yyvsp[(4) - (4)].result);
+	      s->output_binding = (yyvsp[(5) - (5)].result);
 	   }
 	;}
     break;
 
-  case 236:
+  case 252:
 
 /* Line 1455 of yacc.c  */
-#line 1745 "program_parse.y"
+#line 1981 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_HPOS;
@@ -4270,10 +4602,10 @@ yyreduce:
 	;}
     break;
 
-  case 237:
+  case 253:
 
 /* Line 1455 of yacc.c  */
-#line 1754 "program_parse.y"
+#line 1990 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_FOGC;
@@ -4284,19 +4616,19 @@ yyreduce:
 	;}
     break;
 
-  case 238:
+  case 254:
 
 /* Line 1455 of yacc.c  */
-#line 1763 "program_parse.y"
+#line 1999 "program_parse.y"
     {
 	   (yyval.result) = (yyvsp[(2) - (2)].result);
 	;}
     break;
 
-  case 239:
+  case 255:
 
 /* Line 1455 of yacc.c  */
-#line 1767 "program_parse.y"
+#line 2003 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_PSIZ;
@@ -4307,10 +4639,10 @@ yyreduce:
 	;}
     break;
 
-  case 240:
+  case 256:
 
 /* Line 1455 of yacc.c  */
-#line 1776 "program_parse.y"
+#line 2012 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_TEX0 + (yyvsp[(3) - (3)].integer);
@@ -4321,10 +4653,10 @@ yyreduce:
 	;}
     break;
 
-  case 241:
+  case 257:
 
 /* Line 1455 of yacc.c  */
-#line 1785 "program_parse.y"
+#line 2021 "program_parse.y"
     {
 	   if (state->mode == ARB_fragment) {
 	      (yyval.result) = FRAG_RESULT_DEPTH;
@@ -4335,19 +4667,19 @@ yyreduce:
 	;}
     break;
 
-  case 242:
+  case 258:
 
 /* Line 1455 of yacc.c  */
-#line 1796 "program_parse.y"
+#line 2032 "program_parse.y"
     {
 	   (yyval.result) = (yyvsp[(2) - (3)].integer) + (yyvsp[(3) - (3)].integer);
 	;}
     break;
 
-  case 243:
+  case 259:
 
 /* Line 1455 of yacc.c  */
-#line 1802 "program_parse.y"
+#line 2038 "program_parse.y"
     {
 	   (yyval.integer) = (state->mode == ARB_vertex)
 	      ? VERT_RESULT_COL0
@@ -4355,10 +4687,10 @@ yyreduce:
 	;}
     break;
 
-  case 244:
+  case 260:
 
 /* Line 1455 of yacc.c  */
-#line 1808 "program_parse.y"
+#line 2044 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = VERT_RESULT_COL0;
@@ -4369,10 +4701,10 @@ yyreduce:
 	;}
     break;
 
-  case 245:
+  case 261:
 
 /* Line 1455 of yacc.c  */
-#line 1817 "program_parse.y"
+#line 2053 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = VERT_RESULT_BFC0;
@@ -4383,19 +4715,19 @@ yyreduce:
 	;}
     break;
 
-  case 246:
+  case 262:
 
 /* Line 1455 of yacc.c  */
-#line 1828 "program_parse.y"
+#line 2064 "program_parse.y"
     {
 	   (yyval.integer) = 0; 
 	;}
     break;
 
-  case 247:
+  case 263:
 
 /* Line 1455 of yacc.c  */
-#line 1832 "program_parse.y"
+#line 2068 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = 0;
@@ -4406,10 +4738,10 @@ yyreduce:
 	;}
     break;
 
-  case 248:
+  case 264:
 
 /* Line 1455 of yacc.c  */
-#line 1841 "program_parse.y"
+#line 2077 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = 1;
@@ -4420,94 +4752,94 @@ yyreduce:
 	;}
     break;
 
-  case 249:
+  case 265:
 
 /* Line 1455 of yacc.c  */
-#line 1851 "program_parse.y"
+#line 2087 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 250:
+  case 266:
 
 /* Line 1455 of yacc.c  */
-#line 1852 "program_parse.y"
+#line 2088 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 251:
+  case 267:
 
 /* Line 1455 of yacc.c  */
-#line 1853 "program_parse.y"
+#line 2089 "program_parse.y"
     { (yyval.integer) = 1; ;}
     break;
 
-  case 252:
+  case 268:
 
 /* Line 1455 of yacc.c  */
-#line 1856 "program_parse.y"
+#line 2092 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 253:
+  case 269:
 
 /* Line 1455 of yacc.c  */
-#line 1857 "program_parse.y"
+#line 2093 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 254:
+  case 270:
 
 /* Line 1455 of yacc.c  */
-#line 1858 "program_parse.y"
+#line 2094 "program_parse.y"
     { (yyval.integer) = 1; ;}
     break;
 
-  case 255:
+  case 271:
 
 /* Line 1455 of yacc.c  */
-#line 1861 "program_parse.y"
+#line 2097 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 256:
+  case 272:
 
 /* Line 1455 of yacc.c  */
-#line 1862 "program_parse.y"
+#line 2098 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;}
     break;
 
-  case 257:
+  case 273:
 
 /* Line 1455 of yacc.c  */
-#line 1865 "program_parse.y"
+#line 2101 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 258:
+  case 274:
 
 /* Line 1455 of yacc.c  */
-#line 1866 "program_parse.y"
+#line 2102 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;}
     break;
 
-  case 259:
+  case 275:
 
 /* Line 1455 of yacc.c  */
-#line 1869 "program_parse.y"
+#line 2105 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
-  case 260:
+  case 276:
 
 /* Line 1455 of yacc.c  */
-#line 1870 "program_parse.y"
+#line 2106 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;}
     break;
 
-  case 261:
+  case 277:
 
 /* Line 1455 of yacc.c  */
-#line 1874 "program_parse.y"
+#line 2110 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureCoordUnits) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid texture coordinate unit selector");
@@ -4518,10 +4850,10 @@ yyreduce:
 	;}
     break;
 
-  case 262:
+  case 278:
 
 /* Line 1455 of yacc.c  */
-#line 1885 "program_parse.y"
+#line 2121 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureImageUnits) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid texture image unit selector");
@@ -4532,10 +4864,10 @@ yyreduce:
 	;}
     break;
 
-  case 263:
+  case 279:
 
 /* Line 1455 of yacc.c  */
-#line 1896 "program_parse.y"
+#line 2132 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureUnits) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid texture unit selector");
@@ -4546,10 +4878,10 @@ yyreduce:
 	;}
     break;
 
-  case 264:
+  case 280:
 
 /* Line 1455 of yacc.c  */
-#line 1907 "program_parse.y"
+#line 2143 "program_parse.y"
     {
 	   struct asm_symbol *exist = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(2) - (4)].string));
@@ -4573,7 +4905,7 @@ yyreduce:
 
 
 /* Line 1455 of yacc.c  */
-#line 4577 "program_parse.tab.c"
+#line 4909 "program_parse.tab.c"
       default: break;
     }
   YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc);
@@ -4792,7 +5124,49 @@ yyreturn:
 
 
 /* Line 1675 of yacc.c  */
-#line 1927 "program_parse.y"
+#line 2167 "program_parse.y"
+
+
+void
+asm_instruction_set_operands(struct asm_instruction *inst,
+			     const struct prog_dst_register *dst,
+			     const struct asm_src_register *src0,
+			     const struct asm_src_register *src1,
+			     const struct asm_src_register *src2)
+{
+   /* In the core ARB extensions only the KIL instruction doesn't have a
+    * destination register.
+    */
+   if (dst == NULL) {
+      init_dst_reg(& inst->Base.DstReg);
+   } else {
+      inst->Base.DstReg = *dst;
+   }
+
+   /* The only instruction that doesn't have any source registers is the
+    * condition-code based KIL instruction added by NV_fragment_program_option.
+    */
+   if (src0 != NULL) {
+      inst->Base.SrcReg[0] = src0->Base;
+      inst->SrcReg[0] = *src0;
+   } else {
+      init_src_reg(& inst->SrcReg[0]);
+   }
+
+   if (src1 != NULL) {
+      inst->Base.SrcReg[1] = src1->Base;
+      inst->SrcReg[1] = *src1;
+   } else {
+      init_src_reg(& inst->SrcReg[1]);
+   }
+
+   if (src2 != NULL) {
+      inst->Base.SrcReg[2] = src2->Base;
+      inst->SrcReg[2] = *src2;
+   } else {
+      init_src_reg(& inst->SrcReg[2]);
+   }
+}
 
 
 struct asm_instruction *
@@ -4802,37 +5176,37 @@ asm_instruction_ctor(gl_inst_opcode op,
 		     const struct asm_src_register *src1,
 		     const struct asm_src_register *src2)
 {
-   struct asm_instruction *inst = calloc(1, sizeof(struct asm_instruction));
+   struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction);
 
    if (inst) {
       _mesa_init_instructions(& inst->Base, 1);
       inst->Base.Opcode = op;
 
-      /* In the core ARB extensions only the KIL instruction doesn't have a
-       * destination register.
-       */
-      if (dst == NULL) {
-	 init_dst_reg(& inst->Base.DstReg);
-      } else {
-	 inst->Base.DstReg = *dst;
-      }
+      asm_instruction_set_operands(inst, dst, src0, src1, src2);
+   }
 
-      inst->Base.SrcReg[0] = src0->Base;
-      inst->SrcReg[0] = *src0;
+   return inst;
+}
 
-      if (src1 != NULL) {
-	 inst->Base.SrcReg[1] = src1->Base;
-	 inst->SrcReg[1] = *src1;
-      } else {
-	 init_src_reg(& inst->SrcReg[1]);
-      }
 
-      if (src2 != NULL) {
-	 inst->Base.SrcReg[2] = src2->Base;
-	 inst->SrcReg[2] = *src2;
-      } else {
-	 init_src_reg(& inst->SrcReg[2]);
-      }
+struct asm_instruction *
+asm_instruction_copy_ctor(const struct prog_instruction *base,
+			  const struct prog_dst_register *dst,
+			  const struct asm_src_register *src0,
+			  const struct asm_src_register *src1,
+			  const struct asm_src_register *src2)
+{
+   struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction);
+
+   if (inst) {
+      _mesa_init_instructions(& inst->Base, 1);
+      inst->Base.Opcode = base->Opcode;
+      inst->Base.CondUpdate = base->CondUpdate;
+      inst->Base.CondDst = base->CondDst;
+      inst->Base.SaturateMode = base->SaturateMode;
+      inst->Base.Precision = base->Precision;
+
+      asm_instruction_set_operands(inst, dst, src0, src1, src2);
    }
 
    return inst;
diff --git a/src/mesa/shader/program_parse.tab.h b/src/mesa/shader/program_parse.tab.h
index dabb3bf1a3..860b6f886e 100644
--- a/src/mesa/shader/program_parse.tab.h
+++ b/src/mesa/shader/program_parse.tab.h
@@ -58,90 +58,92 @@
      ARL = 274,
      KIL = 275,
      SWZ = 276,
-     INTEGER = 277,
-     REAL = 278,
-     AMBIENT = 279,
-     ATTENUATION = 280,
-     BACK = 281,
-     CLIP = 282,
-     COLOR = 283,
-     DEPTH = 284,
-     DIFFUSE = 285,
-     DIRECTION = 286,
-     EMISSION = 287,
-     ENV = 288,
-     EYE = 289,
-     FOG = 290,
-     FOGCOORD = 291,
-     FRAGMENT = 292,
-     FRONT = 293,
-     HALF = 294,
-     INVERSE = 295,
-     INVTRANS = 296,
-     LIGHT = 297,
-     LIGHTMODEL = 298,
-     LIGHTPROD = 299,
-     LOCAL = 300,
-     MATERIAL = 301,
-     MAT_PROGRAM = 302,
-     MATRIX = 303,
-     MATRIXINDEX = 304,
-     MODELVIEW = 305,
-     MVP = 306,
-     NORMAL = 307,
-     OBJECT = 308,
-     PALETTE = 309,
-     PARAMS = 310,
-     PLANE = 311,
-     POINT_TOK = 312,
-     POINTSIZE = 313,
-     POSITION = 314,
-     PRIMARY = 315,
-     PROGRAM = 316,
-     PROJECTION = 317,
-     RANGE = 318,
-     RESULT = 319,
-     ROW = 320,
-     SCENECOLOR = 321,
-     SECONDARY = 322,
-     SHININESS = 323,
-     SIZE_TOK = 324,
-     SPECULAR = 325,
-     SPOT = 326,
-     STATE = 327,
-     TEXCOORD = 328,
-     TEXENV = 329,
-     TEXGEN = 330,
-     TEXGEN_Q = 331,
-     TEXGEN_R = 332,
-     TEXGEN_S = 333,
-     TEXGEN_T = 334,
-     TEXTURE = 335,
-     TRANSPOSE = 336,
-     TEXTURE_UNIT = 337,
-     TEX_1D = 338,
-     TEX_2D = 339,
-     TEX_3D = 340,
-     TEX_CUBE = 341,
-     TEX_RECT = 342,
-     TEX_SHADOW1D = 343,
-     TEX_SHADOW2D = 344,
-     TEX_SHADOWRECT = 345,
-     TEX_ARRAY1D = 346,
-     TEX_ARRAY2D = 347,
-     TEX_ARRAYSHADOW1D = 348,
-     TEX_ARRAYSHADOW2D = 349,
-     VERTEX = 350,
-     VTXATTRIB = 351,
-     WEIGHT = 352,
-     IDENTIFIER = 353,
-     MASK4 = 354,
-     MASK3 = 355,
-     MASK2 = 356,
-     MASK1 = 357,
-     SWIZZLE = 358,
-     DOT_DOT = 359,
-     DOT = 360
+     TXD_OP = 277,
+     INTEGER = 278,
+     REAL = 279,
+     AMBIENT = 280,
+     ATTENUATION = 281,
+     BACK = 282,
+     CLIP = 283,
+     COLOR = 284,
+     DEPTH = 285,
+     DIFFUSE = 286,
+     DIRECTION = 287,
+     EMISSION = 288,
+     ENV = 289,
+     EYE = 290,
+     FOG = 291,
+     FOGCOORD = 292,
+     FRAGMENT = 293,
+     FRONT = 294,
+     HALF = 295,
+     INVERSE = 296,
+     INVTRANS = 297,
+     LIGHT = 298,
+     LIGHTMODEL = 299,
+     LIGHTPROD = 300,
+     LOCAL = 301,
+     MATERIAL = 302,
+     MAT_PROGRAM = 303,
+     MATRIX = 304,
+     MATRIXINDEX = 305,
+     MODELVIEW = 306,
+     MVP = 307,
+     NORMAL = 308,
+     OBJECT = 309,
+     PALETTE = 310,
+     PARAMS = 311,
+     PLANE = 312,
+     POINT_TOK = 313,
+     POINTSIZE = 314,
+     POSITION = 315,
+     PRIMARY = 316,
+     PROGRAM = 317,
+     PROJECTION = 318,
+     RANGE = 319,
+     RESULT = 320,
+     ROW = 321,
+     SCENECOLOR = 322,
+     SECONDARY = 323,
+     SHININESS = 324,
+     SIZE_TOK = 325,
+     SPECULAR = 326,
+     SPOT = 327,
+     STATE = 328,
+     TEXCOORD = 329,
+     TEXENV = 330,
+     TEXGEN = 331,
+     TEXGEN_Q = 332,
+     TEXGEN_R = 333,
+     TEXGEN_S = 334,
+     TEXGEN_T = 335,
+     TEXTURE = 336,
+     TRANSPOSE = 337,
+     TEXTURE_UNIT = 338,
+     TEX_1D = 339,
+     TEX_2D = 340,
+     TEX_3D = 341,
+     TEX_CUBE = 342,
+     TEX_RECT = 343,
+     TEX_SHADOW1D = 344,
+     TEX_SHADOW2D = 345,
+     TEX_SHADOWRECT = 346,
+     TEX_ARRAY1D = 347,
+     TEX_ARRAY2D = 348,
+     TEX_ARRAYSHADOW1D = 349,
+     TEX_ARRAYSHADOW2D = 350,
+     VERTEX = 351,
+     VTXATTRIB = 352,
+     WEIGHT = 353,
+     IDENTIFIER = 354,
+     USED_IDENTIFIER = 355,
+     MASK4 = 356,
+     MASK3 = 357,
+     MASK2 = 358,
+     MASK1 = 359,
+     SWIZZLE = 360,
+     DOT_DOT = 361,
+     DOT = 362
    };
 #endif
 
@@ -152,7 +154,7 @@ typedef union YYSTYPE
 {
 
 /* Line 1676 of yacc.c  */
-#line 107 "program_parse.y"
+#line 116 "program_parse.y"
 
    struct asm_instruction *inst;
    struct asm_symbol *sym;
@@ -181,7 +183,7 @@ typedef union YYSTYPE
 
 
 /* Line 1676 of yacc.c  */
-#line 185 "program_parse.tab.h"
+#line 187 "program_parse.tab.h"
 } YYSTYPE;
 # define YYSTYPE_IS_TRIVIAL 1
 # define yystype YYSTYPE /* obsolescent; will be withdrawn */
diff --git a/src/mesa/shader/program_parse.y b/src/mesa/shader/program_parse.y
index 06c1915fbe..9703e8e670 100644
--- a/src/mesa/shader/program_parse.y
+++ b/src/mesa/shader/program_parse.y
@@ -68,10 +68,19 @@ static void init_dst_reg(struct prog_dst_register *r);
 
 static void init_src_reg(struct asm_src_register *r);
 
+static void asm_instruction_set_operands(struct asm_instruction *inst,
+    const struct prog_dst_register *dst, const struct asm_src_register *src0,
+    const struct asm_src_register *src1, const struct asm_src_register *src2);
+
 static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
     const struct prog_dst_register *dst, const struct asm_src_register *src0,
     const struct asm_src_register *src1, const struct asm_src_register *src2);
 
+static struct asm_instruction *asm_instruction_copy_ctor(
+    const struct prog_instruction *base, const struct prog_dst_register *dst,
+    const struct asm_src_register *src0, const struct asm_src_register *src1,
+    const struct asm_src_register *src2);
+
 #ifndef FALSE
 #define FALSE 0
 #define TRUE (!FALSE)
@@ -142,7 +151,7 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
 
  /* Tokens for instructions */
 %token <temp_inst> BIN_OP BINSC_OP SAMPLE_OP SCALAR_OP TRI_OP VECTOR_OP
-%token <temp_inst> ARL KIL SWZ
+%token <temp_inst> ARL KIL SWZ TXD_OP
 
 %token <integer> INTEGER
 %token <real> REAL
@@ -169,7 +178,8 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
 %token VERTEX VTXATTRIB
 %token WEIGHT
 
-%token <string> IDENTIFIER
+%token <string> IDENTIFIER USED_IDENTIFIER
+%type <string> string
 %token <swiz_mask> MASK4 MASK3 MASK2 MASK1 SWIZZLE
 %token DOT_DOT
 %token DOT
@@ -177,11 +187,11 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
 %type <inst> instruction ALU_instruction TexInstruction
 %type <inst> ARL_instruction VECTORop_instruction
 %type <inst> SCALARop_instruction BINSCop_instruction BINop_instruction
-%type <inst> TRIop_instruction SWZ_instruction SAMPLE_instruction
+%type <inst> TRIop_instruction TXD_instruction SWZ_instruction SAMPLE_instruction
 %type <inst> KIL_instruction
 
 %type <dst_reg> dstReg maskedDstReg maskedAddrReg
-%type <src_reg> srcReg scalarSrcReg swizzleSrcReg
+%type <src_reg> srcReg scalarUse scalarSrcReg swizzleSrcReg
 %type <swiz_mask> scalarSuffix swizzleSuffix extendedSwizzle
 %type <ext_swizzle> extSwizComp extSwizSel
 %type <swiz_mask> optionalMask
@@ -192,6 +202,8 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op,
 %type <sym> addrReg
 %type <swiz_mask> addrComponent addrWriteMask
 
+%type <dst_reg> ccMaskRule ccTest ccMaskRule2 ccTest2 optionalCcMask
+
 %type <result> resultBinding resultColBinding
 %type <integer> optFaceType optColorType
 %type <integer> optResultFaceType optResultColorType
@@ -280,7 +292,7 @@ optionSequence: optionSequence option
 	|
 	;
 
-option: OPTION IDENTIFIER ';'
+option: OPTION string ';'
 	{
 	   int valid = 0;
 
@@ -347,6 +359,7 @@ ALU_instruction: ARL_instruction
 
 TexInstruction: SAMPLE_instruction
 	| KIL_instruction
+	| TXD_instruction
 	;
 
 ARL_instruction: ARL maskedAddrReg ',' scalarSrcReg
@@ -357,51 +370,45 @@ ARL_instruction: ARL maskedAddrReg ',' scalarSrcReg
 
 VECTORop_instruction: VECTOR_OP maskedDstReg ',' swizzleSrcReg
 	{
-	   $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, NULL, NULL);
-	   $$->Base.SaturateMode = $1.SaturateMode;
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL);
 	}
 	;
 
 SCALARop_instruction: SCALAR_OP maskedDstReg ',' scalarSrcReg
 	{
-	   $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, NULL, NULL);
-	   $$->Base.SaturateMode = $1.SaturateMode;
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL);
 	}
 	;
 
 BINSCop_instruction: BINSC_OP maskedDstReg ',' scalarSrcReg ',' scalarSrcReg
 	{
-	   $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, & $6, NULL);
-	   $$->Base.SaturateMode = $1.SaturateMode;
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, NULL);
 	}
 	;
 
 
 BINop_instruction: BIN_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg
 	{
-	   $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, & $6, NULL);
-	   $$->Base.SaturateMode = $1.SaturateMode;
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, NULL);
 	}
 	;
 
 TRIop_instruction: TRI_OP maskedDstReg ','
                    swizzleSrcReg ',' swizzleSrcReg ',' swizzleSrcReg
 	{
-	   $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, & $6, & $8);
-	   $$->Base.SaturateMode = $1.SaturateMode;
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, & $8);
 	}
 	;
 
 SAMPLE_instruction: SAMPLE_OP maskedDstReg ',' swizzleSrcReg ',' texImageUnit ',' texTarget
 	{
-	   $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, NULL, NULL);
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL);
 	   if ($$ != NULL) {
 	      const GLbitfield tex_mask = (1U << $6);
 	      GLbitfield shadow_tex = 0;
 	      GLbitfield target_mask = 0;
 
 
-	      $$->Base.SaturateMode = $1.SaturateMode;
 	      $$->Base.TexSrcUnit = $6;
 
 	      if ($8 < 0) {
@@ -442,6 +449,58 @@ KIL_instruction: KIL swizzleSrcReg
 	   $$ = asm_instruction_ctor(OPCODE_KIL, NULL, & $2, NULL, NULL);
 	   state->fragment.UsesKill = 1;
 	}
+	| KIL ccTest
+	{
+	   $$ = asm_instruction_ctor(OPCODE_KIL_NV, NULL, NULL, NULL, NULL);
+	   $$->Base.DstReg.CondMask = $2.CondMask;
+	   $$->Base.DstReg.CondSwizzle = $2.CondSwizzle;
+	   $$->Base.DstReg.CondSrc = $2.CondSrc;
+	   state->fragment.UsesKill = 1;
+	}
+	;
+
+TXD_instruction: TXD_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg ',' swizzleSrcReg ',' texImageUnit ',' texTarget
+	{
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, & $8);
+	   if ($$ != NULL) {
+	      const GLbitfield tex_mask = (1U << $10);
+	      GLbitfield shadow_tex = 0;
+	      GLbitfield target_mask = 0;
+
+
+	      $$->Base.TexSrcUnit = $10;
+
+	      if ($12 < 0) {
+		 shadow_tex = tex_mask;
+
+		 $$->Base.TexSrcTarget = -$12;
+		 $$->Base.TexShadow = 1;
+	      } else {
+		 $$->Base.TexSrcTarget = $12;
+	      }
+
+	      target_mask = (1U << $$->Base.TexSrcTarget);
+
+	      /* If this texture unit was previously accessed and that access
+	       * had a different texture target, generate an error.
+	       *
+	       * If this texture unit was previously accessed and that access
+	       * had a different shadow mode, generate an error.
+	       */
+	      if ((state->prog->TexturesUsed[$10] != 0)
+		  && ((state->prog->TexturesUsed[$10] != target_mask)
+		      || ((state->prog->ShadowSamplers & tex_mask)
+			  != shadow_tex))) {
+		 yyerror(& @12, state,
+			 "multiple targets used on one texture image unit");
+		 YYERROR;
+	      }
+
+
+	      state->prog->TexturesUsed[$10] |= target_mask;
+	      state->prog->ShadowSamplers |= shadow_tex;
+	   }
+	}
 	;
 
 texImageUnit: TEXTURE_UNIT optTexImageUnitNum
@@ -472,21 +531,58 @@ SWZ_instruction: SWZ maskedDstReg ',' srcReg ',' extendedSwizzle
 	   $4.Base.Swizzle = $6.swizzle;
 	   $4.Base.Negate = $6.mask;
 
-	   $$ = asm_instruction_ctor(OPCODE_SWZ, & $2, & $4, NULL, NULL);
-	   $$->Base.SaturateMode = $1.SaturateMode;
+	   $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL);
 	}
 	;
 
-scalarSrcReg: optionalSign srcReg scalarSuffix
+scalarSrcReg: optionalSign scalarUse
 	{
 	   $$ = $2;
 
 	   if ($1) {
 	      $$.Base.Negate = ~$$.Base.Negate;
 	   }
+	}
+	| optionalSign '|' scalarUse '|'
+	{
+	   $$ = $3;
+
+	   if (!state->option.NV_fragment) {
+	      yyerror(& @2, state, "unexpected character '|'");
+	      YYERROR;
+	   }
+
+	   if ($1) {
+	      $$.Base.Negate = ~$$.Base.Negate;
+	   }
+
+	   $$.Base.Abs = 1;
+	}
+	;
+
+scalarUse:  srcReg scalarSuffix
+	{
+	   $$ = $1;
 
 	   $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle,
-						    $3.swizzle);
+						    $2.swizzle);
+	}
+	| paramConstScalarUse
+	{
+	   struct asm_symbol temp_sym;
+
+	   if (!state->option.NV_fragment) {
+	      yyerror(& @1, state, "expected scalar suffix");
+	      YYERROR;
+	   }
+
+	   memset(& temp_sym, 0, sizeof(temp_sym));
+	   temp_sym.param_binding_begin = ~0;
+	   initialize_symbol_from_const(state->prog, & temp_sym, & $1);
+
+	   init_src_reg(& $$);
+	   $$.Base.File = PROGRAM_CONSTANT;
+	   $$.Base.Index = temp_sym.param_binding_begin;
 	}
 	;
 
@@ -501,12 +597,33 @@ swizzleSrcReg: optionalSign srcReg swizzleSuffix
 	   $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle,
 						    $3.swizzle);
 	}
+	| optionalSign '|' srcReg swizzleSuffix '|'
+	{
+	   $$ = $3;
+
+	   if (!state->option.NV_fragment) {
+	      yyerror(& @2, state, "unexpected character '|'");
+	      YYERROR;
+	   }
+
+	   if ($1) {
+	      $$.Base.Negate = ~$$.Base.Negate;
+	   }
+
+	   $$.Base.Abs = 1;
+	   $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle,
+						    $4.swizzle);
+	}
+
 	;
 
-maskedDstReg: dstReg optionalMask
+maskedDstReg: dstReg optionalMask optionalCcMask
 	{
 	   $$ = $1;
 	   $$.WriteMask = $2.mask;
+	   $$.CondMask = $3.CondMask;
+	   $$.CondSwizzle = $3.CondSwizzle;
+	   $$.CondSrc = $3.CondSrc;
 
 	   if ($$.File == PROGRAM_OUTPUT) {
 	      /* Technically speaking, this should check that it is in
@@ -589,7 +706,7 @@ extSwizSel: INTEGER
 	   $$.xyzw_valid = 1;
 	   $$.rgba_valid = 1;
 	}
-	| IDENTIFIER
+	| string
 	{
 	   if (strlen($1) > 1) {
 	      yyerror(& @1, state, "invalid extended swizzle selector");
@@ -639,7 +756,7 @@ extSwizSel: INTEGER
 	}
 	;
 
-srcReg: IDENTIFIER /* temporaryReg | progParamSingle */
+srcReg: USED_IDENTIFIER /* temporaryReg | progParamSingle */
 	{
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, $1);
@@ -729,7 +846,7 @@ dstReg: resultBinding
 	   $$.File = PROGRAM_OUTPUT;
 	   $$.Index = $1;
 	}
-	| IDENTIFIER /* temporaryReg | vertexResultReg */
+	| USED_IDENTIFIER /* temporaryReg | vertexResultReg */
 	{
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, $1);
@@ -760,7 +877,7 @@ dstReg: resultBinding
 	}
 	;
 
-progParamArray: IDENTIFIER
+progParamArray: USED_IDENTIFIER
 	{
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, $1);
@@ -806,8 +923,10 @@ addrRegRelOffset:              { $$ = 0; }
 addrRegPosOffset: INTEGER
 	{
 	   if (($1 < 0) || ($1 > 63)) {
-	      yyerror(& @1, state,
-		      "relative address offset too large (positive)");
+              char s[100];
+              _mesa_snprintf(s, sizeof(s),
+                             "relative address offset too large (%d)", $1);
+	      yyerror(& @1, state, s);
 	      YYERROR;
 	   } else {
 	      $$ = $1;
@@ -818,8 +937,10 @@ addrRegPosOffset: INTEGER
 addrRegNegOffset: INTEGER
 	{
 	   if (($1 < 0) || ($1 > 64)) {
-	      yyerror(& @1, state,
-		      "relative address offset too large (negative)");
+              char s[100];
+              _mesa_snprintf(s, sizeof(s),
+                             "relative address offset too large (%d)", $1);
+	      yyerror(& @1, state, s);
 	      YYERROR;
 	   } else {
 	      $$ = $1;
@@ -827,7 +948,7 @@ addrRegNegOffset: INTEGER
 	}
 	;
 
-addrReg: IDENTIFIER
+addrReg: USED_IDENTIFIER
 	{
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, $1);
@@ -880,6 +1001,82 @@ optionalMask: MASK4 | MASK3 | MASK2 | MASK1
 	|              { $$.swizzle = SWIZZLE_NOOP; $$.mask = WRITEMASK_XYZW; }
 	;
 
+optionalCcMask: '(' ccTest ')'
+	{
+	   $$ = $2;
+	}
+	| '(' ccTest2 ')'
+	{
+	   $$ = $2;
+	}
+	|
+	{
+	   $$.CondMask = COND_TR;
+	   $$.CondSwizzle = SWIZZLE_NOOP;
+	   $$.CondSrc = 0;
+	}
+	;
+
+ccTest: ccMaskRule swizzleSuffix
+	{
+	   $$ = $1;
+	   $$.CondSwizzle = $2.swizzle;
+	}
+	;
+
+ccTest2: ccMaskRule2 swizzleSuffix
+	{
+	   $$ = $1;
+	   $$.CondSwizzle = $2.swizzle;
+	}
+	;
+
+ccMaskRule: IDENTIFIER
+	{
+	   const int cond = _mesa_parse_cc($1);
+	   if ((cond == 0) || ($1[2] != '\0')) {
+	      char *const err_str =
+		 make_error_string("invalid condition code \"%s\"", $1);
+
+	      yyerror(& @1, state, (err_str != NULL)
+		      ? err_str : "invalid condition code");
+
+	      if (err_str != NULL) {
+		 _mesa_free(err_str);
+	      }
+
+	      YYERROR;
+	   }
+
+	   $$.CondMask = cond;
+	   $$.CondSwizzle = SWIZZLE_NOOP;
+	   $$.CondSrc = 0;
+	}
+	;
+
+ccMaskRule2: USED_IDENTIFIER
+	{
+	   const int cond = _mesa_parse_cc($1);
+	   if ((cond == 0) || ($1[2] != '\0')) {
+	      char *const err_str =
+		 make_error_string("invalid condition code \"%s\"", $1);
+
+	      yyerror(& @1, state, (err_str != NULL)
+		      ? err_str : "invalid condition code");
+
+	      if (err_str != NULL) {
+		 _mesa_free(err_str);
+	      }
+
+	      YYERROR;
+	   }
+
+	   $$.CondMask = cond;
+	   $$.CondSwizzle = SWIZZLE_NOOP;
+	   $$.CondSrc = 0;
+	}
+	;
+
 namingStatement: ATTRIB_statement
 	| PARAM_statement
 	| TEMP_statement
@@ -1708,7 +1905,46 @@ optionalSign: '+'        { $$ = FALSE; }
 	|                { $$ = FALSE; }
 	;
 
-TEMP_statement: TEMP { $<integer>$ = $1; } varNameList
+TEMP_statement: optVarSize TEMP { $<integer>$ = $2; } varNameList
+	;
+
+optVarSize: string
+	{
+	   /* NV_fragment_program_option defines the size qualifiers in a
+	    * fairly broken way.  "SHORT" or "LONG" can optionally be used
+	    * before TEMP or OUTPUT.  However, neither is a reserved word!
+	    * This means that we have to parse it as an identifier, then check
+	    * to make sure it's one of the valid values.  *sigh*
+	    *
+	    * In addition, the grammar in the extension spec does *not* allow
+	    * the size specifier to be optional, but all known implementations
+	    * do.
+	    */
+	   if (!state->option.NV_fragment) {
+	      yyerror(& @1, state, "unexpected IDENTIFIER");
+	      YYERROR;
+	   }
+
+	   if (strcmp("SHORT", $1) == 0) {
+	   } else if (strcmp("LONG", $1) == 0) {
+	   } else {
+	      char *const err_str =
+		 make_error_string("invalid storage size specifier \"%s\"",
+				   $1);
+
+	      yyerror(& @1, state, (err_str != NULL)
+		      ? err_str : "invalid storage size specifier");
+
+	      if (err_str != NULL) {
+		 _mesa_free(err_str);
+	      }
+
+	      YYERROR;
+	   }
+	}
+	|
+	{
+	}
 	;
 
 ADDRESS_statement: ADDRESS { $<integer>$ = $1; } varNameList
@@ -1728,15 +1964,15 @@ varNameList: varNameList ',' IDENTIFIER
 	}
 	;
 
-OUTPUT_statement: OUTPUT IDENTIFIER '=' resultBinding
+OUTPUT_statement: optVarSize OUTPUT IDENTIFIER '=' resultBinding
 	{
 	   struct asm_symbol *const s =
-	      declare_variable(state, $2, at_output, & @2);
+	      declare_variable(state, $3, at_output, & @3);
 
 	   if (s == NULL) {
 	      YYERROR;
 	   } else {
-	      s->output_binding = $4;
+	      s->output_binding = $5;
 	   }
 	}
 	;
@@ -1903,7 +2139,7 @@ legacyTexUnitNum: INTEGER
 	}
 	;
 
-ALIAS_statement: ALIAS IDENTIFIER '=' IDENTIFIER
+ALIAS_statement: ALIAS IDENTIFIER '=' USED_IDENTIFIER
 	{
 	   struct asm_symbol *exist = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, $2);
@@ -1924,8 +2160,54 @@ ALIAS_statement: ALIAS IDENTIFIER '=' IDENTIFIER
 	}
 	;
 
+string: IDENTIFIER
+	| USED_IDENTIFIER
+	;
+
 %%
 
+void
+asm_instruction_set_operands(struct asm_instruction *inst,
+			     const struct prog_dst_register *dst,
+			     const struct asm_src_register *src0,
+			     const struct asm_src_register *src1,
+			     const struct asm_src_register *src2)
+{
+   /* In the core ARB extensions only the KIL instruction doesn't have a
+    * destination register.
+    */
+   if (dst == NULL) {
+      init_dst_reg(& inst->Base.DstReg);
+   } else {
+      inst->Base.DstReg = *dst;
+   }
+
+   /* The only instruction that doesn't have any source registers is the
+    * condition-code based KIL instruction added by NV_fragment_program_option.
+    */
+   if (src0 != NULL) {
+      inst->Base.SrcReg[0] = src0->Base;
+      inst->SrcReg[0] = *src0;
+   } else {
+      init_src_reg(& inst->SrcReg[0]);
+   }
+
+   if (src1 != NULL) {
+      inst->Base.SrcReg[1] = src1->Base;
+      inst->SrcReg[1] = *src1;
+   } else {
+      init_src_reg(& inst->SrcReg[1]);
+   }
+
+   if (src2 != NULL) {
+      inst->Base.SrcReg[2] = src2->Base;
+      inst->SrcReg[2] = *src2;
+   } else {
+      init_src_reg(& inst->SrcReg[2]);
+   }
+}
+
+
 struct asm_instruction *
 asm_instruction_ctor(gl_inst_opcode op,
 		     const struct prog_dst_register *dst,
@@ -1933,37 +2215,37 @@ asm_instruction_ctor(gl_inst_opcode op,
 		     const struct asm_src_register *src1,
 		     const struct asm_src_register *src2)
 {
-   struct asm_instruction *inst = calloc(1, sizeof(struct asm_instruction));
+   struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction);
 
    if (inst) {
       _mesa_init_instructions(& inst->Base, 1);
       inst->Base.Opcode = op;
 
-      /* In the core ARB extensions only the KIL instruction doesn't have a
-       * destination register.
-       */
-      if (dst == NULL) {
-	 init_dst_reg(& inst->Base.DstReg);
-      } else {
-	 inst->Base.DstReg = *dst;
-      }
+      asm_instruction_set_operands(inst, dst, src0, src1, src2);
+   }
 
-      inst->Base.SrcReg[0] = src0->Base;
-      inst->SrcReg[0] = *src0;
+   return inst;
+}
 
-      if (src1 != NULL) {
-	 inst->Base.SrcReg[1] = src1->Base;
-	 inst->SrcReg[1] = *src1;
-      } else {
-	 init_src_reg(& inst->SrcReg[1]);
-      }
 
-      if (src2 != NULL) {
-	 inst->Base.SrcReg[2] = src2->Base;
-	 inst->SrcReg[2] = *src2;
-      } else {
-	 init_src_reg(& inst->SrcReg[2]);
-      }
+struct asm_instruction *
+asm_instruction_copy_ctor(const struct prog_instruction *base,
+			  const struct prog_dst_register *dst,
+			  const struct asm_src_register *src0,
+			  const struct asm_src_register *src1,
+			  const struct asm_src_register *src2)
+{
+   struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction);
+
+   if (inst) {
+      _mesa_init_instructions(& inst->Base, 1);
+      inst->Base.Opcode = base->Opcode;
+      inst->Base.CondUpdate = base->CondUpdate;
+      inst->Base.CondDst = base->CondDst;
+      inst->Base.SaturateMode = base->SaturateMode;
+      inst->Base.Precision = base->Precision;
+
+      asm_instruction_set_operands(inst, dst, src0, src1, src2);
    }
 
    return inst;
diff --git a/src/mesa/shader/program_parse_extra.c b/src/mesa/shader/program_parse_extra.c
index 8e4be606f1..0656c5eaa8 100644
--- a/src/mesa/shader/program_parse_extra.c
+++ b/src/mesa/shader/program_parse_extra.c
@@ -34,6 +34,121 @@
  */
 
 int
+_mesa_parse_instruction_suffix(const struct asm_parser_state *state,
+			       const char *suffix,
+			       struct prog_instruction *inst)
+{
+   inst->CondUpdate = 0;
+   inst->CondDst = 0;
+   inst->SaturateMode = SATURATE_OFF;
+   inst->Precision = FLOAT32;
+
+
+   /* The first possible suffix element is the precision specifier from
+    * NV_fragment_program_option.
+    */
+   if (state->option.NV_fragment) {
+      switch (suffix[0]) {
+      case 'H':
+	 inst->Precision = FLOAT16;
+	 suffix++;
+	 break;
+      case 'R':
+	 inst->Precision = FLOAT32;
+	 suffix++;
+	 break;
+      case 'X':
+	 inst->Precision = FIXED12;
+	 suffix++;
+	 break;
+      default:
+	 break;
+      }
+   }
+
+   /* The next possible suffix element is the condition code modifier selection
+    * from NV_fragment_program_option.
+    */
+   if (state->option.NV_fragment) {
+      if (suffix[0] == 'C') {
+	 inst->CondUpdate = 1;
+	 suffix++;
+      }
+   }
+
+
+   /* The final possible suffix element is the saturation selector from
+    * ARB_fragment_program.
+    */
+   if (state->mode == ARB_fragment) {
+      if (strcmp(suffix, "_SAT") == 0) {
+	 inst->SaturateMode = SATURATE_ZERO_ONE;
+	 suffix += 4;
+      }
+   }
+
+
+   /* It is an error for all of the suffix string not to be consumed.
+    */
+   return suffix[0] == '\0';
+}
+
+
+int
+_mesa_parse_cc(const char *s)
+{
+   int cond = 0;
+
+   switch (s[0]) {
+   case 'E':
+      if (s[1] == 'Q') {
+	 cond = COND_EQ;
+      }
+      break;
+
+   case 'F':
+      if (s[1] == 'L') {
+	 cond = COND_FL;
+      }
+      break;
+
+   case 'G':
+      if (s[1] == 'E') {
+	 cond = COND_GE;
+      } else if (s[1] == 'T') {
+	 cond = COND_GT;
+      }
+      break;
+
+   case 'L':
+      if (s[1] == 'E') {
+	 cond = COND_LE;
+      } else if (s[1] == 'T') {
+	 cond = COND_LT;
+      }
+      break;
+
+   case 'N':
+      if (s[1] == 'E') {
+	 cond = COND_NE;
+      }
+      break;
+
+   case 'T':
+      if (s[1] == 'R') {
+	 cond = COND_TR;
+      }
+      break;
+
+   default:
+      break;
+   }
+
+   return ((cond == 0) || (s[2] != '\0')) ? 0 : cond;
+}
+
+
+int
 _mesa_ARBvp_parse_option(struct asm_parser_state *state, const char *option)
 {
    if (strcmp(option, "ARB_position_invariant") == 0) {
@@ -102,6 +217,17 @@ _mesa_ARBfp_parse_option(struct asm_parser_state *state, const char *option)
 	    return 1;
 	 }
       }
+   } else if (strncmp(option, "NV_fragment_program", 19) == 0) {
+      option += 19;
+
+      /* Other NV_fragment_program strings may be supported later.
+       */
+      if (option[0] == '\0') {
+	 if (state->ctx->Extensions.NV_fragment_program_option) {
+	    state->option.NV_fragment = 1;
+	    return 1;
+	 }
+      }
    } else if (strncmp(option, "MESA_", 5) == 0) {
       option += 5;
 
diff --git a/src/mesa/shader/program_parser.h b/src/mesa/shader/program_parser.h
index fa47d84565..bce6041381 100644
--- a/src/mesa/shader/program_parser.h
+++ b/src/mesa/shader/program_parser.h
@@ -202,6 +202,7 @@ struct asm_parser_state {
       unsigned Shadow:1;
       unsigned TexRect:1;
       unsigned TexArray:1;
+      unsigned NV_fragment:1;
    } option;
 
    struct {
@@ -263,4 +264,31 @@ extern int _mesa_ARBvp_parse_option(struct asm_parser_state *state,
 extern int _mesa_ARBfp_parse_option(struct asm_parser_state *state,
     const char *option);
 
+/**
+ * Parses and processes instruction suffixes
+ *
+ * Instruction suffixes, such as \c _SAT, are processed.  The relevant bits
+ * are set in \c inst.  If suffixes are encountered that are either not known
+ * or not supported by the modes and options set in \c state, zero will be
+ * returned.
+ *
+ * \return
+ * Non-zero on success, zero on failure.
+ */
+extern int _mesa_parse_instruction_suffix(const struct asm_parser_state *state,
+    const char *suffix, struct prog_instruction *inst);
+
+/**
+ * Parses a condition code name
+ *
+ * The condition code names (e.g., \c LT, \c GT, \c NE) were added to assembly
+ * shaders with the \c GL_NV_fragment_program_option extension.  This function
+ * converts a string representation into one of the \c COND_ macros.
+ *
+ * \return
+ * One of the \c COND_ macros defined in prog_instruction.h on success or zero
+ * on failure.
+ */
+extern int _mesa_parse_cc(const char *s);
+
 /*@}*/
diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c
index f70c75cec8..3b8529592d 100644
--- a/src/mesa/shader/programopt.c
+++ b/src/mesa/shader/programopt.c
@@ -573,3 +573,94 @@ _mesa_remove_output_reads(struct gl_program *prog, gl_register_file type)
       }
    }
 }
+
+
+/**
+ * Make the given fragment program into a "no-op" shader.
+ * Actually, just copy the incoming fragment color (or texcoord)
+ * to the output color.
+ * This is for debug/test purposes.
+ */
+void
+_mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog)
+{
+   struct prog_instruction *inst;
+   GLuint inputAttr;
+
+   inst = _mesa_alloc_instructions(2);
+   if (!inst) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "_mesa_nop_fragment_program");
+      return;
+   }
+
+   _mesa_init_instructions(inst, 2);
+
+   inst[0].Opcode = OPCODE_MOV;
+   inst[0].DstReg.File = PROGRAM_OUTPUT;
+   inst[0].DstReg.Index = FRAG_RESULT_COLOR;
+   inst[0].SrcReg[0].File = PROGRAM_INPUT;
+   if (prog->Base.InputsRead & FRAG_BIT_COL0)
+      inputAttr = FRAG_ATTRIB_COL0;
+   else
+      inputAttr = FRAG_ATTRIB_TEX0;
+   inst[0].SrcReg[0].Index = inputAttr;
+
+   inst[1].Opcode = OPCODE_END;
+
+   _mesa_free_instructions(prog->Base.Instructions,
+                           prog->Base.NumInstructions);
+
+   prog->Base.Instructions = inst;
+   prog->Base.NumInstructions = 2;
+   prog->Base.InputsRead = 1 << inputAttr;
+   prog->Base.OutputsWritten = 1 << FRAG_RESULT_COLOR;
+}
+
+
+/**
+ * \sa _mesa_nop_fragment_program
+ * Replace the given vertex program with a "no-op" program that just
+ * transforms vertex position and emits color.
+ */
+void
+_mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog)
+{
+   struct prog_instruction *inst;
+   GLuint inputAttr;
+
+   /*
+    * Start with a simple vertex program that emits color.
+    */
+   inst = _mesa_alloc_instructions(2);
+   if (!inst) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "_mesa_nop_vertex_program");
+      return;
+   }
+
+   _mesa_init_instructions(inst, 2);
+
+   inst[0].Opcode = OPCODE_MOV;
+   inst[0].DstReg.File = PROGRAM_OUTPUT;
+   inst[0].DstReg.Index = VERT_RESULT_COL0;
+   inst[0].SrcReg[0].File = PROGRAM_INPUT;
+   if (prog->Base.InputsRead & VERT_BIT_COLOR0)
+      inputAttr = VERT_ATTRIB_COLOR0;
+   else
+      inputAttr = VERT_ATTRIB_TEX0;
+   inst[0].SrcReg[0].Index = inputAttr;
+
+   inst[1].Opcode = OPCODE_END;
+
+   _mesa_free_instructions(prog->Base.Instructions,
+                           prog->Base.NumInstructions);
+
+   prog->Base.Instructions = inst;
+   prog->Base.NumInstructions = 2;
+   prog->Base.InputsRead = 1 << inputAttr;
+   prog->Base.OutputsWritten = 1 << VERT_RESULT_COL0;
+
+   /*
+    * Now insert code to do standard modelview/projection transformation.
+    */
+   _mesa_insert_mvp_code(ctx, prog);
+}
diff --git a/src/mesa/shader/programopt.h b/src/mesa/shader/programopt.h
index 96acaf9566..21fac07849 100644
--- a/src/mesa/shader/programopt.h
+++ b/src/mesa/shader/programopt.h
@@ -42,4 +42,11 @@ _mesa_count_texture_instructions(struct gl_program *prog);
 extern void
 _mesa_remove_output_reads(struct gl_program *prog, gl_register_file type);
 
+extern void
+_mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog);
+
+extern void
+_mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog);
+
+
 #endif /* PROGRAMOPT_H */
diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c
index b282d7af60..f473bd1173 100644
--- a/src/mesa/shader/shader_api.c
+++ b/src/mesa/shader/shader_api.c
@@ -380,12 +380,18 @@ get_shader_flags(void)
          flags |= GLSL_DUMP;
       if (_mesa_strstr(env, "log"))
          flags |= GLSL_LOG;
+      if (_mesa_strstr(env, "nopvert"))
+         flags |= GLSL_NOP_VERT;
+      if (_mesa_strstr(env, "nopfrag"))
+         flags |= GLSL_NOP_FRAG;
       if (_mesa_strstr(env, "nopt"))
          flags |= GLSL_NO_OPT;
       else if (_mesa_strstr(env, "opt"))
          flags |= GLSL_OPT;
       if (_mesa_strstr(env, "uniform"))
          flags |= GLSL_UNIFORMS;
+      if (_mesa_strstr(env, "useprog"))
+         flags |= GLSL_USE_PROG;
    }
 
    return flags;
@@ -1520,19 +1526,32 @@ _mesa_use_program(GLcontext *ctx, GLuint program)
       }
 
       /* debug code */
-      if (0) {
+      if (ctx->Shader.Flags & GLSL_USE_PROG) {
          GLuint i;
-         _mesa_printf("Use Shader %u\n", shProg->Name);
+         _mesa_printf("Mesa: glUseProgram(%u)\n", shProg->Name);
          for (i = 0; i < shProg->NumShaders; i++) {
-            _mesa_printf(" shader %u, type 0x%x, checksum %u\n",
+            const char *s;
+            switch (shProg->Shaders[i]->Type) {
+            case GL_VERTEX_SHADER:
+               s = "vertex";
+               break;
+            case GL_FRAGMENT_SHADER:
+               s = "fragment";
+               break;
+            case GL_GEOMETRY_SHADER:
+               s = "geometry";
+               break;
+            default:
+               s = "";
+            }
+            _mesa_printf("  %s shader %u, checksum %u\n", s, 
                          shProg->Shaders[i]->Name,
-                         shProg->Shaders[i]->Type,
                          shProg->Shaders[i]->SourceChecksum);
          }
          if (shProg->VertexProgram)
-            printf(" vert prog %u\n", shProg->VertexProgram->Base.Id);
+            _mesa_printf("  vert prog %u\n", shProg->VertexProgram->Base.Id);
          if (shProg->FragmentProgram)
-            printf(" frag prog %u\n", shProg->FragmentProgram->Base.Id);
+            _mesa_printf("  frag prog %u\n", shProg->FragmentProgram->Base.Id);
       }
    }
    else {
diff --git a/src/mesa/shader/slang/slang_compile.c b/src/mesa/shader/slang/slang_compile.c
index c1b97c7cb7..a270888443 100644
--- a/src/mesa/shader/slang/slang_compile.c
+++ b/src/mesa/shader/slang/slang_compile.c
@@ -2814,6 +2814,16 @@ _slang_compile(GLcontext *ctx, struct gl_shader *shader)
           (ctx->Shader.Flags & GLSL_NO_OPT) == 0) {
          _mesa_optimize_program(ctx, shader->Program);
       }
+      if ((ctx->Shader.Flags & GLSL_NOP_VERT) &&
+          shader->Program->Target == GL_VERTEX_PROGRAM_ARB) {
+         _mesa_nop_vertex_program(ctx,
+                                  (struct gl_vertex_program *) shader->Program);
+      }
+      if ((ctx->Shader.Flags & GLSL_NOP_FRAG) &&
+          shader->Program->Target == GL_FRAGMENT_PROGRAM_ARB) {
+         _mesa_nop_fragment_program(ctx,
+                                (struct gl_fragment_program *) shader->Program);
+      }
    }
 
    if (ctx->Shader.Flags & GLSL_LOG) {
diff --git a/src/mesa/sources.mak b/src/mesa/sources.mak
index fa2a6307a4..7107538cee 100644
--- a/src/mesa/sources.mak
+++ b/src/mesa/sources.mak
@@ -115,7 +115,6 @@ SWRAST_SOURCES = \
 	swrast/s_feedback.c \
 	swrast/s_fog.c \
 	swrast/s_fragprog.c \
-	swrast/s_imaging.c \
 	swrast/s_lines.c \
 	swrast/s_logic.c \
 	swrast/s_masking.c \
@@ -125,7 +124,6 @@ SWRAST_SOURCES = \
 	swrast/s_stencil.c \
 	swrast/s_texcombine.c \
 	swrast/s_texfilter.c \
-	swrast/s_texstore.c \
 	swrast/s_triangle.c \
 	swrast/s_zoom.c
 
diff --git a/src/mesa/sparc/clip.S b/src/mesa/sparc/clip.S
index 208843c606..dc239171ff 100644
--- a/src/mesa/sparc/clip.S
+++ b/src/mesa/sparc/clip.S
@@ -58,7 +58,8 @@ clip_table:
 	.byte	31, 29, 31, 30, 27, 25, 27, 26
 
 /* GLvector4f *clip_vec, GLvector4f *proj_vec, 
-   GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask */
+   GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask,
+   GLboolean viewport_z_enable */
 
 	.align		64
 __pc_tramp:
diff --git a/src/mesa/sparc/glapi_sparc.S b/src/mesa/sparc/glapi_sparc.S
index 5b2c9e4a9a..aaa17e6a3b 100644
--- a/src/mesa/sparc/glapi_sparc.S
+++ b/src/mesa/sparc/glapi_sparc.S
@@ -771,23 +771,26 @@ gl_dispatch_functions_start:
 	GL_STUB(glGetSynciv, _gloffset_GetSynciv)
 	GL_STUB(glIsSync, _gloffset_IsSync)
 	GL_STUB(glWaitSync, _gloffset_WaitSync)
+	GL_STUB(glDrawElementsBaseVertex, _gloffset_DrawElementsBaseVertex)
+	GL_STUB(glDrawRangeElementsBaseVertex, _gloffset_DrawRangeElementsBaseVertex)
+	GL_STUB(glMultiDrawElementsBaseVertex, _gloffset_MultiDrawElementsBaseVertex)
 	GL_STUB(glPolygonOffsetEXT, _gloffset_PolygonOffsetEXT)
-	GL_STUB(gl_dispatch_stub_575, _gloffset_GetPixelTexGenParameterfvSGIS)
-	HIDDEN(gl_dispatch_stub_575)
-	GL_STUB(gl_dispatch_stub_576, _gloffset_GetPixelTexGenParameterivSGIS)
-	HIDDEN(gl_dispatch_stub_576)
-	GL_STUB(gl_dispatch_stub_577, _gloffset_PixelTexGenParameterfSGIS)
-	HIDDEN(gl_dispatch_stub_577)
-	GL_STUB(gl_dispatch_stub_578, _gloffset_PixelTexGenParameterfvSGIS)
+	GL_STUB(gl_dispatch_stub_578, _gloffset_GetPixelTexGenParameterfvSGIS)
 	HIDDEN(gl_dispatch_stub_578)
-	GL_STUB(gl_dispatch_stub_579, _gloffset_PixelTexGenParameteriSGIS)
+	GL_STUB(gl_dispatch_stub_579, _gloffset_GetPixelTexGenParameterivSGIS)
 	HIDDEN(gl_dispatch_stub_579)
-	GL_STUB(gl_dispatch_stub_580, _gloffset_PixelTexGenParameterivSGIS)
+	GL_STUB(gl_dispatch_stub_580, _gloffset_PixelTexGenParameterfSGIS)
 	HIDDEN(gl_dispatch_stub_580)
-	GL_STUB(gl_dispatch_stub_581, _gloffset_SampleMaskSGIS)
+	GL_STUB(gl_dispatch_stub_581, _gloffset_PixelTexGenParameterfvSGIS)
 	HIDDEN(gl_dispatch_stub_581)
-	GL_STUB(gl_dispatch_stub_582, _gloffset_SamplePatternSGIS)
+	GL_STUB(gl_dispatch_stub_582, _gloffset_PixelTexGenParameteriSGIS)
 	HIDDEN(gl_dispatch_stub_582)
+	GL_STUB(gl_dispatch_stub_583, _gloffset_PixelTexGenParameterivSGIS)
+	HIDDEN(gl_dispatch_stub_583)
+	GL_STUB(gl_dispatch_stub_584, _gloffset_SampleMaskSGIS)
+	HIDDEN(gl_dispatch_stub_584)
+	GL_STUB(gl_dispatch_stub_585, _gloffset_SamplePatternSGIS)
+	HIDDEN(gl_dispatch_stub_585)
 	GL_STUB(glColorPointerEXT, _gloffset_ColorPointerEXT)
 	GL_STUB(glEdgeFlagPointerEXT, _gloffset_EdgeFlagPointerEXT)
 	GL_STUB(glIndexPointerEXT, _gloffset_IndexPointerEXT)
@@ -798,10 +801,10 @@ gl_dispatch_functions_start:
 	GL_STUB(glPointParameterfvEXT, _gloffset_PointParameterfvEXT)
 	GL_STUB(glLockArraysEXT, _gloffset_LockArraysEXT)
 	GL_STUB(glUnlockArraysEXT, _gloffset_UnlockArraysEXT)
-	GL_STUB(gl_dispatch_stub_593, _gloffset_CullParameterdvEXT)
-	HIDDEN(gl_dispatch_stub_593)
-	GL_STUB(gl_dispatch_stub_594, _gloffset_CullParameterfvEXT)
-	HIDDEN(gl_dispatch_stub_594)
+	GL_STUB(gl_dispatch_stub_596, _gloffset_CullParameterdvEXT)
+	HIDDEN(gl_dispatch_stub_596)
+	GL_STUB(gl_dispatch_stub_597, _gloffset_CullParameterfvEXT)
+	HIDDEN(gl_dispatch_stub_597)
 	GL_STUB(glSecondaryColor3bEXT, _gloffset_SecondaryColor3bEXT)
 	GL_STUB(glSecondaryColor3bvEXT, _gloffset_SecondaryColor3bvEXT)
 	GL_STUB(glSecondaryColor3dEXT, _gloffset_SecondaryColor3dEXT)
@@ -826,8 +829,8 @@ gl_dispatch_functions_start:
 	GL_STUB(glFogCoorddvEXT, _gloffset_FogCoorddvEXT)
 	GL_STUB(glFogCoordfEXT, _gloffset_FogCoordfEXT)
 	GL_STUB(glFogCoordfvEXT, _gloffset_FogCoordfvEXT)
-	GL_STUB(gl_dispatch_stub_619, _gloffset_PixelTexGenSGIX)
-	HIDDEN(gl_dispatch_stub_619)
+	GL_STUB(gl_dispatch_stub_622, _gloffset_PixelTexGenSGIX)
+	HIDDEN(gl_dispatch_stub_622)
 	GL_STUB(glBlendFuncSeparateEXT, _gloffset_BlendFuncSeparateEXT)
 	GL_STUB(glFlushVertexArrayRangeNV, _gloffset_FlushVertexArrayRangeNV)
 	GL_STUB(glVertexArrayRangeNV, _gloffset_VertexArrayRangeNV)
@@ -869,24 +872,24 @@ gl_dispatch_functions_start:
 	GL_STUB(glWindowPos4ivMESA, _gloffset_WindowPos4ivMESA)
 	GL_STUB(glWindowPos4sMESA, _gloffset_WindowPos4sMESA)
 	GL_STUB(glWindowPos4svMESA, _gloffset_WindowPos4svMESA)
-	GL_STUB(gl_dispatch_stub_661, _gloffset_MultiModeDrawArraysIBM)
-	HIDDEN(gl_dispatch_stub_661)
-	GL_STUB(gl_dispatch_stub_662, _gloffset_MultiModeDrawElementsIBM)
-	HIDDEN(gl_dispatch_stub_662)
-	GL_STUB(gl_dispatch_stub_663, _gloffset_DeleteFencesNV)
-	HIDDEN(gl_dispatch_stub_663)
-	GL_STUB(gl_dispatch_stub_664, _gloffset_FinishFenceNV)
+	GL_STUB(gl_dispatch_stub_664, _gloffset_MultiModeDrawArraysIBM)
 	HIDDEN(gl_dispatch_stub_664)
-	GL_STUB(gl_dispatch_stub_665, _gloffset_GenFencesNV)
+	GL_STUB(gl_dispatch_stub_665, _gloffset_MultiModeDrawElementsIBM)
 	HIDDEN(gl_dispatch_stub_665)
-	GL_STUB(gl_dispatch_stub_666, _gloffset_GetFenceivNV)
+	GL_STUB(gl_dispatch_stub_666, _gloffset_DeleteFencesNV)
 	HIDDEN(gl_dispatch_stub_666)
-	GL_STUB(gl_dispatch_stub_667, _gloffset_IsFenceNV)
+	GL_STUB(gl_dispatch_stub_667, _gloffset_FinishFenceNV)
 	HIDDEN(gl_dispatch_stub_667)
-	GL_STUB(gl_dispatch_stub_668, _gloffset_SetFenceNV)
+	GL_STUB(gl_dispatch_stub_668, _gloffset_GenFencesNV)
 	HIDDEN(gl_dispatch_stub_668)
-	GL_STUB(gl_dispatch_stub_669, _gloffset_TestFenceNV)
+	GL_STUB(gl_dispatch_stub_669, _gloffset_GetFenceivNV)
 	HIDDEN(gl_dispatch_stub_669)
+	GL_STUB(gl_dispatch_stub_670, _gloffset_IsFenceNV)
+	HIDDEN(gl_dispatch_stub_670)
+	GL_STUB(gl_dispatch_stub_671, _gloffset_SetFenceNV)
+	HIDDEN(gl_dispatch_stub_671)
+	GL_STUB(gl_dispatch_stub_672, _gloffset_TestFenceNV)
+	HIDDEN(gl_dispatch_stub_672)
 	GL_STUB(glAreProgramsResidentNV, _gloffset_AreProgramsResidentNV)
 	GL_STUB(glBindProgramNV, _gloffset_BindProgramNV)
 	GL_STUB(glDeleteProgramsNV, _gloffset_DeleteProgramsNV)
@@ -967,26 +970,26 @@ gl_dispatch_functions_start:
 	GL_STUB(glSetFragmentShaderConstantATI, _gloffset_SetFragmentShaderConstantATI)
 	GL_STUB(glPointParameteriNV, _gloffset_PointParameteriNV)
 	GL_STUB(glPointParameterivNV, _gloffset_PointParameterivNV)
-	GL_STUB(gl_dispatch_stub_750, _gloffset_ActiveStencilFaceEXT)
-	HIDDEN(gl_dispatch_stub_750)
-	GL_STUB(gl_dispatch_stub_751, _gloffset_BindVertexArrayAPPLE)
-	HIDDEN(gl_dispatch_stub_751)
-	GL_STUB(gl_dispatch_stub_752, _gloffset_DeleteVertexArraysAPPLE)
-	HIDDEN(gl_dispatch_stub_752)
-	GL_STUB(gl_dispatch_stub_753, _gloffset_GenVertexArraysAPPLE)
+	GL_STUB(gl_dispatch_stub_753, _gloffset_ActiveStencilFaceEXT)
 	HIDDEN(gl_dispatch_stub_753)
-	GL_STUB(gl_dispatch_stub_754, _gloffset_IsVertexArrayAPPLE)
+	GL_STUB(gl_dispatch_stub_754, _gloffset_BindVertexArrayAPPLE)
 	HIDDEN(gl_dispatch_stub_754)
+	GL_STUB(gl_dispatch_stub_755, _gloffset_DeleteVertexArraysAPPLE)
+	HIDDEN(gl_dispatch_stub_755)
+	GL_STUB(gl_dispatch_stub_756, _gloffset_GenVertexArraysAPPLE)
+	HIDDEN(gl_dispatch_stub_756)
+	GL_STUB(gl_dispatch_stub_757, _gloffset_IsVertexArrayAPPLE)
+	HIDDEN(gl_dispatch_stub_757)
 	GL_STUB(glGetProgramNamedParameterdvNV, _gloffset_GetProgramNamedParameterdvNV)
 	GL_STUB(glGetProgramNamedParameterfvNV, _gloffset_GetProgramNamedParameterfvNV)
 	GL_STUB(glProgramNamedParameter4dNV, _gloffset_ProgramNamedParameter4dNV)
 	GL_STUB(glProgramNamedParameter4dvNV, _gloffset_ProgramNamedParameter4dvNV)
 	GL_STUB(glProgramNamedParameter4fNV, _gloffset_ProgramNamedParameter4fNV)
 	GL_STUB(glProgramNamedParameter4fvNV, _gloffset_ProgramNamedParameter4fvNV)
-	GL_STUB(gl_dispatch_stub_761, _gloffset_DepthBoundsEXT)
-	HIDDEN(gl_dispatch_stub_761)
-	GL_STUB(gl_dispatch_stub_762, _gloffset_BlendEquationSeparateEXT)
-	HIDDEN(gl_dispatch_stub_762)
+	GL_STUB(gl_dispatch_stub_764, _gloffset_DepthBoundsEXT)
+	HIDDEN(gl_dispatch_stub_764)
+	GL_STUB(gl_dispatch_stub_765, _gloffset_BlendEquationSeparateEXT)
+	HIDDEN(gl_dispatch_stub_765)
 	GL_STUB(glBindFramebufferEXT, _gloffset_BindFramebufferEXT)
 	GL_STUB(glBindRenderbufferEXT, _gloffset_BindRenderbufferEXT)
 	GL_STUB(glCheckFramebufferStatusEXT, _gloffset_CheckFramebufferStatusEXT)
@@ -1004,28 +1007,28 @@ gl_dispatch_functions_start:
 	GL_STUB(glIsFramebufferEXT, _gloffset_IsFramebufferEXT)
 	GL_STUB(glIsRenderbufferEXT, _gloffset_IsRenderbufferEXT)
 	GL_STUB(glRenderbufferStorageEXT, _gloffset_RenderbufferStorageEXT)
-	GL_STUB(gl_dispatch_stub_780, _gloffset_BlitFramebufferEXT)
-	HIDDEN(gl_dispatch_stub_780)
-	GL_STUB(gl_dispatch_stub_781, _gloffset_BufferParameteriAPPLE)
-	HIDDEN(gl_dispatch_stub_781)
-	GL_STUB(gl_dispatch_stub_782, _gloffset_FlushMappedBufferRangeAPPLE)
-	HIDDEN(gl_dispatch_stub_782)
+	GL_STUB(gl_dispatch_stub_783, _gloffset_BlitFramebufferEXT)
+	HIDDEN(gl_dispatch_stub_783)
+	GL_STUB(gl_dispatch_stub_784, _gloffset_BufferParameteriAPPLE)
+	HIDDEN(gl_dispatch_stub_784)
+	GL_STUB(gl_dispatch_stub_785, _gloffset_FlushMappedBufferRangeAPPLE)
+	HIDDEN(gl_dispatch_stub_785)
 	GL_STUB(glFramebufferTextureLayerEXT, _gloffset_FramebufferTextureLayerEXT)
 	GL_STUB(glProvokingVertexEXT, _gloffset_ProvokingVertexEXT)
-	GL_STUB(gl_dispatch_stub_785, _gloffset_GetTexParameterPointervAPPLE)
-	HIDDEN(gl_dispatch_stub_785)
-	GL_STUB(gl_dispatch_stub_786, _gloffset_TextureRangeAPPLE)
-	HIDDEN(gl_dispatch_stub_786)
-	GL_STUB(gl_dispatch_stub_787, _gloffset_StencilFuncSeparateATI)
-	HIDDEN(gl_dispatch_stub_787)
-	GL_STUB(gl_dispatch_stub_788, _gloffset_ProgramEnvParameters4fvEXT)
+	GL_STUB(gl_dispatch_stub_788, _gloffset_GetTexParameterPointervAPPLE)
 	HIDDEN(gl_dispatch_stub_788)
-	GL_STUB(gl_dispatch_stub_789, _gloffset_ProgramLocalParameters4fvEXT)
+	GL_STUB(gl_dispatch_stub_789, _gloffset_TextureRangeAPPLE)
 	HIDDEN(gl_dispatch_stub_789)
-	GL_STUB(gl_dispatch_stub_790, _gloffset_GetQueryObjecti64vEXT)
+	GL_STUB(gl_dispatch_stub_790, _gloffset_StencilFuncSeparateATI)
 	HIDDEN(gl_dispatch_stub_790)
-	GL_STUB(gl_dispatch_stub_791, _gloffset_GetQueryObjectui64vEXT)
+	GL_STUB(gl_dispatch_stub_791, _gloffset_ProgramEnvParameters4fvEXT)
 	HIDDEN(gl_dispatch_stub_791)
+	GL_STUB(gl_dispatch_stub_792, _gloffset_ProgramLocalParameters4fvEXT)
+	HIDDEN(gl_dispatch_stub_792)
+	GL_STUB(gl_dispatch_stub_793, _gloffset_GetQueryObjecti64vEXT)
+	HIDDEN(gl_dispatch_stub_793)
+	GL_STUB(gl_dispatch_stub_794, _gloffset_GetQueryObjectui64vEXT)
+	HIDDEN(gl_dispatch_stub_794)
 	GL_STUB_ALIAS(glArrayElementEXT, glArrayElement)
 	GL_STUB_ALIAS(glBindTextureEXT, glBindTexture)
 	GL_STUB_ALIAS(glDrawArraysEXT, glDrawArrays)
@@ -1359,6 +1362,7 @@ gl_dispatch_functions_start:
 	GL_STUB_ALIAS(glIsRenderbuffer, glIsRenderbufferEXT)
 	GL_STUB_ALIAS(glRenderbufferStorage, glRenderbufferStorageEXT)
 	GL_STUB_ALIAS(glFramebufferTextureLayer, glFramebufferTextureLayerEXT)
+	GL_STUB_ALIAS(glProvokingVertex, glProvokingVertexEXT)
 
 	.globl	gl_dispatch_functions_end
 	HIDDEN(gl_dispatch_functions_end)
diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c
index d2286a2c83..cea0c7cecf 100644
--- a/src/mesa/sparc/sparc.c
+++ b/src/mesa/sparc/sparc.c
@@ -78,13 +78,15 @@ extern GLvector4f  *_mesa_sparc_cliptest_points4(GLvector4f *clip_vec,
 						 GLvector4f *proj_vec,
 						 GLubyte clipMask[],
 						 GLubyte *orMask,
-						 GLubyte *andMask);
+						 GLubyte *andMask,
+						 GLboolean viewport_z_clip);
 
 extern GLvector4f  *_mesa_sparc_cliptest_points4_np(GLvector4f *clip_vec,
 						    GLvector4f *proj_vec,
 						    GLubyte clipMask[],
 						    GLubyte *orMask,
-						    GLubyte *andMask);
+						    GLubyte *andMask,
+						    GLboolean viewport_z_clip);
 
 #define NORM_ARGS	const GLmatrix *mat,				\
 			GLfloat scale,					\
diff --git a/src/mesa/state_tracker/st_atom_constbuf.c b/src/mesa/state_tracker/st_atom_constbuf.c
index 5d4d8eee02..77153889b6 100644
--- a/src/mesa/state_tracker/st_atom_constbuf.c
+++ b/src/mesa/state_tracker/st_atom_constbuf.c
@@ -39,6 +39,7 @@
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
 
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_atom.h"
 #include "st_atom_constbuf.h"
@@ -75,7 +76,7 @@ void st_upload_constants( struct st_context *st,
                                         PIPE_BUFFER_USAGE_CONSTANT,
 					paramBytes );
 
-      if (0) {
+      if (ST_DEBUG & DEBUG_CONSTANTS) {
 	 debug_printf("%s(shader=%d, numParams=%d, stateFlags=0x%x)\n", 
                       __FUNCTION__, shader_type, params->NumParameters,
                       params->StateFlags);
diff --git a/src/mesa/state_tracker/st_atom_depth.c b/src/mesa/state_tracker/st_atom_depth.c
index 0aa128f947..88b80a07fc 100644
--- a/src/mesa/state_tracker/st_atom_depth.c
+++ b/src/mesa/state_tracker/st_atom_depth.c
@@ -104,10 +104,6 @@ update_depth_stencil_alpha(struct st_context *st)
       dsa->depth.func = st_compare_func_to_pipe(ctx->Depth.Func);
    }
 
-   if (ctx->Query.CurrentOcclusionObject &&
-       ctx->Query.CurrentOcclusionObject->Active)
-      dsa->depth.occlusion_count = 1;
-
    if (ctx->Stencil.Enabled && ctx->DrawBuffer->Visual.stencilBits > 0) {
       dsa->stencil[0].enabled = 1;
       dsa->stencil[0].func = st_compare_func_to_pipe(ctx->Stencil.Function[0]);
diff --git a/src/mesa/state_tracker/st_atom_framebuffer.c b/src/mesa/state_tracker/st_atom_framebuffer.c
index 5209a6a0c9..e18c0f6e0a 100644
--- a/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -39,6 +39,7 @@
 #include "pipe/p_context.h"
 #include "pipe/p_inlines.h"
 #include "cso_cache/cso_context.h"
+#include "util/u_rect.h"
 
 
 
@@ -162,10 +163,17 @@ update_framebuffer_state( struct st_context *st )
          (void) st_get_framebuffer_surface(stfb, ST_SURFACE_FRONT_LEFT, &surf_front);
          (void) st_get_framebuffer_surface(stfb, ST_SURFACE_BACK_LEFT, &surf_back);
 
-         st->pipe->surface_copy(st->pipe,
-                                surf_front, 0, 0,  /* dest */
-                                surf_back, 0, 0,   /* src */
-                                fb->Width, fb->Height);
+         if (st->pipe->surface_copy) {
+            st->pipe->surface_copy(st->pipe,
+                                   surf_front, 0, 0,  /* dest */
+                                   surf_back, 0, 0,   /* src */
+                                   fb->Width, fb->Height);
+         } else {
+            util_surface_copy(st->pipe, FALSE,
+                              surf_front, 0, 0,
+                              surf_back, 0, 0,
+                              fb->Width, fb->Height);
+         }
       }
       /* we're assuming we'll really draw to the front buffer */
       st->frontbuffer_status = FRONT_STATUS_DIRTY;
diff --git a/src/mesa/state_tracker/st_atom_pixeltransfer.c b/src/mesa/state_tracker/st_atom_pixeltransfer.c
index eff3666ca8..babfcc87b4 100644
--- a/src/mesa/state_tracker/st_atom_pixeltransfer.c
+++ b/src/mesa/state_tracker/st_atom_pixeltransfer.c
@@ -122,7 +122,8 @@ create_color_map_texture(GLcontext *ctx)
    const uint texSize = 256; /* simple, and usually perfect */
 
    /* find an RGBA texture format */
-   format = st_choose_format(pipe, GL_RGBA, PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_SAMPLER);
+   format = st_choose_format(pipe->screen, GL_RGBA,
+                             PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_SAMPLER);
 
    /* create texture for color map/table */
    pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0,
diff --git a/src/mesa/state_tracker/st_cb_accum.c b/src/mesa/state_tracker/st_cb_accum.c
index 95181578f6..a6b9765452 100644
--- a/src/mesa/state_tracker/st_cb_accum.c
+++ b/src/mesa/state_tracker/st_cb_accum.c
@@ -34,6 +34,7 @@
 #include "main/image.h"
 #include "main/macros.h"
 
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_cb_accum.h"
 #include "st_cb_fbo.h"
@@ -136,6 +137,9 @@ accum_accum(struct st_context *st, GLfloat value,
    GLubyte *data = acc_strb->data;
    GLfloat *buf;
 
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
+
    color_trans = st_cond_flush_get_tex_transfer(st, color_strb->texture,
 						0, 0, 0,
 						PIPE_TRANSFER_READ, xpos, ypos,
@@ -181,6 +185,10 @@ accum_load(struct st_context *st, GLfloat value,
    GLubyte *data = acc_strb->data;
    GLfloat *buf;
 
+
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
+
    color_trans = st_cond_flush_get_tex_transfer(st, color_strb->texture,
 						0, 0, 0,
 						PIPE_TRANSFER_READ, xpos, ypos,
@@ -228,6 +236,9 @@ accum_return(GLcontext *ctx, GLfloat value,
    const GLubyte *data = acc_strb->data;
    GLfloat *buf;
 
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
+
    buf = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat));
 
    if (!colormask[0] || !colormask[1] || !colormask[2] || !colormask[3])
@@ -241,7 +252,7 @@ accum_return(GLcontext *ctx, GLfloat value,
 						xpos, ypos,
 						width, height);
 
-   if (usage != PIPE_TRANSFER_WRITE)
+   if (usage & PIPE_TRANSFER_READ)
       pipe_get_tile_rgba(color_trans, 0, 0, width, height, buf);
 
    switch (acc_strb->format) {
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index 902fb38d1a..a22fa68299 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -330,7 +330,18 @@ setup_bitmap_vertex_data(struct st_context *st,
    const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0);
    const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0);
    const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0);
-   const GLuint max_slots = 4096 / sizeof(st->bitmap.vertices);
+
+   /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as
+    * no_flush) updates to buffers where we know there is no conflict
+    * with previous data.  Currently using max_slots > 1 will cause
+    * synchronous rendering if the driver flushes its command buffers
+    * between one bitmap and the next.  Our flush hook below isn't
+    * sufficient to catch this as the driver doesn't tell us when it
+    * flushes its own command buffers.  Until this gets fixed, pay the
+    * price of allocating a new buffer for each bitmap cache-flush to
+    * avoid synchronous rendering.
+    */
+   const GLuint max_slots = 1; /* 4096 / sizeof(st->bitmap.vertices); */
    GLuint i;
 
    if (st->bitmap.vbuf_slot >= max_slots) {
diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c
index c741940bcf..5626e25323 100644
--- a/src/mesa/state_tracker/st_cb_blit.c
+++ b/src/mesa/state_tracker/st_cb_blit.c
@@ -39,6 +39,7 @@
 #include "shader/prog_print.h"
 
 #include "st_context.h"
+#include "st_texture.h"
 #include "st_program.h"
 #include "st_cb_blit.h"
 #include "st_cb_fbo.h"
@@ -110,17 +111,50 @@ st_BlitFramebuffer(GLcontext *ctx,
    }
 
    if (mask & GL_COLOR_BUFFER_BIT) {
-      struct st_renderbuffer *srcRb = 
-         st_renderbuffer(readFB->_ColorReadBuffer);
-      struct st_renderbuffer *dstRb = 
-         st_renderbuffer(drawFB->_ColorDrawBuffers[0]);
-      struct pipe_surface *srcSurf = srcRb->surface;
-      struct pipe_surface *dstSurf = dstRb->surface;
-
-      util_blit_pixels(st->blit,
-                       srcSurf, srcX0, srcY0, srcX1, srcY1,
-                       dstSurf, dstX0, dstY0, dstX1, dstY1,
-                       0.0, pFilter);
+      struct gl_renderbuffer_attachment *srcAtt =
+         &readFB->Attachment[readFB->_ColorReadBufferIndex];
+
+      if(srcAtt->Type == GL_TEXTURE) {
+         struct pipe_screen *screen = ctx->st->pipe->screen;
+         const struct st_texture_object *srcObj =
+            st_texture_object(srcAtt->Texture);
+         struct st_renderbuffer *dstRb =
+            st_renderbuffer(drawFB->_ColorDrawBuffers[0]);
+         struct pipe_surface *srcSurf;
+         struct pipe_surface *dstSurf = dstRb->surface;
+
+         if (!srcObj->pt)
+            return;
+
+         srcSurf = screen->get_tex_surface(screen,
+                                           srcObj->pt,
+                                           srcAtt->CubeMapFace,
+                                           srcAtt->TextureLevel,
+                                           srcAtt->Zoffset,
+                                           PIPE_BUFFER_USAGE_GPU_READ);
+         if(!srcSurf)
+            return;
+
+         util_blit_pixels(st->blit,
+                          srcSurf, srcX0, srcY0, srcX1, srcY1,
+                          dstSurf, dstX0, dstY0, dstX1, dstY1,
+                          0.0, pFilter);
+
+         pipe_surface_reference(&srcSurf, NULL);
+      }
+      else {
+         struct st_renderbuffer *srcRb =
+            st_renderbuffer(readFB->_ColorReadBuffer);
+         struct st_renderbuffer *dstRb =
+            st_renderbuffer(drawFB->_ColorDrawBuffers[0]);
+         struct pipe_surface *srcSurf = srcRb->surface;
+         struct pipe_surface *dstSurf = dstRb->surface;
+
+         util_blit_pixels(st->blit,
+                          srcSurf, srcX0, srcY0, srcX1, srcY1,
+                          dstSurf, dstX0, dstY0, dstX1, dstY1,
+                          0.0, pFilter);
+      }
    }
 
    if (mask & depthStencil) {
diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index 8a8c99f7e1..36510720a4 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -116,7 +116,18 @@ draw_quad(GLcontext *ctx,
 {
    struct st_context *st = ctx->st;
    struct pipe_context *pipe = st->pipe;
-   const GLuint max_slots = 1024 / sizeof(st->clear.vertices);
+
+   /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as
+    * no_flush) updates to buffers where we know there is no conflict
+    * with previous data.  Currently using max_slots > 1 will cause
+    * synchronous rendering if the driver flushes its command buffers
+    * between one bitmap and the next.  Our flush hook below isn't
+    * sufficient to catch this as the driver doesn't tell us when it
+    * flushes its own command buffers.  Until this gets fixed, pay the
+    * price of allocating a new buffer for each bitmap cache-flush to
+    * avoid synchronous rendering.
+    */
+   const GLuint max_slots = 1; /* 1024 / sizeof(st->clear.vertices); */
    GLuint i;
 
    if (st->clear.vbuf_slot >= max_slots) {
diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c
index c655690603..be44577117 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -40,6 +40,7 @@
 #include "shader/prog_parameter.h"
 #include "shader/prog_print.h"
 
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_atom.h"
 #include "st_atom_constbuf.h"
@@ -61,6 +62,7 @@
 #include "util/u_tile.h"
 #include "util/u_draw_quad.h"
 #include "util/u_math.h"
+#include "util/u_rect.h"
 #include "shader/prog_instruction.h"
 #include "cso_cache/cso_context.h"
 
@@ -98,7 +100,7 @@ is_passthrough_program(const struct gl_fragment_program *prog)
 static struct st_fragment_program *
 combined_drawpix_fragment_program(GLcontext *ctx)
 {
-   struct st_context *st = ctx->st;
+   struct st_context *st = st_context(ctx);
    struct st_fragment_program *stfp;
 
    if (st->pixel_xfer.program->serialNo == st->pixel_xfer.xfer_prog_sn
@@ -445,8 +447,8 @@ draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z,
           GLfloat x1, GLfloat y1, const GLfloat *color,
           GLboolean invertTex, GLfloat maxXcoord, GLfloat maxYcoord)
 {
-   struct st_context *st = ctx->st;
-   struct pipe_context *pipe = ctx->st->pipe;
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
    GLfloat verts[4][3][4]; /* four verts, three attribs, XYZW */
 
    /* setup vertex data */
@@ -540,9 +542,9 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z,
                    const GLfloat *color,
                    GLboolean invertTex)
 {
-   struct st_context *st = ctx->st;
-   struct pipe_context *pipe = ctx->st->pipe;
-   struct cso_context *cso = ctx->st->cso_context;
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+   struct cso_context *cso = st->cso_context;
    GLfloat x0, y0, x1, y1;
    GLsizei maxSize;
 
@@ -652,7 +654,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y,
                     const struct gl_pixelstore_attrib *unpack,
                     const GLvoid *pixels)
 {
-   struct st_context *st = ctx->st;
+   struct st_context *st = st_context(ctx);
    struct pipe_context *pipe = st->pipe;
    struct pipe_screen *screen = pipe->screen;
    struct st_renderbuffer *strb;
@@ -793,7 +795,7 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
 {
    struct st_fragment_program *stfp;
    struct st_vertex_program *stvp;
-   struct st_context *st = ctx->st;
+   struct st_context *st = st_context(ctx);
    struct pipe_surface *ps;
    const GLfloat *color;
 
@@ -811,21 +813,21 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
 
    if (format == GL_DEPTH_COMPONENT) {
       ps = st->state.framebuffer.zsbuf;
-      stfp = make_fragment_shader_z(ctx->st);
-      stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE);
+      stfp = make_fragment_shader_z(st);
+      stvp = st_make_passthrough_vertex_shader(st, GL_TRUE);
       color = ctx->Current.RasterColor;
    }
    else {
       ps = st->state.framebuffer.cbufs[0];
       stfp = combined_drawpix_fragment_program(ctx);
-      stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE);
+      stvp = st_make_passthrough_vertex_shader(st, GL_FALSE);
       color = NULL;
    }
 
    /* draw with textured quad */
    {
       struct pipe_texture *pt
-         = make_texture(ctx->st, width, height, format, type, unpack, pixels);
+         = make_texture(st, width, height, format, type, unpack, pixels);
       if (pt) {
          draw_textured_quad(ctx, x, y, ctx->Current.RasterPos[2],
                             width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY,
@@ -942,7 +944,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
               GLsizei width, GLsizei height,
               GLint dstx, GLint dsty, GLenum type)
 {
-   struct st_context *st = ctx->st;
+   struct st_context *st = st_context(ctx);
    struct pipe_context *pipe = st->pipe;
    struct pipe_screen *screen = pipe->screen;
    struct st_renderbuffer *rbRead;
@@ -995,14 +997,14 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
       rbRead = st_get_color_read_renderbuffer(ctx);
       color = NULL;
       stfp = combined_drawpix_fragment_program(ctx);
-      stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE);
+      stvp = st_make_passthrough_vertex_shader(st, GL_FALSE);
    }
    else {
       assert(type == GL_DEPTH);
       rbRead = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer);
       color = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
-      stfp = make_fragment_shader_z(ctx->st);
-      stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE);
+      stfp = make_fragment_shader_z(st);
+      stvp = st_make_passthrough_vertex_shader(st, GL_TRUE);
    }
 
    srcFormat = rbRead->texture->format;
@@ -1014,13 +1016,14 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
    else {
       /* srcFormat can't be used as a texture format */
       if (type == GL_DEPTH) {
-         texFormat = st_choose_format(pipe, GL_DEPTH_COMPONENT, PIPE_TEXTURE_2D, 
+         texFormat = st_choose_format(screen, GL_DEPTH_COMPONENT,
+                                      PIPE_TEXTURE_2D, 
                                       PIPE_TEXTURE_USAGE_DEPTH_STENCIL);
          assert(texFormat != PIPE_FORMAT_NONE); /* XXX no depth texture formats??? */
       }
       else {
          /* default color format */
-         texFormat = st_choose_format(pipe, GL_RGBA, PIPE_TEXTURE_2D, 
+         texFormat = st_choose_format(screen, GL_RGBA, PIPE_TEXTURE_2D, 
                                       PIPE_TEXTURE_USAGE_SAMPLER);
          assert(texFormat != PIPE_FORMAT_NONE);
       }
@@ -1059,7 +1062,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
       assert(pth <= maxSize);
    }
 
-   pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, texFormat, 0,
+   pt = st_texture_create(st, PIPE_TEXTURE_2D, texFormat, 0,
                           ptw, pth, 1,
                           PIPE_TEXTURE_USAGE_SAMPLER);
    if (!pt)
@@ -1073,11 +1076,19 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
                                        PIPE_BUFFER_USAGE_GPU_READ);
       struct pipe_surface *psTex = screen->get_tex_surface(screen, pt, 0, 0, 0, 
                                       PIPE_BUFFER_USAGE_GPU_WRITE );
-      pipe->surface_copy(pipe,
-			 psTex, /* dest */
-			 0, 0, /* destx/y */
-			 psRead,
-			 srcx, srcy, width, height);
+      if (pipe->surface_copy) {
+         pipe->surface_copy(pipe,
+                            psTex, /* dest */
+                            0, 0, /* destx/y */
+                            psRead,
+                            srcx, srcy, width, height);
+      } else {
+         util_surface_copy(pipe, FALSE,
+                           psTex,
+                           0, 0,
+                           psRead,
+                           srcx, srcy, width, height);
+      }
       pipe_surface_reference(&psRead, NULL); 
       pipe_surface_reference(&psTex, NULL);
    }
@@ -1090,6 +1101,9 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
       struct pipe_transfer *ptTex;
       enum pipe_transfer_usage transfer_usage;
 
+      if (ST_DEBUG & DEBUG_FALLBACK)
+         debug_printf("%s: fallback processing\n", __FUNCTION__);
+
       if (type == GL_DEPTH && pf_is_depth_and_stencil(pt->format))
          transfer_usage = PIPE_TRANSFER_READ_WRITE;
       else
diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c
index e8399ded7b..73aa65955b 100644
--- a/src/mesa/state_tracker/st_cb_fbo.c
+++ b/src/mesa/state_tracker/st_cb_fbo.c
@@ -49,6 +49,7 @@
 #include "st_public.h"
 #include "st_texture.h"
 
+#include "util/u_rect.h"
 
 
 /**
@@ -93,7 +94,7 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    if (strb->format != PIPE_FORMAT_NONE)
       format = strb->format;
    else
-      format = st_choose_renderbuffer_format(pipe, internalFormat);
+      format = st_choose_renderbuffer_format(pipe->screen, internalFormat);
       
    /* init renderbuffer fields */
    strb->Base.Width  = width;
@@ -384,6 +385,7 @@ st_render_texture(GLcontext *ctx,
 
    rb->Width = texImage->Width2;
    rb->Height = texImage->Height2;
+   rb->_BaseFormat = texImage->_BaseFormat;
    /*printf("***** render to texture level %d: %d x %d\n", att->TextureLevel, rb->Width, rb->Height);*/
 
    /*printf("***** pipe texture %d x %d\n", pt->width[0], pt->height[0]);*/
@@ -537,10 +539,17 @@ copy_back_to_front(struct st_context *st,
    (void) st_get_framebuffer_surface(stfb, backIndex, &surf_back);
 
    if (surf_front && surf_back) {
-      st->pipe->surface_copy(st->pipe,
-                             surf_front, 0, 0,  /* dest */
-                             surf_back, 0, 0,   /* src */
-                             fb->Width, fb->Height);
+      if (st->pipe->surface_copy) {
+         st->pipe->surface_copy(st->pipe,
+                                surf_front, 0, 0,  /* dest */
+                                surf_back, 0, 0,   /* src */
+                                fb->Width, fb->Height);
+      } else {
+         util_surface_copy(st->pipe, FALSE,
+                           surf_front, 0, 0,
+                           surf_back, 0, 0,
+                           fb->Width, fb->Height);
+      }
    }
 }
 
diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c
index 4398ab2839..b2d5c39a3a 100644
--- a/src/mesa/state_tracker/st_cb_program.c
+++ b/src/mesa/state_tracker/st_cb_program.c
@@ -45,6 +45,7 @@
 #include "st_context.h"
 #include "st_program.h"
 #include "st_atom_shader.h"
+#include "st_mesa_to_tgsi.h"
 #include "st_cb_program.h"
 
 
@@ -152,7 +153,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog)
          }
 
          if (stvp->state.tokens) {
-            _mesa_free((void *) stvp->state.tokens);
+            st_free_tokens(stvp->state.tokens);
             stvp->state.tokens = NULL;
          }
       }
@@ -167,7 +168,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog)
          }
          
          if (stfp->state.tokens) {
-            _mesa_free((void *) stfp->state.tokens);
+            st_free_tokens(stfp->state.tokens);
             stfp->state.tokens = NULL;
          }
 
@@ -214,7 +215,7 @@ static void st_program_string_notify( GLcontext *ctx,
       }
 
       if (stfp->state.tokens) {
-         _mesa_free((void *) stfp->state.tokens);
+         st_free_tokens(stfp->state.tokens);
          stfp->state.tokens = NULL;
       }
 
@@ -242,7 +243,7 @@ static void st_program_string_notify( GLcontext *ctx,
       }
 
       if (stvp->state.tokens) {
-         _mesa_free((void *) stvp->state.tokens);
+         st_free_tokens(stvp->state.tokens);
          stvp->state.tokens = NULL;
       }
 
diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c
index 75424aa2e7..772bb3bb69 100644
--- a/src/mesa/state_tracker/st_cb_readpixels.c
+++ b/src/mesa/state_tracker/st_cb_readpixels.c
@@ -43,6 +43,7 @@
 #include "pipe/p_inlines.h"
 #include "util/u_tile.h"
 
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_cb_bitmap.h"
 #include "st_cb_readpixels.h"
@@ -416,6 +417,9 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
       yStep = 1;
    }
 
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
+
    /*
     * Copy pixels from pipe_transfer to user memory
     */
diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c
index 50675b5896..a1953342b4 100644
--- a/src/mesa/state_tracker/st_cb_texture.c
+++ b/src/mesa/state_tracker/st_cb_texture.c
@@ -43,6 +43,7 @@
 #include "main/texobj.h"
 #include "main/texstore.h"
 
+#include "state_tracker/st_debug.h"
 #include "state_tracker/st_context.h"
 #include "state_tracker/st_cb_fbo.h"
 #include "state_tracker/st_cb_texture.h"
@@ -328,10 +329,13 @@ guess_and_alloc_texture(struct st_context *st,
         stObj->base.MinFilter == GL_LINEAR ||
         stImage->base._BaseFormat == GL_DEPTH_COMPONENT ||
         stImage->base._BaseFormat == GL_DEPTH_STENCIL_EXT) &&
+       !stObj->base.GenerateMipmap &&
        stImage->level == firstLevel) {
+      /* only alloc space for a single mipmap level */
       lastLevel = firstLevel;
    }
    else {
+      /* alloc space for a full mipmap */
       GLuint l2width = util_logbase2(width);
       GLuint l2height = util_logbase2(height);
       GLuint l2depth = util_logbase2(depth);
@@ -768,10 +772,6 @@ done:
       st_texture_image_unmap(ctx->st, stImage);
       texImage->Data = NULL;
    }
-
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
 }
 
 
@@ -904,6 +904,9 @@ decompress_with_blit(GLcontext * ctx, GLenum target, GLint level,
          GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width,
                                               height, format, type, row, 0);
 
+         if (ST_DEBUG & DEBUG_FALLBACK)
+            debug_printf("%s: fallback format translation\n", __FUNCTION__);
+
          /* get float[4] rgba row from surface */
          pipe_get_tile_rgba(tex_xfer, 0, row, width, 1, rgba);
 
@@ -1142,10 +1145,6 @@ done:
       st_texture_image_unmap(ctx->st, stImage);
       texImage->Data = NULL;
    }
-
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
 }
 
 
@@ -1299,6 +1298,9 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level,
    struct pipe_transfer *src_trans;
    GLvoid *texDest;
    enum pipe_transfer_usage transfer_usage;
+   
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
 
    assert(width <= MAX_WIDTH);
 
@@ -1424,6 +1426,12 @@ compatible_src_dst_formats(const struct gl_renderbuffer *src,
       return TGSI_WRITEMASK_XYZ; /* A ==> 1.0 */
    }
    else {
+      if (ST_DEBUG & DEBUG_FALLBACK)
+         debug_printf("%s failed for src %s, dst %s\n",
+                      __FUNCTION__, 
+                      _mesa_lookup_enum_by_nr(srcFormat),
+                      _mesa_lookup_enum_by_nr(dstLogicalFormat));
+
       /* Otherwise fail.
        */
       return 0;
@@ -1538,7 +1546,8 @@ st_copy_texsubimage(GLcontext *ctx,
 
    if (ctx->_ImageTransferState == 0x0) {
 
-      if (matching_base_formats && 
+      if (pipe->surface_copy &&
+          matching_base_formats &&
           src_format == dest_format &&
           !do_flip) 
       {
@@ -1609,10 +1618,6 @@ st_copy_texsubimage(GLcontext *ctx,
                                 destX, destY, destZ,
                                 srcX, srcY, width, height);
    }
-
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
 }
 
 
diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c
index 8514b6b375..f0eddafd33 100644
--- a/src/mesa/state_tracker/st_context.c
+++ b/src/mesa/state_tracker/st_context.c
@@ -36,6 +36,7 @@
 #include "shader/shader_api.h"
 #include "glapi/glapi.h"
 #include "st_public.h"
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_cb_accum.h"
 #include "st_cb_bitmap.h"
@@ -113,6 +114,9 @@ st_create_context_priv( GLcontext *ctx, struct pipe_context *pipe )
    st->ctx = ctx;
    st->pipe = pipe;
 
+   /* XXX: this is one-off, per-screen init: */
+   st_debug_init();
+   
    /* state tracker needs the VBO module */
    _vbo_CreateContext(ctx);
 
@@ -332,7 +336,7 @@ void st_init_driver_functions(struct dd_function_table *functions)
    st_init_feedback_functions(functions);
 #endif
    st_init_program_functions(functions);
-#if FEATURE_ARB_occlusion_query
+#if FEATURE_queryobj
    st_init_query_functions(functions);
 #endif
    st_init_readpixels_functions(functions);
diff --git a/src/mesa/state_tracker/st_debug.c b/src/mesa/state_tracker/st_debug.c
index c7d26ce33c..3009cde9d5 100644
--- a/src/mesa/state_tracker/st_debug.c
+++ b/src/mesa/state_tracker/st_debug.c
@@ -41,6 +41,32 @@
 
 
 
+#ifdef DEBUG
+int ST_DEBUG = 0;
+
+static const struct debug_named_value st_debug_flags[] = {
+   { "mesa",     DEBUG_MESA },
+   { "tgsi",     DEBUG_TGSI },
+   { "pipe",     DEBUG_PIPE },
+   { "tex",      DEBUG_TEX },
+   { "fallback", DEBUG_FALLBACK },
+   { "screen",   DEBUG_SCREEN },
+   { "query",    DEBUG_QUERY },
+   {NULL, 0}
+};
+#endif
+
+
+void
+st_debug_init(void)
+{
+#ifdef DEBUG
+   ST_DEBUG = debug_get_flags_option("ST_DEBUG", st_debug_flags, 0 );
+#endif
+}
+
+
+
 /**
  * Print current state.  May be called from inside gdb to see currently
  * bound vertex/fragment shaders and associated constants.
@@ -68,3 +94,5 @@ st_print_current(void)
    if (st->fp->Base.Base.Parameters)
       _mesa_print_parameter_list(st->fp->Base.Base.Parameters);
 }
+
+
diff --git a/src/mesa/state_tracker/st_debug.h b/src/mesa/state_tracker/st_debug.h
index 49d752e1b2..4a060d7759 100644
--- a/src/mesa/state_tracker/st_debug.h
+++ b/src/mesa/state_tracker/st_debug.h
@@ -29,8 +29,44 @@
 #ifndef ST_DEBUG_H
 #define ST_DEBUG_H
 
+#include "pipe/p_compiler.h"
+#include "util/u_debug.h"
+
 extern void
 st_print_current(void);
 
 
+#define DEBUG_MESA      0x1
+#define DEBUG_TGSI      0x2
+#define DEBUG_CONSTANTS 0x4
+#define DEBUG_PIPE      0x8
+#define DEBUG_TEX       0x10
+#define DEBUG_FALLBACK  0x20
+#define DEBUG_QUERY     0x40
+#define DEBUG_SCREEN    0x80
+
+#ifdef DEBUG
+extern int ST_DEBUG;
+#define DBSTR(x) x
+#else
+#define ST_DEBUG 0
+#define DBSTR(x) ""
+#endif
+
+void st_debug_init( void );
+
+static INLINE void
+ST_DBG( unsigned flag, const char *fmt, ... )
+{
+    if (ST_DEBUG & flag)
+    {
+        va_list args;
+
+        va_start( args, fmt );
+        debug_vprintf( fmt, args );
+        va_end( args );
+    }
+}
+
+
 #endif /* ST_DEBUG_H */
diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c
index dcb90a3107..3e0db37414 100644
--- a/src/mesa/state_tracker/st_format.c
+++ b/src/mesa/state_tracker/st_format.c
@@ -392,10 +392,9 @@ default_depth_format(struct pipe_screen *screen,
  *                   or PIPE_TEXTURE_USAGE_SAMPLER
  */
 enum pipe_format
-st_choose_format(struct pipe_context *pipe, GLenum internalFormat,
+st_choose_format(struct pipe_screen *screen, GLenum internalFormat,
                  enum pipe_texture_target target, unsigned tex_usage)
 {
-   struct pipe_screen *screen = pipe->screen;
    unsigned geom_flags = 0;
 
    switch (internalFormat) {
@@ -618,14 +617,15 @@ is_depth_or_stencil_format(GLenum internalFormat)
  * Called by FBO code to choose a PIPE_FORMAT_ for drawing surfaces.
  */
 enum pipe_format
-st_choose_renderbuffer_format(struct pipe_context *pipe, GLenum internalFormat)
+st_choose_renderbuffer_format(struct pipe_screen *screen,
+                              GLenum internalFormat)
 {
    uint usage;
    if (is_depth_or_stencil_format(internalFormat))
       usage = PIPE_TEXTURE_USAGE_DEPTH_STENCIL;
    else
       usage = PIPE_TEXTURE_USAGE_RENDER_TARGET;
-   return st_choose_format(pipe, internalFormat, PIPE_TEXTURE_2D, usage);
+   return st_choose_format(screen, internalFormat, PIPE_TEXTURE_2D, usage);
 }
 
 
@@ -713,8 +713,8 @@ st_ChooseTextureFormat(GLcontext *ctx, GLint internalFormat,
    (void) format;
    (void) type;
 
-   pFormat = st_choose_format(ctx->st->pipe, internalFormat, PIPE_TEXTURE_2D, 
-                              PIPE_TEXTURE_USAGE_SAMPLER);
+   pFormat = st_choose_format(ctx->st->pipe->screen, internalFormat,
+                              PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_SAMPLER);
    if (pFormat == PIPE_FORMAT_NONE)
       return NULL;
 
diff --git a/src/mesa/state_tracker/st_format.h b/src/mesa/state_tracker/st_format.h
index 9d9e02fe9b..e4a788c89b 100644
--- a/src/mesa/state_tracker/st_format.h
+++ b/src/mesa/state_tracker/st_format.h
@@ -64,11 +64,12 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat);
 
 
 extern enum pipe_format
-st_choose_format(struct pipe_context *pipe, GLenum internalFormat,
+st_choose_format(struct pipe_screen *screen, GLenum internalFormat,
                  enum pipe_texture_target target, unsigned tex_usage);
 
 extern enum pipe_format
-st_choose_renderbuffer_format(struct pipe_context *pipe, GLenum internalFormat);
+st_choose_renderbuffer_format(struct pipe_screen *screen,
+                              GLenum internalFormat);
 
 
 extern const struct gl_texture_format *
diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c
index f75b2348b8..16ca2771b0 100644
--- a/src/mesa/state_tracker/st_gen_mipmap.c
+++ b/src/mesa/state_tracker/st_gen_mipmap.c
@@ -42,6 +42,7 @@
 #include "cso_cache/cso_cache.h"
 #include "cso_cache/cso_context.h"
 
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_draw.h"
 #include "st_gen_mipmap.h"
@@ -113,6 +114,9 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target,
    uint dstLevel;
    GLenum datatype;
    GLuint comps;
+   
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
 
    assert(target != GL_TEXTURE_3D); /* not done yet */
 
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 2ab12d3cf3..70d7c4fee2 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -27,836 +27,649 @@
 
 /*
  * \author
- * Michal Krol
+ * Michal Krol,
+ * Keith Whitwell
  */
 
 #include "pipe/p_compiler.h"
 #include "pipe/p_shader_tokens.h"
-#include "tgsi/tgsi_parse.h"
-#include "tgsi/tgsi_build.h"
-#include "tgsi/tgsi_util.h"
-#include "tgsi/tgsi_dump.h"
-#include "tgsi/tgsi_sanity.h"
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_ureg.h"
 #include "st_mesa_to_tgsi.h"
 #include "shader/prog_instruction.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_print.h"
 #include "util/u_debug.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+struct label {
+   unsigned branch_target;
+   unsigned token;
+};
+
+struct st_translate {
+   struct ureg_program *ureg;
+
+   struct ureg_dst temps[MAX_PROGRAM_TEMPS];
+   struct ureg_src *constants;
+   struct ureg_dst outputs[PIPE_MAX_SHADER_OUTPUTS];
+   struct ureg_src inputs[PIPE_MAX_SHADER_INPUTS];
+   struct ureg_dst address[1];
+   struct ureg_src samplers[PIPE_MAX_SAMPLERS];
+
+   const GLuint *inputMapping;
+   const GLuint *outputMapping;
+
+   /* For every instruction that contains a label (eg CALL), keep
+    * details so that we can go back afterwards and emit the correct
+    * tgsi instruction number for each label.
+    */
+   struct label *labels;
+   unsigned labels_size;
+   unsigned labels_count;
+
+   /* Keep a record of the tgsi instruction number that each mesa
+    * instruction starts at, will be used to fix up labels after
+    * translation.
+    */
+   unsigned *insn;
+   unsigned insn_size;
+   unsigned insn_count;
+
+   unsigned procType;  /**< TGSI_PROCESSOR_VERTEX/FRAGMENT */
+
+   boolean error;
+};
+
+
+static unsigned *get_label( struct st_translate *t,
+                            unsigned branch_target )
+{
+   unsigned i;
+
+   if (t->labels_count + 1 >= t->labels_size) {
+      unsigned old_size = t->labels_size;
+      t->labels_size = 1 << (util_logbase2(t->labels_size) + 1);
+      t->labels = REALLOC( t->labels, 
+                           old_size * sizeof t->labels[0], 
+                           t->labels_size * sizeof t->labels[0] );
+      if (t->labels == NULL) {
+         static unsigned dummy;
+         t->error = TRUE;
+         return &dummy;
+      }
+   }
+
+   i = t->labels_count++;
+   t->labels[i].branch_target = branch_target;
+   return &t->labels[i].token;
+}
+
+
+static void set_insn_start( struct st_translate *t,
+                            unsigned start )
+{
+   if (t->insn_count + 1 >= t->insn_size) {
+      unsigned old_size = t->insn_size;
+      t->insn_size = 1 << (util_logbase2(t->insn_size) + 1);
+      t->insn = REALLOC( t->insn, 
+                         old_size * sizeof t->insn[0], 
+                         t->insn_size * sizeof t->insn[0] );
+      if (t->insn == NULL) {
+         t->error = TRUE;
+         return;
+      }
+   }
+
+   t->insn[t->insn_count++] = start;
+}
+
 
 /*
  * Map mesa register file to TGSI register file.
  */
-static GLuint
-map_register_file(
-   gl_register_file file,
-   GLuint index,
-   const GLuint immediateMapping[],
-   GLboolean indirectAccess )
+static struct ureg_dst
+dst_register( struct st_translate *t,
+              gl_register_file file,
+              GLuint index )
 {
    switch( file ) {
    case PROGRAM_UNDEFINED:
-      return TGSI_FILE_NULL;
+      return ureg_dst_undef();
+
    case PROGRAM_TEMPORARY:
-      return TGSI_FILE_TEMPORARY;
-   /*case PROGRAM_LOCAL_PARAM:*/
-   /*case PROGRAM_ENV_PARAM:*/
-
-      /* Because of the longstanding problem with mesa arb shaders
-       * where constants, immediates and state variables are all
-       * bundled together as PROGRAM_STATE_VAR, we can't tell from the
-       * mesa register file whether this is a CONSTANT or an
-       * IMMEDIATE, hence we need all the other information.
-       */
-   case PROGRAM_STATE_VAR:
-   case PROGRAM_NAMED_PARAM:
-   case PROGRAM_UNIFORM:
-      if (!indirectAccess && immediateMapping && immediateMapping[index] != ~0)
-         return TGSI_FILE_IMMEDIATE;
-      else
-	 return TGSI_FILE_CONSTANT;
-   case PROGRAM_CONSTANT:
-      if (indirectAccess)
-         return TGSI_FILE_CONSTANT;
-      assert(immediateMapping[index] != ~0);
-      return TGSI_FILE_IMMEDIATE;
-   case PROGRAM_INPUT:
-      return TGSI_FILE_INPUT;
+      if (ureg_dst_is_undef(t->temps[index]))
+         t->temps[index] = ureg_DECL_temporary( t->ureg );
+
+      return t->temps[index];
+
    case PROGRAM_OUTPUT:
-      return TGSI_FILE_OUTPUT;
+      return t->outputs[t->outputMapping[index]];
+
    case PROGRAM_ADDRESS:
-      return TGSI_FILE_ADDRESS;
+      return t->address[index];
+
    default:
-      assert( 0 );
-      return TGSI_FILE_NULL;
+      debug_assert( 0 );
+      return ureg_dst_undef();
    }
 }
 
-/**
- * Map mesa register file index to TGSI index.
- * Take special care when processing input and output indices.
- * \param file  one of TGSI_FILE_x
- * \param index  the mesa register file index
- * \param inputMapping  maps Mesa input indexes to TGSI input indexes
- * \param outputMapping  maps Mesa output indexes to TGSI output indexes
- */
-static GLuint
-map_register_file_index(
-   GLuint procType,
-   GLuint file,
-   GLuint index,
-   GLuint *swizzle,
-   const GLuint inputMapping[],
-   const GLuint outputMapping[],
-   const GLuint immediateMapping[],
-   GLboolean indirectAccess )
+
+static struct ureg_src
+src_register( struct st_translate *t,
+              gl_register_file file,
+              GLuint index )
 {
    switch( file ) {
-   case TGSI_FILE_INPUT:
-      /* inputs are mapped according to the user-defined map */
-      return inputMapping[index];
+   case PROGRAM_UNDEFINED:
+      return ureg_src_undef();
+
+   case PROGRAM_TEMPORARY:
+      if (ureg_dst_is_undef(t->temps[index]))
+         t->temps[index] = ureg_DECL_temporary( t->ureg );
+      return ureg_src(t->temps[index]);
+
+   case PROGRAM_STATE_VAR:
+   case PROGRAM_NAMED_PARAM:
+   case PROGRAM_ENV_PARAM:
+   case PROGRAM_UNIFORM:
+   case PROGRAM_CONSTANT:       /* ie, immediate */
+      return t->constants[index];
 
-   case TGSI_FILE_OUTPUT:
-      return outputMapping[index];
+   case PROGRAM_INPUT:
+      return t->inputs[t->inputMapping[index]];
 
-   case TGSI_FILE_IMMEDIATE:
-      if (indirectAccess)
-         return index;
-      assert(immediateMapping[index] != ~0);
-      return immediateMapping[index];
+   case PROGRAM_OUTPUT:
+      return ureg_src(t->outputs[t->outputMapping[index]]); /* not needed? */
+
+   case PROGRAM_ADDRESS:
+      return ureg_src(t->address[index]);
 
    default:
-      return index;
+      debug_assert( 0 );
+      return ureg_src_undef();
    }
 }
 
-/*
+
+/**
  * Map mesa texture target to TGSI texture target.
  */
-static GLuint
-map_texture_target(
-    GLuint textarget,
-    GLboolean shadow )
+static unsigned
+translate_texture_target( GLuint textarget,
+                          GLboolean shadow )
 {
-   switch( textarget ) {
-   case TEXTURE_1D_INDEX:
-      if (shadow)
-         return TGSI_TEXTURE_SHADOW1D;
-      else
-         return TGSI_TEXTURE_1D;
-   case TEXTURE_2D_INDEX:
-      if (shadow)
-         return TGSI_TEXTURE_SHADOW2D;
-      else
-         return TGSI_TEXTURE_2D;
-   case TEXTURE_3D_INDEX:
-      return TGSI_TEXTURE_3D;
-   case TEXTURE_CUBE_INDEX:
-      return TGSI_TEXTURE_CUBE;
-   case TEXTURE_RECT_INDEX:
-      if (shadow)
-         return TGSI_TEXTURE_SHADOWRECT;
-      else
-         return TGSI_TEXTURE_RECT;
-   default:
-      assert( 0 );
+   if (shadow) {
+      switch( textarget ) {
+      case TEXTURE_1D_INDEX:   return TGSI_TEXTURE_SHADOW1D;
+      case TEXTURE_2D_INDEX:   return TGSI_TEXTURE_SHADOW2D;
+      case TEXTURE_RECT_INDEX: return TGSI_TEXTURE_SHADOWRECT;
+      default: break;
+      }
    }
 
-   return TGSI_TEXTURE_1D;
-}
-
-static GLuint
-convert_sat(
-   GLuint sat )
-{
-   switch( sat ) {
-   case SATURATE_OFF:
-      return TGSI_SAT_NONE;
-   case SATURATE_ZERO_ONE:
-      return TGSI_SAT_ZERO_ONE;
-   case SATURATE_PLUS_MINUS_ONE:
-      return TGSI_SAT_MINUS_PLUS_ONE;
+   switch( textarget ) {
+   case TEXTURE_1D_INDEX:   return TGSI_TEXTURE_1D;
+   case TEXTURE_2D_INDEX:   return TGSI_TEXTURE_2D;
+   case TEXTURE_3D_INDEX:   return TGSI_TEXTURE_3D;
+   case TEXTURE_CUBE_INDEX: return TGSI_TEXTURE_CUBE;
+   case TEXTURE_RECT_INDEX: return TGSI_TEXTURE_RECT;
    default:
-      assert( 0 );
-      return TGSI_SAT_NONE;
+      debug_assert( 0 );
+      return TGSI_TEXTURE_1D;
    }
 }
 
-static GLuint
-convert_writemask(
-   GLuint writemask )
+
+static struct ureg_dst
+translate_dst( struct st_translate *t,
+               const struct prog_dst_register *DstReg,
+               boolean saturate )
 {
-   assert( WRITEMASK_X == TGSI_WRITEMASK_X );
-   assert( WRITEMASK_Y == TGSI_WRITEMASK_Y );
-   assert( WRITEMASK_Z == TGSI_WRITEMASK_Z );
-   assert( WRITEMASK_W == TGSI_WRITEMASK_W );
-   assert( (writemask & ~TGSI_WRITEMASK_XYZW) == 0 );
+   struct ureg_dst dst = dst_register( t, 
+                                       DstReg->File,
+                                       DstReg->Index );
+
+   dst = ureg_writemask( dst, 
+                         DstReg->WriteMask );
+   
+   if (saturate)
+      dst = ureg_saturate( dst );
 
-   return writemask;
+   if (DstReg->RelAddr)
+      dst = ureg_dst_indirect( dst, ureg_src(t->address[0]) );
+
+   return dst;
 }
 
-static struct tgsi_full_immediate
-make_immediate(const float *value, uint size)
+
+static struct ureg_src
+translate_src( struct st_translate *t,
+               const struct prog_src_register *SrcReg )
 {
-   struct tgsi_full_immediate imm;
-   unsigned i;
+   struct ureg_src src = src_register( t, SrcReg->File, SrcReg->Index );
+
+   src = ureg_swizzle( src,
+                       GET_SWZ( SrcReg->Swizzle, 0 ) & 0x3,
+                       GET_SWZ( SrcReg->Swizzle, 1 ) & 0x3,
+                       GET_SWZ( SrcReg->Swizzle, 2 ) & 0x3,
+                       GET_SWZ( SrcReg->Swizzle, 3 ) & 0x3);
 
-   imm = tgsi_default_full_immediate();
-   imm.Immediate.NrTokens += size;
-   imm.Immediate.DataType = TGSI_IMM_FLOAT32;
+   if (SrcReg->Negate == NEGATE_XYZW)
+      src = ureg_negate(src);
 
-   for (i = 0; i < size; i++)
-      imm.u[i].Float = value[i];
+   if (SrcReg->Abs) 
+      src = ureg_abs(src);
 
-   return imm;
+   if (SrcReg->RelAddr) 
+      src = ureg_src_indirect( src, ureg_src(t->address[0]));
+   
+   return src;
 }
 
-static void
-compile_instruction(
-   const struct prog_instruction *inst,
-   struct tgsi_full_instruction *fullinst,
-   const GLuint inputMapping[],
-   const GLuint outputMapping[],
-   const GLuint immediateMapping[],
-   GLboolean indirectAccess,
-   GLuint preamble_size,
-   GLuint procType,
-   GLboolean *insideSubroutine,
-   GLint wposTemp)
+
+static struct ureg_src swizzle_4v( struct ureg_src src,
+                                   const unsigned *swz )
 {
-   GLuint i;
-   struct tgsi_full_dst_register *fulldst;
-   struct tgsi_full_src_register *fullsrc;
-
-   *fullinst = tgsi_default_full_instruction();
-
-   fullinst->Instruction.Saturate = convert_sat( inst->SaturateMode );
-   fullinst->Instruction.NumDstRegs = _mesa_num_inst_dst_regs( inst->Opcode );
-   fullinst->Instruction.NumSrcRegs = _mesa_num_inst_src_regs( inst->Opcode );
-
-   fulldst = &fullinst->FullDstRegisters[0];
-   fulldst->DstRegister.File = map_register_file( inst->DstReg.File, 0, NULL, GL_FALSE );
-   fulldst->DstRegister.Index = map_register_file_index(
-      procType,
-      fulldst->DstRegister.File,
-      inst->DstReg.Index,
-      NULL,
-      inputMapping,
-      outputMapping,
-      NULL,
-      GL_FALSE );
-   fulldst->DstRegister.WriteMask = convert_writemask( inst->DstReg.WriteMask );
-   if (inst->DstReg.RelAddr) {
-      fulldst->DstRegister.Indirect = 1;
-      fulldst->DstRegisterInd.File = TGSI_FILE_ADDRESS;
-      fulldst->DstRegisterInd.Index = 0;
+   return ureg_swizzle( src, swz[0], swz[1], swz[2], swz[3] );
+}
+
+
+/**
+ * Translate SWZ instructions into a single MAD.  EG:
+ *
+ *   SWZ dst, src.x-y10 
+ * 
+ * becomes:
+ *
+ *   MAD dst {1,-1,0,0}, src.xyxx, {0,0,1,0}
+ */
+static void emit_swz( struct st_translate *t,
+                      struct ureg_dst dst,
+                      const struct prog_src_register *SrcReg )
+{
+   struct ureg_program *ureg = t->ureg;
+   struct ureg_src src = src_register( t, SrcReg->File, SrcReg->Index );
+
+   unsigned negate_mask =  SrcReg->Negate;
+
+   unsigned one_mask = ((GET_SWZ(SrcReg->Swizzle, 0) == SWIZZLE_ONE) << 0 |
+                        (GET_SWZ(SrcReg->Swizzle, 1) == SWIZZLE_ONE) << 1 |
+                        (GET_SWZ(SrcReg->Swizzle, 2) == SWIZZLE_ONE) << 2 |
+                        (GET_SWZ(SrcReg->Swizzle, 3) == SWIZZLE_ONE) << 3);
+
+   unsigned zero_mask = ((GET_SWZ(SrcReg->Swizzle, 0) == SWIZZLE_ZERO) << 0 |
+                         (GET_SWZ(SrcReg->Swizzle, 1) == SWIZZLE_ZERO) << 1 |
+                         (GET_SWZ(SrcReg->Swizzle, 2) == SWIZZLE_ZERO) << 2 |
+                         (GET_SWZ(SrcReg->Swizzle, 3) == SWIZZLE_ZERO) << 3);
+
+   unsigned negative_one_mask = one_mask & negate_mask;
+   unsigned positive_one_mask = one_mask & ~negate_mask;
+   
+   struct ureg_src imm;
+   unsigned i;
+   unsigned mul_swizzle[4] = {0,0,0,0};
+   unsigned add_swizzle[4] = {0,0,0,0};
+   unsigned src_swizzle[4] = {0,0,0,0};
+   boolean need_add = FALSE;
+   boolean need_mul = FALSE;
+
+   if (dst.WriteMask == 0)
+      return;
+
+   /* Is this just a MOV?
+    */
+   if (zero_mask == 0 &&
+       one_mask == 0 &&
+       (negate_mask == 0 || negate_mask == TGSI_WRITEMASK_XYZW)) 
+   {
+      ureg_MOV( ureg, dst, translate_src( t, SrcReg ));
+      return;
    }
 
-   for (i = 0; i < fullinst->Instruction.NumSrcRegs; i++) {
-      GLuint j;
-      GLuint swizzle = inst->SrcReg[i].Swizzle;
+#define IMM_ZERO    0
+#define IMM_ONE     1
+#define IMM_NEG_ONE 2
 
-      fullsrc = &fullinst->FullSrcRegisters[i];
+   imm = ureg_imm3f( ureg, 0, 1, -1 );
 
-      if (procType == TGSI_PROCESSOR_FRAGMENT &&
-          inst->SrcReg[i].File == PROGRAM_INPUT &&
-          inst->SrcReg[i].Index == FRAG_ATTRIB_WPOS) {
-         /* special case of INPUT[WPOS] */
-         fullsrc->SrcRegister.File = TGSI_FILE_TEMPORARY;
-         fullsrc->SrcRegister.Index = wposTemp;
-      }
-      else {
-         /* any other src register */
-         fullsrc->SrcRegister.File = map_register_file(
-            inst->SrcReg[i].File,
-            inst->SrcReg[i].Index,
-            immediateMapping,
-            indirectAccess );
-         fullsrc->SrcRegister.Index = map_register_file_index(
-            procType,
-            fullsrc->SrcRegister.File,
-            inst->SrcReg[i].Index,
-            &swizzle,
-            inputMapping,
-            outputMapping,
-            immediateMapping,
-            indirectAccess );
-      }
+   for (i = 0; i < 4; i++) {
+      unsigned bit = 1 << i;
 
-      /* swizzle (ext swizzle also depends on negation) */
-      {
-         GLuint swz[4];
-         GLboolean extended = (inst->SrcReg[i].Negate != NEGATE_NONE &&
-                               inst->SrcReg[i].Negate != NEGATE_XYZW);
-         for( j = 0; j < 4; j++ ) {
-            swz[j] = GET_SWZ( swizzle, j );
-            if (swz[j] > SWIZZLE_W)
-               extended = GL_TRUE;
+      if (dst.WriteMask & bit) {
+         if (positive_one_mask & bit) {
+            mul_swizzle[i] = IMM_ZERO;
+            add_swizzle[i] = IMM_ONE;
+            need_add = TRUE;
          }
-         if (extended) {
-            for (j = 0; j < 4; j++) {
-               tgsi_util_set_src_register_extswizzle(&fullsrc->SrcRegisterExtSwz,
-                                                     swz[j], j);
-            }
+         else if (negative_one_mask & bit) {
+            mul_swizzle[i] = IMM_ZERO;
+            add_swizzle[i] = IMM_NEG_ONE;
+            need_add = TRUE;
+         }
+         else if (zero_mask & bit) {
+            mul_swizzle[i] = IMM_ZERO;
+            add_swizzle[i] = IMM_ZERO;
+            need_add = TRUE;
          }
          else {
-            for (j = 0; j < 4; j++) {
-               tgsi_util_set_src_register_swizzle(&fullsrc->SrcRegister,
-                                                  swz[j], j);
+            add_swizzle[i] = IMM_ZERO;
+            src_swizzle[i] = GET_SWZ(SrcReg->Swizzle, i);
+            need_mul = TRUE;
+            if (negate_mask & bit) {
+               mul_swizzle[i] = IMM_NEG_ONE;
+            }
+            else {
+               mul_swizzle[i] = IMM_ONE;
             }
          }
       }
+   }
 
-      if( inst->SrcReg[i].Negate == NEGATE_XYZW ) {
-         fullsrc->SrcRegister.Negate = 1;
-      }
-      else if( inst->SrcReg[i].Negate != NEGATE_NONE ) {
-         if( inst->SrcReg[i].Negate & NEGATE_X ) {
-            fullsrc->SrcRegisterExtSwz.NegateX = 1;
-         }
-         if( inst->SrcReg[i].Negate & NEGATE_Y ) {
-            fullsrc->SrcRegisterExtSwz.NegateY = 1;
-         }
-         if( inst->SrcReg[i].Negate & NEGATE_Z ) {
-            fullsrc->SrcRegisterExtSwz.NegateZ = 1;
-         }
-         if( inst->SrcReg[i].Negate & NEGATE_W ) {
-            fullsrc->SrcRegisterExtSwz.NegateW = 1;
-         }
-      }
+   if (need_mul && need_add) {
+      ureg_MAD( ureg, 
+                dst,
+                swizzle_4v( src, src_swizzle ),
+                swizzle_4v( imm, mul_swizzle ),
+                swizzle_4v( imm, add_swizzle ) );
+   }
+   else if (need_mul) {
+      ureg_MUL( ureg, 
+                dst,
+                swizzle_4v( src, src_swizzle ),
+                swizzle_4v( imm, mul_swizzle ) );
+   }
+   else if (need_add) {
+      ureg_MOV( ureg, 
+                dst,
+                swizzle_4v( imm, add_swizzle ) );
+   }
+   else {
+      debug_assert(0);
+   }
 
-      if( inst->SrcReg[i].Abs ) {
-         fullsrc->SrcRegisterExtMod.Absolute = 1;
-      }
+#undef IMM_ZERO
+#undef IMM_ONE
+#undef IMM_NEG_ONE
+}
 
-      if( inst->SrcReg[i].RelAddr ) {
-         fullsrc->SrcRegister.Indirect = 1;
 
-         fullsrc->SrcRegisterInd.File = TGSI_FILE_ADDRESS;
-         fullsrc->SrcRegisterInd.Index = 0;
-      }
-   }
 
-   switch( inst->Opcode ) {
+static unsigned
+translate_opcode( unsigned op )
+{
+   switch( op ) {
    case OPCODE_ARL:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ARL;
-      break;
+      return TGSI_OPCODE_ARL;
    case OPCODE_ABS:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ABS;
-      break;
+      return TGSI_OPCODE_ABS;
    case OPCODE_ADD:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ADD;
-      break;
+      return TGSI_OPCODE_ADD;
    case OPCODE_BGNLOOP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_BGNLOOP;
-      fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
-      break;
+      return TGSI_OPCODE_BGNLOOP;
    case OPCODE_BGNSUB:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_BGNSUB;
-      *insideSubroutine = GL_TRUE;
-      break;
+      return TGSI_OPCODE_BGNSUB;
    case OPCODE_BRA:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_BRA;
-      break;
+      return TGSI_OPCODE_BRA;
    case OPCODE_BRK:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_BRK;
-      break;
+      return TGSI_OPCODE_BRK;
    case OPCODE_CAL:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_CAL;
-      fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
-      break;
+      return TGSI_OPCODE_CAL;
    case OPCODE_CMP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_CMP;
-      break;
+      return TGSI_OPCODE_CMP;
    case OPCODE_CONT:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_CONT;
-      break;
+      return TGSI_OPCODE_CONT;
    case OPCODE_COS:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_COS;
-      break;
+      return TGSI_OPCODE_COS;
    case OPCODE_DDX:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DDX;
-      break;
+      return TGSI_OPCODE_DDX;
    case OPCODE_DDY:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DDY;
-      break;
+      return TGSI_OPCODE_DDY;
    case OPCODE_DP2:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DP2;
-      break;
+      return TGSI_OPCODE_DP2;
    case OPCODE_DP2A:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DP2A;
-      break;
+      return TGSI_OPCODE_DP2A;
    case OPCODE_DP3:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DP3;
-      break;
+      return TGSI_OPCODE_DP3;
    case OPCODE_DP4:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DP4;
-      break;
+      return TGSI_OPCODE_DP4;
    case OPCODE_DPH:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DPH;
-      break;
+      return TGSI_OPCODE_DPH;
    case OPCODE_DST:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_DST;
-      break;
+      return TGSI_OPCODE_DST;
    case OPCODE_ELSE:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ELSE;
-      fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
-      break;
+      return TGSI_OPCODE_ELSE;
    case OPCODE_ENDIF:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ENDIF;
-      break;
+      return TGSI_OPCODE_ENDIF;
    case OPCODE_ENDLOOP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ENDLOOP;
-      fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
-      break;
+      return TGSI_OPCODE_ENDLOOP;
    case OPCODE_ENDSUB:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ENDSUB;
-      *insideSubroutine = GL_FALSE;
-      break;
+      return TGSI_OPCODE_ENDSUB;
    case OPCODE_EX2:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_EX2;
-      break;
+      return TGSI_OPCODE_EX2;
    case OPCODE_EXP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_EXP;
-      break;
+      return TGSI_OPCODE_EXP;
    case OPCODE_FLR:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_FLR;
-      break;
+      return TGSI_OPCODE_FLR;
    case OPCODE_FRC:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_FRC;
-      break;
+      return TGSI_OPCODE_FRC;
    case OPCODE_IF:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_IF;
-      fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
-      break;
+      return TGSI_OPCODE_IF;
    case OPCODE_TRUNC:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_TRUNC;
-      break;
+      return TGSI_OPCODE_TRUNC;
    case OPCODE_KIL:
-      /* conditional */
-      fullinst->Instruction.Opcode = TGSI_OPCODE_KIL;
-      break;
+      return TGSI_OPCODE_KIL;
    case OPCODE_KIL_NV:
-      /* predicated */
-      assert(inst->DstReg.CondMask == COND_TR);
-      fullinst->Instruction.Opcode = TGSI_OPCODE_KILP;
-      break;
+      return TGSI_OPCODE_KILP;
    case OPCODE_LG2:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_LG2;
-      break;
+      return TGSI_OPCODE_LG2;
    case OPCODE_LOG:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_LOG;
-      break;
+      return TGSI_OPCODE_LOG;
    case OPCODE_LIT:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_LIT;
-      break;
+      return TGSI_OPCODE_LIT;
    case OPCODE_LRP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_LRP;
-      break;
+      return TGSI_OPCODE_LRP;
    case OPCODE_MAD:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_MAD;
-      break;
+      return TGSI_OPCODE_MAD;
    case OPCODE_MAX:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_MAX;
-      break;
+      return TGSI_OPCODE_MAX;
    case OPCODE_MIN:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_MIN;
-      break;
+      return TGSI_OPCODE_MIN;
    case OPCODE_MOV:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_MOV;
-      break;
+      return TGSI_OPCODE_MOV;
    case OPCODE_MUL:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_MUL;
-      break;
+      return TGSI_OPCODE_MUL;
    case OPCODE_NOISE1:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE1;
-      break;
+      return TGSI_OPCODE_NOISE1;
    case OPCODE_NOISE2:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE2;
-      break;
+      return TGSI_OPCODE_NOISE2;
    case OPCODE_NOISE3:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE3;
-      break;
+      return TGSI_OPCODE_NOISE3;
    case OPCODE_NOISE4:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NOISE4;
-      break;
+      return TGSI_OPCODE_NOISE4;
    case OPCODE_NOP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NOP;
-      break;
+      return TGSI_OPCODE_NOP;
    case OPCODE_NRM3:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NRM;
-      break;
+      return TGSI_OPCODE_NRM;
    case OPCODE_NRM4:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_NRM4;
-      break;
+      return TGSI_OPCODE_NRM4;
    case OPCODE_POW:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_POW;
-      break;
+      return TGSI_OPCODE_POW;
    case OPCODE_RCP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_RCP;
-      break;
+      return TGSI_OPCODE_RCP;
    case OPCODE_RET:
-      /* If RET is used inside main (not a real subroutine) we may want
-       * to execute END instead of RET.  TBD...
-       */
-      if (1 /*  *insideSubroutine */) {
-         fullinst->Instruction.Opcode = TGSI_OPCODE_RET;
-      }
-      else {
-         /* inside main() pseudo-function */
-         fullinst->Instruction.Opcode = TGSI_OPCODE_END;
-      }
-      break;
+      return TGSI_OPCODE_RET;
    case OPCODE_RSQ:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_RSQ;
-      break;
+      return TGSI_OPCODE_RSQ;
    case OPCODE_SCS:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SCS;
-      fulldst->DstRegister.WriteMask &= TGSI_WRITEMASK_XY;
-      break;
+      return TGSI_OPCODE_SCS;
    case OPCODE_SEQ:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SEQ;
-      break;
+      return TGSI_OPCODE_SEQ;
    case OPCODE_SGE:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SGE;
-      break;
+      return TGSI_OPCODE_SGE;
    case OPCODE_SGT:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SGT;
-      break;
+      return TGSI_OPCODE_SGT;
    case OPCODE_SIN:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SIN;
-      break;
+      return TGSI_OPCODE_SIN;
    case OPCODE_SLE:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SLE;
-      break;
+      return TGSI_OPCODE_SLE;
    case OPCODE_SLT:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SLT;
-      break;
+      return TGSI_OPCODE_SLT;
    case OPCODE_SNE:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SNE;
-      break;
+      return TGSI_OPCODE_SNE;
    case OPCODE_SSG:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SSG;
-      break;
+      return TGSI_OPCODE_SSG;
    case OPCODE_SUB:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SUB;
-      break;
+      return TGSI_OPCODE_SUB;
    case OPCODE_SWZ:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_SWZ;
-      break;
+      return TGSI_OPCODE_SWZ;
    case OPCODE_TEX:
-      /* ordinary texture lookup */
-      fullinst->Instruction.Opcode = TGSI_OPCODE_TEX;
-      fullinst->Instruction.NumSrcRegs = 2;
-      fullinst->InstructionExtTexture.Texture =
-         map_texture_target( inst->TexSrcTarget, inst->TexShadow );
-      fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
-      fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit;
-      break;
+      return TGSI_OPCODE_TEX;
    case OPCODE_TXB:
-      /* texture lookup with LOD bias */
-      fullinst->Instruction.Opcode = TGSI_OPCODE_TXB;
-      fullinst->Instruction.NumSrcRegs = 2;
-      fullinst->InstructionExtTexture.Texture =
-         map_texture_target( inst->TexSrcTarget, inst->TexShadow );
-      fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
-      fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit;
-      break;
+      return TGSI_OPCODE_TXB;
    case OPCODE_TXD:
-      /* texture lookup with explicit partial derivatives */
-      fullinst->Instruction.Opcode = TGSI_OPCODE_TXD;
-      fullinst->Instruction.NumSrcRegs = 4;
-      fullinst->InstructionExtTexture.Texture =
-         map_texture_target( inst->TexSrcTarget, inst->TexShadow );
-      /* src[0] = coord, src[1] = d[strq]/dx, src[2] = d[strq]/dy */
-      fullinst->FullSrcRegisters[3].SrcRegister.File = TGSI_FILE_SAMPLER;
-      fullinst->FullSrcRegisters[3].SrcRegister.Index = inst->TexSrcUnit;
-      break;
+      return TGSI_OPCODE_TXD;
    case OPCODE_TXL:
-      /* texture lookup with explicit LOD */
-      fullinst->Instruction.Opcode = TGSI_OPCODE_TXL;
-      fullinst->Instruction.NumSrcRegs = 2;
-      fullinst->InstructionExtTexture.Texture =
-         map_texture_target( inst->TexSrcTarget, inst->TexShadow );
-      fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
-      fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit;
-      break;
+      return TGSI_OPCODE_TXL;
    case OPCODE_TXP:
-      /* texture lookup with divide by Q component */
-      /* convert to TEX w/ special flag for division */
-      fullinst->Instruction.Opcode = TGSI_OPCODE_TXP;
-      fullinst->Instruction.NumSrcRegs = 2;
-      fullinst->InstructionExtTexture.Texture =
-         map_texture_target( inst->TexSrcTarget, inst->TexShadow );
-      fullinst->FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
-      fullinst->FullSrcRegisters[1].SrcRegister.Index = inst->TexSrcUnit;
-      break;
+      return TGSI_OPCODE_TXP;
    case OPCODE_XPD:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_XPD;
-      fulldst->DstRegister.WriteMask &= TGSI_WRITEMASK_XYZ;
-      break;
+      return TGSI_OPCODE_XPD;
    case OPCODE_END:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_END;
-      break;
+      return TGSI_OPCODE_END;
    default:
-      assert( 0 );
+      debug_assert( 0 );
+      return TGSI_OPCODE_NOP;
    }
 }
 
-/**
- * \param usage_mask  bitfield of TGSI_WRITEMASK_{XYZW} tokens
- */
-static struct tgsi_full_declaration
-make_input_decl(
-   GLuint index,
-   GLboolean interpolate_info,
-   GLuint interpolate,
-   GLuint usage_mask,
-   GLboolean semantic_info,
-   GLuint semantic_name,
-   GLbitfield semantic_index,
-   GLbitfield input_flags)
-{
-   struct tgsi_full_declaration decl;
-
-   assert(semantic_name < TGSI_SEMANTIC_COUNT);
-
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_INPUT;
-   decl.Declaration.UsageMask = usage_mask;
-   decl.Declaration.Semantic = semantic_info;
-   decl.DeclarationRange.First = index;
-   decl.DeclarationRange.Last = index;
-   if (semantic_info) {
-      decl.Semantic.SemanticName = semantic_name;
-      decl.Semantic.SemanticIndex = semantic_index;
-   }
-   if (interpolate_info) {
-      decl.Declaration.Interpolate = interpolate;
-   }
-   if (input_flags & PROG_PARAM_BIT_CENTROID)
-      decl.Declaration.Centroid = 1;
-   if (input_flags & PROG_PARAM_BIT_INVARIANT)
-      decl.Declaration.Invariant = 1;
-
-   return decl;
-}
-
-/**
- * \param usage_mask  bitfield of TGSI_WRITEMASK_{XYZW} tokens
- */
-static struct tgsi_full_declaration
-make_output_decl(
-   GLuint index,
-   GLuint semantic_name,
-   GLuint semantic_index,
-   GLuint usage_mask,
-   GLbitfield output_flags)
-{
-   struct tgsi_full_declaration decl;
-
-   assert(semantic_name < TGSI_SEMANTIC_COUNT);
-
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_OUTPUT;
-   decl.Declaration.UsageMask = usage_mask;
-   decl.Declaration.Semantic = 1;
-   decl.DeclarationRange.First = index;
-   decl.DeclarationRange.Last = index;
-   decl.Semantic.SemanticName = semantic_name;
-   decl.Semantic.SemanticIndex = semantic_index;
-   if (output_flags & PROG_PARAM_BIT_CENTROID)
-      decl.Declaration.Centroid = 1;
-   if (output_flags & PROG_PARAM_BIT_INVARIANT)
-      decl.Declaration.Invariant = 1;
-
-   return decl;
-}
-
 
-static struct tgsi_full_declaration
-make_temp_decl(
-   GLuint start_index,
-   GLuint end_index )
+static void
+compile_instruction(
+   struct st_translate *t,
+   const struct prog_instruction *inst )
 {
-   struct tgsi_full_declaration decl;
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_TEMPORARY;
-   decl.DeclarationRange.First = start_index;
-   decl.DeclarationRange.Last = end_index;
-   return decl;
-}
+   struct ureg_program *ureg = t->ureg;
+   GLuint i;
+   struct ureg_dst dst[1];
+   struct ureg_src src[4];
+   unsigned num_dst;
+   unsigned num_src;
 
-static struct tgsi_full_declaration
-make_addr_decl(
-   GLuint start_index,
-   GLuint end_index )
-{
-   struct tgsi_full_declaration decl;
+   num_dst = _mesa_num_inst_dst_regs( inst->Opcode );
+   num_src = _mesa_num_inst_src_regs( inst->Opcode );
 
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_ADDRESS;
-   decl.DeclarationRange.First = start_index;
-   decl.DeclarationRange.Last = end_index;
-   return decl;
-}
+   if (num_dst) 
+      dst[0] = translate_dst( t, 
+                              &inst->DstReg,
+                              inst->SaturateMode );
 
-static struct tgsi_full_declaration
-make_sampler_decl(GLuint index)
-{
-   struct tgsi_full_declaration decl;
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_SAMPLER;
-   decl.DeclarationRange.First = index;
-   decl.DeclarationRange.Last = index;
-   return decl;
-}
+   for (i = 0; i < num_src; i++) 
+      src[i] = translate_src( t, &inst->SrcReg[i] );
 
-/** Reference into a constant buffer */
-static struct tgsi_full_declaration
-make_constant_decl(GLuint first, GLuint last)
-{
-   struct tgsi_full_declaration decl;
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_CONSTANT;
-   decl.DeclarationRange.First = first;
-   decl.DeclarationRange.Last = last;
-   return decl;
-}
+   switch( inst->Opcode ) {
+   case OPCODE_SWZ:
+      emit_swz( t, dst[0], &inst->SrcReg[0] );
+      return;
 
+   case OPCODE_BGNLOOP:
+   case OPCODE_CAL:
+   case OPCODE_ELSE:
+   case OPCODE_ENDLOOP:
+   case OPCODE_IF:
+      debug_assert(num_dst == 0);
+      ureg_label_insn( ureg,
+                       translate_opcode( inst->Opcode ),
+                       src, num_src,
+                       get_label( t, inst->BranchTarget ));
+      return;
 
+   case OPCODE_TEX:
+   case OPCODE_TXB:
+   case OPCODE_TXD:
+   case OPCODE_TXL:
+   case OPCODE_TXP:
+      src[num_src++] = t->samplers[inst->TexSrcUnit];
+      ureg_tex_insn( ureg,
+                     translate_opcode( inst->Opcode ),
+                     dst, num_dst, 
+                     translate_texture_target( inst->TexSrcTarget,
+                                               inst->TexShadow ),
+                     src, num_src );
+      return;
 
-/**
- * Find the temporaries which are used in the given program.
- */
-static void
-find_temporaries(const struct gl_program *program,
-                 GLboolean tempsUsed[MAX_PROGRAM_TEMPS])
-{
-   GLuint i, j;
+   case OPCODE_SCS:
+      dst[0] = ureg_writemask(dst[0], TGSI_WRITEMASK_XY );
+      ureg_insn( ureg, 
+                 translate_opcode( inst->Opcode ), 
+                 dst, num_dst, 
+                 src, num_src );
+      break;
 
-   for (i = 0; i < MAX_PROGRAM_TEMPS; i++)
-      tempsUsed[i] = GL_FALSE;
+   case OPCODE_XPD:
+      dst[0] = ureg_writemask(dst[0], TGSI_WRITEMASK_XYZ );
+      ureg_insn( ureg, 
+                 translate_opcode( inst->Opcode ), 
+                 dst, num_dst, 
+                 src, num_src );
+      break;
 
-   for (i = 0; i < program->NumInstructions; i++) {
-      const struct prog_instruction *inst = program->Instructions + i;
-      const GLuint n = _mesa_num_inst_src_regs( inst->Opcode );
-      for (j = 0; j < n; j++) {
-         if (inst->SrcReg[j].File == PROGRAM_TEMPORARY)
-            tempsUsed[inst->SrcReg[j].Index] = GL_TRUE;
-         if (inst->DstReg.File == PROGRAM_TEMPORARY)
-            tempsUsed[inst->DstReg.Index] = GL_TRUE;
-      }
+   default:
+      ureg_insn( ureg, 
+                 translate_opcode( inst->Opcode ), 
+                 dst, num_dst, 
+                 src, num_src );
+      break;
    }
 }
 
 
 /**
- * Find an unused temporary in the tempsUsed array.
+ * Emit the TGSI instructions for inverting the WPOS y coordinate.
  */
-static int
-find_free_temporary(GLboolean tempsUsed[MAX_PROGRAM_TEMPS])
-{
-   int i;
-   for (i = 0; i < MAX_PROGRAM_TEMPS; i++) {
-      if (!tempsUsed[i]) {
-         tempsUsed[i] = GL_TRUE;
-         return i;
-      }
-   }
-   return -1;
-}
-
-
-/** helper for building simple TGSI instruction, one src register */
-static void
-build_tgsi_instruction1(struct tgsi_full_instruction *inst,
-                        int opcode,
-                        int dstFile, int dstIndex, int writemask,
-                        int srcFile1, int srcIndex1)
-{
-   *inst = tgsi_default_full_instruction();
-
-   inst->Instruction.Opcode = opcode;
-
-   inst->Instruction.NumDstRegs = 1;
-   inst->FullDstRegisters[0].DstRegister.File = dstFile;
-   inst->FullDstRegisters[0].DstRegister.Index = dstIndex;
-   inst->FullDstRegisters[0].DstRegister.WriteMask = writemask;
-
-   inst->Instruction.NumSrcRegs = 1;
-   inst->FullSrcRegisters[0].SrcRegister.File = srcFile1;
-   inst->FullSrcRegisters[0].SrcRegister.Index = srcIndex1;
-}
-
-
-/** helper for building simple TGSI instruction, two src registers */
 static void
-build_tgsi_instruction2(struct tgsi_full_instruction *inst,
-                        int opcode,
-                        int dstFile, int dstIndex, int writemask,
-                        int srcFile1, int srcIndex1,
-                        int srcFile2, int srcIndex2)
+emit_inverted_wpos( struct st_translate *t,
+                    const struct gl_program *program )
 {
-   *inst = tgsi_default_full_instruction();
+   struct ureg_program *ureg = t->ureg;
 
-   inst->Instruction.Opcode = opcode;
-
-   inst->Instruction.NumDstRegs = 1;
-   inst->FullDstRegisters[0].DstRegister.File = dstFile;
-   inst->FullDstRegisters[0].DstRegister.Index = dstIndex;
-   inst->FullDstRegisters[0].DstRegister.WriteMask = writemask;
+   /* Fragment program uses fragment position input.
+    * Need to replace instances of INPUT[WPOS] with temp T
+    * where T = INPUT[WPOS] by y is inverted.
+    */
+   static const gl_state_index winSizeState[STATE_LENGTH]
+      = { STATE_INTERNAL, STATE_FB_SIZE, 0, 0, 0 };
+   
+   /* XXX: note we are modifying the incoming shader here!  Need to
+    * do this before emitting the constant decls below, or this
+    * will be missed:
+    */
+   unsigned winHeightConst = _mesa_add_state_reference(program->Parameters,
+                                                       winSizeState);
 
-   inst->Instruction.NumSrcRegs = 2;
-   inst->FullSrcRegisters[0].SrcRegister.File = srcFile1;
-   inst->FullSrcRegisters[0].SrcRegister.Index = srcIndex1;
-   inst->FullSrcRegisters[1].SrcRegister.File = srcFile2;
-   inst->FullSrcRegisters[1].SrcRegister.Index = srcIndex2;
-}
+   struct ureg_src winsize = ureg_DECL_constant( ureg, winHeightConst );
+   struct ureg_dst wpos_temp = ureg_DECL_temporary( ureg );
+   struct ureg_src wpos_input = t->inputs[t->inputMapping[FRAG_ATTRIB_WPOS]];
 
+   /* MOV wpos_temp, input[wpos]
+    */
+   ureg_MOV( ureg, wpos_temp, wpos_input );
 
+   /* SUB wpos_temp.y, winsize_const, wpos_input
+    */
+   ureg_SUB( ureg,
+             ureg_writemask(wpos_temp, TGSI_WRITEMASK_Y ),
+             winsize,
+             wpos_input);
 
-/**
- * Emit the TGSI instructions for inverting the WPOS y coordinate.
- */
-static int
-emit_inverted_wpos(struct tgsi_token *tokens,
-                   int wpos_temp,
-                   int winsize_const,
-                   int wpos_input,
-                   struct tgsi_header *header, int maxTokens)
-{
-   struct tgsi_full_instruction fullinst;
-   int ti = 0;
-
-   /* MOV wpos_temp.xzw, input[wpos]; */
-   build_tgsi_instruction1(&fullinst,
-                           TGSI_OPCODE_MOV,
-                           TGSI_FILE_TEMPORARY, wpos_temp, WRITEMASK_XZW,
-                           TGSI_FILE_INPUT, 0);
-
-   ti += tgsi_build_full_instruction(&fullinst,
-                                     &tokens[ti],
-                                     header,
-                                     maxTokens - ti);
-
-   /* SUB wpos_temp.y, const[winsize_const] - input[wpos_input]; */
-   build_tgsi_instruction2(&fullinst,
-                           TGSI_OPCODE_SUB,
-                           TGSI_FILE_TEMPORARY, wpos_temp, WRITEMASK_Y,
-                           TGSI_FILE_CONSTANT, winsize_const,
-                           TGSI_FILE_INPUT, wpos_input);
-
-   ti += tgsi_build_full_instruction(&fullinst,
-                                     &tokens[ti],
-                                     header,
-                                     maxTokens - ti);
-
-   return ti;
+   /* Use wpos_temp as position input from here on:
+    */
+   t->inputs[t->inputMapping[FRAG_ATTRIB_WPOS]] = ureg_src(wpos_temp);
 }
 
 
-
-
 /**
  * Translate Mesa program to TGSI format.
  * \param program  the program to translate
@@ -864,20 +677,19 @@ emit_inverted_wpos(struct tgsi_token *tokens,
  * \param inputMapping  maps Mesa fragment program inputs to TGSI generic
  *                      input indexes
  * \param inputSemanticName  the TGSI_SEMANTIC flag for each input
- * \param inputSemanticIndex  the semantic index (ex: which texcoord) for each input
+ * \param inputSemanticIndex  the semantic index (ex: which texcoord) for
+ *                            each input
  * \param interpMode  the TGSI_INTERPOLATE_LINEAR/PERSP mode for each input
-
  * \param numOutputs  number of output registers used
  * \param outputMapping  maps Mesa fragment program outputs to TGSI
  *                       generic outputs
  * \param outputSemanticName  the TGSI_SEMANTIC flag for each output
- * \param outputSemanticIndex  the semantic index (ex: which texcoord) for each output
- * \param tokens  array to store translated tokens in
- * \param maxTokens  size of the tokens array
+ * \param outputSemanticIndex  the semantic index (ex: which texcoord) for
+ *                             each output
  *
- * \return number of tokens placed in 'tokens' buffer, or zero if error
+ * \return  array of translated tokens, caller's responsibility to free
  */
-GLuint
+const struct tgsi_token *
 st_translate_mesa_program(
    GLcontext *ctx,
    uint procType,
@@ -892,252 +704,124 @@ st_translate_mesa_program(
    const GLuint outputMapping[],
    const ubyte outputSemanticName[],
    const ubyte outputSemanticIndex[],
-   const GLbitfield outputFlags[],
-   struct tgsi_token *tokens,
-   GLuint maxTokens )
+   const GLbitfield outputFlags[] )
 {
-   GLuint i;
-   GLuint ti;  /* token index */
-   struct tgsi_header *header;
-   struct tgsi_processor *processor;
-   GLuint preamble_size = 0;
-   GLuint immediates[1000];
-   GLuint numImmediates = 0;
-   GLboolean insideSubroutine = GL_FALSE;
-   GLboolean indirectAccess = GL_FALSE;
-   GLboolean tempsUsed[MAX_PROGRAM_TEMPS + 1];
-   GLint wposTemp = -1, winHeightConst = -1;
-
-   assert(procType == TGSI_PROCESSOR_FRAGMENT ||
-          procType == TGSI_PROCESSOR_VERTEX);
-
-   find_temporaries(program, tempsUsed);
-
-   if (procType == TGSI_PROCESSOR_FRAGMENT) {
-      if (program->InputsRead & FRAG_BIT_WPOS) {
-         /* Fragment program uses fragment position input.
-          * Need to replace instances of INPUT[WPOS] with temp T
-          * where T = INPUT[WPOS] by y is inverted.
-          */
-         static const gl_state_index winSizeState[STATE_LENGTH]
-            = { STATE_INTERNAL, STATE_FB_SIZE, 0, 0, 0 };
-         winHeightConst = _mesa_add_state_reference(program->Parameters,
-                                                    winSizeState);
-         wposTemp = find_free_temporary(tempsUsed);
-      }
-   }
-
+   struct st_translate translate, *t;
+   struct ureg_program *ureg;
+   const struct tgsi_token *tokens = NULL;
+   unsigned i;
 
-   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   t = &translate;
+   memset(t, 0, sizeof *t);
 
-   header = (struct tgsi_header *) &tokens[1];
-   *header = tgsi_build_header();
+   t->procType = procType;
+   t->inputMapping = inputMapping;
+   t->outputMapping = outputMapping;
+   t->ureg = ureg_create( procType );
+   if (t->ureg == NULL)
+      return NULL;
 
-   processor = (struct tgsi_processor *) &tokens[2];
-   *processor = tgsi_build_processor( procType, header );
+   ureg = t->ureg;
 
-   ti = 3;
+   /*_mesa_print_program(program);*/
 
    /*
     * Declare input attributes.
     */
    if (procType == TGSI_PROCESSOR_FRAGMENT) {
       for (i = 0; i < numInputs; i++) {
-         struct tgsi_full_declaration fulldecl;
-         fulldecl = make_input_decl(i,
-                                    GL_TRUE, interpMode[i],
-                                    TGSI_WRITEMASK_XYZW,
-                                    GL_TRUE, inputSemanticName[i],
-                                    inputSemanticIndex[i],
-                                    inputFlags[i]);
-         ti += tgsi_build_full_declaration(&fulldecl,
-                                           &tokens[ti],
-                                           header,
-                                           maxTokens - ti );
+         t->inputs[i] = ureg_DECL_fs_input(ureg,
+                                           inputSemanticName[i],
+                                           inputSemanticIndex[i],
+                                           interpMode[i]);
       }
-   }
-   else {
-      /* vertex prog */
-      /* XXX: this could probaby be merged with the clause above.
-       * the only difference is the semantic tags.
-       */
-      for (i = 0; i < numInputs; i++) {
-         struct tgsi_full_declaration fulldecl;
-         fulldecl = make_input_decl(i,
-                                    GL_FALSE, 0,
-                                    TGSI_WRITEMASK_XYZW,
-                                    GL_FALSE, 0, 0,
-                                    inputFlags[i]);
-         ti += tgsi_build_full_declaration(&fulldecl,
-                                           &tokens[ti],
-                                           header,
-                                           maxTokens - ti );
+
+      if (program->InputsRead & FRAG_BIT_WPOS) {
+         /* Must do this after setting up t->inputs, and before
+          * emitting constant references, below:
+          */
+         emit_inverted_wpos( t, program );
       }
-   }
 
-   /*
-    * Declare output attributes.
-    */
-   if (procType == TGSI_PROCESSOR_FRAGMENT) {
+      /*
+       * Declare output attributes.
+       */
       for (i = 0; i < numOutputs; i++) {
-         struct tgsi_full_declaration fulldecl;
          switch (outputSemanticName[i]) {
          case TGSI_SEMANTIC_POSITION:
-            fulldecl = make_output_decl(i,
-                                        TGSI_SEMANTIC_POSITION, /* Z / Depth */
-                                        outputSemanticIndex[i],
-                                        TGSI_WRITEMASK_Z,
-                                        outputFlags[i]);
+            t->outputs[i] = ureg_DECL_output( ureg,
+                                              TGSI_SEMANTIC_POSITION, /* Z / Depth */
+                                              outputSemanticIndex[i] );
+
+            t->outputs[i] = ureg_writemask( t->outputs[i],
+                                            TGSI_WRITEMASK_Z );
             break;
          case TGSI_SEMANTIC_COLOR:
-            fulldecl = make_output_decl(i,
-                                        TGSI_SEMANTIC_COLOR,
-                                        outputSemanticIndex[i],
-                                        TGSI_WRITEMASK_XYZW,
-                                        outputFlags[i]);
+            t->outputs[i] = ureg_DECL_output( ureg,
+                                              TGSI_SEMANTIC_COLOR,
+                                              outputSemanticIndex[i] );
             break;
          default:
-            assert(0);
+            debug_assert(0);
             return 0;
          }
-         ti += tgsi_build_full_declaration(&fulldecl,
-                                           &tokens[ti],
-                                           header,
-                                           maxTokens - ti );
       }
    }
    else {
-      /* vertex prog */
-      for (i = 0; i < numOutputs; i++) {
-         struct tgsi_full_declaration fulldecl;
-         fulldecl = make_output_decl(i,
-                                     outputSemanticName[i],
-                                     outputSemanticIndex[i],
-                                     TGSI_WRITEMASK_XYZW,
-                                     outputFlags[i]);
-         ti += tgsi_build_full_declaration(&fulldecl,
-                                           &tokens[ti],
-                                           header,
-                                           maxTokens - ti );
+      for (i = 0; i < numInputs; i++) {
+         t->inputs[i] = ureg_DECL_vs_input(ureg, i);
       }
-   }
 
-   /* temporary decls */
-   {
-      GLboolean inside_range = GL_FALSE;
-      GLuint start_range = 0;
-
-      tempsUsed[MAX_PROGRAM_TEMPS] = GL_FALSE;
-      for (i = 0; i < MAX_PROGRAM_TEMPS + 1; i++) {
-         if (tempsUsed[i] && !inside_range) {
-            inside_range = GL_TRUE;
-            start_range = i;
-         }
-         else if (!tempsUsed[i] && inside_range) {
-            struct tgsi_full_declaration fulldecl;
-
-            inside_range = GL_FALSE;
-            fulldecl = make_temp_decl( start_range, i - 1 );
-            ti += tgsi_build_full_declaration(
-               &fulldecl,
-               &tokens[ti],
-               header,
-               maxTokens - ti );
-         }
+      for (i = 0; i < numOutputs; i++) {
+         t->outputs[i] = ureg_DECL_output( ureg,
+                                           outputSemanticName[i],
+                                           outputSemanticIndex[i] );
       }
    }
 
    /* Declare address register.
-   */
+    */
    if (program->NumAddressRegs > 0) {
-      struct tgsi_full_declaration fulldecl;
-
-      assert( program->NumAddressRegs == 1 );
-
-      fulldecl = make_addr_decl( 0, 0 );
-      ti += tgsi_build_full_declaration(
-         &fulldecl,
-         &tokens[ti],
-         header,
-         maxTokens - ti );
-
-      indirectAccess = GL_TRUE;
+      debug_assert( program->NumAddressRegs == 1 );
+      t->address[0] = ureg_DECL_address( ureg );
    }
 
-   /* immediates/literals */
-   memset(immediates, ~0, sizeof(immediates));
 
-   /* Emit immediates only when there is no address register in use.
-    * FIXME: Be smarter and recognize param arrays -- indirect addressing is
-    *        only valid within the referenced array.
+   /* Emit constants and immediates.  Mesa uses a single index space
+    * for these, so we put all the translated regs in t->constants.
     */
-   if (program->Parameters && !indirectAccess) {
-      for (i = 0; i < program->Parameters->NumParameters; i++) {
-         if (program->Parameters->Parameters[i].Type == PROGRAM_CONSTANT) {
-            struct tgsi_full_immediate fullimm;
-
-            fullimm = make_immediate( program->Parameters->ParameterValues[i], 4 );
-            ti += tgsi_build_full_immediate(
-               &fullimm,
-               &tokens[ti],
-               header,
-               maxTokens - ti );
-            immediates[i] = numImmediates;
-            numImmediates++;
-         }
-      }
-   }
-
-   /* constant buffer refs */
    if (program->Parameters) {
-      GLint start = -1, end = -1;
-
+      
+      t->constants = CALLOC( program->Parameters->NumParameters,
+                             sizeof t->constants[0] );
+      if (t->constants == NULL)
+         goto out;
+      
       for (i = 0; i < program->Parameters->NumParameters; i++) {
-         GLboolean emit = (i == program->Parameters->NumParameters - 1);
-         GLboolean matches;
-
          switch (program->Parameters->Parameters[i].Type) {
          case PROGRAM_ENV_PARAM:
          case PROGRAM_STATE_VAR:
          case PROGRAM_NAMED_PARAM:
          case PROGRAM_UNIFORM:
-            matches = GL_TRUE;
+            t->constants[i] = ureg_DECL_constant( ureg, i );
             break;
+
+            /* Emit immediates only when there is no address register
+             * in use.  FIXME: Be smarter and recognize param arrays:
+             * indirect addressing is only valid within the referenced
+             * array.
+             */
          case PROGRAM_CONSTANT:
-            matches = indirectAccess;
+            if (program->NumAddressRegs > 0) 
+               t->constants[i] = ureg_DECL_constant( ureg, i );
+            else
+               t->constants[i] = 
+                  ureg_DECL_immediate( ureg,
+                                       program->Parameters->ParameterValues[i],
+                                       4 );
             break;
          default:
-            matches = GL_FALSE;
-         }
-
-         if (matches) {
-            if (start == -1) {
-               /* begin a sequence */
-               start = i;
-               end = i;
-            }
-            else {
-               /* continue sequence */
-               end = i;
-            }
-         }
-         else {
-            if (start != -1) {
-               /* end of sequence */
-               emit = GL_TRUE;
-            }
-         }
-
-         if (emit && start >= 0) {
-            struct tgsi_full_declaration fulldecl;
-
-            fulldecl = make_constant_decl( start, end );
-            ti += tgsi_build_full_declaration(
-               &fulldecl,
-               &tokens[ti],
-               header,
-               maxTokens - ti );
-            start = end = -1;
+            break;
          }
       }
    }
@@ -1145,58 +829,55 @@ st_translate_mesa_program(
    /* texture samplers */
    for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
       if (program->SamplersUsed & (1 << i)) {
-         struct tgsi_full_declaration fulldecl;
-
-         fulldecl = make_sampler_decl( i );
-         ti += tgsi_build_full_declaration(
-            &fulldecl,
-            &tokens[ti],
-            header,
-            maxTokens - ti );
+         t->samplers[i] = ureg_DECL_sampler( ureg, i );
       }
    }
 
-   /* invert WPOS fragment input */
-   if (wposTemp >= 0) {
-      ti += emit_inverted_wpos(&tokens[ti], wposTemp, winHeightConst,
-                               inputMapping[FRAG_ATTRIB_WPOS],
-                               header, maxTokens - ti);
-      preamble_size = 2; /* two instructions added */
+   /* Emit each instruction in turn:
+    */
+   for (i = 0; i < program->NumInstructions; i++) {
+      set_insn_start( t, ureg_get_instruction_number( ureg ));
+      compile_instruction( t, &program->Instructions[i] );
    }
 
-   for (i = 0; i < program->NumInstructions; i++) {
-      struct tgsi_full_instruction fullinst;
-
-      compile_instruction(
-         &program->Instructions[i],
-         &fullinst,
-         inputMapping,
-         outputMapping,
-         immediates,
-         indirectAccess,
-         preamble_size,
-         procType,
-         &insideSubroutine,
-         wposTemp);
-
-      ti += tgsi_build_full_instruction(
-         &fullinst,
-         &tokens[ti],
-         header,
-         maxTokens - ti );
+   /* Fix up all emitted labels:
+    */
+   for (i = 0; i < t->labels_count; i++) {
+      ureg_fixup_label( ureg,
+                        t->labels[i].token,
+                        t->insn[t->labels[i].branch_target] );
    }
 
-#if DEBUG
-   if(!tgsi_sanity_check(tokens)) {
-      debug_printf("Due to sanity check failure(s) above the following shader program is invalid:\n");
-      debug_printf("\nOriginal program:\n%s", program->String);
-      debug_printf("\nMesa program:\n");
+   tokens = ureg_get_tokens( ureg, NULL );
+   ureg_destroy( ureg );
+
+out:
+   FREE(t->insn);
+   FREE(t->labels);
+   FREE(t->constants);
+
+   if (t->error) {
+      debug_printf("%s: translate error flag set\n", __FUNCTION__);
+      FREE((void *)tokens);
+      tokens = NULL;
+   }
+
+   if (!tokens) {
+      debug_printf("%s: failed to translate Mesa program:\n", __FUNCTION__);
       _mesa_print_program(program);
-      debug_printf("\nTGSI program:\n");
-      tgsi_dump(tokens, 0);
-      assert(0);
+      debug_assert(0);
    }
-#endif
 
-   return ti;
+   return tokens;
+}
+
+
+/**
+ * Tokens cannot be free with _mesa_free otherwise the builtin gallium
+ * malloc debugging will get confused.
+ */
+void
+st_free_tokens(const struct tgsi_token *tokens)
+{
+   FREE((void *)tokens);
 }
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h
index b465b3bddc..c0d1ff59e1 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.h
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h
@@ -39,7 +39,7 @@ extern "C" {
 struct tgsi_token;
 struct gl_program;
 
-GLuint
+const struct tgsi_token *
 st_translate_mesa_program(
    GLcontext *ctx,
    uint procType,
@@ -54,9 +54,10 @@ st_translate_mesa_program(
    const GLuint outputMapping[],
    const ubyte outputSemanticName[],
    const ubyte outputSemanticIndex[],
-   const GLbitfield outputFlags[],
-   struct tgsi_token *tokens,
-   GLuint maxTokens );
+   const GLbitfield outputFlags[] );
+
+void
+st_free_tokens(const struct tgsi_token *tokens);
 
 
 #if defined __cplusplus
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index d2da20ae42..a9be80ce8f 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -42,6 +42,7 @@
 #include "draw/draw_context.h"
 #include "tgsi/tgsi_dump.h"
 
+#include "st_debug.h"
 #include "st_context.h"
 #include "st_atom.h"
 #include "st_program.h"
@@ -49,12 +50,6 @@
 #include "cso_cache/cso_context.h"
 
 
-#define ST_MAX_SHADER_TOKENS (8 * 1024)
-
-
-#define TGSI_DEBUG 0
-
-
 /**
  * Translate a Mesa vertex shader into a TGSI shader.
  * \param outputMapping  to map vertex program output registers (VERT_RESULT_x)
@@ -70,12 +65,9 @@ st_translate_vertex_program(struct st_context *st,
                             const ubyte *outputSemanticIndex)
 {
    struct pipe_context *pipe = st->pipe;
-   struct tgsi_token *tokens;
    GLuint defaultOutputMapping[VERT_RESULT_MAX];
-   struct pipe_shader_state vs;
    GLuint attr, i;
    GLuint num_generic = 0;
-   GLuint num_tokens;
 
    ubyte vs_input_semantic_name[PIPE_MAX_SHADER_INPUTS];
    ubyte vs_input_semantic_index[PIPE_MAX_SHADER_INPUTS];
@@ -88,14 +80,7 @@ st_translate_vertex_program(struct st_context *st,
    GLbitfield input_flags[MAX_PROGRAM_INPUTS];
    GLbitfield output_flags[MAX_PROGRAM_OUTPUTS];
 
-   tokens =  (struct tgsi_token *)MALLOC(ST_MAX_SHADER_TOKENS * sizeof *tokens);
-   if(!tokens) {
-      /* FIXME: propagate error to the caller */
-      assert(0);
-      return;
-   }
-
-   memset(&vs, 0, sizeof(vs));
+//   memset(&vs, 0, sizeof(vs));
    memset(input_flags, 0, sizeof(input_flags));
    memset(output_flags, 0, sizeof(output_flags));
 
@@ -330,7 +315,7 @@ st_translate_vertex_program(struct st_context *st,
 
    /* free old shader state, if any */
    if (stvp->state.tokens) {
-      _mesa_free((void *) stvp->state.tokens);
+      st_free_tokens(stvp->state.tokens);
       stvp->state.tokens = NULL;
    }
    if (stvp->driver_shader) {
@@ -338,43 +323,36 @@ st_translate_vertex_program(struct st_context *st,
       stvp->driver_shader = NULL;
    }
 
-   /* XXX: fix static allocation of tokens:
-    */
-   num_tokens = st_translate_mesa_program(st->ctx,
-                                          TGSI_PROCESSOR_VERTEX,
-                                          &stvp->Base.Base,
-                                          /* inputs */
-                                          vs_num_inputs,
-                                          stvp->input_to_index,
-                                          vs_input_semantic_name,
-                                          vs_input_semantic_index,
-                                          NULL,
-                                          input_flags,
-                                          /* outputs */
-                                          vs_num_outputs,
-                                          outputMapping,
-                                          vs_output_semantic_name,
-                                          vs_output_semantic_index,
-                                          output_flags,
-                                          /* tokenized result */
-                                          tokens, ST_MAX_SHADER_TOKENS);
-
-   assert(num_tokens < ST_MAX_SHADER_TOKENS);
-
-   vs.tokens = (struct tgsi_token *)
-      _mesa_realloc(tokens,
-                    ST_MAX_SHADER_TOKENS * sizeof *tokens,
-                    num_tokens * sizeof *tokens);
+   stvp->state.tokens = 
+      st_translate_mesa_program(st->ctx,
+                                TGSI_PROCESSOR_VERTEX,
+                                &stvp->Base.Base,
+                                /* inputs */
+                                vs_num_inputs,
+                                stvp->input_to_index,
+                                vs_input_semantic_name,
+                                vs_input_semantic_index,
+                                NULL,
+                                input_flags,
+                                /* outputs */
+                                vs_num_outputs,
+                                outputMapping,
+                                vs_output_semantic_name,
+                                vs_output_semantic_index,
+                                output_flags );
 
    stvp->num_inputs = vs_num_inputs;
-   stvp->state = vs; /* struct copy */
-   stvp->driver_shader = pipe->create_vs_state(pipe, &vs);
+   stvp->driver_shader = pipe->create_vs_state(pipe, &stvp->state);
 
-   if (0)
+   if ((ST_DEBUG & DEBUG_TGSI) && (ST_DEBUG & DEBUG_MESA)) {
       _mesa_print_program(&stvp->Base.Base);
+      debug_printf("\n");
+   }
 
-   if (TGSI_DEBUG)
-      tgsi_dump( vs.tokens, 0 );
+   if (ST_DEBUG & DEBUG_TGSI) {
+      tgsi_dump( stvp->state.tokens, 0 );
+      debug_printf("\n");
+   }
 }
 
 
@@ -383,7 +361,6 @@ st_translate_vertex_program(struct st_context *st,
  * Translate a Mesa fragment shader into a TGSI shader.
  * \param inputMapping  to map fragment program input registers to TGSI
  *                      input slots
- * \param tokensOut  destination for TGSI tokens
  * \return  pointer to cached pipe_shader object.
  */
 void
@@ -392,16 +369,13 @@ st_translate_fragment_program(struct st_context *st,
                               const GLuint inputMapping[])
 {
    struct pipe_context *pipe = st->pipe;
-   struct tgsi_token *tokens;
    GLuint outputMapping[FRAG_RESULT_MAX];
    GLuint defaultInputMapping[FRAG_ATTRIB_MAX];
-   struct pipe_shader_state fs;
    GLuint interpMode[16];  /* XXX size? */
    GLuint attr;
    const GLbitfield inputsRead = stfp->Base.Base.InputsRead;
    GLuint vslot = 0;
    GLuint num_generic = 0;
-   GLuint num_tokens;
 
    uint fs_num_inputs = 0;
 
@@ -412,14 +386,7 @@ st_translate_fragment_program(struct st_context *st,
    GLbitfield input_flags[MAX_PROGRAM_INPUTS];
    GLbitfield output_flags[MAX_PROGRAM_OUTPUTS];
 
-   tokens =  (struct tgsi_token *)MALLOC(ST_MAX_SHADER_TOKENS * sizeof *tokens);
-   if(!tokens) {
-      /* FIXME: propagate error to the caller */
-      assert(0);
-      return;
-   }
-
-   memset(&fs, 0, sizeof(fs));
+//   memset(&fs, 0, sizeof(fs));
    memset(input_flags, 0, sizeof(input_flags));
    memset(output_flags, 0, sizeof(output_flags));
 
@@ -541,42 +508,35 @@ st_translate_fragment_program(struct st_context *st,
    if (!inputMapping)
       inputMapping = defaultInputMapping;
 
-   /* XXX: fix static allocation of tokens:
-    */
-   num_tokens = st_translate_mesa_program(st->ctx,
-                                          TGSI_PROCESSOR_FRAGMENT,
-                                          &stfp->Base.Base,
-                                          /* inputs */
-                                          fs_num_inputs,
-                                          inputMapping,
-                                          stfp->input_semantic_name,
-                                          stfp->input_semantic_index,
-                                          interpMode,
-                                          input_flags,
-                                          /* outputs */
-                                          fs_num_outputs,
-                                          outputMapping,
-                                          fs_output_semantic_name,
-                                          fs_output_semantic_index,
-                                          output_flags,
-                                          /* tokenized result */
-                                          tokens, ST_MAX_SHADER_TOKENS);
-
-   assert(num_tokens < ST_MAX_SHADER_TOKENS);
-
-   fs.tokens = (struct tgsi_token *)
-      _mesa_realloc(tokens,
-                    ST_MAX_SHADER_TOKENS * sizeof *tokens,
-                    num_tokens * sizeof *tokens);
-
-   stfp->state = fs; /* struct copy */
-   stfp->driver_shader = pipe->create_fs_state(pipe, &fs);
-
-   if (0)
+   stfp->state.tokens = 
+      st_translate_mesa_program(st->ctx,
+                                TGSI_PROCESSOR_FRAGMENT,
+                                &stfp->Base.Base,
+                                /* inputs */
+                                fs_num_inputs,
+                                inputMapping,
+                                stfp->input_semantic_name,
+                                stfp->input_semantic_index,
+                                interpMode,
+                                input_flags,
+                                /* outputs */
+                                fs_num_outputs,
+                                outputMapping,
+                                fs_output_semantic_name,
+                                fs_output_semantic_index,
+                                output_flags );
+
+   stfp->driver_shader = pipe->create_fs_state(pipe, &stfp->state);
+
+   if ((ST_DEBUG & DEBUG_TGSI) && (ST_DEBUG & DEBUG_MESA)) {
       _mesa_print_program(&stfp->Base.Base);
+      debug_printf("\n");
+   }
 
-   if (TGSI_DEBUG)
-      tgsi_dump( fs.tokens, 0/*TGSI_DUMP_VERBOSE*/ );
+   if (ST_DEBUG & DEBUG_TGSI) {
+      tgsi_dump( stfp->state.tokens, 0/*TGSI_DUMP_VERBOSE*/ );
+      debug_printf("\n");
+   }
 }
 
 
diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c
index bbc2830e69..ba8d1e8cc1 100644
--- a/src/mesa/state_tracker/st_texture.c
+++ b/src/mesa/state_tracker/st_texture.c
@@ -577,7 +577,6 @@ st_teximage_flush_before_map(struct st_context *st,
       pipe->is_texture_referenced(pipe, pt, face, level);
 
    if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) ||
-		      usage == PIPE_TRANSFER_WRITE ||
-		      usage == PIPE_TRANSFER_READ_WRITE))
+		      (usage & PIPE_TRANSFER_WRITE)))
       st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL);
 }
diff --git a/src/mesa/swrast/s_depth.c b/src/mesa/swrast/s_depth.c
index 7086cae0c8..17e00dda4f 100644
--- a/src/mesa/swrast/s_depth.c
+++ b/src/mesa/swrast/s_depth.c
@@ -497,6 +497,33 @@ depth_test_span32( GLcontext *ctx, GLuint n,
    return passed;
 }
 
+/* Apply ARB_depth_clamp to span of fragments. */
+void
+_swrast_depth_clamp_span( GLcontext *ctx, SWspan *span )
+{
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   struct gl_renderbuffer *rb = fb->_DepthBuffer;
+   const GLuint count = span->end;
+   GLuint *zValues = span->array->z;
+   GLuint near, far;
+   int i;
+
+   if (rb->DataType == GL_UNSIGNED_SHORT) {
+      near = FLOAT_TO_UINT(ctx->Viewport.Near);
+      far = FLOAT_TO_UINT(ctx->Viewport.Far);
+   } else {
+      assert(rb->DataType == GL_UNSIGNED_INT);
+      CLAMPED_FLOAT_TO_USHORT(near, ctx->Viewport.Near);
+      CLAMPED_FLOAT_TO_USHORT(far, ctx->Viewport.Far);
+   }
+   for (i = 0; i < count; i++) {
+      if (zValues[i] < near)
+	 zValues[i] = near;
+      if (zValues[i] > far)
+	 zValues[i] = far;
+   }
+}
+
 
 
 /*
diff --git a/src/mesa/swrast/s_depth.h b/src/mesa/swrast/s_depth.h
index 3688625683..7eae366742 100644
--- a/src/mesa/swrast/s_depth.h
+++ b/src/mesa/swrast/s_depth.h
@@ -33,6 +33,8 @@
 extern GLuint
 _swrast_depth_test_span( GLcontext *ctx, SWspan *span);
 
+extern void
+_swrast_depth_clamp_span( GLcontext *ctx, SWspan *span );
 
 extern GLboolean
 _swrast_depth_bounds_test( GLcontext *ctx, SWspan *span );
diff --git a/src/mesa/swrast/s_imaging.c b/src/mesa/swrast/s_imaging.c
deleted file mode 100644
index 3578b713f6..0000000000
--- a/src/mesa/swrast/s_imaging.c
+++ /dev/null
@@ -1,196 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5
- *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/* KW:  Moved these here to remove knowledge of swrast from core mesa.
- * Should probably pull the entire software implementation of these
- * extensions into either swrast or a sister module.  
- */
-
-#include "main/glheader.h"
-#include "main/colortab.h"
-#include "main/convolve.h"
-#include "s_context.h"
-#include "s_span.h"
-
-
-void
-_swrast_CopyColorTable( GLcontext *ctx, 
-			GLenum target, GLenum internalformat,
-			GLint x, GLint y, GLsizei width)
-{
-   GLchan data[MAX_WIDTH][4];
-   struct gl_buffer_object *bufferSave;
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no readbuffer - OK */
-      return;
-   }
-
-   if (width > MAX_WIDTH)
-      width = MAX_WIDTH;
-
-   swrast_render_start(ctx);
-
-   /* read the data from framebuffer */
-   _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
-                           width, x, y, CHAN_TYPE, data );
-
-   swrast_render_finish(ctx);
-
-   /* save PBO binding */
-   bufferSave = ctx->Unpack.BufferObj;
-   ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj;
-
-   _mesa_ColorTable(target, internalformat, width, GL_RGBA, CHAN_TYPE, data);
-
-   /* restore PBO binding */
-   ctx->Unpack.BufferObj = bufferSave;
-}
-
-
-void
-_swrast_CopyColorSubTable( GLcontext *ctx,GLenum target, GLsizei start,
-			   GLint x, GLint y, GLsizei width)
-{
-   GLchan data[MAX_WIDTH][4];
-   struct gl_buffer_object *bufferSave;
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no readbuffer - OK */
-      return;
-   }
-
-   if (width > MAX_WIDTH)
-      width = MAX_WIDTH;
-
-   swrast_render_start(ctx);
-
-   /* read the data from framebuffer */
-   _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
-                           width, x, y, CHAN_TYPE, data );
-
-   swrast_render_finish(ctx);
-
-   /* save PBO binding */
-   bufferSave = ctx->Unpack.BufferObj;
-   ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj;
-
-   _mesa_ColorSubTable(target, start, width, GL_RGBA, CHAN_TYPE, data);
-
-   /* restore PBO binding */
-   ctx->Unpack.BufferObj = bufferSave;
-}
-
-
-void
-_swrast_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, 
-				GLenum internalFormat, 
-				GLint x, GLint y, GLsizei width)
-{
-   GLchan rgba[MAX_CONVOLUTION_WIDTH][4];
-   struct gl_buffer_object *bufferSave;
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no readbuffer - OK */
-      return;
-   }
-
-   swrast_render_start(ctx);
-
-   /* read the data from framebuffer */
-   _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
-                           width, x, y, CHAN_TYPE, rgba );
-   
-   swrast_render_finish(ctx);
-
-   /* save PBO binding */
-   bufferSave = ctx->Unpack.BufferObj;
-   ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj;
-
-   /* store as convolution filter */
-   _mesa_ConvolutionFilter1D(target, internalFormat, width,
-                             GL_RGBA, CHAN_TYPE, rgba);
-
-   /* restore PBO binding */
-   ctx->Unpack.BufferObj = bufferSave;
-}
-
-
-void
-_swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, 
-				GLenum internalFormat, 
-				GLint x, GLint y, GLsizei width, GLsizei height)
-{
-   struct gl_pixelstore_attrib packSave;
-   GLchan rgba[MAX_CONVOLUTION_HEIGHT][MAX_CONVOLUTION_WIDTH][4];
-   GLint i;
-   struct gl_buffer_object *bufferSave;
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no readbuffer - OK */
-      return;
-   }
-
-   swrast_render_start(ctx);
-   
-   /* read pixels from framebuffer */
-   for (i = 0; i < height; i++) {
-      _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
-                              width, x, y + i, CHAN_TYPE, rgba[i] );
-   }
-
-   swrast_render_finish(ctx);
-
-   /*
-    * HACK: save & restore context state so we can store this as a
-    * convolution filter via the GL api.  Doesn't call any callbacks
-    * hanging off ctx->Unpack statechanges.
-    */
-
-   packSave = ctx->Unpack;  /* save pixel packing params */
-
-   ctx->Unpack.Alignment = 1;
-   ctx->Unpack.RowLength = MAX_CONVOLUTION_WIDTH;
-   ctx->Unpack.SkipPixels = 0;
-   ctx->Unpack.SkipRows = 0;
-   ctx->Unpack.ImageHeight = 0;
-   ctx->Unpack.SkipImages = 0;
-   ctx->Unpack.SwapBytes = GL_FALSE;
-   ctx->Unpack.LsbFirst = GL_FALSE;
-   ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj;
-   ctx->NewState |= _NEW_PACKUNPACK;
-
-   /* save PBO binding */
-   bufferSave = ctx->Unpack.BufferObj;
-   ctx->Unpack.BufferObj = ctx->Shared->NullBufferObj;
-
-   _mesa_ConvolutionFilter2D(target, internalFormat, width, height,
-                             GL_RGBA, CHAN_TYPE, rgba);
-
-   /* restore PBO binding */
-   ctx->Unpack.BufferObj = bufferSave;
-
-   ctx->Unpack = packSave;  /* restore pixel packing params */
-   ctx->NewState |= _NEW_PACKUNPACK; 
-}
diff --git a/src/mesa/swrast/s_readpix.c b/src/mesa/swrast/s_readpix.c
index 48b9408d24..a1aeb2e01f 100644
--- a/src/mesa/swrast/s_readpix.c
+++ b/src/mesa/swrast/s_readpix.c
@@ -555,15 +555,15 @@ _swrast_ReadPixels( GLcontext *ctx,
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_pixelstore_attrib clippedPacking = *packing;
 
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
+
    /* Need to do swrast_render_start() before clipping or anything else
     * since this is where a driver may grab the hw lock and get an updated
     * window size.
     */
    swrast_render_start(ctx);
 
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
    if (swrast->NewState)
       _swrast_validate_derived( ctx );
 
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index 0e2793b474..704230d1d7 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -880,6 +880,9 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
       stipple_polygon_span(ctx, span);
    }
 
+   if (ctx->Transform.DepthClamp)
+      _swrast_depth_clamp_span(ctx, span);
+
    /* Stencil and Z testing */
    if (ctx->Stencil._Enabled || ctx->Depth.Test) {
       if (!(span->arrayMask & SPAN_Z))
@@ -901,7 +904,6 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
       }
    }
 
-#if FEATURE_ARB_occlusion_query
    if (ctx->Query.CurrentOcclusionObject) {
       /* update count of 'passed' fragments */
       struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;
@@ -909,7 +911,6 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
       for (i = 0; i < span->end; i++)
          q->Result += span->array->mask[i];
    }
-#endif
 
    /* we have to wait until after occlusion to do this test */
    if (ctx->Color.IndexMask == 0) {
@@ -1373,7 +1374,6 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
       }
    }
 
-#if FEATURE_ARB_occlusion_query
    if (ctx->Query.CurrentOcclusionObject) {
       /* update count of 'passed' fragments */
       struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;
@@ -1381,7 +1381,6 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
       for (i = 0; i < span->end; i++)
          q->Result += span->array->mask[i];
    }
-#endif
 
    /* We had to wait until now to check for glColorMask(0,0,0,0) because of
     * the occlusion test.
diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c
index efe6f23474..004d4e05ae 100644
--- a/src/mesa/swrast/s_texfilter.c
+++ b/src/mesa/swrast/s_texfilter.c
@@ -1905,8 +1905,12 @@ choose_cube_face(const struct gl_texture_object *texObj,
       }
    }
 
-   newCoord[0] = ( sc / ma + 1.0F ) * 0.5F;
-   newCoord[1] = ( tc / ma + 1.0F ) * 0.5F;
+   { 
+      const float ima = 1.0F / ma;
+      newCoord[0] = ( sc * ima + 1.0F ) * 0.5F;
+      newCoord[1] = ( tc * ima + 1.0F ) * 0.5F;
+   }
+
    return (const struct gl_texture_image **) texObj->Image[face];
 }
 
diff --git a/src/mesa/swrast/s_texstore.c b/src/mesa/swrast/s_texstore.c
deleted file mode 100644
index f9ff9ad6a4..0000000000
--- a/src/mesa/swrast/s_texstore.c
+++ /dev/null
@@ -1,601 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.2
- *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/*
- * Authors:
- *   Brian Paul
- */
-
-
-/*
- * The functions in this file are mostly related to software texture fallbacks.
- * This includes texture image transfer/packing and texel fetching.
- * Hardware drivers will likely override most of this.
- */
-
-
-
-#include "main/glheader.h"
-#include "main/imports.h"
-#include "main/colormac.h"
-#include "main/context.h"
-#include "main/convolve.h"
-#include "main/image.h"
-#include "main/macros.h"
-#include "main/mipmap.h"
-#include "main/texformat.h"
-#include "main/teximage.h"
-#include "main/texstore.h"
-
-#include "s_context.h"
-#include "s_depth.h"
-#include "s_span.h"
-
-
-/**
- * Read an RGBA image from the frame buffer.
- * This is used by glCopyTex[Sub]Image[12]D().
- * \param x  window source x
- * \param y  window source y
- * \param width  image width
- * \param height  image height
- * \param type  datatype for returned GL_RGBA image
- * \return pointer to image
- */
-static GLvoid *
-read_color_image( GLcontext *ctx, GLint x, GLint y, GLenum type,
-                  GLsizei width, GLsizei height )
-{
-   struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorReadBuffer;
-   const GLint pixelSize = _mesa_bytes_per_pixel(GL_RGBA, type);
-   const GLint stride = width * pixelSize;
-   GLint row;
-   GLubyte *image, *dst;
-
-   image = (GLubyte *) _mesa_malloc(width * height * pixelSize);
-   if (!image)
-      return NULL;
-
-   swrast_render_start(ctx);
-
-   dst = image;
-   for (row = 0; row < height; row++) {
-      _swrast_read_rgba_span(ctx, rb, width, x, y + row, type, dst);
-      dst += stride;
-   }
-
-   swrast_render_finish(ctx);
-
-   return image;
-}
-
-
-/**
- * As above, but read data from depth buffer.  Returned as GLuints.
- * \sa read_color_image
- */
-static GLuint *
-read_depth_image( GLcontext *ctx, GLint x, GLint y,
-                  GLsizei width, GLsizei height )
-{
-   struct gl_renderbuffer *rb = ctx->ReadBuffer->_DepthBuffer;
-   GLuint *image, *dst;
-   GLint i;
-
-   image = (GLuint *) _mesa_malloc(width * height * sizeof(GLuint));
-   if (!image)
-      return NULL;
-
-   swrast_render_start(ctx);
-
-   dst = image;
-   for (i = 0; i < height; i++) {
-      _swrast_read_depth_span_uint(ctx, rb, width, x, y + i, dst);
-      dst += width;
-   }
-
-   swrast_render_finish(ctx);
-
-   return image;
-}
-
-
-/**
- * As above, but read data from depth+stencil buffers.
- */
-static GLuint *
-read_depth_stencil_image(GLcontext *ctx, GLint x, GLint y,
-                         GLsizei width, GLsizei height)
-{
-   struct gl_renderbuffer *depthRb = ctx->ReadBuffer->_DepthBuffer;
-   struct gl_renderbuffer *stencilRb = ctx->ReadBuffer->_StencilBuffer;
-   GLuint *image, *dst;
-   GLint i;
-
-   ASSERT(depthRb);
-   ASSERT(stencilRb);
-
-   image = (GLuint *) _mesa_malloc(width * height * sizeof(GLuint));
-   if (!image)
-      return NULL;
-
-   swrast_render_start(ctx);
-
-   /* read from depth buffer */
-   dst = image;
-   if (depthRb->DataType == GL_UNSIGNED_INT) {
-      for (i = 0; i < height; i++) {
-         _swrast_get_row(ctx, depthRb, width, x, y + i, dst, sizeof(GLuint));
-         dst += width;
-      }
-   }
-   else {
-      GLushort z16[MAX_WIDTH];
-      ASSERT(depthRb->DataType == GL_UNSIGNED_SHORT);
-      for (i = 0; i < height; i++) {
-         GLint j;
-         _swrast_get_row(ctx, depthRb, width, x, y + i, z16, sizeof(GLushort));
-         /* convert GLushorts to GLuints */
-         for (j = 0; j < width; j++) {
-            dst[j] = z16[j];
-         }
-         dst += width;
-      }
-   }
-
-   /* put depth values into bits 0xffffff00 */
-   if (ctx->ReadBuffer->Visual.depthBits == 24) {
-      GLint j;
-      for (j = 0; j < width * height; j++) {
-         image[j] <<= 8;
-      }
-   }
-   else if (ctx->ReadBuffer->Visual.depthBits == 16) {
-      GLint j;
-      for (j = 0; j < width * height; j++) {
-         image[j] = (image[j] << 16) | (image[j] & 0xff00);
-      }      
-   }
-   else {
-      /* this handles arbitrary depthBits >= 12 */
-      const GLint rShift = ctx->ReadBuffer->Visual.depthBits;
-      const GLint lShift = 32 - rShift;
-      GLint j;
-      for (j = 0; j < width * height; j++) {
-         GLuint z = (image[j] << lShift);
-         image[j] = z | (z >> rShift);
-      }
-   }
-
-   /* read stencil values and interleave into image array */
-   dst = image;
-   for (i = 0; i < height; i++) {
-      GLstencil stencil[MAX_WIDTH];
-      GLint j;
-      ASSERT(8 * sizeof(GLstencil) == stencilRb->StencilBits);
-      _swrast_get_row(ctx, stencilRb, width, x, y + i,
-                      stencil, sizeof(GLstencil));
-      for (j = 0; j < width; j++) {
-         dst[j] = (dst[j] & 0xffffff00) | (stencil[j] & 0xff);
-      }
-      dst += width;
-   }
-
-   swrast_render_finish(ctx);
-
-   return image;
-}
-
-
-static GLboolean
-is_depth_format(GLenum format)
-{
-   switch (format) {
-      case GL_DEPTH_COMPONENT:
-      case GL_DEPTH_COMPONENT16:
-      case GL_DEPTH_COMPONENT24:
-      case GL_DEPTH_COMPONENT32:
-         return GL_TRUE;
-      default:
-         return GL_FALSE;
-   }
-}
-
-
-static GLboolean
-is_depth_stencil_format(GLenum format)
-{
-   switch (format) {
-      case GL_DEPTH_STENCIL_EXT:
-      case GL_DEPTH24_STENCIL8_EXT:
-         return GL_TRUE;
-      default:
-         return GL_FALSE;
-   }
-}
-
-
-/*
- * Fallback for Driver.CopyTexImage1D().
- */
-void
-_swrast_copy_teximage1d( GLcontext *ctx, GLenum target, GLint level,
-                         GLenum internalFormat,
-                         GLint x, GLint y, GLsizei width, GLint border )
-{
-   struct gl_texture_unit *texUnit;
-   struct gl_texture_object *texObj;
-   struct gl_texture_image *texImage;
-
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-   texObj = _mesa_select_tex_object(ctx, texUnit, target);
-   ASSERT(texObj);
-   texImage = _mesa_select_tex_image(ctx, texObj, target, level);
-   ASSERT(texImage);
-
-   ASSERT(ctx->Driver.TexImage1D);
-
-   if (is_depth_format(internalFormat)) {
-      /* read depth image from framebuffer */
-      GLuint *image = read_depth_image(ctx, x, y, width, 1);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D");
-         return;
-      }
-      /* call glTexImage1D to redefine the texture */
-      ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
-                             width, border,
-                             GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image,
-                             &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else if (is_depth_stencil_format(internalFormat)) {
-      /* read depth/stencil image from framebuffer */
-      GLuint *image = read_depth_stencil_image(ctx, x, y, width, 1);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D");
-         return;
-      }
-      /* call glTexImage1D to redefine the texture */
-      ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
-                             width, border,
-                             GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT,
-                             image, &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else {
-      /* read RGBA image from framebuffer */
-      const GLenum format = GL_RGBA;
-      const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType;
-      GLvoid *image = read_color_image(ctx, x, y, type, width, 1);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage1D");
-         return;
-      }
-      /* call glTexImage1D to redefine the texture */
-      ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
-                             width, border, format, type, image,
-                             &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-}
-
-
-/**
- * Fallback for Driver.CopyTexImage2D().
- *
- * We implement CopyTexImage by reading the image from the framebuffer
- * then passing it to the ctx->Driver.TexImage2D() function.
- *
- * Device drivers should try to implement direct framebuffer->texture copies.
- */
-void
-_swrast_copy_teximage2d( GLcontext *ctx, GLenum target, GLint level,
-                         GLenum internalFormat,
-                         GLint x, GLint y, GLsizei width, GLsizei height,
-                         GLint border )
-{
-   struct gl_texture_unit *texUnit;
-   struct gl_texture_object *texObj;
-   struct gl_texture_image *texImage;
-
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-   texObj = _mesa_select_tex_object(ctx, texUnit, target);
-   ASSERT(texObj);
-   texImage = _mesa_select_tex_image(ctx, texObj, target, level);
-   ASSERT(texImage);
-
-   ASSERT(ctx->Driver.TexImage2D);
-
-   if (is_depth_format(internalFormat)) {
-      /* read depth image from framebuffer */
-      GLuint *image = read_depth_image(ctx, x, y, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D");
-         return;
-      }
-      /* call glTexImage2D to redefine the texture */
-      ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
-                             width, height, border,
-                             GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image,
-                             &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else if (is_depth_stencil_format(internalFormat)) {
-      GLuint *image = read_depth_stencil_image(ctx, x, y, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D");
-         return;
-      }
-      /* call glTexImage2D to redefine the texture */
-      ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
-                             width, height, border,
-                             GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT,
-                             image, &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else {
-      /* read RGBA image from framebuffer */
-      const GLenum format = GL_RGBA;
-      const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType;
-      GLvoid *image = read_color_image(ctx, x, y, type, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage2D");
-         return;
-      }
-      /* call glTexImage2D to redefine the texture */
-      ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
-                             width, height, border, format, type, image,
-                             &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-}
-
-
-/*
- * Fallback for Driver.CopyTexSubImage1D().
- */
-void
-_swrast_copy_texsubimage1d( GLcontext *ctx, GLenum target, GLint level,
-                            GLint xoffset, GLint x, GLint y, GLsizei width )
-{
-   struct gl_texture_unit *texUnit;
-   struct gl_texture_object *texObj;
-   struct gl_texture_image *texImage;
-
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-   texObj = _mesa_select_tex_object(ctx, texUnit, target);
-   ASSERT(texObj);
-   texImage = _mesa_select_tex_image(ctx, texObj, target, level);
-   ASSERT(texImage);
-
-   ASSERT(ctx->Driver.TexImage1D);
-
-   if (texImage->_BaseFormat == GL_DEPTH_COMPONENT) {
-      /* read depth image from framebuffer */
-      GLuint *image = read_depth_image(ctx, x, y, width, 1);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage1D");
-         return;
-      }
-
-      /* call glTexSubImage1D to redefine the texture */
-      ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width,
-                                GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image,
-                                &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else if (texImage->_BaseFormat == GL_DEPTH_STENCIL_EXT) {
-      /* read depth/stencil image from framebuffer */
-      GLuint *image = read_depth_stencil_image(ctx, x, y, width, 1);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage1D");
-         return;
-      }
-      /* call glTexImage1D to redefine the texture */
-      ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width,
-                                GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT,
-                                image, &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else {
-      /* read RGBA image from framebuffer */
-      const GLenum format = GL_RGBA;
-      const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType;
-      GLvoid *image = read_color_image(ctx, x, y, type, width, 1);
-      if (!image) {
-         _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage1D" );
-         return;
-      }
-      /* now call glTexSubImage1D to do the real work */
-      ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, width,
-                                format, type, image,
-                                &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-}
-
-
-/**
- * Fallback for Driver.CopyTexSubImage2D().
- *
- * Read the image from the framebuffer then hand it
- * off to ctx->Driver.TexSubImage2D().
- */
-void
-_swrast_copy_texsubimage2d( GLcontext *ctx,
-                            GLenum target, GLint level,
-                            GLint xoffset, GLint yoffset,
-                            GLint x, GLint y, GLsizei width, GLsizei height )
-{
-   struct gl_texture_unit *texUnit;
-   struct gl_texture_object *texObj;
-   struct gl_texture_image *texImage;
-
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-   texObj = _mesa_select_tex_object(ctx, texUnit, target);
-   ASSERT(texObj);
-   texImage = _mesa_select_tex_image(ctx, texObj, target, level);
-   ASSERT(texImage);
-
-   ASSERT(ctx->Driver.TexImage2D);
-
-   if (texImage->_BaseFormat == GL_DEPTH_COMPONENT) {
-      /* read depth image from framebuffer */
-      GLuint *image = read_depth_image(ctx, x, y, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage2D");
-         return;
-      }
-      /* call glTexImage2D to redefine the texture */
-      ctx->Driver.TexSubImage2D(ctx, target, level,
-                                xoffset, yoffset, width, height,
-                                GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image,
-                                &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else if (texImage->_BaseFormat == GL_DEPTH_STENCIL_EXT) {
-      /* read depth/stencil image from framebuffer */
-      GLuint *image = read_depth_stencil_image(ctx, x, y, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage2D");
-         return;
-      }
-      /* call glTexImage2D to redefine the texture */
-      ctx->Driver.TexSubImage2D(ctx, target, level,
-                                xoffset, yoffset, width, height,
-                                GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT,
-                                image, &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else {
-      /* read RGBA image from framebuffer */
-      const GLenum format = GL_RGBA;
-      const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType;
-      GLvoid *image = read_color_image(ctx, x, y, type, width, height);
-      if (!image) {
-         _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage2D" );
-         return;
-      }
-      /* now call glTexSubImage2D to do the real work */
-      ctx->Driver.TexSubImage2D(ctx, target, level,
-                                xoffset, yoffset, width, height,
-                                format, type, image,
-                                &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-}
-
-
-/*
- * Fallback for Driver.CopyTexSubImage3D().
- */
-void
-_swrast_copy_texsubimage3d( GLcontext *ctx,
-                            GLenum target, GLint level,
-                            GLint xoffset, GLint yoffset, GLint zoffset,
-                            GLint x, GLint y, GLsizei width, GLsizei height )
-{
-   struct gl_texture_unit *texUnit;
-   struct gl_texture_object *texObj;
-   struct gl_texture_image *texImage;
-
-   texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-   texObj = _mesa_select_tex_object(ctx, texUnit, target);
-   ASSERT(texObj);
-   texImage = _mesa_select_tex_image(ctx, texObj, target, level);
-   ASSERT(texImage);
-
-   ASSERT(ctx->Driver.TexImage3D);
-
-   if (texImage->_BaseFormat == GL_DEPTH_COMPONENT) {
-      /* read depth image from framebuffer */
-      GLuint *image = read_depth_image(ctx, x, y, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage3D");
-         return;
-      }
-      /* call glTexImage3D to redefine the texture */
-      ctx->Driver.TexSubImage3D(ctx, target, level,
-                                xoffset, yoffset, zoffset, width, height, 1,
-                                GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, image,
-                                &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else if (texImage->_BaseFormat == GL_DEPTH_STENCIL_EXT) {
-      /* read depth/stencil image from framebuffer */
-      GLuint *image = read_depth_stencil_image(ctx, x, y, width, height);
-      if (!image) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage3D");
-         return;
-      }
-      /* call glTexImage3D to redefine the texture */
-      ctx->Driver.TexSubImage3D(ctx, target, level,
-                                xoffset, yoffset, zoffset, width, height, 1,
-                                GL_DEPTH_STENCIL_EXT, GL_UNSIGNED_INT_24_8_EXT,
-                                image, &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-   else {
-      /* read RGBA image from framebuffer */
-      const GLenum format = GL_RGBA;
-      const GLenum type = ctx->ReadBuffer->_ColorReadBuffer->DataType;
-      GLvoid *image = read_color_image(ctx, x, y, type, width, height);
-      if (!image) {
-         _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage3D" );
-         return;
-      }
-      /* now call glTexSubImage3D to do the real work */
-      ctx->Driver.TexSubImage3D(ctx, target, level,
-                                xoffset, yoffset, zoffset, width, height, 1,
-                                format, type, image,
-                                &ctx->DefaultPacking, texObj, texImage);
-      _mesa_free(image);
-   }
-
-   /* GL_SGIS_generate_mipmap */
-   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      ctx->Driver.GenerateMipmap(ctx, target, texObj);
-   }
-}
diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h
index c319ca62f9..c183b315b6 100644
--- a/src/mesa/swrast/swrast.h
+++ b/src/mesa/swrast/swrast.h
@@ -207,60 +207,6 @@ extern void
 _swrast_print_vertex( GLcontext *ctx, const SWvertex *v );
 
 
-/*
- * Imaging fallbacks (a better solution should be found, perhaps
- * moving all the imaging fallback code to a new module) 
- */
-extern void
-_swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, 
-				GLenum internalFormat, 
-				GLint x, GLint y, GLsizei width, 
-				GLsizei height);
-extern void
-_swrast_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, 
-				GLenum internalFormat, 
-				GLint x, GLint y, GLsizei width);
-extern void
-_swrast_CopyColorSubTable( GLcontext *ctx,GLenum target, GLsizei start,
-			   GLint x, GLint y, GLsizei width);
-extern void
-_swrast_CopyColorTable( GLcontext *ctx, 
-			GLenum target, GLenum internalformat,
-			GLint x, GLint y, GLsizei width);
-
-
-/*
- * Texture fallbacks.  Could also live in a new module
- * with the rest of the texture store fallbacks?
- */
-extern void
-_swrast_copy_teximage1d(GLcontext *ctx, GLenum target, GLint level,
-                        GLenum internalFormat,
-                        GLint x, GLint y, GLsizei width, GLint border);
-
-extern void
-_swrast_copy_teximage2d(GLcontext *ctx, GLenum target, GLint level,
-                        GLenum internalFormat,
-                        GLint x, GLint y, GLsizei width, GLsizei height,
-                        GLint border);
-
-
-extern void
-_swrast_copy_texsubimage1d(GLcontext *ctx, GLenum target, GLint level,
-                           GLint xoffset, GLint x, GLint y, GLsizei width);
-
-extern void
-_swrast_copy_texsubimage2d(GLcontext *ctx,
-                           GLenum target, GLint level,
-                           GLint xoffset, GLint yoffset,
-                           GLint x, GLint y, GLsizei width, GLsizei height);
-
-extern void
-_swrast_copy_texsubimage3d(GLcontext *ctx,
-                           GLenum target, GLint level,
-                           GLint xoffset, GLint yoffset, GLint zoffset,
-                           GLint x, GLint y, GLsizei width, GLsizei height);
-
 
 extern void
 _swrast_eject_texture_images(GLcontext *ctx);
diff --git a/src/mesa/tnl/t_context.h b/src/mesa/tnl/t_context.h
index c19eb3df3c..ca4edcfcb9 100644
--- a/src/mesa/tnl/t_context.h
+++ b/src/mesa/tnl/t_context.h
@@ -207,6 +207,7 @@ struct vertex_buffer
    GLvector4f  *EyePtr;		                /* _TNL_BIT_POS */
    GLvector4f  *ClipPtr;	                /* _TNL_BIT_POS */
    GLvector4f  *NdcPtr;                         /* _TNL_BIT_POS */
+   GLfloat     *ClipDistancePtr[MAX_CLIP_PLANES]; /* _TNL_BIT_POS */
    GLubyte     ClipOrMask;	                /* _TNL_BIT_POS */
    GLubyte     ClipAndMask;	                /* _TNL_BIT_POS */
    GLubyte     *ClipMask;		        /* _TNL_BIT_POS */
@@ -548,4 +549,8 @@ typedef struct
 #define MAX_TYPES TYPE_IDX(GL_DOUBLE)+1      /* 0xa + 1 */
 
 
+extern void
+tnl_clip_prepare(GLcontext *ctx);
+
+
 #endif
diff --git a/src/mesa/tnl/t_draw.c b/src/mesa/tnl/t_draw.c
index c64c2c2077..04fa106300 100644
--- a/src/mesa/tnl/t_draw.c
+++ b/src/mesa/tnl/t_draw.c
@@ -316,22 +316,27 @@ static void bind_indices( GLcontext *ctx,
 
    ptr = ADD_POINTERS(ib->obj->Pointer, ib->ptr);
 
-   if (ib->type == GL_UNSIGNED_INT) {
+   if (ib->type == GL_UNSIGNED_INT && VB->Primitive[0].basevertex == 0) {
       VB->Elts = (GLuint *) ptr;
    }
    else {
       GLuint *elts = (GLuint *)get_space(ctx, ib->count * sizeof(GLuint));
       VB->Elts = elts;
 
-      if (ib->type == GL_UNSIGNED_SHORT) {
+      if (ib->type == GL_UNSIGNED_INT) {
+	 const GLuint *in = (GLuint *)ptr;
+	 for (i = 0; i < ib->count; i++)
+	    *elts++ = (GLuint)(*in++) + VB->Primitive[0].basevertex;
+      }
+      else if (ib->type == GL_UNSIGNED_SHORT) {
 	 const GLushort *in = (GLushort *)ptr;
 	 for (i = 0; i < ib->count; i++) 
-	    *elts++ = (GLuint)(*in++);
+	    *elts++ = (GLuint)(*in++) + VB->Primitive[0].basevertex;
       }
       else {
 	 const GLubyte *in = (GLubyte *)ptr;
 	 for (i = 0; i < ib->count; i++) 
-	    *elts++ = (GLuint)(*in++);
+	    *elts++ = (GLuint)(*in++) + VB->Primitive[0].basevertex;
       }
    }
 }
@@ -390,10 +395,14 @@ void _tnl_draw_prims( GLcontext *ctx,
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    const GLuint TEST_SPLIT = 0;
    const GLint max = TEST_SPLIT ? 8 : tnl->vb.Size - MAX_CLIPPED_VERTICES;
+   GLuint max_basevertex = prim->basevertex;
+   GLuint i;
+
+   for (i = 1; i < nr_prims; i++)
+      max_basevertex = MAX2(max_basevertex, prim[i].basevertex);
 
    if (0)
    {
-      GLuint i;
       _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index);
       for (i = 0; i < nr_prims; i++)
 	 _mesa_printf("prim %d: %s start %d count %d\n", i, 
@@ -410,7 +419,7 @@ void _tnl_draw_prims( GLcontext *ctx,
 			_tnl_vbo_draw_prims );
       return;
    }
-   else if (max_index > max) {
+   else if (max_index + max_basevertex > max) {
       /* The software TNL pipeline has a fixed amount of storage for
        * vertices and it is necessary to split incoming drawing commands
        * if they exceed that limit.
@@ -424,7 +433,7 @@ void _tnl_draw_prims( GLcontext *ctx,
        * recursively call back into this function.
        */
       vbo_split_prims( ctx, arrays, prim, nr_prims, ib, 
-		       0, max_index,
+		       0, max_index + prim->basevertex,
 		       _tnl_vbo_draw_prims,
 		       &limits );
    }
@@ -435,17 +444,34 @@ void _tnl_draw_prims( GLcontext *ctx,
       struct gl_buffer_object *bo[VERT_ATTRIB_MAX + 1];
       GLuint nr_bo = 0;
 
-      /* Binding inputs may imply mapping some vertex buffer objects.
-       * They will need to be unmapped below.
-       */
-      bind_inputs(ctx, arrays, max_index+1, bo, &nr_bo);
-      bind_indices(ctx, ib, bo, &nr_bo);
-      bind_prims(ctx, prim, nr_prims );
+      for (i = 0; i < nr_prims;) {
+	 GLuint this_nr_prims;
+
+	 /* Our SW TNL pipeline doesn't handle basevertex yet, so bind_indices
+	  * will rebase the elements to the basevertex, and we'll only
+	  * emit strings of prims with the same basevertex in one draw call.
+	  */
+	 for (this_nr_prims = 1; i + this_nr_prims < nr_prims;
+	      this_nr_prims++) {
+	    if (prim[i].basevertex != prim[i + this_nr_prims].basevertex)
+	       break;
+	 }
+
+	 /* Binding inputs may imply mapping some vertex buffer objects.
+	  * They will need to be unmapped below.
+	  */
+	 bind_prims(ctx, &prim[i], this_nr_prims);
+	 bind_inputs(ctx, arrays, max_index + prim[i].basevertex + 1,
+		     bo, &nr_bo);
+	 bind_indices(ctx, ib, bo, &nr_bo);
 
-      TNL_CONTEXT(ctx)->Driver.RunPipeline(ctx);
+	 TNL_CONTEXT(ctx)->Driver.RunPipeline(ctx);
 
-      unmap_vbos(ctx, bo, nr_bo);
-      free_space(ctx);
+	 unmap_vbos(ctx, bo, nr_bo);
+	 free_space(ctx);
+
+	 i += this_nr_prims;
+      }
    }
 }
 
diff --git a/src/mesa/tnl/t_rasterpos.c b/src/mesa/tnl/t_rasterpos.c
index f1fdddf0f5..99b6787455 100644
--- a/src/mesa/tnl/t_rasterpos.c
+++ b/src/mesa/tnl/t_rasterpos.c
@@ -46,11 +46,10 @@
  * \return zero if outside view volume, or one if inside.
  */
 static GLuint
-viewclip_point( const GLfloat v[] )
+viewclip_point_xy( const GLfloat v[] )
 {
    if (   v[0] > v[3] || v[0] < -v[3]
-       || v[1] > v[3] || v[1] < -v[3]
-       || v[2] > v[3] || v[2] < -v[3] ) {
+       || v[1] > v[3] || v[1] < -v[3] ) {
       return 0;
    }
    else {
@@ -408,18 +407,18 @@ _tnl_RasterPos(GLcontext *ctx, const GLfloat vObj[4])
       /* apply projection matrix:  clip = Proj * eye */
       TRANSFORM_POINT( clip, ctx->ProjectionMatrixStack.Top->m, eye );
 
-      /* clip to view volume */
-      if (ctx->Transform.RasterPositionUnclipped) {
-         /* GL_IBM_rasterpos_clip: only clip against Z */
+      /* clip to view volume. */
+      if (!ctx->Transform.DepthClamp) {
          if (viewclip_point_z(clip) == 0) {
             ctx->Current.RasterPosValid = GL_FALSE;
             return;
          }
       }
-      else if (viewclip_point(clip) == 0) {
-         /* Normal OpenGL behaviour */
-         ctx->Current.RasterPosValid = GL_FALSE;
-         return;
+      if (!ctx->Transform.RasterPositionUnclipped) {
+         if (viewclip_point_xy(clip) == 0) {
+            ctx->Current.RasterPosValid = GL_FALSE;
+            return;
+         }
       }
 
       /* clip to user clipping planes */
@@ -443,6 +442,12 @@ _tnl_RasterPos(GLcontext *ctx, const GLfloat vObj[4])
                                   / ctx->DrawBuffer->_DepthMaxF;
       ctx->Current.RasterPos[3] = clip[3];
 
+      if (ctx->Transform.DepthClamp) {
+	 ctx->Current.RasterPos[3] = CLAMP(ctx->Current.RasterPos[3],
+					   ctx->Viewport.Near,
+					   ctx->Viewport.Far);
+      }
+
       /* compute raster distance */
       if (ctx->Fog.FogCoordinateSource == GL_FOG_COORDINATE_EXT)
          ctx->Current.RasterDistance = ctx->Current.Attrib[VERT_ATTRIB_FOG][0];
diff --git a/src/mesa/tnl/t_vb_cliptmp.h b/src/mesa/tnl/t_vb_cliptmp.h
index 618b8b3130..0d2183a9e6 100644
--- a/src/mesa/tnl/t_vb_cliptmp.h
+++ b/src/mesa/tnl/t_vb_cliptmp.h
@@ -80,6 +80,58 @@ do {									\
 } while (0)
 
 
+#define POLY_USERCLIP(PLANE)						\
+do {									\
+   if (mask & CLIP_USER_BIT) {						\
+      GLuint idxPrev = inlist[0];					\
+      GLfloat dpPrev = VB->ClipDistancePtr[PLANE][idxPrev];		\
+      GLuint outcount = 0;						\
+      GLuint i;								\
+									\
+      inlist[n] = inlist[0]; /* prevent rotation of vertices */		\
+      for (i = 1; i <= n; i++) {					\
+	 GLuint idx = inlist[i];					\
+	 GLfloat dp = VB->ClipDistancePtr[PLANE][idx];			\
+									\
+	 if (!IS_NEGATIVE(dpPrev)) {					\
+	    outlist[outcount++] = idxPrev;				\
+	 }								\
+									\
+	 if (DIFFERENT_SIGNS(dp, dpPrev)) {				\
+	    if (IS_NEGATIVE(dp)) {					\
+	       /* Going out of bounds.  Avoid division by zero as we	\
+		* know dp != dpPrev from DIFFERENT_SIGNS, above.	\
+		*/							\
+	       GLfloat t = dp / (dp - dpPrev);				\
+               INTERP_4F( t, coord[newvert], coord[idx], coord[idxPrev]); \
+	       interp( ctx, t, newvert, idx, idxPrev, GL_TRUE );	\
+	    } else {							\
+	       /* Coming back in.					\
+		*/							\
+	       GLfloat t = dpPrev / (dpPrev - dp);			\
+               INTERP_4F( t, coord[newvert], coord[idxPrev], coord[idx]); \
+	       interp( ctx, t, newvert, idxPrev, idx, GL_FALSE );	\
+	    }								\
+            outlist[outcount++] = newvert++;				\
+	 }								\
+									\
+	 idxPrev = idx;							\
+	 dpPrev = dp;							\
+      }									\
+									\
+      if (outcount < 3)							\
+	 return;							\
+									\
+      {									\
+	 GLuint *tmp = inlist;						\
+	 inlist = outlist;						\
+	 outlist = tmp;							\
+	 n = outcount;							\
+      }									\
+   }									\
+} while (0)
+
+
 #define LINE_CLIP(PLANE_BIT, A, B, C, D )				\
 do {									\
    if (mask & PLANE_BIT) {						\
@@ -111,6 +163,37 @@ do {									\
 } while (0)
 
 
+#define LINE_USERCLIP(PLANE)						\
+do {									\
+   if (mask & CLIP_USER_BIT) {						\
+      const GLfloat dp0 = VB->ClipDistancePtr[PLANE][v0];		\
+      const GLfloat dp1 = VB->ClipDistancePtr[PLANE][v1];		\
+      const GLboolean neg_dp0 = IS_NEGATIVE(dp0);			\
+      const GLboolean neg_dp1 = IS_NEGATIVE(dp1);			\
+									\
+      /* For regular clipping, we know from the clipmask that one	\
+       * (or both) of these must be negative (otherwise we wouldn't	\
+       * be here).							\
+       * For userclip, there is only a single bit for all active	\
+       * planes, so we can end up here when there is nothing to do,	\
+       * hence the second IS_NEGATIVE() test:				\
+       */								\
+      if (neg_dp0 && neg_dp1)						\
+         return; /* both vertices outside clip plane: discard */	\
+									\
+      if (neg_dp1) {							\
+	 GLfloat t = dp1 / (dp1 - dp0);					\
+	 if (t > t1) t1 = t;						\
+      } else if (neg_dp0) {						\
+	 GLfloat t = dp0 / (dp0 - dp1);					\
+	 if (t > t0) t0 = t;						\
+      }									\
+      if (t0 + t1 >= 1.0)						\
+	 return; /* discard */						\
+   }									\
+} while (0)
+
+
 
 /* Clip a line against the viewport and user clip planes.
  */
@@ -139,11 +222,7 @@ TAG(clip_line)( GLcontext *ctx, GLuint v0, GLuint v1, GLubyte mask )
    if (mask & CLIP_USER_BIT) {
       for (p = 0; p < ctx->Const.MaxClipPlanes; p++) {
 	 if (ctx->Transform.ClipPlanesEnabled & (1 << p)) {
-            const GLfloat a = ctx->Transform._ClipUserPlane[p][0];
-            const GLfloat b = ctx->Transform._ClipUserPlane[p][1];
-            const GLfloat c = ctx->Transform._ClipUserPlane[p][2];
-            const GLfloat d = ctx->Transform._ClipUserPlane[p][3];
-	    LINE_CLIP( CLIP_USER_BIT, a, b, c, d );
+	    LINE_USERCLIP(p);
 	 }
       }
    }
@@ -228,11 +307,7 @@ TAG(clip_tri)( GLcontext *ctx, GLuint v0, GLuint v1, GLuint v2, GLubyte mask )
    if (mask & CLIP_USER_BIT) {
       for (p = 0; p < ctx->Const.MaxClipPlanes; p++) {
          if (ctx->Transform.ClipPlanesEnabled & (1 << p)) {
-            const GLfloat a = ctx->Transform._ClipUserPlane[p][0];
-            const GLfloat b = ctx->Transform._ClipUserPlane[p][1];
-            const GLfloat c = ctx->Transform._ClipUserPlane[p][2];
-            const GLfloat d = ctx->Transform._ClipUserPlane[p][3];
-            POLY_CLIP( CLIP_USER_BIT, a, b, c, d );
+            POLY_USERCLIP(p);
          }
       }
    }
@@ -291,11 +366,7 @@ TAG(clip_quad)( GLcontext *ctx, GLuint v0, GLuint v1, GLuint v2, GLuint v3,
    if (mask & CLIP_USER_BIT) {
       for (p = 0; p < ctx->Const.MaxClipPlanes; p++) {
 	 if (ctx->Transform.ClipPlanesEnabled & (1 << p)) {
-            const GLfloat a = ctx->Transform._ClipUserPlane[p][0];
-            const GLfloat b = ctx->Transform._ClipUserPlane[p][1];
-            const GLfloat c = ctx->Transform._ClipUserPlane[p][2];
-            const GLfloat d = ctx->Transform._ClipUserPlane[p][3];
-	    POLY_CLIP( CLIP_USER_BIT, a, b, c, d );
+	    POLY_USERCLIP(p);
 	 }
       }
    }
@@ -317,4 +388,6 @@ TAG(clip_quad)( GLcontext *ctx, GLuint v0, GLuint v1, GLuint v2, GLuint v3,
 #undef SIZE
 #undef TAG
 #undef POLY_CLIP
+#undef POLY_USERCLIP
 #undef LINE_CLIP
+#undef LINE_USERCLIP
diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c
index dc954bcba1..5fb83c2b01 100644
--- a/src/mesa/tnl/t_vb_program.c
+++ b/src/mesa/tnl/t_vb_program.c
@@ -66,6 +66,7 @@ struct vp_stage_data {
    GLvector4f results[VERT_RESULT_MAX];
 
    GLvector4f ndcCoords;              /**< normalized device coords */
+   GLfloat *clipdistance[MAX_CLIP_PLANES];
    GLubyte *clipmask;                 /**< clip flags */
    GLubyte ormask, andmask;           /**< for clipping */
 };
@@ -77,6 +78,7 @@ struct vp_stage_data {
 static void
 userclip( GLcontext *ctx,
           GLvector4f *clip,
+          GLfloat *clipdistance[MAX_CLIP_PLANES],
           GLubyte *clipmask,
           GLubyte *clipormask,
           GLubyte *clipandmask )
@@ -105,6 +107,8 @@ userclip( GLcontext *ctx,
 	       clipmask[i] |= CLIP_USER_BIT;
 	    }
 
+	    clipdistance[p][i] = dp;
+
 	    STRIDE_F(coord, stride);
 	 }
 
@@ -131,13 +135,16 @@ do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store)
    store->ormask = 0;
    store->andmask = CLIP_FRUSTUM_BITS;
 
+   tnl_clip_prepare(ctx);
+
    if (tnl->NeedNdcCoords) {
       VB->NdcPtr =
          _mesa_clip_tab[VB->ClipPtr->size]( VB->ClipPtr,
                                             &store->ndcCoords,
                                             store->clipmask,
                                             &store->ormask,
-                                            &store->andmask );
+                                            &store->andmask,
+					    !ctx->Transform.DepthClamp );
    }
    else {
       VB->NdcPtr = NULL;
@@ -145,7 +152,8 @@ do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store)
                                             NULL,
                                             store->clipmask,
                                             &store->ormask,
-                                            &store->andmask );
+                                            &store->andmask,
+					    !ctx->Transform.DepthClamp );
    }
 
    if (store->andmask) {
@@ -160,6 +168,7 @@ do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store)
       ctx->VertexProgram.Current->IsPositionInvariant)) {
       userclip( ctx,
 		VB->ClipPtr,
+		store->clipdistance,
 		store->clipmask,
 		&store->ormask,
 		&store->andmask );
@@ -167,6 +176,9 @@ do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store)
       if (store->andmask) {
 	 return GL_FALSE;
       }
+
+      memcpy(VB->ClipDistancePtr, store->clipdistance,
+	     sizeof(store->clipdistance));
    }
 
    VB->ClipAndMask = store->andmask;
@@ -510,6 +522,10 @@ init_vp(GLcontext *ctx, struct tnl_pipeline_stage *stage)
    _mesa_vector4f_alloc( &store->ndcCoords, 0, size, 32 );
    store->clipmask = (GLubyte *) ALIGN_MALLOC(sizeof(GLubyte)*size, 32 );
 
+   for (i = 0; i < MAX_CLIP_PLANES; i++)
+      store->clipdistance[i] =
+	 (GLfloat *) ALIGN_MALLOC(sizeof(GLfloat) * size, 32);
+
    return GL_TRUE;
 }
 
@@ -533,6 +549,9 @@ dtr(struct tnl_pipeline_stage *stage)
       _mesa_vector4f_free( &store->ndcCoords );
       ALIGN_FREE( store->clipmask );
 
+      for (i = 0; i < MAX_CLIP_PLANES; i++)
+	 ALIGN_FREE(store->clipdistance[i]);
+
       FREE( store );
       stage->privatePtr = NULL;
    }
diff --git a/src/mesa/tnl/t_vb_vertex.c b/src/mesa/tnl/t_vb_vertex.c
index 30aa7c4086..2a61ff1177 100644
--- a/src/mesa/tnl/t_vb_vertex.c
+++ b/src/mesa/tnl/t_vb_vertex.c
@@ -44,6 +44,7 @@ struct vertex_stage_data {
    GLvector4f eye;
    GLvector4f clip;
    GLvector4f proj;
+   GLfloat *clipdistance[MAX_CLIP_PLANES];
    GLubyte *clipmask;
    GLubyte ormask;
    GLubyte andmask;
@@ -56,11 +57,12 @@ struct vertex_stage_data {
 
 /* This function implements cliptesting for user-defined clip planes.
  * The clipping of primitives to these planes is implemented in
- * t_render_clip.h.
+ * t_vp_cliptmp.h.
  */
 #define USER_CLIPTEST(NAME, SZ)					\
 static void NAME( GLcontext *ctx,				\
 		  GLvector4f *clip,				\
+		  GLfloat *clipdistances[MAX_CLIP_PLANES],	\
 		  GLubyte *clipmask,				\
 		  GLubyte *clipormask,				\
 		  GLubyte *clipandmask )			\
@@ -88,6 +90,8 @@ static void NAME( GLcontext *ctx,				\
 	       clipmask[i] |= CLIP_USER_BIT;			\
 	    }							\
 								\
+	    clipdistances[p][i] = dp;				\
+								\
 	    STRIDE_F(coord, stride);				\
 	 }							\
 								\
@@ -107,8 +111,9 @@ USER_CLIPTEST(userclip3, 3)
 USER_CLIPTEST(userclip4, 4)
 
 static void (*(usercliptab[5]))( GLcontext *,
-				 GLvector4f *, GLubyte *,
-				 GLubyte *, GLubyte * ) =
+				 GLvector4f *,
+				 GLfloat *[MAX_CLIP_PLANES],
+				 GLubyte *, GLubyte *, GLubyte * ) =
 {
    NULL,
    NULL,
@@ -118,6 +123,22 @@ static void (*(usercliptab[5]))( GLcontext *,
 };
 
 
+void
+tnl_clip_prepare(GLcontext *ctx)
+{
+   /* Neither the x86 nor sparc asm cliptest functions have been updated
+    * for ARB_depth_clamp, so force the C paths.
+    */
+   if (ctx->Transform.DepthClamp) {
+      static GLboolean c_funcs_installed = GL_FALSE;
+      if (!c_funcs_installed) {
+         init_c_cliptest();
+         c_funcs_installed = GL_TRUE;
+      }
+   }
+}
+
+
 
 static GLboolean run_vertex_stage( GLcontext *ctx,
 				   struct tnl_pipeline_stage *stage )
@@ -129,6 +150,8 @@ static GLboolean run_vertex_stage( GLcontext *ctx,
    if (ctx->VertexProgram._Current) 
       return GL_TRUE;
 
+   tnl_clip_prepare(ctx);
+
    if (ctx->_NeedEyeCoords) {
       /* Separate modelview transformation:
        * Use combined ModelProject to avoid some depth artifacts
@@ -173,7 +196,8 @@ static GLboolean run_vertex_stage( GLcontext *ctx,
 					    &store->proj,
 					    store->clipmask,
 					    &store->ormask,
-					    &store->andmask );
+					    &store->andmask,
+					    !ctx->Transform.DepthClamp );
    }
    else {
       VB->NdcPtr = NULL;
@@ -181,7 +205,8 @@ static GLboolean run_vertex_stage( GLcontext *ctx,
 					    NULL,
 					    store->clipmask,
 					    &store->ormask,
-					    &store->andmask );
+					    &store->andmask,
+					    !ctx->Transform.DepthClamp );
    }
 
    if (store->andmask)
@@ -194,12 +219,16 @@ static GLboolean run_vertex_stage( GLcontext *ctx,
    if (ctx->Transform.ClipPlanesEnabled) {
       usercliptab[VB->ClipPtr->size]( ctx,
 				      VB->ClipPtr,
+				      store->clipdistance,
 				      store->clipmask,
 				      &store->ormask,
 				      &store->andmask );
 
       if (store->andmask)
 	 return GL_FALSE;
+
+      memcpy(VB->ClipDistancePtr, store->clipdistance,
+	     sizeof(store->clipdistance));
    }
 
    VB->ClipAndMask = store->andmask;
@@ -216,6 +245,7 @@ static GLboolean init_vertex_stage( GLcontext *ctx,
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
    struct vertex_stage_data *store;
    GLuint size = VB->Size;
+   unsigned i;
 
    stage->privatePtr = CALLOC(sizeof(*store));
    store = VERTEX_STAGE_DATA(stage);
@@ -227,8 +257,17 @@ static GLboolean init_vertex_stage( GLcontext *ctx,
    _mesa_vector4f_alloc( &store->proj, 0, size, 32 );
 
    store->clipmask = (GLubyte *) ALIGN_MALLOC(sizeof(GLubyte)*size, 32 );
+   for (i = 0; i < MAX_CLIP_PLANES; i++)
+      store->clipdistance[i] =
+	 (GLfloat *) ALIGN_MALLOC(sizeof(GLfloat) * size, 32);
 
    if (!store->clipmask ||
+       !store->clipdistance[0] ||
+       !store->clipdistance[1] ||
+       !store->clipdistance[2] ||
+       !store->clipdistance[3] ||
+       !store->clipdistance[4] ||
+       !store->clipdistance[5] ||
        !store->eye.data ||
        !store->clip.data ||
        !store->proj.data)
@@ -242,10 +281,16 @@ static void dtr( struct tnl_pipeline_stage *stage )
    struct vertex_stage_data *store = VERTEX_STAGE_DATA(stage);
 
    if (store) {
+      unsigned i;
+
       _mesa_vector4f_free( &store->eye );
       _mesa_vector4f_free( &store->clip );
       _mesa_vector4f_free( &store->proj );
       ALIGN_FREE( store->clipmask );
+
+      for (i = 0; i < MAX_CLIP_PLANES; i++)
+	 ALIGN_FREE(store->clipdistance[i]);
+
       FREE(store);
       stage->privatePtr = NULL;
       stage->run = init_vertex_stage;
diff --git a/src/mesa/vbo/vbo.h b/src/mesa/vbo/vbo.h
index 5986e93576..b24ecfd7cd 100644
--- a/src/mesa/vbo/vbo.h
+++ b/src/mesa/vbo/vbo.h
@@ -44,6 +44,7 @@ struct _mesa_prim {
 
    GLuint start;
    GLuint count;
+   GLint basevertex;
 };
 
 /* Would like to call this a "vbo_index_buffer", but this would be
diff --git a/src/mesa/vbo/vbo_exec.h b/src/mesa/vbo/vbo_exec.h
index e0f44892cf..516be0995b 100644
--- a/src/mesa/vbo/vbo_exec.h
+++ b/src/mesa/vbo/vbo_exec.h
@@ -161,8 +161,26 @@ void vbo_exec_array_destroy( struct vbo_exec_context *exec );
 
 void vbo_exec_vtx_init( struct vbo_exec_context *exec );
 void vbo_exec_vtx_destroy( struct vbo_exec_context *exec );
+
+#if FEATURE_beginend
+
 void vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap );
 void vbo_exec_vtx_map( struct vbo_exec_context *exec );
+
+#else /* FEATURE_beginend */
+
+static INLINE void
+vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap )
+{
+}
+
+static INLINE void
+vbo_exec_vtx_map( struct vbo_exec_context *exec )
+{
+}
+
+#endif /* FEATURE_beginend */
+
 void vbo_exec_vtx_wrap( struct vbo_exec_context *exec );
 
 void vbo_exec_eval_update( struct vbo_exec_context *exec );
diff --git a/src/mesa/vbo/vbo_exec_api.c b/src/mesa/vbo/vbo_exec_api.c
index 387d4ee3d4..f88df5aac7 100644
--- a/src/mesa/vbo/vbo_exec_api.c
+++ b/src/mesa/vbo/vbo_exec_api.c
@@ -35,9 +35,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/context.h"
 #include "main/macros.h"
 #include "main/vtxfmt.h"
-#if FEATURE_dlist
 #include "main/dlist.h"
-#endif
+#include "main/eval.h"
 #include "main/state.h"
 #include "main/light.h"
 #include "main/api_arrayelt.h"
@@ -58,7 +57,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 static void reset_attrfv( struct vbo_exec_context *exec );
 
 
-/* Close off the last primitive, execute the buffer, restart the
+/**
+ * Close off the last primitive, execute the buffer, restart the
  * primitive.  
  */
 static void vbo_exec_wrap_buffers( struct vbo_exec_context *exec )
@@ -107,7 +107,8 @@ static void vbo_exec_wrap_buffers( struct vbo_exec_context *exec )
 }
 
 
-/* Deal with buffer wrapping where provoked by the vertex buffer
+/**
+ * Deal with buffer wrapping where provoked by the vertex buffer
  * filling up, as opposed to upgrade_vertex().
  */
 void vbo_exec_vtx_wrap( struct vbo_exec_context *exec )
@@ -136,7 +137,7 @@ void vbo_exec_vtx_wrap( struct vbo_exec_context *exec )
 }
 
 
-/*
+/**
  * Copy the active vertex's values to the ctx->Current fields.
  */
 static void vbo_exec_copy_to_current( struct vbo_exec_context *exec )
@@ -209,7 +210,8 @@ static void vbo_exec_copy_from_current( struct vbo_exec_context *exec )
 }
 
 
-/* Flush existing data, set new attrib size, replay copied vertices.
+/**
+ * Flush existing data, set new attrib size, replay copied vertices.
  */ 
 static void vbo_exec_wrap_upgrade_vertex( struct vbo_exec_context *exec,
 					  GLuint attr,
@@ -349,7 +351,7 @@ static void vbo_exec_fixup_vertex( GLcontext *ctx,
 }
 
 
-
+#if FEATURE_beginend
 
 /* 
  */
@@ -392,8 +394,8 @@ do {								\
 
 
 
-/* Eval
- */
+#if FEATURE_evaluators
+
 static void GLAPIENTRY vbo_exec_EvalCoord1f( GLfloat u )
 {
    GET_CURRENT_CONTEXT( ctx );
@@ -485,9 +487,15 @@ static void GLAPIENTRY vbo_exec_EvalPoint2( GLint i, GLint j )
    vbo_exec_EvalCoord2f( u, v );
 }
 
+/* use noop eval mesh */
+#define vbo_exec_EvalMesh1 _mesa_noop_EvalMesh1
+#define vbo_exec_EvalMesh2 _mesa_noop_EvalMesh2
+
+#endif /* FEATURE_evaluators */
 
-/* Build a list of primitives on the fly.  Keep
- * ctx->Driver.CurrentExecPrimitive uptodate as well.
+
+/**
+ * Called via glBegin.
  */
 static void GLAPIENTRY vbo_exec_Begin( GLenum mode )
 {
@@ -531,6 +539,10 @@ static void GLAPIENTRY vbo_exec_Begin( GLenum mode )
       
 }
 
+
+/**
+ * Called via glEnd.
+ */
 static void GLAPIENTRY vbo_exec_End( void )
 {
    GET_CURRENT_CONTEXT( ctx ); 
@@ -557,24 +569,15 @@ static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec )
 {
    GLvertexformat *vfmt = &exec->vtxfmt;
 
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
+   _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_);
+
    vfmt->Begin = vbo_exec_Begin;
-#if FEATURE_dlist
-   vfmt->CallList = _mesa_CallList;
-   vfmt->CallLists = _mesa_CallLists;
-#endif
    vfmt->End = vbo_exec_End;
-   vfmt->EvalCoord1f = vbo_exec_EvalCoord1f;
-   vfmt->EvalCoord1fv = vbo_exec_EvalCoord1fv;
-   vfmt->EvalCoord2f = vbo_exec_EvalCoord2f;
-   vfmt->EvalCoord2fv = vbo_exec_EvalCoord2fv;
-   vfmt->EvalPoint1 = vbo_exec_EvalPoint1;
-   vfmt->EvalPoint2 = vbo_exec_EvalPoint2;
 
-   vfmt->Rectf = _mesa_noop_Rectf;
-   vfmt->EvalMesh1 = _mesa_noop_EvalMesh1;
-   vfmt->EvalMesh2 = _mesa_noop_EvalMesh2;
+   _MESA_INIT_DLIST_VTXFMT(vfmt, _mesa_);
+   _MESA_INIT_EVAL_VTXFMT(vfmt, vbo_exec_);
 
+   vfmt->Rectf = _mesa_noop_Rectf;
 
    /* from attrib_tmp.h:
     */
@@ -638,6 +641,98 @@ static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec )
 }
 
 
+#else /* FEATURE_beginend */
+
+
+#define ATTR( A, N, V0, V1, V2, V3 )				\
+do {								\
+   struct vbo_exec_context *exec = &vbo_context(ctx)->exec;	\
+								\
+   /* FLUSH_UPDATE_CURRENT needs to be set manually */		\
+   exec->ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;		\
+								\
+   if (exec->vtx.active_sz[A] != N)				\
+      vbo_exec_fixup_vertex(ctx, A, N);				\
+								\
+   {								\
+      GLfloat *dest = exec->vtx.attrptr[A];			\
+      if (N>0) dest[0] = V0;					\
+      if (N>1) dest[1] = V1;					\
+      if (N>2) dest[2] = V2;					\
+      if (N>3) dest[3] = V3;					\
+   }								\
+} while (0)
+
+#define ERROR() _mesa_error( ctx, GL_INVALID_ENUM, __FUNCTION__ )
+#define TAG(x) vbo_##x
+
+#include "vbo_attrib_tmp.h"
+
+static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec )
+{
+   /* silence warnings */
+   (void) vbo_Color3f;
+   (void) vbo_Color3fv;
+   (void) vbo_Color4f;
+   (void) vbo_Color4fv;
+   (void) vbo_FogCoordfEXT;
+   (void) vbo_FogCoordfvEXT;
+   (void) vbo_MultiTexCoord1f;
+   (void) vbo_MultiTexCoord1fv;
+   (void) vbo_MultiTexCoord2f;
+   (void) vbo_MultiTexCoord2fv;
+   (void) vbo_MultiTexCoord3f;
+   (void) vbo_MultiTexCoord3fv;
+   (void) vbo_MultiTexCoord4f;
+   (void) vbo_MultiTexCoord4fv;
+   (void) vbo_Normal3f;
+   (void) vbo_Normal3fv;
+   (void) vbo_SecondaryColor3fEXT;
+   (void) vbo_SecondaryColor3fvEXT;
+   (void) vbo_TexCoord1f;
+   (void) vbo_TexCoord1fv;
+   (void) vbo_TexCoord2f;
+   (void) vbo_TexCoord2fv;
+   (void) vbo_TexCoord3f;
+   (void) vbo_TexCoord3fv;
+   (void) vbo_TexCoord4f;
+   (void) vbo_TexCoord4fv;
+   (void) vbo_Vertex2f;
+   (void) vbo_Vertex2fv;
+   (void) vbo_Vertex3f;
+   (void) vbo_Vertex3fv;
+   (void) vbo_Vertex4f;
+   (void) vbo_Vertex4fv;
+
+   (void) vbo_VertexAttrib1fARB;
+   (void) vbo_VertexAttrib1fvARB;
+   (void) vbo_VertexAttrib2fARB;
+   (void) vbo_VertexAttrib2fvARB;
+   (void) vbo_VertexAttrib3fARB;
+   (void) vbo_VertexAttrib3fvARB;
+   (void) vbo_VertexAttrib4fARB;
+   (void) vbo_VertexAttrib4fvARB;
+
+   (void) vbo_VertexAttrib1fNV;
+   (void) vbo_VertexAttrib1fvNV;
+   (void) vbo_VertexAttrib2fNV;
+   (void) vbo_VertexAttrib2fvNV;
+   (void) vbo_VertexAttrib3fNV;
+   (void) vbo_VertexAttrib3fvNV;
+   (void) vbo_VertexAttrib4fNV;
+   (void) vbo_VertexAttrib4fvNV;
+
+   (void) vbo_Materialfv;
+
+   (void) vbo_EdgeFlag;
+   (void) vbo_Indexf;
+   (void) vbo_Indexfv;
+}
+
+
+#endif /* FEATURE_beginend */
+
+
 /**
  * Tell the VBO module to use a real OpenGL vertex buffer object to
  * store accumulated immediate-mode vertex data.
diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c
index 39c2957631..fd70b57b72 100644
--- a/src/mesa/vbo/vbo_exec_array.c
+++ b/src/mesa/vbo/vbo_exec_array.c
@@ -33,6 +33,7 @@
 #include "main/api_noop.h"
 #include "main/varray.h"
 #include "main/bufferobj.h"
+#include "main/enums.h"
 #include "main/macros.h"
 #include "glapi/dispatch.h"
 
@@ -53,10 +54,9 @@ vbo_get_minmax_index(GLcontext *ctx,
    const void *indices;
 
    if (_mesa_is_bufferobj(ib->obj)) {
-      const GLvoid *map = ctx->Driver.MapBuffer(ctx,
-                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
-                                                GL_READ_ONLY,
-                                                ib->obj);
+      const GLvoid *map =
+         ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB,
+                               GL_READ_ONLY, ib->obj);
       indices = ADD_POINTERS(map, ib->ptr);
    } else {
       indices = ib->ptr;
@@ -105,9 +105,7 @@ vbo_get_minmax_index(GLcontext *ctx,
    }
 
    if (_mesa_is_bufferobj(ib->obj)) {
-      ctx->Driver.UnmapBuffer(ctx,
-			      GL_ELEMENT_ARRAY_BUFFER_ARB,
-			      ib->obj);
+      ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, ib->obj);
    }
 }
 
@@ -125,10 +123,9 @@ check_array_data(GLcontext *ctx, struct gl_client_array *array,
       if (_mesa_is_bufferobj(array->BufferObj)) {
          if (!array->BufferObj->Pointer) {
             /* need to map now */
-            array->BufferObj->Pointer = ctx->Driver.MapBuffer(ctx,
-                                                              GL_ARRAY_BUFFER_ARB,
-                                                              GL_READ_ONLY,
-                                                              array->BufferObj);
+            array->BufferObj->Pointer =
+               ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER_ARB,
+                                     GL_READ_ONLY, array->BufferObj);
          }
          data = ADD_POINTERS(data, array->BufferObj->Pointer);
       }
@@ -169,9 +166,7 @@ unmap_array_buffer(GLcontext *ctx, struct gl_client_array *array)
    if (array->Enabled &&
        _mesa_is_bufferobj(array->BufferObj) &&
        _mesa_bufferobj_mapped(array->BufferObj)) {
-      ctx->Driver.UnmapBuffer(ctx,
-                              GL_ARRAY_BUFFER_ARB,
-                              array->BufferObj);
+      ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER_ARB, array->BufferObj);
    }
 }
 
@@ -181,7 +176,7 @@ unmap_array_buffer(GLcontext *ctx, struct gl_client_array *array)
  */
 static void
 check_draw_elements_data(GLcontext *ctx, GLsizei count, GLenum elemType,
-                         const void *elements)
+                         const void *elements, GLint basevertex)
 {
    struct gl_array_object *arrayObj = ctx->Array.ArrayObj;
    const void *elemMap;
@@ -222,13 +217,13 @@ check_draw_elements_data(GLcontext *ctx, GLsizei count, GLenum elemType,
          check_array_data(ctx, &arrayObj->TexCoord[k], VERT_ATTRIB_TEX0 + k, j);
       }
       for (k = 0; k < Elements(arrayObj->VertexAttrib); k++) {
-         check_array_data(ctx, &arrayObj->VertexAttrib[k], VERT_ATTRIB_GENERIC0 + k, j);
+         check_array_data(ctx, &arrayObj->VertexAttrib[k],
+                          VERT_ATTRIB_GENERIC0 + k, j);
       }
    }
 
    if (_mesa_is_bufferobj(ctx->Array.ElementArrayBufferObj)) {
-      ctx->Driver.UnmapBuffer(ctx,
-			      GL_ELEMENT_ARRAY_BUFFER_ARB,
+      ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB,
 			      ctx->Array.ElementArrayBufferObj);
    }
 
@@ -459,7 +454,6 @@ bind_arrays(GLcontext *ctx)
    }
    else if (exec->array.program_mode != get_program_mode(ctx) ||
 	    exec->array.enabled_flags != ctx->Array.ArrayObj->_Enabled) {
-      
       recalculate_input_bindings(ctx);
    }
 #else
@@ -482,6 +476,10 @@ vbo_exec_DrawArrays(GLenum mode, GLint start, GLsizei count)
    struct vbo_exec_context *exec = &vbo->exec;
    struct _mesa_prim prim[1];
 
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawArrays(%s, %d, %d)\n",
+                  _mesa_lookup_enum_by_nr(mode), start, count);
+
    if (!_mesa_validate_DrawArrays( ctx, mode, start, count ))
       return;
 
@@ -518,6 +516,7 @@ vbo_exec_DrawArrays(GLenum mode, GLint start, GLsizei count)
    prim[0].start = start;
    prim[0].count = count;
    prim[0].indexed = 0;
+   prim[0].basevertex = 0;
 
    vbo->draw_prims( ctx, exec->array.inputs, prim, 1, NULL,
                     GL_TRUE, start, start + count - 1 );
@@ -581,18 +580,19 @@ dump_element_buffer(GLcontext *ctx, GLenum type)
       ;
    }
 
-   ctx->Driver.UnmapBuffer(ctx,
-                           GL_ELEMENT_ARRAY_BUFFER_ARB,
+   ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB,
                            ctx->Array.ElementArrayBufferObj);
 }
 
+
 /* Inner support for both _mesa_DrawElements and _mesa_DrawRangeElements */
 static void
 vbo_validated_drawrangeelements(GLcontext *ctx, GLenum mode,
 				GLboolean index_bounds_valid,
 				GLuint start, GLuint end,
 				GLsizei count, GLenum type,
-				const GLvoid *indices)
+				const GLvoid *indices,
+				GLint basevertex)
 {
    struct vbo_context *vbo = vbo_context(ctx);
    struct vbo_exec_context *exec = &vbo->exec;
@@ -626,6 +626,7 @@ vbo_validated_drawrangeelements(GLcontext *ctx, GLenum mode,
    prim[0].start = 0;
    prim[0].count = count;
    prim[0].indexed = 1;
+   prim[0].basevertex = basevertex;
 
    /* Need to give special consideration to rendering a range of
     * indices starting somewhere above zero.  Typically the
@@ -663,17 +664,31 @@ vbo_validated_drawrangeelements(GLcontext *ctx, GLenum mode,
 }
 
 static void GLAPIENTRY
-vbo_exec_DrawRangeElements(GLenum mode,
-			   GLuint start, GLuint end,
-			   GLsizei count, GLenum type, const GLvoid *indices)
+vbo_exec_DrawRangeElementsBaseVertex(GLenum mode,
+				     GLuint start, GLuint end,
+				     GLsizei count, GLenum type,
+				     const GLvoid *indices,
+				     GLint basevertex)
 {
    static GLuint warnCount = 0;
    GET_CURRENT_CONTEXT(ctx);
 
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                "glDrawRangeElementsBaseVertex(%s, %u, %u, %d, %s, %p, %d)\n",
+                _mesa_lookup_enum_by_nr(mode), start, end, count,
+                _mesa_lookup_enum_by_nr(type), indices, basevertex);
+
    if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count,
-                                          type, indices ))
+                                          type, indices, basevertex ))
       return;
 
+   /* NOTE: It's important that 'end' is a reasonable value.
+    * in _tnl_draw_prims(), we use end to determine how many vertices
+    * to transform.  If it's too large, we can unnecessarily split prims
+    * or we can read/write out of memory in several different places!
+    */
+
    if (end >= ctx->Array.ArrayObj->_MaxElement) {
       /* the max element is out of bounds of one or more enabled arrays */
       warnCount++;
@@ -726,10 +741,12 @@ vbo_exec_DrawRangeElements(GLenum mode,
 #endif
    }
    else if (0) {
-      _mesa_printf("glDraw[Range]Elements"
-                   "(start %u, end %u, type 0x%x, count %d) ElemBuf %u\n",
+      _mesa_printf("glDraw[Range]Elements{,BaseVertex}"
+                   "(start %u, end %u, type 0x%x, count %d) ElemBuf %u, "
+		   "base %d\n",
                    start, end, type, count,
-                   ctx->Array.ElementArrayBufferObj->Name);
+                   ctx->Array.ElementArrayBufferObj->Name,
+		   basevertex);
    }
 
 #if 0
@@ -739,7 +756,24 @@ vbo_exec_DrawRangeElements(GLenum mode,
 #endif
 
    vbo_validated_drawrangeelements(ctx, mode, GL_TRUE, start, end,
-				   count, type, indices);
+				   count, type, indices, basevertex);
+}
+
+
+static void GLAPIENTRY
+vbo_exec_DrawRangeElements(GLenum mode, GLuint start, GLuint end,
+                           GLsizei count, GLenum type, const GLvoid *indices)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glDrawRangeElements(%s, %u, %u, %d, %s, %p)\n",
+                  _mesa_lookup_enum_by_nr(mode), start, end, count,
+                  _mesa_lookup_enum_by_nr(type), indices);
+
+   vbo_exec_DrawRangeElementsBaseVertex(mode, start, end, count, type,
+					indices, 0);
 }
 
 
@@ -749,18 +783,45 @@ vbo_exec_DrawElements(GLenum mode, GLsizei count, GLenum type,
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawElements(%s, %u, %s, %p)\n",
+                  _mesa_lookup_enum_by_nr(mode), count,
+                  _mesa_lookup_enum_by_nr(type), indices);
+
+   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices, 0 ))
       return;
 
    vbo_validated_drawrangeelements(ctx, mode, GL_FALSE, ~0, ~0,
-				   count, type, indices);
+				   count, type, indices, 0);
 }
 
-/* Inner support for both _mesa_DrawElements and _mesa_DrawRangeElements */
+
+static void GLAPIENTRY
+vbo_exec_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+				const GLvoid *indices, GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawElementsBaseVertex(%s, %d, %s, %p, %d)\n",
+                  _mesa_lookup_enum_by_nr(mode), count,
+                  _mesa_lookup_enum_by_nr(type), indices, basevertex);
+
+   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices,
+				     basevertex ))
+      return;
+
+   vbo_validated_drawrangeelements(ctx, mode, GL_FALSE, ~0, ~0,
+				   count, type, indices, basevertex);
+}
+
+
+/** Inner support for both _mesa_DrawElements and _mesa_DrawRangeElements */
 static void
 vbo_validated_multidrawelements(GLcontext *ctx, GLenum mode,
 				const GLsizei *count, GLenum type,
-				const GLvoid **indices, GLsizei primcount)
+				const GLvoid **indices, GLsizei primcount,
+				const GLint *basevertex)
 {
    struct vbo_context *vbo = vbo_context(ctx);
    struct vbo_exec_context *exec = &vbo->exec;
@@ -854,6 +915,10 @@ vbo_validated_multidrawelements(GLcontext *ctx, GLenum mode,
 	 prim[i].start = ((uintptr_t)indices[i] - min_index_ptr) / index_type_size;
 	 prim[i].count = count[i];
 	 prim[i].indexed = 1;
+	 if (basevertex != NULL)
+	    prim[i].basevertex = basevertex[i];
+	 else
+	    prim[i].basevertex = 0;
       }
 
       vbo->draw_prims(ctx, exec->array.inputs, prim, primcount, &ib,
@@ -874,6 +939,10 @@ vbo_validated_multidrawelements(GLcontext *ctx, GLenum mode,
 	 prim[0].start = 0;
 	 prim[0].count = count[i];
 	 prim[0].indexed = 1;
+	 if (basevertex != NULL)
+	    prim[0].basevertex = basevertex[i];
+	 else
+	    prim[0].basevertex = 0;
       }
 
       vbo->draw_prims(ctx, exec->array.inputs, prim, 1, &ib,
@@ -882,6 +951,7 @@ vbo_validated_multidrawelements(GLcontext *ctx, GLenum mode,
    _mesa_free(prim);
 }
 
+
 static void GLAPIENTRY
 vbo_exec_MultiDrawElements(GLenum mode,
 			   const GLsizei *count, GLenum type,
@@ -894,14 +964,38 @@ vbo_exec_MultiDrawElements(GLenum mode,
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
    for (i = 0; i < primcount; i++) {
-      if (!_mesa_validate_DrawElements( ctx, mode, count[i], type, indices[i] ))
+      if (!_mesa_validate_DrawElements(ctx, mode, count[i], type, indices[i],
+				       0))
 	 return;
    }
 
-   vbo_validated_multidrawelements(ctx, mode, count, type, indices, primcount);
+   vbo_validated_multidrawelements(ctx, mode, count, type, indices, primcount,
+				   NULL);
 }
 
 
+static void GLAPIENTRY
+vbo_exec_MultiDrawElementsBaseVertex(GLenum mode,
+				     const GLsizei *count, GLenum type,
+				     const GLvoid **indices,
+				     GLsizei primcount,
+				     const GLsizei *basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
+
+   for (i = 0; i < primcount; i++) {
+      if (!_mesa_validate_DrawElements(ctx, mode, count[i], type, indices[i],
+				       basevertex[i]))
+	 return;
+   }
+
+   vbo_validated_multidrawelements(ctx, mode, count, type, indices, primcount,
+				   basevertex);
+}
+
 
 /***********************************************************************
  * Initialization
@@ -915,11 +1009,17 @@ vbo_exec_array_init( struct vbo_exec_context *exec )
    exec->vtxfmt.DrawElements = vbo_exec_DrawElements;
    exec->vtxfmt.DrawRangeElements = vbo_exec_DrawRangeElements;
    exec->vtxfmt.MultiDrawElementsEXT = vbo_exec_MultiDrawElements;
+   exec->vtxfmt.DrawElementsBaseVertex = vbo_exec_DrawElementsBaseVertex;
+   exec->vtxfmt.DrawRangeElementsBaseVertex = vbo_exec_DrawRangeElementsBaseVertex;
+   exec->vtxfmt.MultiDrawElementsBaseVertex = vbo_exec_MultiDrawElementsBaseVertex;
 #else
    exec->vtxfmt.DrawArrays = _mesa_noop_DrawArrays;
    exec->vtxfmt.DrawElements = _mesa_noop_DrawElements;
    exec->vtxfmt.DrawRangeElements = _mesa_noop_DrawRangeElements;
    exec->vtxfmt.MultiDrawElementsEXT = _mesa_noop_MultiDrawElements;
+   exec->vtxfmt.DrawElementsBaseVertex = _mesa_noop_DrawElementsBaseVertex;
+   exec->vtxfmt.DrawRangeElementsBaseVertex = _mesa_noop_DrawRangeElementsBaseVertex;
+   exec->vtxfmt.MultiDrawElementsBaseVertex = _mesa_noop_MultiDrawElementsBaseVertex;
 #endif
 }
 
@@ -947,6 +1047,13 @@ _mesa_DrawElements(GLenum mode, GLsizei count, GLenum type,
    vbo_exec_DrawElements(mode, count, type, indices);
 }
 
+void GLAPIENTRY
+_mesa_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+			     const GLvoid *indices, GLint basevertex)
+{
+   vbo_exec_DrawElementsBaseVertex(mode, count, type, indices, basevertex);
+}
+
 
 /* This API entrypoint is not ordinarily used */
 void GLAPIENTRY
@@ -956,6 +1063,17 @@ _mesa_DrawRangeElements(GLenum mode, GLuint start, GLuint end, GLsizei count,
    vbo_exec_DrawRangeElements(mode, start, end, count, type, indices);
 }
 
+
+void GLAPIENTRY
+_mesa_DrawRangeElementsBaseVertex(GLenum mode, GLuint start, GLuint end,
+				  GLsizei count, GLenum type,
+				  const GLvoid *indices, GLint basevertex)
+{
+   vbo_exec_DrawRangeElementsBaseVertex(mode, start, end, count, type,
+					indices, basevertex);
+}
+
+
 /* GL_EXT_multi_draw_arrays */
 void GLAPIENTRY
 _mesa_MultiDrawElementsEXT(GLenum mode, const GLsizei *count, GLenum type,
@@ -963,3 +1081,14 @@ _mesa_MultiDrawElementsEXT(GLenum mode, const GLsizei *count, GLenum type,
 {
    vbo_exec_MultiDrawElements(mode, count, type, indices, primcount);
 }
+
+
+void GLAPIENTRY
+_mesa_MultiDrawElementsBaseVertex(GLenum mode,
+				  const GLsizei *count, GLenum type,
+				  const GLvoid **indices, GLsizei primcount,
+				  const GLint *basevertex)
+{
+   vbo_exec_MultiDrawElementsBaseVertex(mode, count, type, indices,
+					primcount, basevertex);
+}
diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c
index 0c258c535e..ee148df4a1 100644
--- a/src/mesa/vbo/vbo_exec_draw.c
+++ b/src/mesa/vbo/vbo_exec_draw.c
@@ -35,6 +35,9 @@
 #include "vbo_context.h"
 
 
+#if FEATURE_beginend
+
+
 static void
 vbo_exec_debug_verts( struct vbo_exec_context *exec )
 {
@@ -411,3 +414,6 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap )
    exec->vtx.prim_count = 0;
    exec->vtx.vert_count = 0;
 }
+
+
+#endif /* FEATURE_beginend */
diff --git a/src/mesa/vbo/vbo_rebase.c b/src/mesa/vbo/vbo_rebase.c
index 3bf7ef580f..55a82ee369 100644
--- a/src/mesa/vbo/vbo_rebase.c
+++ b/src/mesa/vbo/vbo_rebase.c
@@ -126,7 +126,23 @@ void vbo_rebase_prims( GLcontext *ctx,
    if (0)
       _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index);
 
-   if (ib) {
+
+   /* XXX this path is disabled for now.
+    * There's rendering corruption in some apps when it's enabled.
+    */
+   if (0 && ib && ctx->Extensions.ARB_draw_elements_base_vertex) {
+      /* If we can just tell the hardware or the TNL to interpret our
+       * indices with a different base, do so.
+       */
+      tmp_prims = (struct _mesa_prim *)_mesa_malloc(sizeof(*prim) * nr_prims);
+
+      for (i = 0; i < nr_prims; i++) {
+	 tmp_prims[i] = prim[i];
+	 tmp_prims[i].basevertex -= min_index;
+      }
+
+      prim = tmp_prims;
+   } else if (ib) {
       /* Unfortunately need to adjust each index individually.
        */
       GLboolean map_ib = ib->obj->Name && !ib->obj->Pointer;
diff --git a/src/mesa/vbo/vbo_save_api.c b/src/mesa/vbo/vbo_save_api.c
index 1771510d84..3f86c68b24 100644
--- a/src/mesa/vbo/vbo_save_api.c
+++ b/src/mesa/vbo/vbo_save_api.c
@@ -72,6 +72,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/context.h"
 #include "main/dlist.h"
 #include "main/enums.h"
+#include "main/eval.h"
 #include "main/macros.h"
 #include "main/api_noop.h"
 #include "main/api_validate.h"
@@ -269,7 +270,7 @@ static void _save_compile_vertex_list( GLcontext *ctx )
     * being compiled.
     */
    node = (struct vbo_save_vertex_list *)
-      _mesa_alloc_instruction(ctx, save->opcode_vertex_list, sizeof(*node));
+      _mesa_dlist_alloc(ctx, save->opcode_vertex_list, sizeof(*node));
 
    if (!node)
       return;
@@ -826,6 +827,33 @@ static void GLAPIENTRY _save_DrawRangeElements(GLenum mode,
    _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glDrawRangeElements" );
 }
 
+static void GLAPIENTRY _save_DrawElementsBaseVertex(GLenum mode,
+						    GLsizei count,
+						    GLenum type,
+						    const GLvoid *indices,
+						    GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   (void) mode; (void) count; (void) type; (void) indices; (void)basevertex;
+
+   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glDrawElements" );
+}
+
+static void GLAPIENTRY _save_DrawRangeElementsBaseVertex(GLenum mode,
+							 GLuint start,
+							 GLuint end,
+							 GLsizei count,
+							 GLenum type,
+							 const GLvoid *indices,
+							 GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   (void) mode; (void) start; (void) end; (void) count; (void) type;
+   (void) indices; (void)basevertex;
+
+   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glDrawRangeElements" );
+}
+
 static void GLAPIENTRY _save_DrawArrays(GLenum mode, GLint start, GLsizei count)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -907,7 +935,7 @@ static void GLAPIENTRY _save_OBE_DrawElements(GLenum mode, GLsizei count, GLenum
    GET_CURRENT_CONTEXT(ctx);
    GLint i;
 
-   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
+   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices, 0 ))
       return;
 
    _ae_map_vbos( ctx );
@@ -948,7 +976,7 @@ static void GLAPIENTRY _save_OBE_DrawRangeElements(GLenum mode,
    GET_CURRENT_CONTEXT(ctx);
    if (_mesa_validate_DrawRangeElements( ctx, mode,
 					 start, end,
-					 count, type, indices ))
+					 count, type, indices, 0 ))
       _save_OBE_DrawElements( mode, count, type, indices );
 }
 
@@ -961,7 +989,8 @@ static void _save_vtxfmt_init( GLcontext *ctx )
    struct vbo_save_context *save = &vbo_context(ctx)->save;
    GLvertexformat *vfmt = &save->vtxfmt;
 
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
+   _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_);
+
    vfmt->Begin = _save_Begin;
    vfmt->Color3f = _save_Color3f;
    vfmt->Color3fv = _save_Color3fv;
@@ -1020,28 +1049,23 @@ static void _save_vtxfmt_init( GLcontext *ctx )
    
    /* This will all require us to fallback to saving the list as opcodes:
     */ 
-   vfmt->CallList = _save_CallList; /* inside begin/end */
-   vfmt->CallLists = _save_CallLists; /* inside begin/end */
-   vfmt->EvalCoord1f = _save_EvalCoord1f;
-   vfmt->EvalCoord1fv = _save_EvalCoord1fv;
-   vfmt->EvalCoord2f = _save_EvalCoord2f;
-   vfmt->EvalCoord2fv = _save_EvalCoord2fv;
-   vfmt->EvalPoint1 = _save_EvalPoint1;
-   vfmt->EvalPoint2 = _save_EvalPoint2;
+   _MESA_INIT_DLIST_VTXFMT(vfmt, _save_); /* inside begin/end */
+
+   _MESA_INIT_EVAL_VTXFMT(vfmt, _save_);
 
    /* These are all errors as we at least know we are in some sort of
     * begin/end pair:
     */
-   vfmt->EvalMesh1 = _save_EvalMesh1;	
-   vfmt->EvalMesh2 = _save_EvalMesh2;
    vfmt->Begin = _save_Begin;
    vfmt->Rectf = _save_Rectf;
    vfmt->DrawArrays = _save_DrawArrays;
    vfmt->DrawElements = _save_DrawElements;
    vfmt->DrawRangeElements = _save_DrawRangeElements;
+   vfmt->DrawElementsBaseVertex = _save_DrawElementsBaseVertex;
+   vfmt->DrawRangeElementsBaseVertex = _save_DrawRangeElementsBaseVertex;
    /* Loops back into vfmt->DrawElements */
    vfmt->MultiDrawElementsEXT = _mesa_noop_MultiDrawElements;
-
+   vfmt->MultiDrawElementsBaseVertex = _mesa_noop_MultiDrawElementsBaseVertex;
 }
 
 
@@ -1209,11 +1233,11 @@ void vbo_save_api_init( struct vbo_save_context *save )
    GLuint i;
 
    save->opcode_vertex_list =
-      _mesa_alloc_opcode( ctx,
-			  sizeof(struct vbo_save_vertex_list),
-			  vbo_save_playback_vertex_list,
-			  vbo_destroy_vertex_list,
-			  vbo_print_vertex_list );
+      _mesa_dlist_alloc_opcode( ctx,
+                                sizeof(struct vbo_save_vertex_list),
+                                vbo_save_playback_vertex_list,
+                                vbo_destroy_vertex_list,
+                                vbo_print_vertex_list );
 
    ctx->Driver.NotifySaveBegin = vbo_save_NotifyBegin;
 
@@ -1233,6 +1257,7 @@ void vbo_save_api_init( struct vbo_save_context *save )
    ctx->ListState.ListVtxfmt.DrawRangeElements = _save_OBE_DrawRangeElements;
    /* loops back into _save_OBE_DrawElements */
    ctx->ListState.ListVtxfmt.MultiDrawElementsEXT = _mesa_noop_MultiDrawElements;
+   ctx->ListState.ListVtxfmt.MultiDrawElementsBaseVertex = _mesa_noop_MultiDrawElementsBaseVertex;
    _mesa_install_save_vtxfmt( ctx, &ctx->ListState.ListVtxfmt );
 }
 
diff --git a/src/mesa/vbo/vbo_split.c b/src/mesa/vbo/vbo_split.c
index 58e879628d..c445acca7d 100644
--- a/src/mesa/vbo/vbo_split.c
+++ b/src/mesa/vbo/vbo_split.c
@@ -50,6 +50,7 @@
 #include "main/glheader.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
+#include "main/macros.h"
 
 #include "vbo_split.h"
 #include "vbo.h"
@@ -107,7 +108,12 @@ void vbo_split_prims( GLcontext *ctx,
 		      vbo_draw_func draw,
 		      const struct split_limits *limits )
 {
-  
+   GLuint max_basevertex = prim->basevertex;
+   GLuint i;
+
+   for (i = 1; i < nr_prims; i++)
+      max_basevertex = MAX2(max_basevertex, prim[i].basevertex);
+
    if (ib) {
       if (limits->max_indices == 0) {
 	 /* Could traverse the indices, re-emitting vertices in turn.
diff --git a/src/mesa/vbo/vbo_split_copy.c b/src/mesa/vbo/vbo_split_copy.c
index 8ec180d550..c45190b9dd 100644
--- a/src/mesa/vbo/vbo_split_copy.c
+++ b/src/mesa/vbo/vbo_split_copy.c
@@ -589,28 +589,40 @@ void vbo_split_copy( GLcontext *ctx,
 		     const struct split_limits *limits )
 {
    struct copy_context copy;
-   GLuint i;
+   GLuint i, this_nr_prims;
+
+   for (i = 0; i < nr_prims;) {
+      /* Our SW TNL pipeline doesn't handle basevertex yet, so bind_indices
+       * will rebase the elements to the basevertex, and we'll only
+       * emit strings of prims with the same basevertex in one draw call.
+       */
+      for (this_nr_prims = 1; i + this_nr_prims < nr_prims;
+	   this_nr_prims++) {
+	 if (prim[i].basevertex != prim[i + this_nr_prims].basevertex)
+	    break;
+      }
 
-   memset(&copy, 0, sizeof(copy));
+      memset(&copy, 0, sizeof(copy));
 
-   /* Require indexed primitives:
-    */
-   assert(ib);
-   
-   copy.ctx = ctx;
-   copy.array = arrays;
-   copy.prim = prim;
-   copy.nr_prims = nr_prims;
-   copy.ib = ib;
-   copy.draw = draw;
-   copy.limits = limits;
+      /* Require indexed primitives:
+       */
+      assert(ib);
 
-   /* Clear the vertex cache:
-    */
-   for (i = 0; i < ELT_TABLE_SIZE; i++)
-      copy.vert_cache[i].in = ~0;
+      copy.ctx = ctx;
+      copy.array = arrays;
+      copy.prim = &prim[i];
+      copy.nr_prims = this_nr_prims;
+      copy.ib = ib;
+      copy.draw = draw;
+      copy.limits = limits;
 
-   replay_init(&copy);
-   replay_elts(&copy);
-   replay_finish(&copy);
+      /* Clear the vertex cache:
+       */
+      for (i = 0; i < ELT_TABLE_SIZE; i++)
+	 copy.vert_cache[i].in = ~0;
+
+      replay_init(&copy);
+      replay_elts(&copy);
+      replay_finish(&copy);
+   }
 }
diff --git a/src/mesa/x86-64/glapi_x86-64.S b/src/mesa/x86-64/glapi_x86-64.S
index db917f3982..72d0532906 100644
--- a/src/mesa/x86-64/glapi_x86-64.S
+++ b/src/mesa/x86-64/glapi_x86-64.S
@@ -21601,21 +21601,25 @@ GL_PREFIX(WaitSync):
 	.size	GL_PREFIX(WaitSync), .-GL_PREFIX(WaitSync)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(PolygonOffsetEXT)
-	.type	GL_PREFIX(PolygonOffsetEXT), @function
-GL_PREFIX(PolygonOffsetEXT):
+	.globl	GL_PREFIX(DrawElementsBaseVertex)
+	.type	GL_PREFIX(DrawElementsBaseVertex), @function
+GL_PREFIX(DrawElementsBaseVertex):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
 	movq	4592(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
-	subq	$24, %rsp
-	movq	%xmm0, (%rsp)
-	movq	%xmm1, 8(%rsp)
+	pushq	%rdi
+	pushq	%rsi
+	pushq	%rdx
+	pushq	%rcx
+	pushq	%r8
 	call	_x86_64_get_dispatch@PLT
-	movq	8(%rsp), %xmm1
-	movq	(%rsp), %xmm0
-	addq	$24, %rsp
+	popq	%r8
+	popq	%rcx
+	popq	%rdx
+	popq	%rsi
+	popq	%rdi
 	movq	4592(%rax), %r11
 	jmp	*%r11
 #else
@@ -21625,23 +21629,26 @@ GL_PREFIX(PolygonOffsetEXT):
 	movq	4592(%rax), %r11
 	jmp	*%r11
 1:
-	subq	$24, %rsp
-	movq	%xmm0, (%rsp)
-	movq	%xmm1, 8(%rsp)
+	pushq	%rdi
+	pushq	%rsi
+	pushq	%rdx
+	pushq	%rcx
+	pushq	%r8
 	call	_glapi_get_dispatch
-	movq	8(%rsp), %xmm1
-	movq	(%rsp), %xmm0
-	addq	$24, %rsp
+	popq	%r8
+	popq	%rcx
+	popq	%rdx
+	popq	%rsi
+	popq	%rdi
 	movq	4592(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(PolygonOffsetEXT), .-GL_PREFIX(PolygonOffsetEXT)
+	.size	GL_PREFIX(DrawElementsBaseVertex), .-GL_PREFIX(DrawElementsBaseVertex)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_575)
-	.type	GL_PREFIX(_dispatch_stub_575), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_575))
-GL_PREFIX(_dispatch_stub_575):
+	.globl	GL_PREFIX(DrawRangeElementsBaseVertex)
+	.type	GL_PREFIX(DrawRangeElementsBaseVertex), @function
+GL_PREFIX(DrawRangeElementsBaseVertex):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
 	movq	4600(%rax), %r11
@@ -21649,9 +21656,17 @@ GL_PREFIX(_dispatch_stub_575):
 #elif defined(PTHREADS)
 	pushq	%rdi
 	pushq	%rsi
+	pushq	%rdx
+	pushq	%rcx
+	pushq	%r8
+	pushq	%r9
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
+	popq	%r9
+	popq	%r8
+	popq	%rcx
+	popq	%rdx
 	popq	%rsi
 	popq	%rdi
 	movq	4600(%rax), %r11
@@ -21665,21 +21680,28 @@ GL_PREFIX(_dispatch_stub_575):
 1:
 	pushq	%rdi
 	pushq	%rsi
+	pushq	%rdx
+	pushq	%rcx
+	pushq	%r8
+	pushq	%r9
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
+	popq	%r9
+	popq	%r8
+	popq	%rcx
+	popq	%rdx
 	popq	%rsi
 	popq	%rdi
 	movq	4600(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_575), .-GL_PREFIX(_dispatch_stub_575)
+	.size	GL_PREFIX(DrawRangeElementsBaseVertex), .-GL_PREFIX(DrawRangeElementsBaseVertex)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_576)
-	.type	GL_PREFIX(_dispatch_stub_576), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_576))
-GL_PREFIX(_dispatch_stub_576):
+	.globl	GL_PREFIX(MultiDrawElementsBaseVertex)
+	.type	GL_PREFIX(MultiDrawElementsBaseVertex), @function
+GL_PREFIX(MultiDrawElementsBaseVertex):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
 	movq	4608(%rax), %r11
@@ -21687,9 +21709,17 @@ GL_PREFIX(_dispatch_stub_576):
 #elif defined(PTHREADS)
 	pushq	%rdi
 	pushq	%rsi
+	pushq	%rdx
+	pushq	%rcx
+	pushq	%r8
+	pushq	%r9
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
+	popq	%r9
+	popq	%r8
+	popq	%rcx
+	popq	%rdx
 	popq	%rsi
 	popq	%rdi
 	movq	4608(%rax), %r11
@@ -21703,32 +21733,39 @@ GL_PREFIX(_dispatch_stub_576):
 1:
 	pushq	%rdi
 	pushq	%rsi
+	pushq	%rdx
+	pushq	%rcx
+	pushq	%r8
+	pushq	%r9
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
+	popq	%r9
+	popq	%r8
+	popq	%rcx
+	popq	%rdx
 	popq	%rsi
 	popq	%rdi
 	movq	4608(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_576), .-GL_PREFIX(_dispatch_stub_576)
+	.size	GL_PREFIX(MultiDrawElementsBaseVertex), .-GL_PREFIX(MultiDrawElementsBaseVertex)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_577)
-	.type	GL_PREFIX(_dispatch_stub_577), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_577))
-GL_PREFIX(_dispatch_stub_577):
+	.globl	GL_PREFIX(PolygonOffsetEXT)
+	.type	GL_PREFIX(PolygonOffsetEXT), @function
+GL_PREFIX(PolygonOffsetEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
 	movq	4616(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
-	movq	%rdi, (%rsp)
-	movq	%xmm0, 8(%rsp)
+	movq	%xmm0, (%rsp)
+	movq	%xmm1, 8(%rsp)
 	call	_x86_64_get_dispatch@PLT
-	movq	8(%rsp), %xmm0
-	movq	(%rsp), %rdi
+	movq	8(%rsp), %xmm1
+	movq	(%rsp), %xmm0
 	addq	$24, %rsp
 	movq	4616(%rax), %r11
 	jmp	*%r11
@@ -21740,16 +21777,16 @@ GL_PREFIX(_dispatch_stub_577):
 	jmp	*%r11
 1:
 	subq	$24, %rsp
-	movq	%rdi, (%rsp)
-	movq	%xmm0, 8(%rsp)
+	movq	%xmm0, (%rsp)
+	movq	%xmm1, 8(%rsp)
 	call	_glapi_get_dispatch
-	movq	8(%rsp), %xmm0
-	movq	(%rsp), %rdi
+	movq	8(%rsp), %xmm1
+	movq	(%rsp), %xmm0
 	addq	$24, %rsp
 	movq	4616(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_577), .-GL_PREFIX(_dispatch_stub_577)
+	.size	GL_PREFIX(PolygonOffsetEXT), .-GL_PREFIX(PolygonOffsetEXT)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(_dispatch_stub_578)
@@ -21837,13 +21874,13 @@ GL_PREFIX(_dispatch_stub_580):
 	movq	4640(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
-	pushq	%rdi
-	pushq	%rsi
-	pushq	%rbp
+	subq	$24, %rsp
+	movq	%rdi, (%rsp)
+	movq	%xmm0, 8(%rsp)
 	call	_x86_64_get_dispatch@PLT
-	popq	%rbp
-	popq	%rsi
-	popq	%rdi
+	movq	8(%rsp), %xmm0
+	movq	(%rsp), %rdi
+	addq	$24, %rsp
 	movq	4640(%rax), %r11
 	jmp	*%r11
 #else
@@ -21853,13 +21890,13 @@ GL_PREFIX(_dispatch_stub_580):
 	movq	4640(%rax), %r11
 	jmp	*%r11
 1:
-	pushq	%rdi
-	pushq	%rsi
-	pushq	%rbp
+	subq	$24, %rsp
+	movq	%rdi, (%rsp)
+	movq	%xmm0, 8(%rsp)
 	call	_glapi_get_dispatch
-	popq	%rbp
-	popq	%rsi
-	popq	%rdi
+	movq	8(%rsp), %xmm0
+	movq	(%rsp), %rdi
+	addq	$24, %rsp
 	movq	4640(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
@@ -21914,7 +21951,11 @@ GL_PREFIX(_dispatch_stub_582):
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
+	pushq	%rsi
+	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
+	popq	%rbp
+	popq	%rsi
 	popq	%rdi
 	movq	4656(%rax), %r11
 	jmp	*%r11
@@ -21926,7 +21967,11 @@ GL_PREFIX(_dispatch_stub_582):
 	jmp	*%r11
 1:
 	pushq	%rdi
+	pushq	%rsi
+	pushq	%rbp
 	call	_glapi_get_dispatch
+	popq	%rbp
+	popq	%rsi
 	popq	%rdi
 	movq	4656(%rax), %r11
 	jmp	*%r11
@@ -21934,12 +21979,118 @@ GL_PREFIX(_dispatch_stub_582):
 	.size	GL_PREFIX(_dispatch_stub_582), .-GL_PREFIX(_dispatch_stub_582)
 
 	.p2align	4,,15
+	.globl	GL_PREFIX(_dispatch_stub_583)
+	.type	GL_PREFIX(_dispatch_stub_583), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_583))
+GL_PREFIX(_dispatch_stub_583):
+#if defined(GLX_USE_TLS)
+	call	_x86_64_get_dispatch@PLT
+	movq	4664(%rax), %r11
+	jmp	*%r11
+#elif defined(PTHREADS)
+	pushq	%rdi
+	pushq	%rsi
+	pushq	%rbp
+	call	_x86_64_get_dispatch@PLT
+	popq	%rbp
+	popq	%rsi
+	popq	%rdi
+	movq	4664(%rax), %r11
+	jmp	*%r11
+#else
+	movq	_glapi_Dispatch(%rip), %rax
+	testq	%rax, %rax
+	je	1f
+	movq	4664(%rax), %r11
+	jmp	*%r11
+1:
+	pushq	%rdi
+	pushq	%rsi
+	pushq	%rbp
+	call	_glapi_get_dispatch
+	popq	%rbp
+	popq	%rsi
+	popq	%rdi
+	movq	4664(%rax), %r11
+	jmp	*%r11
+#endif /* defined(GLX_USE_TLS) */
+	.size	GL_PREFIX(_dispatch_stub_583), .-GL_PREFIX(_dispatch_stub_583)
+
+	.p2align	4,,15
+	.globl	GL_PREFIX(_dispatch_stub_584)
+	.type	GL_PREFIX(_dispatch_stub_584), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_584))
+GL_PREFIX(_dispatch_stub_584):
+#if defined(GLX_USE_TLS)
+	call	_x86_64_get_dispatch@PLT
+	movq	4672(%rax), %r11
+	jmp	*%r11
+#elif defined(PTHREADS)
+	pushq	%rdi
+	pushq	%rsi
+	pushq	%rbp
+	call	_x86_64_get_dispatch@PLT
+	popq	%rbp
+	popq	%rsi
+	popq	%rdi
+	movq	4672(%rax), %r11
+	jmp	*%r11
+#else
+	movq	_glapi_Dispatch(%rip), %rax
+	testq	%rax, %rax
+	je	1f
+	movq	4672(%rax), %r11
+	jmp	*%r11
+1:
+	pushq	%rdi
+	pushq	%rsi
+	pushq	%rbp
+	call	_glapi_get_dispatch
+	popq	%rbp
+	popq	%rsi
+	popq	%rdi
+	movq	4672(%rax), %r11
+	jmp	*%r11
+#endif /* defined(GLX_USE_TLS) */
+	.size	GL_PREFIX(_dispatch_stub_584), .-GL_PREFIX(_dispatch_stub_584)
+
+	.p2align	4,,15
+	.globl	GL_PREFIX(_dispatch_stub_585)
+	.type	GL_PREFIX(_dispatch_stub_585), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_585))
+GL_PREFIX(_dispatch_stub_585):
+#if defined(GLX_USE_TLS)
+	call	_x86_64_get_dispatch@PLT
+	movq	4680(%rax), %r11
+	jmp	*%r11
+#elif defined(PTHREADS)
+	pushq	%rdi
+	call	_x86_64_get_dispatch@PLT
+	popq	%rdi
+	movq	4680(%rax), %r11
+	jmp	*%r11
+#else
+	movq	_glapi_Dispatch(%rip), %rax
+	testq	%rax, %rax
+	je	1f
+	movq	4680(%rax), %r11
+	jmp	*%r11
+1:
+	pushq	%rdi
+	call	_glapi_get_dispatch
+	popq	%rdi
+	movq	4680(%rax), %r11
+	jmp	*%r11
+#endif /* defined(GLX_USE_TLS) */
+	.size	GL_PREFIX(_dispatch_stub_585), .-GL_PREFIX(_dispatch_stub_585)
+
+	.p2align	4,,15
 	.globl	GL_PREFIX(ColorPointerEXT)
 	.type	GL_PREFIX(ColorPointerEXT), @function
 GL_PREFIX(ColorPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4664(%rax), %r11
+	movq	4688(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -21953,13 +22104,13 @@ GL_PREFIX(ColorPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4664(%rax), %r11
+	movq	4688(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4664(%rax), %r11
+	movq	4688(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -21973,7 +22124,7 @@ GL_PREFIX(ColorPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4664(%rax), %r11
+	movq	4688(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ColorPointerEXT), .-GL_PREFIX(ColorPointerEXT)
@@ -21984,7 +22135,7 @@ GL_PREFIX(ColorPointerEXT):
 GL_PREFIX(EdgeFlagPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4672(%rax), %r11
+	movq	4696(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -21994,13 +22145,13 @@ GL_PREFIX(EdgeFlagPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4672(%rax), %r11
+	movq	4696(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4672(%rax), %r11
+	movq	4696(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22010,7 +22161,7 @@ GL_PREFIX(EdgeFlagPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4672(%rax), %r11
+	movq	4696(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(EdgeFlagPointerEXT), .-GL_PREFIX(EdgeFlagPointerEXT)
@@ -22021,7 +22172,7 @@ GL_PREFIX(EdgeFlagPointerEXT):
 GL_PREFIX(IndexPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4680(%rax), %r11
+	movq	4704(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22035,13 +22186,13 @@ GL_PREFIX(IndexPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4680(%rax), %r11
+	movq	4704(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4680(%rax), %r11
+	movq	4704(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22055,7 +22206,7 @@ GL_PREFIX(IndexPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4680(%rax), %r11
+	movq	4704(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(IndexPointerEXT), .-GL_PREFIX(IndexPointerEXT)
@@ -22066,7 +22217,7 @@ GL_PREFIX(IndexPointerEXT):
 GL_PREFIX(NormalPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4688(%rax), %r11
+	movq	4712(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22080,13 +22231,13 @@ GL_PREFIX(NormalPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4688(%rax), %r11
+	movq	4712(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4688(%rax), %r11
+	movq	4712(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22100,7 +22251,7 @@ GL_PREFIX(NormalPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4688(%rax), %r11
+	movq	4712(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(NormalPointerEXT), .-GL_PREFIX(NormalPointerEXT)
@@ -22111,7 +22262,7 @@ GL_PREFIX(NormalPointerEXT):
 GL_PREFIX(TexCoordPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4696(%rax), %r11
+	movq	4720(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22125,13 +22276,13 @@ GL_PREFIX(TexCoordPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4696(%rax), %r11
+	movq	4720(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4696(%rax), %r11
+	movq	4720(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22145,7 +22296,7 @@ GL_PREFIX(TexCoordPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4696(%rax), %r11
+	movq	4720(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(TexCoordPointerEXT), .-GL_PREFIX(TexCoordPointerEXT)
@@ -22156,7 +22307,7 @@ GL_PREFIX(TexCoordPointerEXT):
 GL_PREFIX(VertexPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4704(%rax), %r11
+	movq	4728(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22170,13 +22321,13 @@ GL_PREFIX(VertexPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4704(%rax), %r11
+	movq	4728(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4704(%rax), %r11
+	movq	4728(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22190,7 +22341,7 @@ GL_PREFIX(VertexPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4704(%rax), %r11
+	movq	4728(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexPointerEXT), .-GL_PREFIX(VertexPointerEXT)
@@ -22201,7 +22352,7 @@ GL_PREFIX(VertexPointerEXT):
 GL_PREFIX(PointParameterfEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4712(%rax), %r11
+	movq	4736(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -22211,13 +22362,13 @@ GL_PREFIX(PointParameterfEXT):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	4712(%rax), %r11
+	movq	4736(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4712(%rax), %r11
+	movq	4736(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -22227,7 +22378,7 @@ GL_PREFIX(PointParameterfEXT):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	4712(%rax), %r11
+	movq	4736(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(PointParameterfEXT), .-GL_PREFIX(PointParameterfEXT)
@@ -22238,7 +22389,7 @@ GL_PREFIX(PointParameterfEXT):
 GL_PREFIX(PointParameterfvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4720(%rax), %r11
+	movq	4744(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22248,13 +22399,13 @@ GL_PREFIX(PointParameterfvEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4720(%rax), %r11
+	movq	4744(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4720(%rax), %r11
+	movq	4744(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22264,7 +22415,7 @@ GL_PREFIX(PointParameterfvEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4720(%rax), %r11
+	movq	4744(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(PointParameterfvEXT), .-GL_PREFIX(PointParameterfvEXT)
@@ -22275,7 +22426,7 @@ GL_PREFIX(PointParameterfvEXT):
 GL_PREFIX(LockArraysEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4728(%rax), %r11
+	movq	4752(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22285,13 +22436,13 @@ GL_PREFIX(LockArraysEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4728(%rax), %r11
+	movq	4752(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4728(%rax), %r11
+	movq	4752(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22301,7 +22452,7 @@ GL_PREFIX(LockArraysEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4728(%rax), %r11
+	movq	4752(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(LockArraysEXT), .-GL_PREFIX(LockArraysEXT)
@@ -22312,37 +22463,37 @@ GL_PREFIX(LockArraysEXT):
 GL_PREFIX(UnlockArraysEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4736(%rax), %r11
+	movq	4760(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
-	movq	4736(%rax), %r11
+	movq	4760(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4736(%rax), %r11
+	movq	4760(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
-	movq	4736(%rax), %r11
+	movq	4760(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(UnlockArraysEXT), .-GL_PREFIX(UnlockArraysEXT)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_593)
-	.type	GL_PREFIX(_dispatch_stub_593), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_593))
-GL_PREFIX(_dispatch_stub_593):
+	.globl	GL_PREFIX(_dispatch_stub_596)
+	.type	GL_PREFIX(_dispatch_stub_596), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_596))
+GL_PREFIX(_dispatch_stub_596):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4744(%rax), %r11
+	movq	4768(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22352,13 +22503,13 @@ GL_PREFIX(_dispatch_stub_593):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4744(%rax), %r11
+	movq	4768(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4744(%rax), %r11
+	movq	4768(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22368,19 +22519,19 @@ GL_PREFIX(_dispatch_stub_593):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4744(%rax), %r11
+	movq	4768(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_593), .-GL_PREFIX(_dispatch_stub_593)
+	.size	GL_PREFIX(_dispatch_stub_596), .-GL_PREFIX(_dispatch_stub_596)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_594)
-	.type	GL_PREFIX(_dispatch_stub_594), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_594))
-GL_PREFIX(_dispatch_stub_594):
+	.globl	GL_PREFIX(_dispatch_stub_597)
+	.type	GL_PREFIX(_dispatch_stub_597), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_597))
+GL_PREFIX(_dispatch_stub_597):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4752(%rax), %r11
+	movq	4776(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22390,13 +22541,13 @@ GL_PREFIX(_dispatch_stub_594):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4752(%rax), %r11
+	movq	4776(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4752(%rax), %r11
+	movq	4776(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22406,10 +22557,10 @@ GL_PREFIX(_dispatch_stub_594):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4752(%rax), %r11
+	movq	4776(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_594), .-GL_PREFIX(_dispatch_stub_594)
+	.size	GL_PREFIX(_dispatch_stub_597), .-GL_PREFIX(_dispatch_stub_597)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(SecondaryColor3bEXT)
@@ -22417,7 +22568,7 @@ GL_PREFIX(_dispatch_stub_594):
 GL_PREFIX(SecondaryColor3bEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4760(%rax), %r11
+	movq	4784(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22427,13 +22578,13 @@ GL_PREFIX(SecondaryColor3bEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4760(%rax), %r11
+	movq	4784(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4760(%rax), %r11
+	movq	4784(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22443,7 +22594,7 @@ GL_PREFIX(SecondaryColor3bEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4760(%rax), %r11
+	movq	4784(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3bEXT), .-GL_PREFIX(SecondaryColor3bEXT)
@@ -22454,25 +22605,25 @@ GL_PREFIX(SecondaryColor3bEXT):
 GL_PREFIX(SecondaryColor3bvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4768(%rax), %r11
+	movq	4792(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4768(%rax), %r11
+	movq	4792(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4768(%rax), %r11
+	movq	4792(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4768(%rax), %r11
+	movq	4792(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3bvEXT), .-GL_PREFIX(SecondaryColor3bvEXT)
@@ -22483,7 +22634,7 @@ GL_PREFIX(SecondaryColor3bvEXT):
 GL_PREFIX(SecondaryColor3dEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4776(%rax), %r11
+	movq	4800(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -22495,13 +22646,13 @@ GL_PREFIX(SecondaryColor3dEXT):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	4776(%rax), %r11
+	movq	4800(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4776(%rax), %r11
+	movq	4800(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -22513,7 +22664,7 @@ GL_PREFIX(SecondaryColor3dEXT):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	4776(%rax), %r11
+	movq	4800(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3dEXT), .-GL_PREFIX(SecondaryColor3dEXT)
@@ -22524,25 +22675,25 @@ GL_PREFIX(SecondaryColor3dEXT):
 GL_PREFIX(SecondaryColor3dvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4784(%rax), %r11
+	movq	4808(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4784(%rax), %r11
+	movq	4808(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4784(%rax), %r11
+	movq	4808(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4784(%rax), %r11
+	movq	4808(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3dvEXT), .-GL_PREFIX(SecondaryColor3dvEXT)
@@ -22553,7 +22704,7 @@ GL_PREFIX(SecondaryColor3dvEXT):
 GL_PREFIX(SecondaryColor3fEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4792(%rax), %r11
+	movq	4816(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -22565,13 +22716,13 @@ GL_PREFIX(SecondaryColor3fEXT):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	4792(%rax), %r11
+	movq	4816(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4792(%rax), %r11
+	movq	4816(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -22583,7 +22734,7 @@ GL_PREFIX(SecondaryColor3fEXT):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	4792(%rax), %r11
+	movq	4816(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3fEXT), .-GL_PREFIX(SecondaryColor3fEXT)
@@ -22594,25 +22745,25 @@ GL_PREFIX(SecondaryColor3fEXT):
 GL_PREFIX(SecondaryColor3fvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4800(%rax), %r11
+	movq	4824(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4800(%rax), %r11
+	movq	4824(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4800(%rax), %r11
+	movq	4824(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4800(%rax), %r11
+	movq	4824(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3fvEXT), .-GL_PREFIX(SecondaryColor3fvEXT)
@@ -22623,7 +22774,7 @@ GL_PREFIX(SecondaryColor3fvEXT):
 GL_PREFIX(SecondaryColor3iEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4808(%rax), %r11
+	movq	4832(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22633,13 +22784,13 @@ GL_PREFIX(SecondaryColor3iEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4808(%rax), %r11
+	movq	4832(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4808(%rax), %r11
+	movq	4832(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22649,7 +22800,7 @@ GL_PREFIX(SecondaryColor3iEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4808(%rax), %r11
+	movq	4832(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3iEXT), .-GL_PREFIX(SecondaryColor3iEXT)
@@ -22660,25 +22811,25 @@ GL_PREFIX(SecondaryColor3iEXT):
 GL_PREFIX(SecondaryColor3ivEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4816(%rax), %r11
+	movq	4840(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4816(%rax), %r11
+	movq	4840(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4816(%rax), %r11
+	movq	4840(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4816(%rax), %r11
+	movq	4840(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3ivEXT), .-GL_PREFIX(SecondaryColor3ivEXT)
@@ -22689,7 +22840,7 @@ GL_PREFIX(SecondaryColor3ivEXT):
 GL_PREFIX(SecondaryColor3sEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4824(%rax), %r11
+	movq	4848(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22699,13 +22850,13 @@ GL_PREFIX(SecondaryColor3sEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4824(%rax), %r11
+	movq	4848(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4824(%rax), %r11
+	movq	4848(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22715,7 +22866,7 @@ GL_PREFIX(SecondaryColor3sEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4824(%rax), %r11
+	movq	4848(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3sEXT), .-GL_PREFIX(SecondaryColor3sEXT)
@@ -22726,25 +22877,25 @@ GL_PREFIX(SecondaryColor3sEXT):
 GL_PREFIX(SecondaryColor3svEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4832(%rax), %r11
+	movq	4856(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4832(%rax), %r11
+	movq	4856(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4832(%rax), %r11
+	movq	4856(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4832(%rax), %r11
+	movq	4856(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3svEXT), .-GL_PREFIX(SecondaryColor3svEXT)
@@ -22755,7 +22906,7 @@ GL_PREFIX(SecondaryColor3svEXT):
 GL_PREFIX(SecondaryColor3ubEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4840(%rax), %r11
+	movq	4864(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22765,13 +22916,13 @@ GL_PREFIX(SecondaryColor3ubEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4840(%rax), %r11
+	movq	4864(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4840(%rax), %r11
+	movq	4864(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22781,7 +22932,7 @@ GL_PREFIX(SecondaryColor3ubEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4840(%rax), %r11
+	movq	4864(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3ubEXT), .-GL_PREFIX(SecondaryColor3ubEXT)
@@ -22792,25 +22943,25 @@ GL_PREFIX(SecondaryColor3ubEXT):
 GL_PREFIX(SecondaryColor3ubvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4848(%rax), %r11
+	movq	4872(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4848(%rax), %r11
+	movq	4872(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4848(%rax), %r11
+	movq	4872(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4848(%rax), %r11
+	movq	4872(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3ubvEXT), .-GL_PREFIX(SecondaryColor3ubvEXT)
@@ -22821,7 +22972,7 @@ GL_PREFIX(SecondaryColor3ubvEXT):
 GL_PREFIX(SecondaryColor3uiEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4856(%rax), %r11
+	movq	4880(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22831,13 +22982,13 @@ GL_PREFIX(SecondaryColor3uiEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4856(%rax), %r11
+	movq	4880(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4856(%rax), %r11
+	movq	4880(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22847,7 +22998,7 @@ GL_PREFIX(SecondaryColor3uiEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4856(%rax), %r11
+	movq	4880(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3uiEXT), .-GL_PREFIX(SecondaryColor3uiEXT)
@@ -22858,25 +23009,25 @@ GL_PREFIX(SecondaryColor3uiEXT):
 GL_PREFIX(SecondaryColor3uivEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4864(%rax), %r11
+	movq	4888(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4864(%rax), %r11
+	movq	4888(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4864(%rax), %r11
+	movq	4888(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4864(%rax), %r11
+	movq	4888(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3uivEXT), .-GL_PREFIX(SecondaryColor3uivEXT)
@@ -22887,7 +23038,7 @@ GL_PREFIX(SecondaryColor3uivEXT):
 GL_PREFIX(SecondaryColor3usEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4872(%rax), %r11
+	movq	4896(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22897,13 +23048,13 @@ GL_PREFIX(SecondaryColor3usEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4872(%rax), %r11
+	movq	4896(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4872(%rax), %r11
+	movq	4896(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22913,7 +23064,7 @@ GL_PREFIX(SecondaryColor3usEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4872(%rax), %r11
+	movq	4896(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3usEXT), .-GL_PREFIX(SecondaryColor3usEXT)
@@ -22924,25 +23075,25 @@ GL_PREFIX(SecondaryColor3usEXT):
 GL_PREFIX(SecondaryColor3usvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4880(%rax), %r11
+	movq	4904(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4880(%rax), %r11
+	movq	4904(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4880(%rax), %r11
+	movq	4904(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4880(%rax), %r11
+	movq	4904(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColor3usvEXT), .-GL_PREFIX(SecondaryColor3usvEXT)
@@ -22953,7 +23104,7 @@ GL_PREFIX(SecondaryColor3usvEXT):
 GL_PREFIX(SecondaryColorPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4888(%rax), %r11
+	movq	4912(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -22967,13 +23118,13 @@ GL_PREFIX(SecondaryColorPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4888(%rax), %r11
+	movq	4912(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4888(%rax), %r11
+	movq	4912(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -22987,7 +23138,7 @@ GL_PREFIX(SecondaryColorPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4888(%rax), %r11
+	movq	4912(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SecondaryColorPointerEXT), .-GL_PREFIX(SecondaryColorPointerEXT)
@@ -22998,7 +23149,7 @@ GL_PREFIX(SecondaryColorPointerEXT):
 GL_PREFIX(MultiDrawArraysEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4896(%rax), %r11
+	movq	4920(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23012,13 +23163,13 @@ GL_PREFIX(MultiDrawArraysEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4896(%rax), %r11
+	movq	4920(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4896(%rax), %r11
+	movq	4920(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23032,7 +23183,7 @@ GL_PREFIX(MultiDrawArraysEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4896(%rax), %r11
+	movq	4920(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(MultiDrawArraysEXT), .-GL_PREFIX(MultiDrawArraysEXT)
@@ -23043,7 +23194,7 @@ GL_PREFIX(MultiDrawArraysEXT):
 GL_PREFIX(MultiDrawElementsEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4904(%rax), %r11
+	movq	4928(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23057,13 +23208,13 @@ GL_PREFIX(MultiDrawElementsEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4904(%rax), %r11
+	movq	4928(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4904(%rax), %r11
+	movq	4928(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23077,7 +23228,7 @@ GL_PREFIX(MultiDrawElementsEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4904(%rax), %r11
+	movq	4928(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(MultiDrawElementsEXT), .-GL_PREFIX(MultiDrawElementsEXT)
@@ -23088,7 +23239,7 @@ GL_PREFIX(MultiDrawElementsEXT):
 GL_PREFIX(FogCoordPointerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4912(%rax), %r11
+	movq	4936(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23098,13 +23249,13 @@ GL_PREFIX(FogCoordPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4912(%rax), %r11
+	movq	4936(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4912(%rax), %r11
+	movq	4936(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23114,7 +23265,7 @@ GL_PREFIX(FogCoordPointerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4912(%rax), %r11
+	movq	4936(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FogCoordPointerEXT), .-GL_PREFIX(FogCoordPointerEXT)
@@ -23125,7 +23276,7 @@ GL_PREFIX(FogCoordPointerEXT):
 GL_PREFIX(FogCoorddEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4920(%rax), %r11
+	movq	4944(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$8, %rsp
@@ -23133,13 +23284,13 @@ GL_PREFIX(FogCoorddEXT):
 	call	_x86_64_get_dispatch@PLT
 	movq	(%rsp), %xmm0
 	addq	$8, %rsp
-	movq	4920(%rax), %r11
+	movq	4944(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4920(%rax), %r11
+	movq	4944(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$8, %rsp
@@ -23147,7 +23298,7 @@ GL_PREFIX(FogCoorddEXT):
 	call	_glapi_get_dispatch
 	movq	(%rsp), %xmm0
 	addq	$8, %rsp
-	movq	4920(%rax), %r11
+	movq	4944(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FogCoorddEXT), .-GL_PREFIX(FogCoorddEXT)
@@ -23158,25 +23309,25 @@ GL_PREFIX(FogCoorddEXT):
 GL_PREFIX(FogCoorddvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4928(%rax), %r11
+	movq	4952(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4928(%rax), %r11
+	movq	4952(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4928(%rax), %r11
+	movq	4952(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4928(%rax), %r11
+	movq	4952(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FogCoorddvEXT), .-GL_PREFIX(FogCoorddvEXT)
@@ -23187,7 +23338,7 @@ GL_PREFIX(FogCoorddvEXT):
 GL_PREFIX(FogCoordfEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4936(%rax), %r11
+	movq	4960(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$8, %rsp
@@ -23195,13 +23346,13 @@ GL_PREFIX(FogCoordfEXT):
 	call	_x86_64_get_dispatch@PLT
 	movq	(%rsp), %xmm0
 	addq	$8, %rsp
-	movq	4936(%rax), %r11
+	movq	4960(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4936(%rax), %r11
+	movq	4960(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$8, %rsp
@@ -23209,7 +23360,7 @@ GL_PREFIX(FogCoordfEXT):
 	call	_glapi_get_dispatch
 	movq	(%rsp), %xmm0
 	addq	$8, %rsp
-	movq	4936(%rax), %r11
+	movq	4960(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FogCoordfEXT), .-GL_PREFIX(FogCoordfEXT)
@@ -23220,58 +23371,58 @@ GL_PREFIX(FogCoordfEXT):
 GL_PREFIX(FogCoordfvEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4944(%rax), %r11
+	movq	4968(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4944(%rax), %r11
+	movq	4968(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4944(%rax), %r11
+	movq	4968(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4944(%rax), %r11
+	movq	4968(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FogCoordfvEXT), .-GL_PREFIX(FogCoordfvEXT)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_619)
-	.type	GL_PREFIX(_dispatch_stub_619), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_619))
-GL_PREFIX(_dispatch_stub_619):
+	.globl	GL_PREFIX(_dispatch_stub_622)
+	.type	GL_PREFIX(_dispatch_stub_622), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_622))
+GL_PREFIX(_dispatch_stub_622):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4952(%rax), %r11
+	movq	4976(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	4952(%rax), %r11
+	movq	4976(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4952(%rax), %r11
+	movq	4976(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	4952(%rax), %r11
+	movq	4976(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_619), .-GL_PREFIX(_dispatch_stub_619)
+	.size	GL_PREFIX(_dispatch_stub_622), .-GL_PREFIX(_dispatch_stub_622)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(BlendFuncSeparateEXT)
@@ -23279,7 +23430,7 @@ GL_PREFIX(_dispatch_stub_619):
 GL_PREFIX(BlendFuncSeparateEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4960(%rax), %r11
+	movq	4984(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23293,13 +23444,13 @@ GL_PREFIX(BlendFuncSeparateEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4960(%rax), %r11
+	movq	4984(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4960(%rax), %r11
+	movq	4984(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23313,7 +23464,7 @@ GL_PREFIX(BlendFuncSeparateEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4960(%rax), %r11
+	movq	4984(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(BlendFuncSeparateEXT), .-GL_PREFIX(BlendFuncSeparateEXT)
@@ -23324,25 +23475,25 @@ GL_PREFIX(BlendFuncSeparateEXT):
 GL_PREFIX(FlushVertexArrayRangeNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4968(%rax), %r11
+	movq	4992(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
-	movq	4968(%rax), %r11
+	movq	4992(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4968(%rax), %r11
+	movq	4992(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
-	movq	4968(%rax), %r11
+	movq	4992(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FlushVertexArrayRangeNV), .-GL_PREFIX(FlushVertexArrayRangeNV)
@@ -23353,7 +23504,7 @@ GL_PREFIX(FlushVertexArrayRangeNV):
 GL_PREFIX(VertexArrayRangeNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4976(%rax), %r11
+	movq	5000(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23363,13 +23514,13 @@ GL_PREFIX(VertexArrayRangeNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4976(%rax), %r11
+	movq	5000(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4976(%rax), %r11
+	movq	5000(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23379,7 +23530,7 @@ GL_PREFIX(VertexArrayRangeNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	4976(%rax), %r11
+	movq	5000(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexArrayRangeNV), .-GL_PREFIX(VertexArrayRangeNV)
@@ -23390,7 +23541,7 @@ GL_PREFIX(VertexArrayRangeNV):
 GL_PREFIX(CombinerInputNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4984(%rax), %r11
+	movq	5008(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23408,13 +23559,13 @@ GL_PREFIX(CombinerInputNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4984(%rax), %r11
+	movq	5008(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4984(%rax), %r11
+	movq	5008(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23432,7 +23583,7 @@ GL_PREFIX(CombinerInputNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4984(%rax), %r11
+	movq	5008(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CombinerInputNV), .-GL_PREFIX(CombinerInputNV)
@@ -23443,7 +23594,7 @@ GL_PREFIX(CombinerInputNV):
 GL_PREFIX(CombinerOutputNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	4992(%rax), %r11
+	movq	5016(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23461,13 +23612,13 @@ GL_PREFIX(CombinerOutputNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4992(%rax), %r11
+	movq	5016(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	4992(%rax), %r11
+	movq	5016(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23485,7 +23636,7 @@ GL_PREFIX(CombinerOutputNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	4992(%rax), %r11
+	movq	5016(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CombinerOutputNV), .-GL_PREFIX(CombinerOutputNV)
@@ -23496,7 +23647,7 @@ GL_PREFIX(CombinerOutputNV):
 GL_PREFIX(CombinerParameterfNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5000(%rax), %r11
+	movq	5024(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -23506,13 +23657,13 @@ GL_PREFIX(CombinerParameterfNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5000(%rax), %r11
+	movq	5024(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5000(%rax), %r11
+	movq	5024(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -23522,7 +23673,7 @@ GL_PREFIX(CombinerParameterfNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5000(%rax), %r11
+	movq	5024(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CombinerParameterfNV), .-GL_PREFIX(CombinerParameterfNV)
@@ -23533,7 +23684,7 @@ GL_PREFIX(CombinerParameterfNV):
 GL_PREFIX(CombinerParameterfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5008(%rax), %r11
+	movq	5032(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23543,13 +23694,13 @@ GL_PREFIX(CombinerParameterfvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5008(%rax), %r11
+	movq	5032(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5008(%rax), %r11
+	movq	5032(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23559,7 +23710,7 @@ GL_PREFIX(CombinerParameterfvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5008(%rax), %r11
+	movq	5032(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CombinerParameterfvNV), .-GL_PREFIX(CombinerParameterfvNV)
@@ -23570,7 +23721,7 @@ GL_PREFIX(CombinerParameterfvNV):
 GL_PREFIX(CombinerParameteriNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5016(%rax), %r11
+	movq	5040(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23580,13 +23731,13 @@ GL_PREFIX(CombinerParameteriNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5016(%rax), %r11
+	movq	5040(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5016(%rax), %r11
+	movq	5040(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23596,7 +23747,7 @@ GL_PREFIX(CombinerParameteriNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5016(%rax), %r11
+	movq	5040(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CombinerParameteriNV), .-GL_PREFIX(CombinerParameteriNV)
@@ -23607,7 +23758,7 @@ GL_PREFIX(CombinerParameteriNV):
 GL_PREFIX(CombinerParameterivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5024(%rax), %r11
+	movq	5048(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23617,13 +23768,13 @@ GL_PREFIX(CombinerParameterivNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5024(%rax), %r11
+	movq	5048(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5024(%rax), %r11
+	movq	5048(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23633,7 +23784,7 @@ GL_PREFIX(CombinerParameterivNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5024(%rax), %r11
+	movq	5048(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CombinerParameterivNV), .-GL_PREFIX(CombinerParameterivNV)
@@ -23644,7 +23795,7 @@ GL_PREFIX(CombinerParameterivNV):
 GL_PREFIX(FinalCombinerInputNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5032(%rax), %r11
+	movq	5056(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23658,13 +23809,13 @@ GL_PREFIX(FinalCombinerInputNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5032(%rax), %r11
+	movq	5056(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5032(%rax), %r11
+	movq	5056(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23678,7 +23829,7 @@ GL_PREFIX(FinalCombinerInputNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5032(%rax), %r11
+	movq	5056(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FinalCombinerInputNV), .-GL_PREFIX(FinalCombinerInputNV)
@@ -23689,7 +23840,7 @@ GL_PREFIX(FinalCombinerInputNV):
 GL_PREFIX(GetCombinerInputParameterfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5040(%rax), %r11
+	movq	5064(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23703,13 +23854,13 @@ GL_PREFIX(GetCombinerInputParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5040(%rax), %r11
+	movq	5064(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5040(%rax), %r11
+	movq	5064(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23723,7 +23874,7 @@ GL_PREFIX(GetCombinerInputParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5040(%rax), %r11
+	movq	5064(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetCombinerInputParameterfvNV), .-GL_PREFIX(GetCombinerInputParameterfvNV)
@@ -23734,7 +23885,7 @@ GL_PREFIX(GetCombinerInputParameterfvNV):
 GL_PREFIX(GetCombinerInputParameterivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5048(%rax), %r11
+	movq	5072(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23748,13 +23899,13 @@ GL_PREFIX(GetCombinerInputParameterivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5048(%rax), %r11
+	movq	5072(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5048(%rax), %r11
+	movq	5072(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23768,7 +23919,7 @@ GL_PREFIX(GetCombinerInputParameterivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5048(%rax), %r11
+	movq	5072(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetCombinerInputParameterivNV), .-GL_PREFIX(GetCombinerInputParameterivNV)
@@ -23779,7 +23930,7 @@ GL_PREFIX(GetCombinerInputParameterivNV):
 GL_PREFIX(GetCombinerOutputParameterfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5056(%rax), %r11
+	movq	5080(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23793,13 +23944,13 @@ GL_PREFIX(GetCombinerOutputParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5056(%rax), %r11
+	movq	5080(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5056(%rax), %r11
+	movq	5080(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23813,7 +23964,7 @@ GL_PREFIX(GetCombinerOutputParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5056(%rax), %r11
+	movq	5080(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetCombinerOutputParameterfvNV), .-GL_PREFIX(GetCombinerOutputParameterfvNV)
@@ -23824,7 +23975,7 @@ GL_PREFIX(GetCombinerOutputParameterfvNV):
 GL_PREFIX(GetCombinerOutputParameterivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5064(%rax), %r11
+	movq	5088(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23838,13 +23989,13 @@ GL_PREFIX(GetCombinerOutputParameterivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5064(%rax), %r11
+	movq	5088(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5064(%rax), %r11
+	movq	5088(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23858,7 +24009,7 @@ GL_PREFIX(GetCombinerOutputParameterivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5064(%rax), %r11
+	movq	5088(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetCombinerOutputParameterivNV), .-GL_PREFIX(GetCombinerOutputParameterivNV)
@@ -23869,7 +24020,7 @@ GL_PREFIX(GetCombinerOutputParameterivNV):
 GL_PREFIX(GetFinalCombinerInputParameterfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5072(%rax), %r11
+	movq	5096(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23879,13 +24030,13 @@ GL_PREFIX(GetFinalCombinerInputParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5072(%rax), %r11
+	movq	5096(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5072(%rax), %r11
+	movq	5096(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23895,7 +24046,7 @@ GL_PREFIX(GetFinalCombinerInputParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5072(%rax), %r11
+	movq	5096(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetFinalCombinerInputParameterfvNV), .-GL_PREFIX(GetFinalCombinerInputParameterfvNV)
@@ -23906,7 +24057,7 @@ GL_PREFIX(GetFinalCombinerInputParameterfvNV):
 GL_PREFIX(GetFinalCombinerInputParameterivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5080(%rax), %r11
+	movq	5104(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -23916,13 +24067,13 @@ GL_PREFIX(GetFinalCombinerInputParameterivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5080(%rax), %r11
+	movq	5104(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5080(%rax), %r11
+	movq	5104(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -23932,7 +24083,7 @@ GL_PREFIX(GetFinalCombinerInputParameterivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5080(%rax), %r11
+	movq	5104(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetFinalCombinerInputParameterivNV), .-GL_PREFIX(GetFinalCombinerInputParameterivNV)
@@ -23943,25 +24094,25 @@ GL_PREFIX(GetFinalCombinerInputParameterivNV):
 GL_PREFIX(ResizeBuffersMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5088(%rax), %r11
+	movq	5112(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
-	movq	5088(%rax), %r11
+	movq	5112(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5088(%rax), %r11
+	movq	5112(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
-	movq	5088(%rax), %r11
+	movq	5112(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ResizeBuffersMESA), .-GL_PREFIX(ResizeBuffersMESA)
@@ -23972,7 +24123,7 @@ GL_PREFIX(ResizeBuffersMESA):
 GL_PREFIX(WindowPos2dMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5096(%rax), %r11
+	movq	5120(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -23982,13 +24133,13 @@ GL_PREFIX(WindowPos2dMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5096(%rax), %r11
+	movq	5120(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5096(%rax), %r11
+	movq	5120(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -23998,7 +24149,7 @@ GL_PREFIX(WindowPos2dMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5096(%rax), %r11
+	movq	5120(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2dMESA), .-GL_PREFIX(WindowPos2dMESA)
@@ -24009,25 +24160,25 @@ GL_PREFIX(WindowPos2dMESA):
 GL_PREFIX(WindowPos2dvMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5104(%rax), %r11
+	movq	5128(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5104(%rax), %r11
+	movq	5128(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5104(%rax), %r11
+	movq	5128(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5104(%rax), %r11
+	movq	5128(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2dvMESA), .-GL_PREFIX(WindowPos2dvMESA)
@@ -24038,7 +24189,7 @@ GL_PREFIX(WindowPos2dvMESA):
 GL_PREFIX(WindowPos2fMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5112(%rax), %r11
+	movq	5136(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -24048,13 +24199,13 @@ GL_PREFIX(WindowPos2fMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5112(%rax), %r11
+	movq	5136(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5112(%rax), %r11
+	movq	5136(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -24064,7 +24215,7 @@ GL_PREFIX(WindowPos2fMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5112(%rax), %r11
+	movq	5136(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2fMESA), .-GL_PREFIX(WindowPos2fMESA)
@@ -24075,25 +24226,25 @@ GL_PREFIX(WindowPos2fMESA):
 GL_PREFIX(WindowPos2fvMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5120(%rax), %r11
+	movq	5144(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5120(%rax), %r11
+	movq	5144(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5120(%rax), %r11
+	movq	5144(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5120(%rax), %r11
+	movq	5144(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2fvMESA), .-GL_PREFIX(WindowPos2fvMESA)
@@ -24104,7 +24255,7 @@ GL_PREFIX(WindowPos2fvMESA):
 GL_PREFIX(WindowPos2iMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5128(%rax), %r11
+	movq	5152(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24114,13 +24265,13 @@ GL_PREFIX(WindowPos2iMESA):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5128(%rax), %r11
+	movq	5152(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5128(%rax), %r11
+	movq	5152(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24130,7 +24281,7 @@ GL_PREFIX(WindowPos2iMESA):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5128(%rax), %r11
+	movq	5152(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2iMESA), .-GL_PREFIX(WindowPos2iMESA)
@@ -24141,25 +24292,25 @@ GL_PREFIX(WindowPos2iMESA):
 GL_PREFIX(WindowPos2ivMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5136(%rax), %r11
+	movq	5160(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5136(%rax), %r11
+	movq	5160(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5136(%rax), %r11
+	movq	5160(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5136(%rax), %r11
+	movq	5160(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2ivMESA), .-GL_PREFIX(WindowPos2ivMESA)
@@ -24170,7 +24321,7 @@ GL_PREFIX(WindowPos2ivMESA):
 GL_PREFIX(WindowPos2sMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5144(%rax), %r11
+	movq	5168(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24180,13 +24331,13 @@ GL_PREFIX(WindowPos2sMESA):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5144(%rax), %r11
+	movq	5168(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5144(%rax), %r11
+	movq	5168(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24196,7 +24347,7 @@ GL_PREFIX(WindowPos2sMESA):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5144(%rax), %r11
+	movq	5168(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2sMESA), .-GL_PREFIX(WindowPos2sMESA)
@@ -24207,25 +24358,25 @@ GL_PREFIX(WindowPos2sMESA):
 GL_PREFIX(WindowPos2svMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5152(%rax), %r11
+	movq	5176(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5152(%rax), %r11
+	movq	5176(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5152(%rax), %r11
+	movq	5176(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5152(%rax), %r11
+	movq	5176(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos2svMESA), .-GL_PREFIX(WindowPos2svMESA)
@@ -24236,7 +24387,7 @@ GL_PREFIX(WindowPos2svMESA):
 GL_PREFIX(WindowPos3dMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5160(%rax), %r11
+	movq	5184(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -24248,13 +24399,13 @@ GL_PREFIX(WindowPos3dMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5160(%rax), %r11
+	movq	5184(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5160(%rax), %r11
+	movq	5184(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -24266,7 +24417,7 @@ GL_PREFIX(WindowPos3dMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5160(%rax), %r11
+	movq	5184(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3dMESA), .-GL_PREFIX(WindowPos3dMESA)
@@ -24277,25 +24428,25 @@ GL_PREFIX(WindowPos3dMESA):
 GL_PREFIX(WindowPos3dvMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5168(%rax), %r11
+	movq	5192(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5168(%rax), %r11
+	movq	5192(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5168(%rax), %r11
+	movq	5192(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5168(%rax), %r11
+	movq	5192(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3dvMESA), .-GL_PREFIX(WindowPos3dvMESA)
@@ -24306,7 +24457,7 @@ GL_PREFIX(WindowPos3dvMESA):
 GL_PREFIX(WindowPos3fMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5176(%rax), %r11
+	movq	5200(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -24318,13 +24469,13 @@ GL_PREFIX(WindowPos3fMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5176(%rax), %r11
+	movq	5200(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5176(%rax), %r11
+	movq	5200(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -24336,7 +24487,7 @@ GL_PREFIX(WindowPos3fMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$24, %rsp
-	movq	5176(%rax), %r11
+	movq	5200(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3fMESA), .-GL_PREFIX(WindowPos3fMESA)
@@ -24347,25 +24498,25 @@ GL_PREFIX(WindowPos3fMESA):
 GL_PREFIX(WindowPos3fvMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5184(%rax), %r11
+	movq	5208(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5184(%rax), %r11
+	movq	5208(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5184(%rax), %r11
+	movq	5208(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5184(%rax), %r11
+	movq	5208(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3fvMESA), .-GL_PREFIX(WindowPos3fvMESA)
@@ -24376,7 +24527,7 @@ GL_PREFIX(WindowPos3fvMESA):
 GL_PREFIX(WindowPos3iMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5192(%rax), %r11
+	movq	5216(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24386,13 +24537,13 @@ GL_PREFIX(WindowPos3iMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5192(%rax), %r11
+	movq	5216(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5192(%rax), %r11
+	movq	5216(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24402,7 +24553,7 @@ GL_PREFIX(WindowPos3iMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5192(%rax), %r11
+	movq	5216(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3iMESA), .-GL_PREFIX(WindowPos3iMESA)
@@ -24413,25 +24564,25 @@ GL_PREFIX(WindowPos3iMESA):
 GL_PREFIX(WindowPos3ivMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5200(%rax), %r11
+	movq	5224(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5200(%rax), %r11
+	movq	5224(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5200(%rax), %r11
+	movq	5224(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5200(%rax), %r11
+	movq	5224(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3ivMESA), .-GL_PREFIX(WindowPos3ivMESA)
@@ -24442,7 +24593,7 @@ GL_PREFIX(WindowPos3ivMESA):
 GL_PREFIX(WindowPos3sMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5208(%rax), %r11
+	movq	5232(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24452,13 +24603,13 @@ GL_PREFIX(WindowPos3sMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5208(%rax), %r11
+	movq	5232(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5208(%rax), %r11
+	movq	5232(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24468,7 +24619,7 @@ GL_PREFIX(WindowPos3sMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5208(%rax), %r11
+	movq	5232(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3sMESA), .-GL_PREFIX(WindowPos3sMESA)
@@ -24479,25 +24630,25 @@ GL_PREFIX(WindowPos3sMESA):
 GL_PREFIX(WindowPos3svMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5216(%rax), %r11
+	movq	5240(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5216(%rax), %r11
+	movq	5240(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5216(%rax), %r11
+	movq	5240(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5216(%rax), %r11
+	movq	5240(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos3svMESA), .-GL_PREFIX(WindowPos3svMESA)
@@ -24508,7 +24659,7 @@ GL_PREFIX(WindowPos3svMESA):
 GL_PREFIX(WindowPos4dMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5224(%rax), %r11
+	movq	5248(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$40, %rsp
@@ -24522,13 +24673,13 @@ GL_PREFIX(WindowPos4dMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$40, %rsp
-	movq	5224(%rax), %r11
+	movq	5248(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5224(%rax), %r11
+	movq	5248(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$40, %rsp
@@ -24542,7 +24693,7 @@ GL_PREFIX(WindowPos4dMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$40, %rsp
-	movq	5224(%rax), %r11
+	movq	5248(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4dMESA), .-GL_PREFIX(WindowPos4dMESA)
@@ -24553,25 +24704,25 @@ GL_PREFIX(WindowPos4dMESA):
 GL_PREFIX(WindowPos4dvMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5232(%rax), %r11
+	movq	5256(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5232(%rax), %r11
+	movq	5256(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5232(%rax), %r11
+	movq	5256(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5232(%rax), %r11
+	movq	5256(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4dvMESA), .-GL_PREFIX(WindowPos4dvMESA)
@@ -24582,7 +24733,7 @@ GL_PREFIX(WindowPos4dvMESA):
 GL_PREFIX(WindowPos4fMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5240(%rax), %r11
+	movq	5264(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$40, %rsp
@@ -24596,13 +24747,13 @@ GL_PREFIX(WindowPos4fMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$40, %rsp
-	movq	5240(%rax), %r11
+	movq	5264(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5240(%rax), %r11
+	movq	5264(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$40, %rsp
@@ -24616,7 +24767,7 @@ GL_PREFIX(WindowPos4fMESA):
 	movq	8(%rsp), %xmm1
 	movq	(%rsp), %xmm0
 	addq	$40, %rsp
-	movq	5240(%rax), %r11
+	movq	5264(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4fMESA), .-GL_PREFIX(WindowPos4fMESA)
@@ -24627,25 +24778,25 @@ GL_PREFIX(WindowPos4fMESA):
 GL_PREFIX(WindowPos4fvMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5248(%rax), %r11
+	movq	5272(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5248(%rax), %r11
+	movq	5272(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5248(%rax), %r11
+	movq	5272(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5248(%rax), %r11
+	movq	5272(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4fvMESA), .-GL_PREFIX(WindowPos4fvMESA)
@@ -24656,7 +24807,7 @@ GL_PREFIX(WindowPos4fvMESA):
 GL_PREFIX(WindowPos4iMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5256(%rax), %r11
+	movq	5280(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24670,13 +24821,13 @@ GL_PREFIX(WindowPos4iMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5256(%rax), %r11
+	movq	5280(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5256(%rax), %r11
+	movq	5280(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24690,7 +24841,7 @@ GL_PREFIX(WindowPos4iMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5256(%rax), %r11
+	movq	5280(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4iMESA), .-GL_PREFIX(WindowPos4iMESA)
@@ -24701,25 +24852,25 @@ GL_PREFIX(WindowPos4iMESA):
 GL_PREFIX(WindowPos4ivMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5264(%rax), %r11
+	movq	5288(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5264(%rax), %r11
+	movq	5288(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5264(%rax), %r11
+	movq	5288(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5264(%rax), %r11
+	movq	5288(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4ivMESA), .-GL_PREFIX(WindowPos4ivMESA)
@@ -24730,7 +24881,7 @@ GL_PREFIX(WindowPos4ivMESA):
 GL_PREFIX(WindowPos4sMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5272(%rax), %r11
+	movq	5296(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24744,13 +24895,13 @@ GL_PREFIX(WindowPos4sMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5272(%rax), %r11
+	movq	5296(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5272(%rax), %r11
+	movq	5296(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24764,7 +24915,7 @@ GL_PREFIX(WindowPos4sMESA):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5272(%rax), %r11
+	movq	5296(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4sMESA), .-GL_PREFIX(WindowPos4sMESA)
@@ -24775,37 +24926,37 @@ GL_PREFIX(WindowPos4sMESA):
 GL_PREFIX(WindowPos4svMESA):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5280(%rax), %r11
+	movq	5304(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5280(%rax), %r11
+	movq	5304(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5280(%rax), %r11
+	movq	5304(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5280(%rax), %r11
+	movq	5304(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(WindowPos4svMESA), .-GL_PREFIX(WindowPos4svMESA)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_661)
-	.type	GL_PREFIX(_dispatch_stub_661), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_661))
-GL_PREFIX(_dispatch_stub_661):
+	.globl	GL_PREFIX(_dispatch_stub_664)
+	.type	GL_PREFIX(_dispatch_stub_664), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_664))
+GL_PREFIX(_dispatch_stub_664):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5288(%rax), %r11
+	movq	5312(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24819,13 +24970,13 @@ GL_PREFIX(_dispatch_stub_661):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5288(%rax), %r11
+	movq	5312(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5288(%rax), %r11
+	movq	5312(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24839,19 +24990,19 @@ GL_PREFIX(_dispatch_stub_661):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5288(%rax), %r11
+	movq	5312(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_661), .-GL_PREFIX(_dispatch_stub_661)
+	.size	GL_PREFIX(_dispatch_stub_664), .-GL_PREFIX(_dispatch_stub_664)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_662)
-	.type	GL_PREFIX(_dispatch_stub_662), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_662))
-GL_PREFIX(_dispatch_stub_662):
+	.globl	GL_PREFIX(_dispatch_stub_665)
+	.type	GL_PREFIX(_dispatch_stub_665), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_665))
+GL_PREFIX(_dispatch_stub_665):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5296(%rax), %r11
+	movq	5320(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24869,13 +25020,13 @@ GL_PREFIX(_dispatch_stub_662):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5296(%rax), %r11
+	movq	5320(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5296(%rax), %r11
+	movq	5320(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24893,19 +25044,19 @@ GL_PREFIX(_dispatch_stub_662):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5296(%rax), %r11
+	movq	5320(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_662), .-GL_PREFIX(_dispatch_stub_662)
+	.size	GL_PREFIX(_dispatch_stub_665), .-GL_PREFIX(_dispatch_stub_665)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_663)
-	.type	GL_PREFIX(_dispatch_stub_663), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_663))
-GL_PREFIX(_dispatch_stub_663):
+	.globl	GL_PREFIX(_dispatch_stub_666)
+	.type	GL_PREFIX(_dispatch_stub_666), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_666))
+GL_PREFIX(_dispatch_stub_666):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5304(%rax), %r11
+	movq	5328(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24915,13 +25066,13 @@ GL_PREFIX(_dispatch_stub_663):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5304(%rax), %r11
+	movq	5328(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5304(%rax), %r11
+	movq	5328(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24931,49 +25082,49 @@ GL_PREFIX(_dispatch_stub_663):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5304(%rax), %r11
+	movq	5328(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_663), .-GL_PREFIX(_dispatch_stub_663)
+	.size	GL_PREFIX(_dispatch_stub_666), .-GL_PREFIX(_dispatch_stub_666)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_664)
-	.type	GL_PREFIX(_dispatch_stub_664), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_664))
-GL_PREFIX(_dispatch_stub_664):
+	.globl	GL_PREFIX(_dispatch_stub_667)
+	.type	GL_PREFIX(_dispatch_stub_667), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_667))
+GL_PREFIX(_dispatch_stub_667):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5312(%rax), %r11
+	movq	5336(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5312(%rax), %r11
+	movq	5336(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5312(%rax), %r11
+	movq	5336(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5312(%rax), %r11
+	movq	5336(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_664), .-GL_PREFIX(_dispatch_stub_664)
+	.size	GL_PREFIX(_dispatch_stub_667), .-GL_PREFIX(_dispatch_stub_667)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_665)
-	.type	GL_PREFIX(_dispatch_stub_665), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_665))
-GL_PREFIX(_dispatch_stub_665):
+	.globl	GL_PREFIX(_dispatch_stub_668)
+	.type	GL_PREFIX(_dispatch_stub_668), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_668))
+GL_PREFIX(_dispatch_stub_668):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5320(%rax), %r11
+	movq	5344(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -24983,13 +25134,13 @@ GL_PREFIX(_dispatch_stub_665):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5320(%rax), %r11
+	movq	5344(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5320(%rax), %r11
+	movq	5344(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -24999,19 +25150,19 @@ GL_PREFIX(_dispatch_stub_665):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5320(%rax), %r11
+	movq	5344(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_665), .-GL_PREFIX(_dispatch_stub_665)
+	.size	GL_PREFIX(_dispatch_stub_668), .-GL_PREFIX(_dispatch_stub_668)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_666)
-	.type	GL_PREFIX(_dispatch_stub_666), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_666))
-GL_PREFIX(_dispatch_stub_666):
+	.globl	GL_PREFIX(_dispatch_stub_669)
+	.type	GL_PREFIX(_dispatch_stub_669), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_669))
+GL_PREFIX(_dispatch_stub_669):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5328(%rax), %r11
+	movq	5352(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25021,13 +25172,13 @@ GL_PREFIX(_dispatch_stub_666):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5328(%rax), %r11
+	movq	5352(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5328(%rax), %r11
+	movq	5352(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25037,49 +25188,49 @@ GL_PREFIX(_dispatch_stub_666):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5328(%rax), %r11
+	movq	5352(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_666), .-GL_PREFIX(_dispatch_stub_666)
+	.size	GL_PREFIX(_dispatch_stub_669), .-GL_PREFIX(_dispatch_stub_669)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_667)
-	.type	GL_PREFIX(_dispatch_stub_667), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_667))
-GL_PREFIX(_dispatch_stub_667):
+	.globl	GL_PREFIX(_dispatch_stub_670)
+	.type	GL_PREFIX(_dispatch_stub_670), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_670))
+GL_PREFIX(_dispatch_stub_670):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5336(%rax), %r11
+	movq	5360(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5336(%rax), %r11
+	movq	5360(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5336(%rax), %r11
+	movq	5360(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5336(%rax), %r11
+	movq	5360(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_667), .-GL_PREFIX(_dispatch_stub_667)
+	.size	GL_PREFIX(_dispatch_stub_670), .-GL_PREFIX(_dispatch_stub_670)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_668)
-	.type	GL_PREFIX(_dispatch_stub_668), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_668))
-GL_PREFIX(_dispatch_stub_668):
+	.globl	GL_PREFIX(_dispatch_stub_671)
+	.type	GL_PREFIX(_dispatch_stub_671), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_671))
+GL_PREFIX(_dispatch_stub_671):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5344(%rax), %r11
+	movq	5368(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25089,13 +25240,13 @@ GL_PREFIX(_dispatch_stub_668):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5344(%rax), %r11
+	movq	5368(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5344(%rax), %r11
+	movq	5368(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25105,40 +25256,40 @@ GL_PREFIX(_dispatch_stub_668):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5344(%rax), %r11
+	movq	5368(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_668), .-GL_PREFIX(_dispatch_stub_668)
+	.size	GL_PREFIX(_dispatch_stub_671), .-GL_PREFIX(_dispatch_stub_671)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_669)
-	.type	GL_PREFIX(_dispatch_stub_669), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_669))
-GL_PREFIX(_dispatch_stub_669):
+	.globl	GL_PREFIX(_dispatch_stub_672)
+	.type	GL_PREFIX(_dispatch_stub_672), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_672))
+GL_PREFIX(_dispatch_stub_672):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5352(%rax), %r11
+	movq	5376(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5352(%rax), %r11
+	movq	5376(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5352(%rax), %r11
+	movq	5376(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5352(%rax), %r11
+	movq	5376(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_669), .-GL_PREFIX(_dispatch_stub_669)
+	.size	GL_PREFIX(_dispatch_stub_672), .-GL_PREFIX(_dispatch_stub_672)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(AreProgramsResidentNV)
@@ -25146,7 +25297,7 @@ GL_PREFIX(_dispatch_stub_669):
 GL_PREFIX(AreProgramsResidentNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5360(%rax), %r11
+	movq	5384(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25156,13 +25307,13 @@ GL_PREFIX(AreProgramsResidentNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5360(%rax), %r11
+	movq	5384(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5360(%rax), %r11
+	movq	5384(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25172,7 +25323,7 @@ GL_PREFIX(AreProgramsResidentNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5360(%rax), %r11
+	movq	5384(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(AreProgramsResidentNV), .-GL_PREFIX(AreProgramsResidentNV)
@@ -25183,7 +25334,7 @@ GL_PREFIX(AreProgramsResidentNV):
 GL_PREFIX(BindProgramNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5368(%rax), %r11
+	movq	5392(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25193,13 +25344,13 @@ GL_PREFIX(BindProgramNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5368(%rax), %r11
+	movq	5392(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5368(%rax), %r11
+	movq	5392(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25209,7 +25360,7 @@ GL_PREFIX(BindProgramNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5368(%rax), %r11
+	movq	5392(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(BindProgramNV), .-GL_PREFIX(BindProgramNV)
@@ -25220,7 +25371,7 @@ GL_PREFIX(BindProgramNV):
 GL_PREFIX(DeleteProgramsNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5376(%rax), %r11
+	movq	5400(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25230,13 +25381,13 @@ GL_PREFIX(DeleteProgramsNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5376(%rax), %r11
+	movq	5400(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5376(%rax), %r11
+	movq	5400(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25246,7 +25397,7 @@ GL_PREFIX(DeleteProgramsNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5376(%rax), %r11
+	movq	5400(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(DeleteProgramsNV), .-GL_PREFIX(DeleteProgramsNV)
@@ -25257,7 +25408,7 @@ GL_PREFIX(DeleteProgramsNV):
 GL_PREFIX(ExecuteProgramNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5384(%rax), %r11
+	movq	5408(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25267,13 +25418,13 @@ GL_PREFIX(ExecuteProgramNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5384(%rax), %r11
+	movq	5408(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5384(%rax), %r11
+	movq	5408(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25283,7 +25434,7 @@ GL_PREFIX(ExecuteProgramNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5384(%rax), %r11
+	movq	5408(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ExecuteProgramNV), .-GL_PREFIX(ExecuteProgramNV)
@@ -25294,7 +25445,7 @@ GL_PREFIX(ExecuteProgramNV):
 GL_PREFIX(GenProgramsNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5392(%rax), %r11
+	movq	5416(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25304,13 +25455,13 @@ GL_PREFIX(GenProgramsNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5392(%rax), %r11
+	movq	5416(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5392(%rax), %r11
+	movq	5416(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25320,7 +25471,7 @@ GL_PREFIX(GenProgramsNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5392(%rax), %r11
+	movq	5416(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GenProgramsNV), .-GL_PREFIX(GenProgramsNV)
@@ -25331,7 +25482,7 @@ GL_PREFIX(GenProgramsNV):
 GL_PREFIX(GetProgramParameterdvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5400(%rax), %r11
+	movq	5424(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25345,13 +25496,13 @@ GL_PREFIX(GetProgramParameterdvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5400(%rax), %r11
+	movq	5424(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5400(%rax), %r11
+	movq	5424(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25365,7 +25516,7 @@ GL_PREFIX(GetProgramParameterdvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5400(%rax), %r11
+	movq	5424(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetProgramParameterdvNV), .-GL_PREFIX(GetProgramParameterdvNV)
@@ -25376,7 +25527,7 @@ GL_PREFIX(GetProgramParameterdvNV):
 GL_PREFIX(GetProgramParameterfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5408(%rax), %r11
+	movq	5432(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25390,13 +25541,13 @@ GL_PREFIX(GetProgramParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5408(%rax), %r11
+	movq	5432(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5408(%rax), %r11
+	movq	5432(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25410,7 +25561,7 @@ GL_PREFIX(GetProgramParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5408(%rax), %r11
+	movq	5432(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetProgramParameterfvNV), .-GL_PREFIX(GetProgramParameterfvNV)
@@ -25421,7 +25572,7 @@ GL_PREFIX(GetProgramParameterfvNV):
 GL_PREFIX(GetProgramStringNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5416(%rax), %r11
+	movq	5440(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25431,13 +25582,13 @@ GL_PREFIX(GetProgramStringNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5416(%rax), %r11
+	movq	5440(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5416(%rax), %r11
+	movq	5440(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25447,7 +25598,7 @@ GL_PREFIX(GetProgramStringNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5416(%rax), %r11
+	movq	5440(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetProgramStringNV), .-GL_PREFIX(GetProgramStringNV)
@@ -25458,7 +25609,7 @@ GL_PREFIX(GetProgramStringNV):
 GL_PREFIX(GetProgramivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5424(%rax), %r11
+	movq	5448(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25468,13 +25619,13 @@ GL_PREFIX(GetProgramivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5424(%rax), %r11
+	movq	5448(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5424(%rax), %r11
+	movq	5448(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25484,7 +25635,7 @@ GL_PREFIX(GetProgramivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5424(%rax), %r11
+	movq	5448(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetProgramivNV), .-GL_PREFIX(GetProgramivNV)
@@ -25495,7 +25646,7 @@ GL_PREFIX(GetProgramivNV):
 GL_PREFIX(GetTrackMatrixivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5432(%rax), %r11
+	movq	5456(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25509,13 +25660,13 @@ GL_PREFIX(GetTrackMatrixivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5432(%rax), %r11
+	movq	5456(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5432(%rax), %r11
+	movq	5456(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25529,7 +25680,7 @@ GL_PREFIX(GetTrackMatrixivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5432(%rax), %r11
+	movq	5456(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetTrackMatrixivNV), .-GL_PREFIX(GetTrackMatrixivNV)
@@ -25540,7 +25691,7 @@ GL_PREFIX(GetTrackMatrixivNV):
 GL_PREFIX(GetVertexAttribPointervNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5440(%rax), %r11
+	movq	5464(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25550,13 +25701,13 @@ GL_PREFIX(GetVertexAttribPointervNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5440(%rax), %r11
+	movq	5464(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5440(%rax), %r11
+	movq	5464(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25566,7 +25717,7 @@ GL_PREFIX(GetVertexAttribPointervNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5440(%rax), %r11
+	movq	5464(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetVertexAttribPointervNV), .-GL_PREFIX(GetVertexAttribPointervNV)
@@ -25577,7 +25728,7 @@ GL_PREFIX(GetVertexAttribPointervNV):
 GL_PREFIX(GetVertexAttribdvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5448(%rax), %r11
+	movq	5472(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25587,13 +25738,13 @@ GL_PREFIX(GetVertexAttribdvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5448(%rax), %r11
+	movq	5472(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5448(%rax), %r11
+	movq	5472(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25603,7 +25754,7 @@ GL_PREFIX(GetVertexAttribdvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5448(%rax), %r11
+	movq	5472(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetVertexAttribdvNV), .-GL_PREFIX(GetVertexAttribdvNV)
@@ -25614,7 +25765,7 @@ GL_PREFIX(GetVertexAttribdvNV):
 GL_PREFIX(GetVertexAttribfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5456(%rax), %r11
+	movq	5480(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25624,13 +25775,13 @@ GL_PREFIX(GetVertexAttribfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5456(%rax), %r11
+	movq	5480(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5456(%rax), %r11
+	movq	5480(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25640,7 +25791,7 @@ GL_PREFIX(GetVertexAttribfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5456(%rax), %r11
+	movq	5480(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetVertexAttribfvNV), .-GL_PREFIX(GetVertexAttribfvNV)
@@ -25651,7 +25802,7 @@ GL_PREFIX(GetVertexAttribfvNV):
 GL_PREFIX(GetVertexAttribivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5464(%rax), %r11
+	movq	5488(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25661,13 +25812,13 @@ GL_PREFIX(GetVertexAttribivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5464(%rax), %r11
+	movq	5488(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5464(%rax), %r11
+	movq	5488(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25677,7 +25828,7 @@ GL_PREFIX(GetVertexAttribivNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5464(%rax), %r11
+	movq	5488(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetVertexAttribivNV), .-GL_PREFIX(GetVertexAttribivNV)
@@ -25688,25 +25839,25 @@ GL_PREFIX(GetVertexAttribivNV):
 GL_PREFIX(IsProgramNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5472(%rax), %r11
+	movq	5496(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5472(%rax), %r11
+	movq	5496(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5472(%rax), %r11
+	movq	5496(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5472(%rax), %r11
+	movq	5496(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(IsProgramNV), .-GL_PREFIX(IsProgramNV)
@@ -25717,7 +25868,7 @@ GL_PREFIX(IsProgramNV):
 GL_PREFIX(LoadProgramNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5480(%rax), %r11
+	movq	5504(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25731,13 +25882,13 @@ GL_PREFIX(LoadProgramNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5480(%rax), %r11
+	movq	5504(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5480(%rax), %r11
+	movq	5504(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25751,7 +25902,7 @@ GL_PREFIX(LoadProgramNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5480(%rax), %r11
+	movq	5504(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(LoadProgramNV), .-GL_PREFIX(LoadProgramNV)
@@ -25762,7 +25913,7 @@ GL_PREFIX(LoadProgramNV):
 GL_PREFIX(ProgramParameters4dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5488(%rax), %r11
+	movq	5512(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25776,13 +25927,13 @@ GL_PREFIX(ProgramParameters4dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5488(%rax), %r11
+	movq	5512(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5488(%rax), %r11
+	movq	5512(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25796,7 +25947,7 @@ GL_PREFIX(ProgramParameters4dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5488(%rax), %r11
+	movq	5512(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProgramParameters4dvNV), .-GL_PREFIX(ProgramParameters4dvNV)
@@ -25807,7 +25958,7 @@ GL_PREFIX(ProgramParameters4dvNV):
 GL_PREFIX(ProgramParameters4fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5496(%rax), %r11
+	movq	5520(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25821,13 +25972,13 @@ GL_PREFIX(ProgramParameters4fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5496(%rax), %r11
+	movq	5520(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5496(%rax), %r11
+	movq	5520(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25841,7 +25992,7 @@ GL_PREFIX(ProgramParameters4fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5496(%rax), %r11
+	movq	5520(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProgramParameters4fvNV), .-GL_PREFIX(ProgramParameters4fvNV)
@@ -25852,7 +26003,7 @@ GL_PREFIX(ProgramParameters4fvNV):
 GL_PREFIX(RequestResidentProgramsNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5504(%rax), %r11
+	movq	5528(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25862,13 +26013,13 @@ GL_PREFIX(RequestResidentProgramsNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5504(%rax), %r11
+	movq	5528(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5504(%rax), %r11
+	movq	5528(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25878,7 +26029,7 @@ GL_PREFIX(RequestResidentProgramsNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5504(%rax), %r11
+	movq	5528(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(RequestResidentProgramsNV), .-GL_PREFIX(RequestResidentProgramsNV)
@@ -25889,7 +26040,7 @@ GL_PREFIX(RequestResidentProgramsNV):
 GL_PREFIX(TrackMatrixNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5512(%rax), %r11
+	movq	5536(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25903,13 +26054,13 @@ GL_PREFIX(TrackMatrixNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5512(%rax), %r11
+	movq	5536(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5512(%rax), %r11
+	movq	5536(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25923,7 +26074,7 @@ GL_PREFIX(TrackMatrixNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5512(%rax), %r11
+	movq	5536(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(TrackMatrixNV), .-GL_PREFIX(TrackMatrixNV)
@@ -25934,7 +26085,7 @@ GL_PREFIX(TrackMatrixNV):
 GL_PREFIX(VertexAttrib1dNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5520(%rax), %r11
+	movq	5544(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -25944,13 +26095,13 @@ GL_PREFIX(VertexAttrib1dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5520(%rax), %r11
+	movq	5544(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5520(%rax), %r11
+	movq	5544(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -25960,7 +26111,7 @@ GL_PREFIX(VertexAttrib1dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5520(%rax), %r11
+	movq	5544(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib1dNV), .-GL_PREFIX(VertexAttrib1dNV)
@@ -25971,7 +26122,7 @@ GL_PREFIX(VertexAttrib1dNV):
 GL_PREFIX(VertexAttrib1dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5528(%rax), %r11
+	movq	5552(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -25981,13 +26132,13 @@ GL_PREFIX(VertexAttrib1dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5528(%rax), %r11
+	movq	5552(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5528(%rax), %r11
+	movq	5552(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -25997,7 +26148,7 @@ GL_PREFIX(VertexAttrib1dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5528(%rax), %r11
+	movq	5552(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib1dvNV), .-GL_PREFIX(VertexAttrib1dvNV)
@@ -26008,7 +26159,7 @@ GL_PREFIX(VertexAttrib1dvNV):
 GL_PREFIX(VertexAttrib1fNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5536(%rax), %r11
+	movq	5560(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -26018,13 +26169,13 @@ GL_PREFIX(VertexAttrib1fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5536(%rax), %r11
+	movq	5560(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5536(%rax), %r11
+	movq	5560(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -26034,7 +26185,7 @@ GL_PREFIX(VertexAttrib1fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5536(%rax), %r11
+	movq	5560(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib1fNV), .-GL_PREFIX(VertexAttrib1fNV)
@@ -26045,7 +26196,7 @@ GL_PREFIX(VertexAttrib1fNV):
 GL_PREFIX(VertexAttrib1fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5544(%rax), %r11
+	movq	5568(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26055,13 +26206,13 @@ GL_PREFIX(VertexAttrib1fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5544(%rax), %r11
+	movq	5568(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5544(%rax), %r11
+	movq	5568(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26071,7 +26222,7 @@ GL_PREFIX(VertexAttrib1fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5544(%rax), %r11
+	movq	5568(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib1fvNV), .-GL_PREFIX(VertexAttrib1fvNV)
@@ -26082,7 +26233,7 @@ GL_PREFIX(VertexAttrib1fvNV):
 GL_PREFIX(VertexAttrib1sNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5552(%rax), %r11
+	movq	5576(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26092,13 +26243,13 @@ GL_PREFIX(VertexAttrib1sNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5552(%rax), %r11
+	movq	5576(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5552(%rax), %r11
+	movq	5576(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26108,7 +26259,7 @@ GL_PREFIX(VertexAttrib1sNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5552(%rax), %r11
+	movq	5576(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib1sNV), .-GL_PREFIX(VertexAttrib1sNV)
@@ -26119,7 +26270,7 @@ GL_PREFIX(VertexAttrib1sNV):
 GL_PREFIX(VertexAttrib1svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5560(%rax), %r11
+	movq	5584(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26129,13 +26280,13 @@ GL_PREFIX(VertexAttrib1svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5560(%rax), %r11
+	movq	5584(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5560(%rax), %r11
+	movq	5584(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26145,7 +26296,7 @@ GL_PREFIX(VertexAttrib1svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5560(%rax), %r11
+	movq	5584(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib1svNV), .-GL_PREFIX(VertexAttrib1svNV)
@@ -26156,7 +26307,7 @@ GL_PREFIX(VertexAttrib1svNV):
 GL_PREFIX(VertexAttrib2dNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5568(%rax), %r11
+	movq	5592(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -26168,13 +26319,13 @@ GL_PREFIX(VertexAttrib2dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5568(%rax), %r11
+	movq	5592(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5568(%rax), %r11
+	movq	5592(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -26186,7 +26337,7 @@ GL_PREFIX(VertexAttrib2dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5568(%rax), %r11
+	movq	5592(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib2dNV), .-GL_PREFIX(VertexAttrib2dNV)
@@ -26197,7 +26348,7 @@ GL_PREFIX(VertexAttrib2dNV):
 GL_PREFIX(VertexAttrib2dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5576(%rax), %r11
+	movq	5600(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26207,13 +26358,13 @@ GL_PREFIX(VertexAttrib2dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5576(%rax), %r11
+	movq	5600(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5576(%rax), %r11
+	movq	5600(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26223,7 +26374,7 @@ GL_PREFIX(VertexAttrib2dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5576(%rax), %r11
+	movq	5600(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib2dvNV), .-GL_PREFIX(VertexAttrib2dvNV)
@@ -26234,7 +26385,7 @@ GL_PREFIX(VertexAttrib2dvNV):
 GL_PREFIX(VertexAttrib2fNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5584(%rax), %r11
+	movq	5608(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$24, %rsp
@@ -26246,13 +26397,13 @@ GL_PREFIX(VertexAttrib2fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5584(%rax), %r11
+	movq	5608(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5584(%rax), %r11
+	movq	5608(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$24, %rsp
@@ -26264,7 +26415,7 @@ GL_PREFIX(VertexAttrib2fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$24, %rsp
-	movq	5584(%rax), %r11
+	movq	5608(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib2fNV), .-GL_PREFIX(VertexAttrib2fNV)
@@ -26275,7 +26426,7 @@ GL_PREFIX(VertexAttrib2fNV):
 GL_PREFIX(VertexAttrib2fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5592(%rax), %r11
+	movq	5616(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26285,13 +26436,13 @@ GL_PREFIX(VertexAttrib2fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5592(%rax), %r11
+	movq	5616(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5592(%rax), %r11
+	movq	5616(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26301,7 +26452,7 @@ GL_PREFIX(VertexAttrib2fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5592(%rax), %r11
+	movq	5616(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib2fvNV), .-GL_PREFIX(VertexAttrib2fvNV)
@@ -26312,7 +26463,7 @@ GL_PREFIX(VertexAttrib2fvNV):
 GL_PREFIX(VertexAttrib2sNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5600(%rax), %r11
+	movq	5624(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26322,13 +26473,13 @@ GL_PREFIX(VertexAttrib2sNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5600(%rax), %r11
+	movq	5624(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5600(%rax), %r11
+	movq	5624(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26338,7 +26489,7 @@ GL_PREFIX(VertexAttrib2sNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5600(%rax), %r11
+	movq	5624(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib2sNV), .-GL_PREFIX(VertexAttrib2sNV)
@@ -26349,7 +26500,7 @@ GL_PREFIX(VertexAttrib2sNV):
 GL_PREFIX(VertexAttrib2svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5608(%rax), %r11
+	movq	5632(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26359,13 +26510,13 @@ GL_PREFIX(VertexAttrib2svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5608(%rax), %r11
+	movq	5632(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5608(%rax), %r11
+	movq	5632(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26375,7 +26526,7 @@ GL_PREFIX(VertexAttrib2svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5608(%rax), %r11
+	movq	5632(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib2svNV), .-GL_PREFIX(VertexAttrib2svNV)
@@ -26386,7 +26537,7 @@ GL_PREFIX(VertexAttrib2svNV):
 GL_PREFIX(VertexAttrib3dNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5616(%rax), %r11
+	movq	5640(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$40, %rsp
@@ -26400,13 +26551,13 @@ GL_PREFIX(VertexAttrib3dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5616(%rax), %r11
+	movq	5640(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5616(%rax), %r11
+	movq	5640(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$40, %rsp
@@ -26420,7 +26571,7 @@ GL_PREFIX(VertexAttrib3dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5616(%rax), %r11
+	movq	5640(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib3dNV), .-GL_PREFIX(VertexAttrib3dNV)
@@ -26431,7 +26582,7 @@ GL_PREFIX(VertexAttrib3dNV):
 GL_PREFIX(VertexAttrib3dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5624(%rax), %r11
+	movq	5648(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26441,13 +26592,13 @@ GL_PREFIX(VertexAttrib3dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5624(%rax), %r11
+	movq	5648(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5624(%rax), %r11
+	movq	5648(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26457,7 +26608,7 @@ GL_PREFIX(VertexAttrib3dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5624(%rax), %r11
+	movq	5648(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib3dvNV), .-GL_PREFIX(VertexAttrib3dvNV)
@@ -26468,7 +26619,7 @@ GL_PREFIX(VertexAttrib3dvNV):
 GL_PREFIX(VertexAttrib3fNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5632(%rax), %r11
+	movq	5656(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$40, %rsp
@@ -26482,13 +26633,13 @@ GL_PREFIX(VertexAttrib3fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5632(%rax), %r11
+	movq	5656(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5632(%rax), %r11
+	movq	5656(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$40, %rsp
@@ -26502,7 +26653,7 @@ GL_PREFIX(VertexAttrib3fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5632(%rax), %r11
+	movq	5656(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib3fNV), .-GL_PREFIX(VertexAttrib3fNV)
@@ -26513,7 +26664,7 @@ GL_PREFIX(VertexAttrib3fNV):
 GL_PREFIX(VertexAttrib3fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5640(%rax), %r11
+	movq	5664(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26523,13 +26674,13 @@ GL_PREFIX(VertexAttrib3fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5640(%rax), %r11
+	movq	5664(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5640(%rax), %r11
+	movq	5664(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26539,7 +26690,7 @@ GL_PREFIX(VertexAttrib3fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5640(%rax), %r11
+	movq	5664(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib3fvNV), .-GL_PREFIX(VertexAttrib3fvNV)
@@ -26550,7 +26701,7 @@ GL_PREFIX(VertexAttrib3fvNV):
 GL_PREFIX(VertexAttrib3sNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5648(%rax), %r11
+	movq	5672(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26564,13 +26715,13 @@ GL_PREFIX(VertexAttrib3sNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5648(%rax), %r11
+	movq	5672(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5648(%rax), %r11
+	movq	5672(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26584,7 +26735,7 @@ GL_PREFIX(VertexAttrib3sNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5648(%rax), %r11
+	movq	5672(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib3sNV), .-GL_PREFIX(VertexAttrib3sNV)
@@ -26595,7 +26746,7 @@ GL_PREFIX(VertexAttrib3sNV):
 GL_PREFIX(VertexAttrib3svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5656(%rax), %r11
+	movq	5680(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26605,13 +26756,13 @@ GL_PREFIX(VertexAttrib3svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5656(%rax), %r11
+	movq	5680(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5656(%rax), %r11
+	movq	5680(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26621,7 +26772,7 @@ GL_PREFIX(VertexAttrib3svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5656(%rax), %r11
+	movq	5680(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib3svNV), .-GL_PREFIX(VertexAttrib3svNV)
@@ -26632,7 +26783,7 @@ GL_PREFIX(VertexAttrib3svNV):
 GL_PREFIX(VertexAttrib4dNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5664(%rax), %r11
+	movq	5688(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$40, %rsp
@@ -26648,13 +26799,13 @@ GL_PREFIX(VertexAttrib4dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5664(%rax), %r11
+	movq	5688(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5664(%rax), %r11
+	movq	5688(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$40, %rsp
@@ -26670,7 +26821,7 @@ GL_PREFIX(VertexAttrib4dNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5664(%rax), %r11
+	movq	5688(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4dNV), .-GL_PREFIX(VertexAttrib4dNV)
@@ -26681,7 +26832,7 @@ GL_PREFIX(VertexAttrib4dNV):
 GL_PREFIX(VertexAttrib4dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5672(%rax), %r11
+	movq	5696(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26691,13 +26842,13 @@ GL_PREFIX(VertexAttrib4dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5672(%rax), %r11
+	movq	5696(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5672(%rax), %r11
+	movq	5696(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26707,7 +26858,7 @@ GL_PREFIX(VertexAttrib4dvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5672(%rax), %r11
+	movq	5696(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4dvNV), .-GL_PREFIX(VertexAttrib4dvNV)
@@ -26718,7 +26869,7 @@ GL_PREFIX(VertexAttrib4dvNV):
 GL_PREFIX(VertexAttrib4fNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5680(%rax), %r11
+	movq	5704(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$40, %rsp
@@ -26734,13 +26885,13 @@ GL_PREFIX(VertexAttrib4fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5680(%rax), %r11
+	movq	5704(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5680(%rax), %r11
+	movq	5704(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$40, %rsp
@@ -26756,7 +26907,7 @@ GL_PREFIX(VertexAttrib4fNV):
 	movq	8(%rsp), %xmm0
 	movq	(%rsp), %rdi
 	addq	$40, %rsp
-	movq	5680(%rax), %r11
+	movq	5704(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4fNV), .-GL_PREFIX(VertexAttrib4fNV)
@@ -26767,7 +26918,7 @@ GL_PREFIX(VertexAttrib4fNV):
 GL_PREFIX(VertexAttrib4fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5688(%rax), %r11
+	movq	5712(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26777,13 +26928,13 @@ GL_PREFIX(VertexAttrib4fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5688(%rax), %r11
+	movq	5712(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5688(%rax), %r11
+	movq	5712(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26793,7 +26944,7 @@ GL_PREFIX(VertexAttrib4fvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5688(%rax), %r11
+	movq	5712(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4fvNV), .-GL_PREFIX(VertexAttrib4fvNV)
@@ -26804,7 +26955,7 @@ GL_PREFIX(VertexAttrib4fvNV):
 GL_PREFIX(VertexAttrib4sNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5696(%rax), %r11
+	movq	5720(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26818,13 +26969,13 @@ GL_PREFIX(VertexAttrib4sNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5696(%rax), %r11
+	movq	5720(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5696(%rax), %r11
+	movq	5720(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26838,7 +26989,7 @@ GL_PREFIX(VertexAttrib4sNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5696(%rax), %r11
+	movq	5720(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4sNV), .-GL_PREFIX(VertexAttrib4sNV)
@@ -26849,7 +27000,7 @@ GL_PREFIX(VertexAttrib4sNV):
 GL_PREFIX(VertexAttrib4svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5704(%rax), %r11
+	movq	5728(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26859,13 +27010,13 @@ GL_PREFIX(VertexAttrib4svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5704(%rax), %r11
+	movq	5728(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5704(%rax), %r11
+	movq	5728(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26875,7 +27026,7 @@ GL_PREFIX(VertexAttrib4svNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5704(%rax), %r11
+	movq	5728(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4svNV), .-GL_PREFIX(VertexAttrib4svNV)
@@ -26886,7 +27037,7 @@ GL_PREFIX(VertexAttrib4svNV):
 GL_PREFIX(VertexAttrib4ubNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5712(%rax), %r11
+	movq	5736(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26900,13 +27051,13 @@ GL_PREFIX(VertexAttrib4ubNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5712(%rax), %r11
+	movq	5736(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5712(%rax), %r11
+	movq	5736(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26920,7 +27071,7 @@ GL_PREFIX(VertexAttrib4ubNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5712(%rax), %r11
+	movq	5736(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4ubNV), .-GL_PREFIX(VertexAttrib4ubNV)
@@ -26931,7 +27082,7 @@ GL_PREFIX(VertexAttrib4ubNV):
 GL_PREFIX(VertexAttrib4ubvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5720(%rax), %r11
+	movq	5744(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26941,13 +27092,13 @@ GL_PREFIX(VertexAttrib4ubvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5720(%rax), %r11
+	movq	5744(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5720(%rax), %r11
+	movq	5744(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -26957,7 +27108,7 @@ GL_PREFIX(VertexAttrib4ubvNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5720(%rax), %r11
+	movq	5744(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttrib4ubvNV), .-GL_PREFIX(VertexAttrib4ubvNV)
@@ -26968,7 +27119,7 @@ GL_PREFIX(VertexAttrib4ubvNV):
 GL_PREFIX(VertexAttribPointerNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5728(%rax), %r11
+	movq	5752(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -26982,13 +27133,13 @@ GL_PREFIX(VertexAttribPointerNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5728(%rax), %r11
+	movq	5752(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5728(%rax), %r11
+	movq	5752(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27002,7 +27153,7 @@ GL_PREFIX(VertexAttribPointerNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5728(%rax), %r11
+	movq	5752(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribPointerNV), .-GL_PREFIX(VertexAttribPointerNV)
@@ -27013,7 +27164,7 @@ GL_PREFIX(VertexAttribPointerNV):
 GL_PREFIX(VertexAttribs1dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5736(%rax), %r11
+	movq	5760(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27023,13 +27174,13 @@ GL_PREFIX(VertexAttribs1dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5736(%rax), %r11
+	movq	5760(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5736(%rax), %r11
+	movq	5760(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27039,7 +27190,7 @@ GL_PREFIX(VertexAttribs1dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5736(%rax), %r11
+	movq	5760(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs1dvNV), .-GL_PREFIX(VertexAttribs1dvNV)
@@ -27050,7 +27201,7 @@ GL_PREFIX(VertexAttribs1dvNV):
 GL_PREFIX(VertexAttribs1fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5744(%rax), %r11
+	movq	5768(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27060,13 +27211,13 @@ GL_PREFIX(VertexAttribs1fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5744(%rax), %r11
+	movq	5768(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5744(%rax), %r11
+	movq	5768(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27076,7 +27227,7 @@ GL_PREFIX(VertexAttribs1fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5744(%rax), %r11
+	movq	5768(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs1fvNV), .-GL_PREFIX(VertexAttribs1fvNV)
@@ -27087,7 +27238,7 @@ GL_PREFIX(VertexAttribs1fvNV):
 GL_PREFIX(VertexAttribs1svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5752(%rax), %r11
+	movq	5776(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27097,13 +27248,13 @@ GL_PREFIX(VertexAttribs1svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5752(%rax), %r11
+	movq	5776(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5752(%rax), %r11
+	movq	5776(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27113,7 +27264,7 @@ GL_PREFIX(VertexAttribs1svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5752(%rax), %r11
+	movq	5776(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs1svNV), .-GL_PREFIX(VertexAttribs1svNV)
@@ -27124,7 +27275,7 @@ GL_PREFIX(VertexAttribs1svNV):
 GL_PREFIX(VertexAttribs2dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5760(%rax), %r11
+	movq	5784(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27134,13 +27285,13 @@ GL_PREFIX(VertexAttribs2dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5760(%rax), %r11
+	movq	5784(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5760(%rax), %r11
+	movq	5784(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27150,7 +27301,7 @@ GL_PREFIX(VertexAttribs2dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5760(%rax), %r11
+	movq	5784(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs2dvNV), .-GL_PREFIX(VertexAttribs2dvNV)
@@ -27161,7 +27312,7 @@ GL_PREFIX(VertexAttribs2dvNV):
 GL_PREFIX(VertexAttribs2fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5768(%rax), %r11
+	movq	5792(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27171,13 +27322,13 @@ GL_PREFIX(VertexAttribs2fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5768(%rax), %r11
+	movq	5792(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5768(%rax), %r11
+	movq	5792(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27187,7 +27338,7 @@ GL_PREFIX(VertexAttribs2fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5768(%rax), %r11
+	movq	5792(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs2fvNV), .-GL_PREFIX(VertexAttribs2fvNV)
@@ -27198,7 +27349,7 @@ GL_PREFIX(VertexAttribs2fvNV):
 GL_PREFIX(VertexAttribs2svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5776(%rax), %r11
+	movq	5800(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27208,13 +27359,13 @@ GL_PREFIX(VertexAttribs2svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5776(%rax), %r11
+	movq	5800(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5776(%rax), %r11
+	movq	5800(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27224,7 +27375,7 @@ GL_PREFIX(VertexAttribs2svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5776(%rax), %r11
+	movq	5800(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs2svNV), .-GL_PREFIX(VertexAttribs2svNV)
@@ -27235,7 +27386,7 @@ GL_PREFIX(VertexAttribs2svNV):
 GL_PREFIX(VertexAttribs3dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5784(%rax), %r11
+	movq	5808(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27245,13 +27396,13 @@ GL_PREFIX(VertexAttribs3dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5784(%rax), %r11
+	movq	5808(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5784(%rax), %r11
+	movq	5808(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27261,7 +27412,7 @@ GL_PREFIX(VertexAttribs3dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5784(%rax), %r11
+	movq	5808(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs3dvNV), .-GL_PREFIX(VertexAttribs3dvNV)
@@ -27272,7 +27423,7 @@ GL_PREFIX(VertexAttribs3dvNV):
 GL_PREFIX(VertexAttribs3fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5792(%rax), %r11
+	movq	5816(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27282,13 +27433,13 @@ GL_PREFIX(VertexAttribs3fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5792(%rax), %r11
+	movq	5816(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5792(%rax), %r11
+	movq	5816(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27298,7 +27449,7 @@ GL_PREFIX(VertexAttribs3fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5792(%rax), %r11
+	movq	5816(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs3fvNV), .-GL_PREFIX(VertexAttribs3fvNV)
@@ -27309,7 +27460,7 @@ GL_PREFIX(VertexAttribs3fvNV):
 GL_PREFIX(VertexAttribs3svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5800(%rax), %r11
+	movq	5824(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27319,13 +27470,13 @@ GL_PREFIX(VertexAttribs3svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5800(%rax), %r11
+	movq	5824(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5800(%rax), %r11
+	movq	5824(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27335,7 +27486,7 @@ GL_PREFIX(VertexAttribs3svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5800(%rax), %r11
+	movq	5824(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs3svNV), .-GL_PREFIX(VertexAttribs3svNV)
@@ -27346,7 +27497,7 @@ GL_PREFIX(VertexAttribs3svNV):
 GL_PREFIX(VertexAttribs4dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5808(%rax), %r11
+	movq	5832(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27356,13 +27507,13 @@ GL_PREFIX(VertexAttribs4dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5808(%rax), %r11
+	movq	5832(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5808(%rax), %r11
+	movq	5832(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27372,7 +27523,7 @@ GL_PREFIX(VertexAttribs4dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5808(%rax), %r11
+	movq	5832(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs4dvNV), .-GL_PREFIX(VertexAttribs4dvNV)
@@ -27383,7 +27534,7 @@ GL_PREFIX(VertexAttribs4dvNV):
 GL_PREFIX(VertexAttribs4fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5816(%rax), %r11
+	movq	5840(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27393,13 +27544,13 @@ GL_PREFIX(VertexAttribs4fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5816(%rax), %r11
+	movq	5840(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5816(%rax), %r11
+	movq	5840(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27409,7 +27560,7 @@ GL_PREFIX(VertexAttribs4fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5816(%rax), %r11
+	movq	5840(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs4fvNV), .-GL_PREFIX(VertexAttribs4fvNV)
@@ -27420,7 +27571,7 @@ GL_PREFIX(VertexAttribs4fvNV):
 GL_PREFIX(VertexAttribs4svNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5824(%rax), %r11
+	movq	5848(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27430,13 +27581,13 @@ GL_PREFIX(VertexAttribs4svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5824(%rax), %r11
+	movq	5848(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5824(%rax), %r11
+	movq	5848(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27446,7 +27597,7 @@ GL_PREFIX(VertexAttribs4svNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5824(%rax), %r11
+	movq	5848(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs4svNV), .-GL_PREFIX(VertexAttribs4svNV)
@@ -27457,7 +27608,7 @@ GL_PREFIX(VertexAttribs4svNV):
 GL_PREFIX(VertexAttribs4ubvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5832(%rax), %r11
+	movq	5856(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27467,13 +27618,13 @@ GL_PREFIX(VertexAttribs4ubvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5832(%rax), %r11
+	movq	5856(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5832(%rax), %r11
+	movq	5856(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27483,7 +27634,7 @@ GL_PREFIX(VertexAttribs4ubvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5832(%rax), %r11
+	movq	5856(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(VertexAttribs4ubvNV), .-GL_PREFIX(VertexAttribs4ubvNV)
@@ -27494,7 +27645,7 @@ GL_PREFIX(VertexAttribs4ubvNV):
 GL_PREFIX(GetTexBumpParameterfvATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5840(%rax), %r11
+	movq	5864(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27504,13 +27655,13 @@ GL_PREFIX(GetTexBumpParameterfvATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5840(%rax), %r11
+	movq	5864(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5840(%rax), %r11
+	movq	5864(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27520,7 +27671,7 @@ GL_PREFIX(GetTexBumpParameterfvATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5840(%rax), %r11
+	movq	5864(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetTexBumpParameterfvATI), .-GL_PREFIX(GetTexBumpParameterfvATI)
@@ -27531,7 +27682,7 @@ GL_PREFIX(GetTexBumpParameterfvATI):
 GL_PREFIX(GetTexBumpParameterivATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5848(%rax), %r11
+	movq	5872(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27541,13 +27692,13 @@ GL_PREFIX(GetTexBumpParameterivATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5848(%rax), %r11
+	movq	5872(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5848(%rax), %r11
+	movq	5872(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27557,7 +27708,7 @@ GL_PREFIX(GetTexBumpParameterivATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5848(%rax), %r11
+	movq	5872(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetTexBumpParameterivATI), .-GL_PREFIX(GetTexBumpParameterivATI)
@@ -27568,7 +27719,7 @@ GL_PREFIX(GetTexBumpParameterivATI):
 GL_PREFIX(TexBumpParameterfvATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5856(%rax), %r11
+	movq	5880(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27578,13 +27729,13 @@ GL_PREFIX(TexBumpParameterfvATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5856(%rax), %r11
+	movq	5880(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5856(%rax), %r11
+	movq	5880(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27594,7 +27745,7 @@ GL_PREFIX(TexBumpParameterfvATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5856(%rax), %r11
+	movq	5880(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(TexBumpParameterfvATI), .-GL_PREFIX(TexBumpParameterfvATI)
@@ -27605,7 +27756,7 @@ GL_PREFIX(TexBumpParameterfvATI):
 GL_PREFIX(TexBumpParameterivATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5864(%rax), %r11
+	movq	5888(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27615,13 +27766,13 @@ GL_PREFIX(TexBumpParameterivATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5864(%rax), %r11
+	movq	5888(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5864(%rax), %r11
+	movq	5888(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27631,7 +27782,7 @@ GL_PREFIX(TexBumpParameterivATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5864(%rax), %r11
+	movq	5888(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(TexBumpParameterivATI), .-GL_PREFIX(TexBumpParameterivATI)
@@ -27642,7 +27793,7 @@ GL_PREFIX(TexBumpParameterivATI):
 GL_PREFIX(AlphaFragmentOp1ATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5872(%rax), %r11
+	movq	5896(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27660,13 +27811,13 @@ GL_PREFIX(AlphaFragmentOp1ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5872(%rax), %r11
+	movq	5896(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5872(%rax), %r11
+	movq	5896(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27684,7 +27835,7 @@ GL_PREFIX(AlphaFragmentOp1ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5872(%rax), %r11
+	movq	5896(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(AlphaFragmentOp1ATI), .-GL_PREFIX(AlphaFragmentOp1ATI)
@@ -27695,7 +27846,7 @@ GL_PREFIX(AlphaFragmentOp1ATI):
 GL_PREFIX(AlphaFragmentOp2ATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5880(%rax), %r11
+	movq	5904(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27713,13 +27864,13 @@ GL_PREFIX(AlphaFragmentOp2ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5880(%rax), %r11
+	movq	5904(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5880(%rax), %r11
+	movq	5904(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27737,7 +27888,7 @@ GL_PREFIX(AlphaFragmentOp2ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5880(%rax), %r11
+	movq	5904(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(AlphaFragmentOp2ATI), .-GL_PREFIX(AlphaFragmentOp2ATI)
@@ -27748,7 +27899,7 @@ GL_PREFIX(AlphaFragmentOp2ATI):
 GL_PREFIX(AlphaFragmentOp3ATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5888(%rax), %r11
+	movq	5912(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27766,13 +27917,13 @@ GL_PREFIX(AlphaFragmentOp3ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5888(%rax), %r11
+	movq	5912(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5888(%rax), %r11
+	movq	5912(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27790,7 +27941,7 @@ GL_PREFIX(AlphaFragmentOp3ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5888(%rax), %r11
+	movq	5912(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(AlphaFragmentOp3ATI), .-GL_PREFIX(AlphaFragmentOp3ATI)
@@ -27801,25 +27952,25 @@ GL_PREFIX(AlphaFragmentOp3ATI):
 GL_PREFIX(BeginFragmentShaderATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5896(%rax), %r11
+	movq	5920(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
-	movq	5896(%rax), %r11
+	movq	5920(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5896(%rax), %r11
+	movq	5920(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
-	movq	5896(%rax), %r11
+	movq	5920(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(BeginFragmentShaderATI), .-GL_PREFIX(BeginFragmentShaderATI)
@@ -27830,25 +27981,25 @@ GL_PREFIX(BeginFragmentShaderATI):
 GL_PREFIX(BindFragmentShaderATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5904(%rax), %r11
+	movq	5928(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5904(%rax), %r11
+	movq	5928(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5904(%rax), %r11
+	movq	5928(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5904(%rax), %r11
+	movq	5928(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(BindFragmentShaderATI), .-GL_PREFIX(BindFragmentShaderATI)
@@ -27859,7 +28010,7 @@ GL_PREFIX(BindFragmentShaderATI):
 GL_PREFIX(ColorFragmentOp1ATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5912(%rax), %r11
+	movq	5936(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27877,13 +28028,13 @@ GL_PREFIX(ColorFragmentOp1ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5912(%rax), %r11
+	movq	5936(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5912(%rax), %r11
+	movq	5936(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27901,7 +28052,7 @@ GL_PREFIX(ColorFragmentOp1ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5912(%rax), %r11
+	movq	5936(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ColorFragmentOp1ATI), .-GL_PREFIX(ColorFragmentOp1ATI)
@@ -27912,7 +28063,7 @@ GL_PREFIX(ColorFragmentOp1ATI):
 GL_PREFIX(ColorFragmentOp2ATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5920(%rax), %r11
+	movq	5944(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27930,13 +28081,13 @@ GL_PREFIX(ColorFragmentOp2ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5920(%rax), %r11
+	movq	5944(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5920(%rax), %r11
+	movq	5944(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -27954,7 +28105,7 @@ GL_PREFIX(ColorFragmentOp2ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5920(%rax), %r11
+	movq	5944(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ColorFragmentOp2ATI), .-GL_PREFIX(ColorFragmentOp2ATI)
@@ -27965,7 +28116,7 @@ GL_PREFIX(ColorFragmentOp2ATI):
 GL_PREFIX(ColorFragmentOp3ATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5928(%rax), %r11
+	movq	5952(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -27983,13 +28134,13 @@ GL_PREFIX(ColorFragmentOp3ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5928(%rax), %r11
+	movq	5952(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5928(%rax), %r11
+	movq	5952(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28007,7 +28158,7 @@ GL_PREFIX(ColorFragmentOp3ATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5928(%rax), %r11
+	movq	5952(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ColorFragmentOp3ATI), .-GL_PREFIX(ColorFragmentOp3ATI)
@@ -28018,25 +28169,25 @@ GL_PREFIX(ColorFragmentOp3ATI):
 GL_PREFIX(DeleteFragmentShaderATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5936(%rax), %r11
+	movq	5960(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5936(%rax), %r11
+	movq	5960(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5936(%rax), %r11
+	movq	5960(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5936(%rax), %r11
+	movq	5960(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(DeleteFragmentShaderATI), .-GL_PREFIX(DeleteFragmentShaderATI)
@@ -28047,25 +28198,25 @@ GL_PREFIX(DeleteFragmentShaderATI):
 GL_PREFIX(EndFragmentShaderATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5944(%rax), %r11
+	movq	5968(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rbp
 	call	_x86_64_get_dispatch@PLT
 	popq	%rbp
-	movq	5944(%rax), %r11
+	movq	5968(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5944(%rax), %r11
+	movq	5968(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rbp
 	call	_glapi_get_dispatch
 	popq	%rbp
-	movq	5944(%rax), %r11
+	movq	5968(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(EndFragmentShaderATI), .-GL_PREFIX(EndFragmentShaderATI)
@@ -28076,25 +28227,25 @@ GL_PREFIX(EndFragmentShaderATI):
 GL_PREFIX(GenFragmentShadersATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5952(%rax), %r11
+	movq	5976(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	5952(%rax), %r11
+	movq	5976(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5952(%rax), %r11
+	movq	5976(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	5952(%rax), %r11
+	movq	5976(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GenFragmentShadersATI), .-GL_PREFIX(GenFragmentShadersATI)
@@ -28105,7 +28256,7 @@ GL_PREFIX(GenFragmentShadersATI):
 GL_PREFIX(PassTexCoordATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5960(%rax), %r11
+	movq	5984(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28115,13 +28266,13 @@ GL_PREFIX(PassTexCoordATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5960(%rax), %r11
+	movq	5984(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5960(%rax), %r11
+	movq	5984(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28131,7 +28282,7 @@ GL_PREFIX(PassTexCoordATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5960(%rax), %r11
+	movq	5984(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(PassTexCoordATI), .-GL_PREFIX(PassTexCoordATI)
@@ -28142,7 +28293,7 @@ GL_PREFIX(PassTexCoordATI):
 GL_PREFIX(SampleMapATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5968(%rax), %r11
+	movq	5992(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28152,13 +28303,13 @@ GL_PREFIX(SampleMapATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5968(%rax), %r11
+	movq	5992(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5968(%rax), %r11
+	movq	5992(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28168,7 +28319,7 @@ GL_PREFIX(SampleMapATI):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	5968(%rax), %r11
+	movq	5992(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SampleMapATI), .-GL_PREFIX(SampleMapATI)
@@ -28179,7 +28330,7 @@ GL_PREFIX(SampleMapATI):
 GL_PREFIX(SetFragmentShaderConstantATI):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5976(%rax), %r11
+	movq	6000(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28189,13 +28340,13 @@ GL_PREFIX(SetFragmentShaderConstantATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5976(%rax), %r11
+	movq	6000(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5976(%rax), %r11
+	movq	6000(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28205,7 +28356,7 @@ GL_PREFIX(SetFragmentShaderConstantATI):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5976(%rax), %r11
+	movq	6000(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(SetFragmentShaderConstantATI), .-GL_PREFIX(SetFragmentShaderConstantATI)
@@ -28216,7 +28367,7 @@ GL_PREFIX(SetFragmentShaderConstantATI):
 GL_PREFIX(PointParameteriNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5984(%rax), %r11
+	movq	6008(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28226,13 +28377,13 @@ GL_PREFIX(PointParameteriNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5984(%rax), %r11
+	movq	6008(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5984(%rax), %r11
+	movq	6008(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28242,7 +28393,7 @@ GL_PREFIX(PointParameteriNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5984(%rax), %r11
+	movq	6008(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(PointParameteriNV), .-GL_PREFIX(PointParameteriNV)
@@ -28253,7 +28404,7 @@ GL_PREFIX(PointParameteriNV):
 GL_PREFIX(PointParameterivNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	5992(%rax), %r11
+	movq	6016(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28263,13 +28414,13 @@ GL_PREFIX(PointParameterivNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5992(%rax), %r11
+	movq	6016(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	5992(%rax), %r11
+	movq	6016(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28279,79 +28430,79 @@ GL_PREFIX(PointParameterivNV):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	5992(%rax), %r11
+	movq	6016(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(PointParameterivNV), .-GL_PREFIX(PointParameterivNV)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_750)
-	.type	GL_PREFIX(_dispatch_stub_750), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_750))
-GL_PREFIX(_dispatch_stub_750):
+	.globl	GL_PREFIX(_dispatch_stub_753)
+	.type	GL_PREFIX(_dispatch_stub_753), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_753))
+GL_PREFIX(_dispatch_stub_753):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6000(%rax), %r11
+	movq	6024(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6000(%rax), %r11
+	movq	6024(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6000(%rax), %r11
+	movq	6024(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6000(%rax), %r11
+	movq	6024(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_750), .-GL_PREFIX(_dispatch_stub_750)
+	.size	GL_PREFIX(_dispatch_stub_753), .-GL_PREFIX(_dispatch_stub_753)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_751)
-	.type	GL_PREFIX(_dispatch_stub_751), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_751))
-GL_PREFIX(_dispatch_stub_751):
+	.globl	GL_PREFIX(_dispatch_stub_754)
+	.type	GL_PREFIX(_dispatch_stub_754), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_754))
+GL_PREFIX(_dispatch_stub_754):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6008(%rax), %r11
+	movq	6032(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6008(%rax), %r11
+	movq	6032(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6008(%rax), %r11
+	movq	6032(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6008(%rax), %r11
+	movq	6032(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_751), .-GL_PREFIX(_dispatch_stub_751)
+	.size	GL_PREFIX(_dispatch_stub_754), .-GL_PREFIX(_dispatch_stub_754)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_752)
-	.type	GL_PREFIX(_dispatch_stub_752), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_752))
-GL_PREFIX(_dispatch_stub_752):
+	.globl	GL_PREFIX(_dispatch_stub_755)
+	.type	GL_PREFIX(_dispatch_stub_755), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_755))
+GL_PREFIX(_dispatch_stub_755):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6016(%rax), %r11
+	movq	6040(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28361,13 +28512,13 @@ GL_PREFIX(_dispatch_stub_752):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6016(%rax), %r11
+	movq	6040(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6016(%rax), %r11
+	movq	6040(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28377,19 +28528,19 @@ GL_PREFIX(_dispatch_stub_752):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6016(%rax), %r11
+	movq	6040(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_752), .-GL_PREFIX(_dispatch_stub_752)
+	.size	GL_PREFIX(_dispatch_stub_755), .-GL_PREFIX(_dispatch_stub_755)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_753)
-	.type	GL_PREFIX(_dispatch_stub_753), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_753))
-GL_PREFIX(_dispatch_stub_753):
+	.globl	GL_PREFIX(_dispatch_stub_756)
+	.type	GL_PREFIX(_dispatch_stub_756), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_756))
+GL_PREFIX(_dispatch_stub_756):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6024(%rax), %r11
+	movq	6048(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28399,13 +28550,13 @@ GL_PREFIX(_dispatch_stub_753):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6024(%rax), %r11
+	movq	6048(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6024(%rax), %r11
+	movq	6048(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28415,40 +28566,40 @@ GL_PREFIX(_dispatch_stub_753):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6024(%rax), %r11
+	movq	6048(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_753), .-GL_PREFIX(_dispatch_stub_753)
+	.size	GL_PREFIX(_dispatch_stub_756), .-GL_PREFIX(_dispatch_stub_756)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_754)
-	.type	GL_PREFIX(_dispatch_stub_754), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_754))
-GL_PREFIX(_dispatch_stub_754):
+	.globl	GL_PREFIX(_dispatch_stub_757)
+	.type	GL_PREFIX(_dispatch_stub_757), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_757))
+GL_PREFIX(_dispatch_stub_757):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6032(%rax), %r11
+	movq	6056(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6032(%rax), %r11
+	movq	6056(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6032(%rax), %r11
+	movq	6056(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6032(%rax), %r11
+	movq	6056(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_754), .-GL_PREFIX(_dispatch_stub_754)
+	.size	GL_PREFIX(_dispatch_stub_757), .-GL_PREFIX(_dispatch_stub_757)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(GetProgramNamedParameterdvNV)
@@ -28456,7 +28607,7 @@ GL_PREFIX(_dispatch_stub_754):
 GL_PREFIX(GetProgramNamedParameterdvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6040(%rax), %r11
+	movq	6064(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28470,13 +28621,13 @@ GL_PREFIX(GetProgramNamedParameterdvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6040(%rax), %r11
+	movq	6064(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6040(%rax), %r11
+	movq	6064(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28490,7 +28641,7 @@ GL_PREFIX(GetProgramNamedParameterdvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6040(%rax), %r11
+	movq	6064(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetProgramNamedParameterdvNV), .-GL_PREFIX(GetProgramNamedParameterdvNV)
@@ -28501,7 +28652,7 @@ GL_PREFIX(GetProgramNamedParameterdvNV):
 GL_PREFIX(GetProgramNamedParameterfvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6048(%rax), %r11
+	movq	6072(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28515,13 +28666,13 @@ GL_PREFIX(GetProgramNamedParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6048(%rax), %r11
+	movq	6072(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6048(%rax), %r11
+	movq	6072(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28535,7 +28686,7 @@ GL_PREFIX(GetProgramNamedParameterfvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6048(%rax), %r11
+	movq	6072(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetProgramNamedParameterfvNV), .-GL_PREFIX(GetProgramNamedParameterfvNV)
@@ -28546,7 +28697,7 @@ GL_PREFIX(GetProgramNamedParameterfvNV):
 GL_PREFIX(ProgramNamedParameter4dNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6056(%rax), %r11
+	movq	6080(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$56, %rsp
@@ -28566,13 +28717,13 @@ GL_PREFIX(ProgramNamedParameter4dNV):
 	movq	8(%rsp), %rsi
 	movq	(%rsp), %rdi
 	addq	$56, %rsp
-	movq	6056(%rax), %r11
+	movq	6080(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6056(%rax), %r11
+	movq	6080(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$56, %rsp
@@ -28592,7 +28743,7 @@ GL_PREFIX(ProgramNamedParameter4dNV):
 	movq	8(%rsp), %rsi
 	movq	(%rsp), %rdi
 	addq	$56, %rsp
-	movq	6056(%rax), %r11
+	movq	6080(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProgramNamedParameter4dNV), .-GL_PREFIX(ProgramNamedParameter4dNV)
@@ -28603,7 +28754,7 @@ GL_PREFIX(ProgramNamedParameter4dNV):
 GL_PREFIX(ProgramNamedParameter4dvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6064(%rax), %r11
+	movq	6088(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28617,13 +28768,13 @@ GL_PREFIX(ProgramNamedParameter4dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6064(%rax), %r11
+	movq	6088(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6064(%rax), %r11
+	movq	6088(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28637,7 +28788,7 @@ GL_PREFIX(ProgramNamedParameter4dvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6064(%rax), %r11
+	movq	6088(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProgramNamedParameter4dvNV), .-GL_PREFIX(ProgramNamedParameter4dvNV)
@@ -28648,7 +28799,7 @@ GL_PREFIX(ProgramNamedParameter4dvNV):
 GL_PREFIX(ProgramNamedParameter4fNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6072(%rax), %r11
+	movq	6096(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	subq	$56, %rsp
@@ -28668,13 +28819,13 @@ GL_PREFIX(ProgramNamedParameter4fNV):
 	movq	8(%rsp), %rsi
 	movq	(%rsp), %rdi
 	addq	$56, %rsp
-	movq	6072(%rax), %r11
+	movq	6096(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6072(%rax), %r11
+	movq	6096(%rax), %r11
 	jmp	*%r11
 1:
 	subq	$56, %rsp
@@ -28694,7 +28845,7 @@ GL_PREFIX(ProgramNamedParameter4fNV):
 	movq	8(%rsp), %rsi
 	movq	(%rsp), %rdi
 	addq	$56, %rsp
-	movq	6072(%rax), %r11
+	movq	6096(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProgramNamedParameter4fNV), .-GL_PREFIX(ProgramNamedParameter4fNV)
@@ -28705,7 +28856,7 @@ GL_PREFIX(ProgramNamedParameter4fNV):
 GL_PREFIX(ProgramNamedParameter4fvNV):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6080(%rax), %r11
+	movq	6104(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28719,13 +28870,13 @@ GL_PREFIX(ProgramNamedParameter4fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6080(%rax), %r11
+	movq	6104(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6080(%rax), %r11
+	movq	6104(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28739,19 +28890,19 @@ GL_PREFIX(ProgramNamedParameter4fvNV):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6080(%rax), %r11
+	movq	6104(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProgramNamedParameter4fvNV), .-GL_PREFIX(ProgramNamedParameter4fvNV)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_761)
-	.type	GL_PREFIX(_dispatch_stub_761), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_761))
-GL_PREFIX(_dispatch_stub_761):
+	.globl	GL_PREFIX(_dispatch_stub_764)
+	.type	GL_PREFIX(_dispatch_stub_764), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_764))
+GL_PREFIX(_dispatch_stub_764):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6088(%rax), %r11
+	movq	6112(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28761,13 +28912,13 @@ GL_PREFIX(_dispatch_stub_761):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6088(%rax), %r11
+	movq	6112(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6088(%rax), %r11
+	movq	6112(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28777,19 +28928,19 @@ GL_PREFIX(_dispatch_stub_761):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6088(%rax), %r11
+	movq	6112(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_761), .-GL_PREFIX(_dispatch_stub_761)
+	.size	GL_PREFIX(_dispatch_stub_764), .-GL_PREFIX(_dispatch_stub_764)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_762)
-	.type	GL_PREFIX(_dispatch_stub_762), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_762))
-GL_PREFIX(_dispatch_stub_762):
+	.globl	GL_PREFIX(_dispatch_stub_765)
+	.type	GL_PREFIX(_dispatch_stub_765), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_765))
+GL_PREFIX(_dispatch_stub_765):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6096(%rax), %r11
+	movq	6120(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28799,13 +28950,13 @@ GL_PREFIX(_dispatch_stub_762):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6096(%rax), %r11
+	movq	6120(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6096(%rax), %r11
+	movq	6120(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28815,10 +28966,10 @@ GL_PREFIX(_dispatch_stub_762):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6096(%rax), %r11
+	movq	6120(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_762), .-GL_PREFIX(_dispatch_stub_762)
+	.size	GL_PREFIX(_dispatch_stub_765), .-GL_PREFIX(_dispatch_stub_765)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(BindFramebufferEXT)
@@ -28826,7 +28977,7 @@ GL_PREFIX(_dispatch_stub_762):
 GL_PREFIX(BindFramebufferEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6104(%rax), %r11
+	movq	6128(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28836,13 +28987,13 @@ GL_PREFIX(BindFramebufferEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6104(%rax), %r11
+	movq	6128(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6104(%rax), %r11
+	movq	6128(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28852,7 +29003,7 @@ GL_PREFIX(BindFramebufferEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6104(%rax), %r11
+	movq	6128(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(BindFramebufferEXT), .-GL_PREFIX(BindFramebufferEXT)
@@ -28863,7 +29014,7 @@ GL_PREFIX(BindFramebufferEXT):
 GL_PREFIX(BindRenderbufferEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6112(%rax), %r11
+	movq	6136(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28873,13 +29024,13 @@ GL_PREFIX(BindRenderbufferEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6112(%rax), %r11
+	movq	6136(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6112(%rax), %r11
+	movq	6136(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28889,7 +29040,7 @@ GL_PREFIX(BindRenderbufferEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6112(%rax), %r11
+	movq	6136(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(BindRenderbufferEXT), .-GL_PREFIX(BindRenderbufferEXT)
@@ -28900,25 +29051,25 @@ GL_PREFIX(BindRenderbufferEXT):
 GL_PREFIX(CheckFramebufferStatusEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6120(%rax), %r11
+	movq	6144(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6120(%rax), %r11
+	movq	6144(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6120(%rax), %r11
+	movq	6144(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6120(%rax), %r11
+	movq	6144(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(CheckFramebufferStatusEXT), .-GL_PREFIX(CheckFramebufferStatusEXT)
@@ -28929,7 +29080,7 @@ GL_PREFIX(CheckFramebufferStatusEXT):
 GL_PREFIX(DeleteFramebuffersEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6128(%rax), %r11
+	movq	6152(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28939,13 +29090,13 @@ GL_PREFIX(DeleteFramebuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6128(%rax), %r11
+	movq	6152(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6128(%rax), %r11
+	movq	6152(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28955,7 +29106,7 @@ GL_PREFIX(DeleteFramebuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6128(%rax), %r11
+	movq	6152(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(DeleteFramebuffersEXT), .-GL_PREFIX(DeleteFramebuffersEXT)
@@ -28966,7 +29117,7 @@ GL_PREFIX(DeleteFramebuffersEXT):
 GL_PREFIX(DeleteRenderbuffersEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6136(%rax), %r11
+	movq	6160(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -28976,13 +29127,13 @@ GL_PREFIX(DeleteRenderbuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6136(%rax), %r11
+	movq	6160(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6136(%rax), %r11
+	movq	6160(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -28992,7 +29143,7 @@ GL_PREFIX(DeleteRenderbuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6136(%rax), %r11
+	movq	6160(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(DeleteRenderbuffersEXT), .-GL_PREFIX(DeleteRenderbuffersEXT)
@@ -29003,7 +29154,7 @@ GL_PREFIX(DeleteRenderbuffersEXT):
 GL_PREFIX(FramebufferRenderbufferEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6144(%rax), %r11
+	movq	6168(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29017,13 +29168,13 @@ GL_PREFIX(FramebufferRenderbufferEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6144(%rax), %r11
+	movq	6168(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6144(%rax), %r11
+	movq	6168(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29037,7 +29188,7 @@ GL_PREFIX(FramebufferRenderbufferEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6144(%rax), %r11
+	movq	6168(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FramebufferRenderbufferEXT), .-GL_PREFIX(FramebufferRenderbufferEXT)
@@ -29048,7 +29199,7 @@ GL_PREFIX(FramebufferRenderbufferEXT):
 GL_PREFIX(FramebufferTexture1DEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6152(%rax), %r11
+	movq	6176(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29062,13 +29213,13 @@ GL_PREFIX(FramebufferTexture1DEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6152(%rax), %r11
+	movq	6176(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6152(%rax), %r11
+	movq	6176(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29082,7 +29233,7 @@ GL_PREFIX(FramebufferTexture1DEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6152(%rax), %r11
+	movq	6176(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FramebufferTexture1DEXT), .-GL_PREFIX(FramebufferTexture1DEXT)
@@ -29093,7 +29244,7 @@ GL_PREFIX(FramebufferTexture1DEXT):
 GL_PREFIX(FramebufferTexture2DEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6160(%rax), %r11
+	movq	6184(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29107,13 +29258,13 @@ GL_PREFIX(FramebufferTexture2DEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6160(%rax), %r11
+	movq	6184(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6160(%rax), %r11
+	movq	6184(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29127,7 +29278,7 @@ GL_PREFIX(FramebufferTexture2DEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6160(%rax), %r11
+	movq	6184(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FramebufferTexture2DEXT), .-GL_PREFIX(FramebufferTexture2DEXT)
@@ -29138,7 +29289,7 @@ GL_PREFIX(FramebufferTexture2DEXT):
 GL_PREFIX(FramebufferTexture3DEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6168(%rax), %r11
+	movq	6192(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29156,13 +29307,13 @@ GL_PREFIX(FramebufferTexture3DEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6168(%rax), %r11
+	movq	6192(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6168(%rax), %r11
+	movq	6192(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29180,7 +29331,7 @@ GL_PREFIX(FramebufferTexture3DEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6168(%rax), %r11
+	movq	6192(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FramebufferTexture3DEXT), .-GL_PREFIX(FramebufferTexture3DEXT)
@@ -29191,7 +29342,7 @@ GL_PREFIX(FramebufferTexture3DEXT):
 GL_PREFIX(GenFramebuffersEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6176(%rax), %r11
+	movq	6200(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29201,13 +29352,13 @@ GL_PREFIX(GenFramebuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6176(%rax), %r11
+	movq	6200(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6176(%rax), %r11
+	movq	6200(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29217,7 +29368,7 @@ GL_PREFIX(GenFramebuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6176(%rax), %r11
+	movq	6200(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GenFramebuffersEXT), .-GL_PREFIX(GenFramebuffersEXT)
@@ -29228,7 +29379,7 @@ GL_PREFIX(GenFramebuffersEXT):
 GL_PREFIX(GenRenderbuffersEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6184(%rax), %r11
+	movq	6208(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29238,13 +29389,13 @@ GL_PREFIX(GenRenderbuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6184(%rax), %r11
+	movq	6208(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6184(%rax), %r11
+	movq	6208(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29254,7 +29405,7 @@ GL_PREFIX(GenRenderbuffersEXT):
 	popq	%rbp
 	popq	%rsi
 	popq	%rdi
-	movq	6184(%rax), %r11
+	movq	6208(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GenRenderbuffersEXT), .-GL_PREFIX(GenRenderbuffersEXT)
@@ -29265,25 +29416,25 @@ GL_PREFIX(GenRenderbuffersEXT):
 GL_PREFIX(GenerateMipmapEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6192(%rax), %r11
+	movq	6216(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6192(%rax), %r11
+	movq	6216(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6192(%rax), %r11
+	movq	6216(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6192(%rax), %r11
+	movq	6216(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GenerateMipmapEXT), .-GL_PREFIX(GenerateMipmapEXT)
@@ -29294,7 +29445,7 @@ GL_PREFIX(GenerateMipmapEXT):
 GL_PREFIX(GetFramebufferAttachmentParameterivEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6200(%rax), %r11
+	movq	6224(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29308,13 +29459,13 @@ GL_PREFIX(GetFramebufferAttachmentParameterivEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6200(%rax), %r11
+	movq	6224(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6200(%rax), %r11
+	movq	6224(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29328,7 +29479,7 @@ GL_PREFIX(GetFramebufferAttachmentParameterivEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6200(%rax), %r11
+	movq	6224(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetFramebufferAttachmentParameterivEXT), .-GL_PREFIX(GetFramebufferAttachmentParameterivEXT)
@@ -29339,7 +29490,7 @@ GL_PREFIX(GetFramebufferAttachmentParameterivEXT):
 GL_PREFIX(GetRenderbufferParameterivEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6208(%rax), %r11
+	movq	6232(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29349,13 +29500,13 @@ GL_PREFIX(GetRenderbufferParameterivEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6208(%rax), %r11
+	movq	6232(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6208(%rax), %r11
+	movq	6232(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29365,7 +29516,7 @@ GL_PREFIX(GetRenderbufferParameterivEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6208(%rax), %r11
+	movq	6232(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(GetRenderbufferParameterivEXT), .-GL_PREFIX(GetRenderbufferParameterivEXT)
@@ -29376,25 +29527,25 @@ GL_PREFIX(GetRenderbufferParameterivEXT):
 GL_PREFIX(IsFramebufferEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6216(%rax), %r11
+	movq	6240(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6216(%rax), %r11
+	movq	6240(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6216(%rax), %r11
+	movq	6240(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6216(%rax), %r11
+	movq	6240(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(IsFramebufferEXT), .-GL_PREFIX(IsFramebufferEXT)
@@ -29405,25 +29556,25 @@ GL_PREFIX(IsFramebufferEXT):
 GL_PREFIX(IsRenderbufferEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6224(%rax), %r11
+	movq	6248(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6224(%rax), %r11
+	movq	6248(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6224(%rax), %r11
+	movq	6248(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6224(%rax), %r11
+	movq	6248(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(IsRenderbufferEXT), .-GL_PREFIX(IsRenderbufferEXT)
@@ -29434,7 +29585,7 @@ GL_PREFIX(IsRenderbufferEXT):
 GL_PREFIX(RenderbufferStorageEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6232(%rax), %r11
+	movq	6256(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29448,13 +29599,13 @@ GL_PREFIX(RenderbufferStorageEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6232(%rax), %r11
+	movq	6256(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6232(%rax), %r11
+	movq	6256(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29468,19 +29619,19 @@ GL_PREFIX(RenderbufferStorageEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6232(%rax), %r11
+	movq	6256(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(RenderbufferStorageEXT), .-GL_PREFIX(RenderbufferStorageEXT)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_780)
-	.type	GL_PREFIX(_dispatch_stub_780), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_780))
-GL_PREFIX(_dispatch_stub_780):
+	.globl	GL_PREFIX(_dispatch_stub_783)
+	.type	GL_PREFIX(_dispatch_stub_783), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_783))
+GL_PREFIX(_dispatch_stub_783):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6240(%rax), %r11
+	movq	6264(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29498,13 +29649,13 @@ GL_PREFIX(_dispatch_stub_780):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6240(%rax), %r11
+	movq	6264(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6240(%rax), %r11
+	movq	6264(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29522,19 +29673,19 @@ GL_PREFIX(_dispatch_stub_780):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6240(%rax), %r11
+	movq	6264(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_780), .-GL_PREFIX(_dispatch_stub_780)
+	.size	GL_PREFIX(_dispatch_stub_783), .-GL_PREFIX(_dispatch_stub_783)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_781)
-	.type	GL_PREFIX(_dispatch_stub_781), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_781))
-GL_PREFIX(_dispatch_stub_781):
+	.globl	GL_PREFIX(_dispatch_stub_784)
+	.type	GL_PREFIX(_dispatch_stub_784), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_784))
+GL_PREFIX(_dispatch_stub_784):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6248(%rax), %r11
+	movq	6272(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29544,13 +29695,13 @@ GL_PREFIX(_dispatch_stub_781):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6248(%rax), %r11
+	movq	6272(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6248(%rax), %r11
+	movq	6272(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29560,19 +29711,19 @@ GL_PREFIX(_dispatch_stub_781):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6248(%rax), %r11
+	movq	6272(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_781), .-GL_PREFIX(_dispatch_stub_781)
+	.size	GL_PREFIX(_dispatch_stub_784), .-GL_PREFIX(_dispatch_stub_784)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_782)
-	.type	GL_PREFIX(_dispatch_stub_782), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_782))
-GL_PREFIX(_dispatch_stub_782):
+	.globl	GL_PREFIX(_dispatch_stub_785)
+	.type	GL_PREFIX(_dispatch_stub_785), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_785))
+GL_PREFIX(_dispatch_stub_785):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6256(%rax), %r11
+	movq	6280(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29582,13 +29733,13 @@ GL_PREFIX(_dispatch_stub_782):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6256(%rax), %r11
+	movq	6280(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6256(%rax), %r11
+	movq	6280(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29598,10 +29749,10 @@ GL_PREFIX(_dispatch_stub_782):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6256(%rax), %r11
+	movq	6280(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_782), .-GL_PREFIX(_dispatch_stub_782)
+	.size	GL_PREFIX(_dispatch_stub_785), .-GL_PREFIX(_dispatch_stub_785)
 
 	.p2align	4,,15
 	.globl	GL_PREFIX(FramebufferTextureLayerEXT)
@@ -29609,7 +29760,7 @@ GL_PREFIX(_dispatch_stub_782):
 GL_PREFIX(FramebufferTextureLayerEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6264(%rax), %r11
+	movq	6288(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29623,13 +29774,13 @@ GL_PREFIX(FramebufferTextureLayerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6264(%rax), %r11
+	movq	6288(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6264(%rax), %r11
+	movq	6288(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29643,7 +29794,7 @@ GL_PREFIX(FramebufferTextureLayerEXT):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6264(%rax), %r11
+	movq	6288(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(FramebufferTextureLayerEXT), .-GL_PREFIX(FramebufferTextureLayerEXT)
@@ -29654,37 +29805,37 @@ GL_PREFIX(FramebufferTextureLayerEXT):
 GL_PREFIX(ProvokingVertexEXT):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6272(%rax), %r11
+	movq	6296(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
 	call	_x86_64_get_dispatch@PLT
 	popq	%rdi
-	movq	6272(%rax), %r11
+	movq	6296(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6272(%rax), %r11
+	movq	6296(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
 	call	_glapi_get_dispatch
 	popq	%rdi
-	movq	6272(%rax), %r11
+	movq	6296(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
 	.size	GL_PREFIX(ProvokingVertexEXT), .-GL_PREFIX(ProvokingVertexEXT)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_785)
-	.type	GL_PREFIX(_dispatch_stub_785), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_785))
-GL_PREFIX(_dispatch_stub_785):
+	.globl	GL_PREFIX(_dispatch_stub_788)
+	.type	GL_PREFIX(_dispatch_stub_788), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_788))
+GL_PREFIX(_dispatch_stub_788):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6280(%rax), %r11
+	movq	6304(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29694,13 +29845,13 @@ GL_PREFIX(_dispatch_stub_785):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6280(%rax), %r11
+	movq	6304(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6280(%rax), %r11
+	movq	6304(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29710,19 +29861,19 @@ GL_PREFIX(_dispatch_stub_785):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6280(%rax), %r11
+	movq	6304(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_785), .-GL_PREFIX(_dispatch_stub_785)
+	.size	GL_PREFIX(_dispatch_stub_788), .-GL_PREFIX(_dispatch_stub_788)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_786)
-	.type	GL_PREFIX(_dispatch_stub_786), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_786))
-GL_PREFIX(_dispatch_stub_786):
+	.globl	GL_PREFIX(_dispatch_stub_789)
+	.type	GL_PREFIX(_dispatch_stub_789), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_789))
+GL_PREFIX(_dispatch_stub_789):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6288(%rax), %r11
+	movq	6312(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29732,13 +29883,13 @@ GL_PREFIX(_dispatch_stub_786):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6288(%rax), %r11
+	movq	6312(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6288(%rax), %r11
+	movq	6312(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29748,19 +29899,19 @@ GL_PREFIX(_dispatch_stub_786):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6288(%rax), %r11
+	movq	6312(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_786), .-GL_PREFIX(_dispatch_stub_786)
+	.size	GL_PREFIX(_dispatch_stub_789), .-GL_PREFIX(_dispatch_stub_789)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_787)
-	.type	GL_PREFIX(_dispatch_stub_787), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_787))
-GL_PREFIX(_dispatch_stub_787):
+	.globl	GL_PREFIX(_dispatch_stub_790)
+	.type	GL_PREFIX(_dispatch_stub_790), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_790))
+GL_PREFIX(_dispatch_stub_790):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6296(%rax), %r11
+	movq	6320(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29774,13 +29925,13 @@ GL_PREFIX(_dispatch_stub_787):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6296(%rax), %r11
+	movq	6320(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6296(%rax), %r11
+	movq	6320(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29794,19 +29945,19 @@ GL_PREFIX(_dispatch_stub_787):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6296(%rax), %r11
+	movq	6320(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_787), .-GL_PREFIX(_dispatch_stub_787)
+	.size	GL_PREFIX(_dispatch_stub_790), .-GL_PREFIX(_dispatch_stub_790)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_788)
-	.type	GL_PREFIX(_dispatch_stub_788), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_788))
-GL_PREFIX(_dispatch_stub_788):
+	.globl	GL_PREFIX(_dispatch_stub_791)
+	.type	GL_PREFIX(_dispatch_stub_791), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_791))
+GL_PREFIX(_dispatch_stub_791):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6304(%rax), %r11
+	movq	6328(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29820,13 +29971,13 @@ GL_PREFIX(_dispatch_stub_788):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6304(%rax), %r11
+	movq	6328(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6304(%rax), %r11
+	movq	6328(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29840,19 +29991,19 @@ GL_PREFIX(_dispatch_stub_788):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6304(%rax), %r11
+	movq	6328(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_788), .-GL_PREFIX(_dispatch_stub_788)
+	.size	GL_PREFIX(_dispatch_stub_791), .-GL_PREFIX(_dispatch_stub_791)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_789)
-	.type	GL_PREFIX(_dispatch_stub_789), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_789))
-GL_PREFIX(_dispatch_stub_789):
+	.globl	GL_PREFIX(_dispatch_stub_792)
+	.type	GL_PREFIX(_dispatch_stub_792), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_792))
+GL_PREFIX(_dispatch_stub_792):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6312(%rax), %r11
+	movq	6336(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29866,13 +30017,13 @@ GL_PREFIX(_dispatch_stub_789):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6312(%rax), %r11
+	movq	6336(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6312(%rax), %r11
+	movq	6336(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29886,19 +30037,19 @@ GL_PREFIX(_dispatch_stub_789):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6312(%rax), %r11
+	movq	6336(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_789), .-GL_PREFIX(_dispatch_stub_789)
+	.size	GL_PREFIX(_dispatch_stub_792), .-GL_PREFIX(_dispatch_stub_792)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_790)
-	.type	GL_PREFIX(_dispatch_stub_790), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_790))
-GL_PREFIX(_dispatch_stub_790):
+	.globl	GL_PREFIX(_dispatch_stub_793)
+	.type	GL_PREFIX(_dispatch_stub_793), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_793))
+GL_PREFIX(_dispatch_stub_793):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6320(%rax), %r11
+	movq	6344(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29908,13 +30059,13 @@ GL_PREFIX(_dispatch_stub_790):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6320(%rax), %r11
+	movq	6344(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6320(%rax), %r11
+	movq	6344(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29924,19 +30075,19 @@ GL_PREFIX(_dispatch_stub_790):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6320(%rax), %r11
+	movq	6344(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_790), .-GL_PREFIX(_dispatch_stub_790)
+	.size	GL_PREFIX(_dispatch_stub_793), .-GL_PREFIX(_dispatch_stub_793)
 
 	.p2align	4,,15
-	.globl	GL_PREFIX(_dispatch_stub_791)
-	.type	GL_PREFIX(_dispatch_stub_791), @function
-	HIDDEN(GL_PREFIX(_dispatch_stub_791))
-GL_PREFIX(_dispatch_stub_791):
+	.globl	GL_PREFIX(_dispatch_stub_794)
+	.type	GL_PREFIX(_dispatch_stub_794), @function
+	HIDDEN(GL_PREFIX(_dispatch_stub_794))
+GL_PREFIX(_dispatch_stub_794):
 #if defined(GLX_USE_TLS)
 	call	_x86_64_get_dispatch@PLT
-	movq	6328(%rax), %r11
+	movq	6352(%rax), %r11
 	jmp	*%r11
 #elif defined(PTHREADS)
 	pushq	%rdi
@@ -29946,13 +30097,13 @@ GL_PREFIX(_dispatch_stub_791):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6328(%rax), %r11
+	movq	6352(%rax), %r11
 	jmp	*%r11
 #else
 	movq	_glapi_Dispatch(%rip), %rax
 	testq	%rax, %rax
 	je	1f
-	movq	6328(%rax), %r11
+	movq	6352(%rax), %r11
 	jmp	*%r11
 1:
 	pushq	%rdi
@@ -29962,10 +30113,10 @@ GL_PREFIX(_dispatch_stub_791):
 	popq	%rdx
 	popq	%rsi
 	popq	%rdi
-	movq	6328(%rax), %r11
+	movq	6352(%rax), %r11
 	jmp	*%r11
 #endif /* defined(GLX_USE_TLS) */
-	.size	GL_PREFIX(_dispatch_stub_791), .-GL_PREFIX(_dispatch_stub_791)
+	.size	GL_PREFIX(_dispatch_stub_794), .-GL_PREFIX(_dispatch_stub_794)
 
 	.globl GL_PREFIX(ArrayElementEXT) ; .set GL_PREFIX(ArrayElementEXT), GL_PREFIX(ArrayElement)
 	.globl GL_PREFIX(BindTextureEXT) ; .set GL_PREFIX(BindTextureEXT), GL_PREFIX(BindTexture)
@@ -30220,9 +30371,9 @@ GL_PREFIX(_dispatch_stub_791):
 	.globl GL_PREFIX(IsProgramARB) ; .set GL_PREFIX(IsProgramARB), GL_PREFIX(IsProgramNV)
 	.globl GL_PREFIX(PointParameteri) ; .set GL_PREFIX(PointParameteri), GL_PREFIX(PointParameteriNV)
 	.globl GL_PREFIX(PointParameteriv) ; .set GL_PREFIX(PointParameteriv), GL_PREFIX(PointParameterivNV)
-	.globl GL_PREFIX(DeleteVertexArrays) ; .set GL_PREFIX(DeleteVertexArrays), GL_PREFIX(_dispatch_stub_752)
-	.globl GL_PREFIX(IsVertexArray) ; .set GL_PREFIX(IsVertexArray), GL_PREFIX(_dispatch_stub_754)
-	.globl GL_PREFIX(BlendEquationSeparate) ; .set GL_PREFIX(BlendEquationSeparate), GL_PREFIX(_dispatch_stub_762)
+	.globl GL_PREFIX(DeleteVertexArrays) ; .set GL_PREFIX(DeleteVertexArrays), GL_PREFIX(_dispatch_stub_755)
+	.globl GL_PREFIX(IsVertexArray) ; .set GL_PREFIX(IsVertexArray), GL_PREFIX(_dispatch_stub_757)
+	.globl GL_PREFIX(BlendEquationSeparate) ; .set GL_PREFIX(BlendEquationSeparate), GL_PREFIX(_dispatch_stub_765)
 	.globl GL_PREFIX(BindFramebuffer) ; .set GL_PREFIX(BindFramebuffer), GL_PREFIX(BindFramebufferEXT)
 	.globl GL_PREFIX(BindRenderbuffer) ; .set GL_PREFIX(BindRenderbuffer), GL_PREFIX(BindRenderbufferEXT)
 	.globl GL_PREFIX(CheckFramebufferStatus) ; .set GL_PREFIX(CheckFramebufferStatus), GL_PREFIX(CheckFramebufferStatusEXT)
@@ -30240,8 +30391,9 @@ GL_PREFIX(_dispatch_stub_791):
 	.globl GL_PREFIX(IsFramebuffer) ; .set GL_PREFIX(IsFramebuffer), GL_PREFIX(IsFramebufferEXT)
 	.globl GL_PREFIX(IsRenderbuffer) ; .set GL_PREFIX(IsRenderbuffer), GL_PREFIX(IsRenderbufferEXT)
 	.globl GL_PREFIX(RenderbufferStorage) ; .set GL_PREFIX(RenderbufferStorage), GL_PREFIX(RenderbufferStorageEXT)
-	.globl GL_PREFIX(BlitFramebuffer) ; .set GL_PREFIX(BlitFramebuffer), GL_PREFIX(_dispatch_stub_780)
+	.globl GL_PREFIX(BlitFramebuffer) ; .set GL_PREFIX(BlitFramebuffer), GL_PREFIX(_dispatch_stub_783)
 	.globl GL_PREFIX(FramebufferTextureLayer) ; .set GL_PREFIX(FramebufferTextureLayer), GL_PREFIX(FramebufferTextureLayerEXT)
+	.globl GL_PREFIX(ProvokingVertex) ; .set GL_PREFIX(ProvokingVertex), GL_PREFIX(ProvokingVertexEXT)
 
 #if defined(GLX_USE_TLS) && defined(__linux__)
 	.section ".note.ABI-tag", "a"
diff --git a/src/mesa/x86/glapi_x86.S b/src/mesa/x86/glapi_x86.S
index 46794adaf7..12c77f434e 100644
--- a/src/mesa/x86/glapi_x86.S
+++ b/src/mesa/x86/glapi_x86.S
@@ -725,23 +725,26 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB(GetSynciv, _gloffset_GetSynciv, GetSynciv@20)
 	GL_STUB(IsSync, _gloffset_IsSync, IsSync@4)
 	GL_STUB(WaitSync, _gloffset_WaitSync, WaitSync@12)
+	GL_STUB(DrawElementsBaseVertex, _gloffset_DrawElementsBaseVertex, DrawElementsBaseVertex@20)
+	GL_STUB(DrawRangeElementsBaseVertex, _gloffset_DrawRangeElementsBaseVertex, DrawRangeElementsBaseVertex@28)
+	GL_STUB(MultiDrawElementsBaseVertex, _gloffset_MultiDrawElementsBaseVertex, MultiDrawElementsBaseVertex@24)
 	GL_STUB(PolygonOffsetEXT, _gloffset_PolygonOffsetEXT, PolygonOffsetEXT@8)
-	GL_STUB(_dispatch_stub_575, _gloffset_GetPixelTexGenParameterfvSGIS, _dispatch_stub_575@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_575, _dispatch_stub_575@8))
-	GL_STUB(_dispatch_stub_576, _gloffset_GetPixelTexGenParameterivSGIS, _dispatch_stub_576@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_576, _dispatch_stub_576@8))
-	GL_STUB(_dispatch_stub_577, _gloffset_PixelTexGenParameterfSGIS, _dispatch_stub_577@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_577, _dispatch_stub_577@8))
-	GL_STUB(_dispatch_stub_578, _gloffset_PixelTexGenParameterfvSGIS, _dispatch_stub_578@8)
+	GL_STUB(_dispatch_stub_578, _gloffset_GetPixelTexGenParameterfvSGIS, _dispatch_stub_578@8)
 	HIDDEN(GL_PREFIX(_dispatch_stub_578, _dispatch_stub_578@8))
-	GL_STUB(_dispatch_stub_579, _gloffset_PixelTexGenParameteriSGIS, _dispatch_stub_579@8)
+	GL_STUB(_dispatch_stub_579, _gloffset_GetPixelTexGenParameterivSGIS, _dispatch_stub_579@8)
 	HIDDEN(GL_PREFIX(_dispatch_stub_579, _dispatch_stub_579@8))
-	GL_STUB(_dispatch_stub_580, _gloffset_PixelTexGenParameterivSGIS, _dispatch_stub_580@8)
+	GL_STUB(_dispatch_stub_580, _gloffset_PixelTexGenParameterfSGIS, _dispatch_stub_580@8)
 	HIDDEN(GL_PREFIX(_dispatch_stub_580, _dispatch_stub_580@8))
-	GL_STUB(_dispatch_stub_581, _gloffset_SampleMaskSGIS, _dispatch_stub_581@8)
+	GL_STUB(_dispatch_stub_581, _gloffset_PixelTexGenParameterfvSGIS, _dispatch_stub_581@8)
 	HIDDEN(GL_PREFIX(_dispatch_stub_581, _dispatch_stub_581@8))
-	GL_STUB(_dispatch_stub_582, _gloffset_SamplePatternSGIS, _dispatch_stub_582@4)
-	HIDDEN(GL_PREFIX(_dispatch_stub_582, _dispatch_stub_582@4))
+	GL_STUB(_dispatch_stub_582, _gloffset_PixelTexGenParameteriSGIS, _dispatch_stub_582@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_582, _dispatch_stub_582@8))
+	GL_STUB(_dispatch_stub_583, _gloffset_PixelTexGenParameterivSGIS, _dispatch_stub_583@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_583, _dispatch_stub_583@8))
+	GL_STUB(_dispatch_stub_584, _gloffset_SampleMaskSGIS, _dispatch_stub_584@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_584, _dispatch_stub_584@8))
+	GL_STUB(_dispatch_stub_585, _gloffset_SamplePatternSGIS, _dispatch_stub_585@4)
+	HIDDEN(GL_PREFIX(_dispatch_stub_585, _dispatch_stub_585@4))
 	GL_STUB(ColorPointerEXT, _gloffset_ColorPointerEXT, ColorPointerEXT@20)
 	GL_STUB(EdgeFlagPointerEXT, _gloffset_EdgeFlagPointerEXT, EdgeFlagPointerEXT@12)
 	GL_STUB(IndexPointerEXT, _gloffset_IndexPointerEXT, IndexPointerEXT@16)
@@ -752,10 +755,10 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB(PointParameterfvEXT, _gloffset_PointParameterfvEXT, PointParameterfvEXT@8)
 	GL_STUB(LockArraysEXT, _gloffset_LockArraysEXT, LockArraysEXT@8)
 	GL_STUB(UnlockArraysEXT, _gloffset_UnlockArraysEXT, UnlockArraysEXT@0)
-	GL_STUB(_dispatch_stub_593, _gloffset_CullParameterdvEXT, _dispatch_stub_593@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_593, _dispatch_stub_593@8))
-	GL_STUB(_dispatch_stub_594, _gloffset_CullParameterfvEXT, _dispatch_stub_594@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_594, _dispatch_stub_594@8))
+	GL_STUB(_dispatch_stub_596, _gloffset_CullParameterdvEXT, _dispatch_stub_596@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_596, _dispatch_stub_596@8))
+	GL_STUB(_dispatch_stub_597, _gloffset_CullParameterfvEXT, _dispatch_stub_597@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_597, _dispatch_stub_597@8))
 	GL_STUB(SecondaryColor3bEXT, _gloffset_SecondaryColor3bEXT, SecondaryColor3bEXT@12)
 	GL_STUB(SecondaryColor3bvEXT, _gloffset_SecondaryColor3bvEXT, SecondaryColor3bvEXT@4)
 	GL_STUB(SecondaryColor3dEXT, _gloffset_SecondaryColor3dEXT, SecondaryColor3dEXT@24)
@@ -780,8 +783,8 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB(FogCoorddvEXT, _gloffset_FogCoorddvEXT, FogCoorddvEXT@4)
 	GL_STUB(FogCoordfEXT, _gloffset_FogCoordfEXT, FogCoordfEXT@4)
 	GL_STUB(FogCoordfvEXT, _gloffset_FogCoordfvEXT, FogCoordfvEXT@4)
-	GL_STUB(_dispatch_stub_619, _gloffset_PixelTexGenSGIX, _dispatch_stub_619@4)
-	HIDDEN(GL_PREFIX(_dispatch_stub_619, _dispatch_stub_619@4))
+	GL_STUB(_dispatch_stub_622, _gloffset_PixelTexGenSGIX, _dispatch_stub_622@4)
+	HIDDEN(GL_PREFIX(_dispatch_stub_622, _dispatch_stub_622@4))
 	GL_STUB(BlendFuncSeparateEXT, _gloffset_BlendFuncSeparateEXT, BlendFuncSeparateEXT@16)
 	GL_STUB(FlushVertexArrayRangeNV, _gloffset_FlushVertexArrayRangeNV, FlushVertexArrayRangeNV@0)
 	GL_STUB(VertexArrayRangeNV, _gloffset_VertexArrayRangeNV, VertexArrayRangeNV@8)
@@ -823,24 +826,24 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB(WindowPos4ivMESA, _gloffset_WindowPos4ivMESA, WindowPos4ivMESA@4)
 	GL_STUB(WindowPos4sMESA, _gloffset_WindowPos4sMESA, WindowPos4sMESA@16)
 	GL_STUB(WindowPos4svMESA, _gloffset_WindowPos4svMESA, WindowPos4svMESA@4)
-	GL_STUB(_dispatch_stub_661, _gloffset_MultiModeDrawArraysIBM, _dispatch_stub_661@20)
-	HIDDEN(GL_PREFIX(_dispatch_stub_661, _dispatch_stub_661@20))
-	GL_STUB(_dispatch_stub_662, _gloffset_MultiModeDrawElementsIBM, _dispatch_stub_662@24)
-	HIDDEN(GL_PREFIX(_dispatch_stub_662, _dispatch_stub_662@24))
-	GL_STUB(_dispatch_stub_663, _gloffset_DeleteFencesNV, _dispatch_stub_663@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_663, _dispatch_stub_663@8))
-	GL_STUB(_dispatch_stub_664, _gloffset_FinishFenceNV, _dispatch_stub_664@4)
-	HIDDEN(GL_PREFIX(_dispatch_stub_664, _dispatch_stub_664@4))
-	GL_STUB(_dispatch_stub_665, _gloffset_GenFencesNV, _dispatch_stub_665@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_665, _dispatch_stub_665@8))
-	GL_STUB(_dispatch_stub_666, _gloffset_GetFenceivNV, _dispatch_stub_666@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_666, _dispatch_stub_666@12))
-	GL_STUB(_dispatch_stub_667, _gloffset_IsFenceNV, _dispatch_stub_667@4)
+	GL_STUB(_dispatch_stub_664, _gloffset_MultiModeDrawArraysIBM, _dispatch_stub_664@20)
+	HIDDEN(GL_PREFIX(_dispatch_stub_664, _dispatch_stub_664@20))
+	GL_STUB(_dispatch_stub_665, _gloffset_MultiModeDrawElementsIBM, _dispatch_stub_665@24)
+	HIDDEN(GL_PREFIX(_dispatch_stub_665, _dispatch_stub_665@24))
+	GL_STUB(_dispatch_stub_666, _gloffset_DeleteFencesNV, _dispatch_stub_666@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_666, _dispatch_stub_666@8))
+	GL_STUB(_dispatch_stub_667, _gloffset_FinishFenceNV, _dispatch_stub_667@4)
 	HIDDEN(GL_PREFIX(_dispatch_stub_667, _dispatch_stub_667@4))
-	GL_STUB(_dispatch_stub_668, _gloffset_SetFenceNV, _dispatch_stub_668@8)
+	GL_STUB(_dispatch_stub_668, _gloffset_GenFencesNV, _dispatch_stub_668@8)
 	HIDDEN(GL_PREFIX(_dispatch_stub_668, _dispatch_stub_668@8))
-	GL_STUB(_dispatch_stub_669, _gloffset_TestFenceNV, _dispatch_stub_669@4)
-	HIDDEN(GL_PREFIX(_dispatch_stub_669, _dispatch_stub_669@4))
+	GL_STUB(_dispatch_stub_669, _gloffset_GetFenceivNV, _dispatch_stub_669@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_669, _dispatch_stub_669@12))
+	GL_STUB(_dispatch_stub_670, _gloffset_IsFenceNV, _dispatch_stub_670@4)
+	HIDDEN(GL_PREFIX(_dispatch_stub_670, _dispatch_stub_670@4))
+	GL_STUB(_dispatch_stub_671, _gloffset_SetFenceNV, _dispatch_stub_671@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_671, _dispatch_stub_671@8))
+	GL_STUB(_dispatch_stub_672, _gloffset_TestFenceNV, _dispatch_stub_672@4)
+	HIDDEN(GL_PREFIX(_dispatch_stub_672, _dispatch_stub_672@4))
 	GL_STUB(AreProgramsResidentNV, _gloffset_AreProgramsResidentNV, AreProgramsResidentNV@12)
 	GL_STUB(BindProgramNV, _gloffset_BindProgramNV, BindProgramNV@8)
 	GL_STUB(DeleteProgramsNV, _gloffset_DeleteProgramsNV, DeleteProgramsNV@8)
@@ -921,26 +924,26 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB(SetFragmentShaderConstantATI, _gloffset_SetFragmentShaderConstantATI, SetFragmentShaderConstantATI@8)
 	GL_STUB(PointParameteriNV, _gloffset_PointParameteriNV, PointParameteriNV@8)
 	GL_STUB(PointParameterivNV, _gloffset_PointParameterivNV, PointParameterivNV@8)
-	GL_STUB(_dispatch_stub_750, _gloffset_ActiveStencilFaceEXT, _dispatch_stub_750@4)
-	HIDDEN(GL_PREFIX(_dispatch_stub_750, _dispatch_stub_750@4))
-	GL_STUB(_dispatch_stub_751, _gloffset_BindVertexArrayAPPLE, _dispatch_stub_751@4)
-	HIDDEN(GL_PREFIX(_dispatch_stub_751, _dispatch_stub_751@4))
-	GL_STUB(_dispatch_stub_752, _gloffset_DeleteVertexArraysAPPLE, _dispatch_stub_752@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_752, _dispatch_stub_752@8))
-	GL_STUB(_dispatch_stub_753, _gloffset_GenVertexArraysAPPLE, _dispatch_stub_753@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_753, _dispatch_stub_753@8))
-	GL_STUB(_dispatch_stub_754, _gloffset_IsVertexArrayAPPLE, _dispatch_stub_754@4)
+	GL_STUB(_dispatch_stub_753, _gloffset_ActiveStencilFaceEXT, _dispatch_stub_753@4)
+	HIDDEN(GL_PREFIX(_dispatch_stub_753, _dispatch_stub_753@4))
+	GL_STUB(_dispatch_stub_754, _gloffset_BindVertexArrayAPPLE, _dispatch_stub_754@4)
 	HIDDEN(GL_PREFIX(_dispatch_stub_754, _dispatch_stub_754@4))
+	GL_STUB(_dispatch_stub_755, _gloffset_DeleteVertexArraysAPPLE, _dispatch_stub_755@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_755, _dispatch_stub_755@8))
+	GL_STUB(_dispatch_stub_756, _gloffset_GenVertexArraysAPPLE, _dispatch_stub_756@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_756, _dispatch_stub_756@8))
+	GL_STUB(_dispatch_stub_757, _gloffset_IsVertexArrayAPPLE, _dispatch_stub_757@4)
+	HIDDEN(GL_PREFIX(_dispatch_stub_757, _dispatch_stub_757@4))
 	GL_STUB(GetProgramNamedParameterdvNV, _gloffset_GetProgramNamedParameterdvNV, GetProgramNamedParameterdvNV@16)
 	GL_STUB(GetProgramNamedParameterfvNV, _gloffset_GetProgramNamedParameterfvNV, GetProgramNamedParameterfvNV@16)
 	GL_STUB(ProgramNamedParameter4dNV, _gloffset_ProgramNamedParameter4dNV, ProgramNamedParameter4dNV@44)
 	GL_STUB(ProgramNamedParameter4dvNV, _gloffset_ProgramNamedParameter4dvNV, ProgramNamedParameter4dvNV@16)
 	GL_STUB(ProgramNamedParameter4fNV, _gloffset_ProgramNamedParameter4fNV, ProgramNamedParameter4fNV@28)
 	GL_STUB(ProgramNamedParameter4fvNV, _gloffset_ProgramNamedParameter4fvNV, ProgramNamedParameter4fvNV@16)
-	GL_STUB(_dispatch_stub_761, _gloffset_DepthBoundsEXT, _dispatch_stub_761@16)
-	HIDDEN(GL_PREFIX(_dispatch_stub_761, _dispatch_stub_761@16))
-	GL_STUB(_dispatch_stub_762, _gloffset_BlendEquationSeparateEXT, _dispatch_stub_762@8)
-	HIDDEN(GL_PREFIX(_dispatch_stub_762, _dispatch_stub_762@8))
+	GL_STUB(_dispatch_stub_764, _gloffset_DepthBoundsEXT, _dispatch_stub_764@16)
+	HIDDEN(GL_PREFIX(_dispatch_stub_764, _dispatch_stub_764@16))
+	GL_STUB(_dispatch_stub_765, _gloffset_BlendEquationSeparateEXT, _dispatch_stub_765@8)
+	HIDDEN(GL_PREFIX(_dispatch_stub_765, _dispatch_stub_765@8))
 	GL_STUB(BindFramebufferEXT, _gloffset_BindFramebufferEXT, BindFramebufferEXT@8)
 	GL_STUB(BindRenderbufferEXT, _gloffset_BindRenderbufferEXT, BindRenderbufferEXT@8)
 	GL_STUB(CheckFramebufferStatusEXT, _gloffset_CheckFramebufferStatusEXT, CheckFramebufferStatusEXT@4)
@@ -958,28 +961,28 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB(IsFramebufferEXT, _gloffset_IsFramebufferEXT, IsFramebufferEXT@4)
 	GL_STUB(IsRenderbufferEXT, _gloffset_IsRenderbufferEXT, IsRenderbufferEXT@4)
 	GL_STUB(RenderbufferStorageEXT, _gloffset_RenderbufferStorageEXT, RenderbufferStorageEXT@16)
-	GL_STUB(_dispatch_stub_780, _gloffset_BlitFramebufferEXT, _dispatch_stub_780@40)
-	HIDDEN(GL_PREFIX(_dispatch_stub_780, _dispatch_stub_780@40))
-	GL_STUB(_dispatch_stub_781, _gloffset_BufferParameteriAPPLE, _dispatch_stub_781@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_781, _dispatch_stub_781@12))
-	GL_STUB(_dispatch_stub_782, _gloffset_FlushMappedBufferRangeAPPLE, _dispatch_stub_782@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_782, _dispatch_stub_782@12))
+	GL_STUB(_dispatch_stub_783, _gloffset_BlitFramebufferEXT, _dispatch_stub_783@40)
+	HIDDEN(GL_PREFIX(_dispatch_stub_783, _dispatch_stub_783@40))
+	GL_STUB(_dispatch_stub_784, _gloffset_BufferParameteriAPPLE, _dispatch_stub_784@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_784, _dispatch_stub_784@12))
+	GL_STUB(_dispatch_stub_785, _gloffset_FlushMappedBufferRangeAPPLE, _dispatch_stub_785@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_785, _dispatch_stub_785@12))
 	GL_STUB(FramebufferTextureLayerEXT, _gloffset_FramebufferTextureLayerEXT, FramebufferTextureLayerEXT@20)
 	GL_STUB(ProvokingVertexEXT, _gloffset_ProvokingVertexEXT, ProvokingVertexEXT@4)
-	GL_STUB(_dispatch_stub_785, _gloffset_GetTexParameterPointervAPPLE, _dispatch_stub_785@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_785, _dispatch_stub_785@12))
-	GL_STUB(_dispatch_stub_786, _gloffset_TextureRangeAPPLE, _dispatch_stub_786@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_786, _dispatch_stub_786@12))
-	GL_STUB(_dispatch_stub_787, _gloffset_StencilFuncSeparateATI, _dispatch_stub_787@16)
-	HIDDEN(GL_PREFIX(_dispatch_stub_787, _dispatch_stub_787@16))
-	GL_STUB(_dispatch_stub_788, _gloffset_ProgramEnvParameters4fvEXT, _dispatch_stub_788@16)
-	HIDDEN(GL_PREFIX(_dispatch_stub_788, _dispatch_stub_788@16))
-	GL_STUB(_dispatch_stub_789, _gloffset_ProgramLocalParameters4fvEXT, _dispatch_stub_789@16)
-	HIDDEN(GL_PREFIX(_dispatch_stub_789, _dispatch_stub_789@16))
-	GL_STUB(_dispatch_stub_790, _gloffset_GetQueryObjecti64vEXT, _dispatch_stub_790@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_790, _dispatch_stub_790@12))
-	GL_STUB(_dispatch_stub_791, _gloffset_GetQueryObjectui64vEXT, _dispatch_stub_791@12)
-	HIDDEN(GL_PREFIX(_dispatch_stub_791, _dispatch_stub_791@12))
+	GL_STUB(_dispatch_stub_788, _gloffset_GetTexParameterPointervAPPLE, _dispatch_stub_788@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_788, _dispatch_stub_788@12))
+	GL_STUB(_dispatch_stub_789, _gloffset_TextureRangeAPPLE, _dispatch_stub_789@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_789, _dispatch_stub_789@12))
+	GL_STUB(_dispatch_stub_790, _gloffset_StencilFuncSeparateATI, _dispatch_stub_790@16)
+	HIDDEN(GL_PREFIX(_dispatch_stub_790, _dispatch_stub_790@16))
+	GL_STUB(_dispatch_stub_791, _gloffset_ProgramEnvParameters4fvEXT, _dispatch_stub_791@16)
+	HIDDEN(GL_PREFIX(_dispatch_stub_791, _dispatch_stub_791@16))
+	GL_STUB(_dispatch_stub_792, _gloffset_ProgramLocalParameters4fvEXT, _dispatch_stub_792@16)
+	HIDDEN(GL_PREFIX(_dispatch_stub_792, _dispatch_stub_792@16))
+	GL_STUB(_dispatch_stub_793, _gloffset_GetQueryObjecti64vEXT, _dispatch_stub_793@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_793, _dispatch_stub_793@12))
+	GL_STUB(_dispatch_stub_794, _gloffset_GetQueryObjectui64vEXT, _dispatch_stub_794@12)
+	HIDDEN(GL_PREFIX(_dispatch_stub_794, _dispatch_stub_794@12))
 	GL_STUB_ALIAS(ArrayElementEXT, _gloffset_ArrayElement, ArrayElementEXT@4, ArrayElement, ArrayElement@4)
 	GL_STUB_ALIAS(BindTextureEXT, _gloffset_BindTexture, BindTextureEXT@8, BindTexture, BindTexture@8)
 	GL_STUB_ALIAS(DrawArraysEXT, _gloffset_DrawArrays, DrawArraysEXT@12, DrawArrays, DrawArrays@12)
@@ -1313,6 +1316,7 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB_ALIAS(IsRenderbuffer, _gloffset_IsRenderbufferEXT, IsRenderbuffer@4, IsRenderbufferEXT, IsRenderbufferEXT@4)
 	GL_STUB_ALIAS(RenderbufferStorage, _gloffset_RenderbufferStorageEXT, RenderbufferStorage@16, RenderbufferStorageEXT, RenderbufferStorageEXT@16)
 	GL_STUB_ALIAS(FramebufferTextureLayer, _gloffset_FramebufferTextureLayerEXT, FramebufferTextureLayer@20, FramebufferTextureLayerEXT, FramebufferTextureLayerEXT@20)
+	GL_STUB_ALIAS(ProvokingVertex, _gloffset_ProvokingVertexEXT, ProvokingVertex@4, ProvokingVertexEXT, ProvokingVertexEXT@4)
 
 		GLOBL	GLNAME(gl_dispatch_functions_end)
 		HIDDEN(GLNAME(gl_dispatch_functions_end))
diff --git a/src/mesa/x86/x86_xform.c b/src/mesa/x86/x86_xform.c
index 16b2b26bcc..52f6b25d81 100644
--- a/src/mesa/x86/x86_xform.c
+++ b/src/mesa/x86/x86_xform.c
@@ -60,21 +60,24 @@ _mesa_x86_cliptest_points4( GLvector4f *clip_vec,
 			    GLvector4f *proj_vec,
 			    GLubyte clipMask[],
 			    GLubyte *orMask,
-			    GLubyte *andMask );
+			    GLubyte *andMask,
+			    GLboolean viewport_z_clip );
 
 extern GLvector4f * _ASMAPI
 _mesa_x86_cliptest_points4_np( GLvector4f *clip_vec,
 			       GLvector4f *proj_vec,
 			       GLubyte clipMask[],
 			       GLubyte *orMask,
-			       GLubyte *andMask );
+			       GLubyte *andMask,
+			       GLboolean viewport_z_clip );
 
 extern void _ASMAPI
 _mesa_v16_x86_cliptest_points4( GLfloat *first_vert,
 				GLfloat *last_vert,
 				GLubyte *or_mask,
 				GLubyte *and_mask,
-				GLubyte *clip_mask );
+				GLubyte *clip_mask,
+				GLboolean viewport_z_clip );
 
 extern void _ASMAPI
 _mesa_v16_x86_general_xform( GLfloat *dest,
diff --git a/src/xvmc/Makefile b/src/xvmc/Makefile
deleted file mode 100644
index 7badcfd264..0000000000
--- a/src/xvmc/Makefile
+++ /dev/null
@@ -1,73 +0,0 @@
-TARGET		= libXvMCg3dvl.so
-SONAME		= libXvMCg3dvl.so.1
-GALLIUMDIR	= ../gallium
-
-OBJECTS         = block.o surface.o context.o subpicture.o attributes.o
-
-ifeq (${DRIVER}, softpipe)
-OBJECTS         += ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o
-endif
-
-CFLAGS	+= -g -fPIC -Wall -Werror=implicit-function-declaration	\
-	   -I${GALLIUMDIR}/state_trackers/g3dvl		\
-	   -I${GALLIUMDIR}/winsys/g3dvl			\
-	   -I${GALLIUMDIR}/include			\
-	   -I${GALLIUMDIR}/auxiliary			\
-	   -I${GALLIUMDIR}/drivers
-
-ifeq (${DRIVER}, softpipe)
-LDFLAGS	+= -L${GALLIUMDIR}/state_trackers/g3dvl		\
-	   -L${GALLIUMDIR}/drivers/softpipe		\
-	   -L${GALLIUMDIR}/auxiliary/tgsi		\
-	   -L${GALLIUMDIR}/auxiliary/draw		\
-	   -L${GALLIUMDIR}/auxiliary/translate		\
-	   -L${GALLIUMDIR}/auxiliary/cso_cache		\
-	   -L${GALLIUMDIR}/auxiliary/util		\
-	   -L${GALLIUMDIR}/auxiliary/rtasm
-else
-LDFLAGS	+= -L${GALLIUMDIR}/state_trackers/g3dvl		\
-   	   -L${GALLIUMDIR}/winsys/g3dvl/nouveau		\
-   	   -L${GALLIUMDIR}/auxiliary/util
-endif
-
-ifeq (${DRIVER}, softpipe)
-LIBS	+= -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lm
-else
-LIBS	+= -lg3dvl -lnouveau_dri -lutil
-endif
-
-#############################################
-
-ifeq (${DRIVER}, softpipe)
-.PHONY	= all clean g3dvl
-else
-.PHONY	= all clean g3dvl nouveau_winsys
-endif
-
-all: ${TARGET}
-
-ifeq (${DRIVER}, softpipe)
-${TARGET}: ${OBJECTS} g3dvl
-	$(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS}
-
-g3dvl:
-	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE}
-
-clean:
-	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean
-	rm -rf ${OBJECTS} ${TARGET}
-else
-${TARGET}: ${OBJECTS} g3dvl nouveau_winsys
-	$(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS}
-
-g3dvl:
-	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE}
-
-nouveau_winsys:
-	cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE}
-
-clean:
-	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean
-	cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} clean
-	rm -rf ${OBJECTS} ${TARGET}
-endif
diff --git a/src/xvmc/attributes.c b/src/xvmc/attributes.c
deleted file mode 100644
index 674524b8b8..0000000000
--- a/src/xvmc/attributes.c
+++ /dev/null
@@ -1,20 +0,0 @@
-#include <assert.h>
-#include <X11/Xlib.h>
-#include <X11/extensions/Xvlib.h>
-#include <X11/extensions/XvMC.h>
-
-XvAttribute* XvMCQueryAttributes(Display *display, XvMCContext *context, int *number)
-{
-	return NULL;
-}
-
-Status XvMCSetAttribute(Display *display, XvMCContext *context, Atom attribute, int value)
-{
-	return BadImplementation;
-}
-
-Status XvMCGetAttribute(Display *display, XvMCContext *context, Atom attribute, int *value)
-{
-	return BadImplementation;
-}
-
diff --git a/src/xvmc/block.c b/src/xvmc/block.c
deleted file mode 100644
index b38a89be09..0000000000
--- a/src/xvmc/block.c
+++ /dev/null
@@ -1,79 +0,0 @@
-#include <assert.h>
-#include <X11/Xlib.h>
-#include <X11/extensions/XvMC.h>
-#include <util/u_memory.h>
-#include <vl_display.h>
-#include <vl_screen.h>
-#include <vl_context.h>
-
-#define BLOCK_SIZE (64 * 2)
-
-Status XvMCCreateBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks)
-{
-	struct vlContext *vl_ctx;
-
-	assert(display);
-
-	if (!context)
-		return XvMCBadContext;
-	if (num_blocks == 0)
-		return BadValue;
-
-	assert(blocks);
-
-	vl_ctx = context->privData;
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
-
-	blocks->context_id = context->context_id;
-	blocks->num_blocks = num_blocks;
-	blocks->blocks = MALLOC(BLOCK_SIZE * num_blocks);
-	/* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */
-	blocks->privData = display;
-
-	return Success;
-}
-
-Status XvMCDestroyBlocks(Display *display, XvMCBlockArray *blocks)
-{
-	assert(display);
-	assert(blocks);
-	assert(display == blocks->privData);
-	FREE(blocks->blocks);
-
-	return Success;
-}
-
-Status XvMCCreateMacroBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks)
-{
-	struct vlContext *vl_ctx;
-
-	assert(display);
-
-	if (!context)
-		return XvMCBadContext;
-	if (num_blocks == 0)
-		return BadValue;
-
-	assert(blocks);
-
-	vl_ctx = context->privData;
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
-
-	blocks->context_id = context->context_id;
-	blocks->num_blocks = num_blocks;
-	blocks->macro_blocks = MALLOC(sizeof(XvMCMacroBlock) * num_blocks);
-	/* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */
-	blocks->privData = display;
-
-	return Success;
-}
-
-Status XvMCDestroyMacroBlocks(Display *display, XvMCMacroBlockArray *blocks)
-{
-	assert(display);
-	assert(blocks);
-	assert(display == blocks->privData);
-	FREE(blocks->macro_blocks);
-
-	return Success;
-}
diff --git a/src/xvmc/context.c b/src/xvmc/context.c
deleted file mode 100644
index 273f658029..0000000000
--- a/src/xvmc/context.c
+++ /dev/null
@@ -1,207 +0,0 @@
-#include <assert.h>
-#include <X11/Xlib.h>
-#include <X11/extensions/XvMClib.h>
-#include <pipe/p_context.h>
-#include <vl_display.h>
-#include <vl_screen.h>
-#include <vl_context.h>
-#include <vl_winsys.h>
-
-static Status Validate
-(
-	Display *display,
-	XvPortID port,
-	int surface_type_id,
-	unsigned int width,
-	unsigned int height,
-	int flags,
-	int *found_port,
-	int *chroma_format,
-	int *mc_type
-)
-{
-	unsigned int	found_surface = 0;
-	XvAdaptorInfo	*adaptor_info;
-	unsigned int	num_adaptors;
-	int		num_types;
-	unsigned int	max_width, max_height;
-	Status		ret;
-	unsigned int	i, j, k;
-
-	assert(display && chroma_format);
-
-	*found_port = 0;
-
-	ret = XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info);
-	if (ret != Success)
-		return ret;
-
-	/* Scan through all adaptors looking for this port and surface */
-	for (i = 0; i < num_adaptors && !*found_port; ++i)
-	{
-		/* Scan through all ports of this adaptor looking for our port */
-		for (j = 0; j < adaptor_info[i].num_ports && !*found_port; ++j)
-		{
-			/* If this is our port, scan through all its surfaces looking for our surface */
-			if (adaptor_info[i].base_id + j == port)
-			{
-				XvMCSurfaceInfo *surface_info;
-
-				*found_port = 1;
-				surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types);
-
-				if (surface_info)
-				{
-					for (k = 0; k < num_types && !found_surface; ++k)
-					{
-						if (surface_info[k].surface_type_id == surface_type_id)
-						{
-							found_surface = 1;
-							max_width = surface_info[k].max_width;
-							max_height = surface_info[k].max_height;
-							*chroma_format = surface_info[k].chroma_format;
-							*mc_type = surface_info[k].mc_type;
-						}
-					}
-
-					XFree(surface_info);
-				}
-				else
-				{
-					XvFreeAdaptorInfo(adaptor_info);
-					return BadAlloc;
-				}
-			}
-		}
-	}
-
-	XvFreeAdaptorInfo(adaptor_info);
-
-	if (!*found_port)
-		return XvBadPort;
-	if (!found_surface)
-		return BadMatch;
-	if (width > max_width || height > max_height)
-		return BadValue;
-	if (flags != XVMC_DIRECT && flags != 0)
-		return BadValue;
-
-	return Success;
-}
-
-static enum vlProfile ProfileToVL(int xvmc_profile)
-{
-	if (xvmc_profile & XVMC_MPEG_1)
-		assert(0);
-	else if (xvmc_profile & XVMC_MPEG_2)
-		return vlProfileMpeg2Main;
-	else if (xvmc_profile & XVMC_H263)
-		assert(0);
-	else if (xvmc_profile & XVMC_MPEG_4)
-		assert(0);
-	else
-		assert(0);
-
-	return -1;
-}
-
-static enum vlEntryPoint EntryToVL(int xvmc_entry)
-{
-	return xvmc_entry & XVMC_IDCT ? vlEntryPointIDCT : vlEntryPointMC;
-}
-
-static enum vlFormat FormatToVL(int xvmc_format)
-{
-	switch (xvmc_format)
-	{
-		case XVMC_CHROMA_FORMAT_420:
-			return vlFormatYCbCr420;
-		case XVMC_CHROMA_FORMAT_422:
-			return vlFormatYCbCr422;
-		case XVMC_CHROMA_FORMAT_444:
-			return vlFormatYCbCr444;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
-Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, int width, int height, int flags, XvMCContext *context)
-{
-	int			found_port;
-	int			chroma_format;
-	int			mc_type;
-	Status			ret;
-	struct vlDisplay	*vl_dpy;
-	struct vlScreen		*vl_scrn;
-	struct vlContext	*vl_ctx;
-	struct pipe_context	*pipe;
-
-	assert(display);
-
-	if (!context)
-		return XvMCBadContext;
-
-	ret = Validate(display, port, surface_type_id, width, height, flags, &found_port, &chroma_format, &mc_type);
-
-	/* XXX: Success and XvBadPort have the same value */
-	if (ret != Success || !found_port)
-		return ret;
-
-	/* XXX: Assumes default screen, should check which screen port is on */
-	pipe = create_pipe_context(display, XDefaultScreen(display));
-
-	assert(pipe);
-
-	vlCreateDisplay(display, &vl_dpy);
-	vlCreateScreen(vl_dpy, XDefaultScreen(display), pipe->screen, &vl_scrn);
-	vlCreateContext
-	(
-		vl_scrn,
-		pipe,
-		width,
-		height,
-		FormatToVL(chroma_format),
-		ProfileToVL(mc_type),
-		EntryToVL(mc_type),
-		&vl_ctx
-	);
-
-	context->context_id = XAllocID(display);
-	context->surface_type_id = surface_type_id;
-	context->width = width;
-	context->height = height;
-	context->flags = flags;
-	context->port = port;
-	context->privData = vl_ctx;
-
-	return Success;
-}
-
-Status XvMCDestroyContext(Display *display, XvMCContext *context)
-{
-	struct vlContext	*vl_ctx;
-	struct vlScreen		*vl_screen;
-	struct vlDisplay	*vl_dpy;
-	struct pipe_context	*pipe;
-
-	assert(display);
-
-	if (!context)
-		return XvMCBadContext;
-
-	vl_ctx = context->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
-
-	pipe = vlGetPipeContext(vl_ctx);
-	vl_screen = vlContextGetScreen(vl_ctx);
-	vl_dpy = vlGetDisplay(vl_screen);
-	vlDestroyContext(vl_ctx);
-	vlDestroyScreen(vl_screen);
-	vlDestroyDisplay(vl_dpy);
-	destroy_pipe_context(pipe);
-
-	return Success;
-}
diff --git a/src/xvmc/subpicture.c b/src/xvmc/subpicture.c
deleted file mode 100644
index c8f70c90d0..0000000000
--- a/src/xvmc/subpicture.c
+++ /dev/null
@@ -1,215 +0,0 @@
-#include <assert.h>
-#include <X11/Xlib.h>
-#include <X11/extensions/Xvlib.h>
-#include <X11/extensions/XvMC.h>
-
-Status XvMCCreateSubpicture
-(
-	Display *display,
-	XvMCContext *context,
-	XvMCSubpicture *subpicture,
-	unsigned short width,
-	unsigned short height,
-	int xvimage_id
-)
-{
-	assert(display);
-	
-	if (!context)
-		return XvMCBadContext;
-	
-	assert(subpicture);
-	
-	if (width > 2048 || height > 2048)
-		return BadValue;
-	
-	if (xvimage_id != 123)
-		return BadMatch;
-	
-	subpicture->subpicture_id = XAllocID(display);
-	subpicture->context_id = context->context_id;
-	subpicture->xvimage_id = xvimage_id;
-	subpicture->width = width;
-	subpicture->height = height;
-	subpicture->num_palette_entries = 0;
-	subpicture->entry_bytes = 0;
-	subpicture->component_order[0] = 0;
-	subpicture->component_order[1] = 0;
-	subpicture->component_order[2] = 0;
-	subpicture->component_order[3] = 0;
-	/* TODO: subpicture->privData = ;*/
-	
-	return Success;
-}
-
-Status XvMCClearSubpicture
-(
-	Display *display,
-	XvMCSubpicture *subpicture,
-	short x,
-	short y,
-	unsigned short width,
-	unsigned short height,
-	unsigned int color
-)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	/* TODO: Assert clear rect is within bounds? Or clip? */
-	
-	return Success;
-}
-
-Status XvMCCompositeSubpicture
-(
-	Display *display,
-	XvMCSubpicture *subpicture,
-	XvImage *image,
-	short srcx,
-	short srcy,
-	unsigned short width,
-	unsigned short height,
-	short dstx,
-	short dsty
-)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	assert(image);
-	
-	if (subpicture->xvimage_id != image->id)
-		return BadMatch;
-	
-	/* TODO: Assert rects are within bounds? Or clip? */
-	
-	return Success;
-}
-
-Status XvMCDestroySubpicture(Display *display, XvMCSubpicture *subpicture)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	return BadImplementation;
-}
-
-Status XvMCSetSubpicturePalette(Display *display, XvMCSubpicture *subpicture, unsigned char *palette)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	assert(palette);
-	
-	/* We don't support paletted subpictures */
-	return BadMatch;
-}
-
-Status XvMCBlendSubpicture
-(
-	Display *display,
-	XvMCSurface *target_surface,
-	XvMCSubpicture *subpicture,
-	short subx,
-	short suby,
-	unsigned short subw,
-	unsigned short subh,
-	short surfx,
-	short surfy,
-	unsigned short surfw,
-	unsigned short surfh
-)
-{
-	assert(display);
-	
-	if (!target_surface)
-		return XvMCBadSurface;
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	if (target_surface->context_id != subpicture->context_id)
-		return BadMatch;
-	
-	/* TODO: Assert rects are within bounds? Or clip? */
-	return Success;
-}
-
-Status XvMCBlendSubpicture2
-(
-	Display *display,
-	XvMCSurface *source_surface,
-	XvMCSurface *target_surface,
-	XvMCSubpicture *subpicture,
-	short subx,
-	short suby,
-	unsigned short subw,
-	unsigned short subh,
-	short surfx,
-	short surfy,
-	unsigned short surfw,
-	unsigned short surfh
-)
-{
-	assert(display);
-	
-	if (!source_surface || !target_surface)
-		return XvMCBadSurface;
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	if (source_surface->context_id != subpicture->context_id)
-		return BadMatch;
-		
-	if (source_surface->context_id != subpicture->context_id)
-		return BadMatch;
-	
-	/* TODO: Assert rects are within bounds? Or clip? */
-	return Success;
-}
-
-Status XvMCSyncSubpicture(Display *display, XvMCSubpicture *subpicture)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	return Success;
-}
-
-Status XvMCFlushSubpicture(Display *display, XvMCSubpicture *subpicture)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	return Success;
-}
-
-Status XvMCGetSubpictureStatus(Display *display, XvMCSubpicture *subpicture, int *status)
-{
-	assert(display);
-	
-	if (!subpicture)
-		return XvMCBadSubpicture;
-	
-	assert(status);
-	
-	/* TODO */
-	*status = 0;
-	
-	return Success;
-}
-
diff --git a/src/xvmc/surface.c b/src/xvmc/surface.c
deleted file mode 100644
index 7c5f45bd34..0000000000
--- a/src/xvmc/surface.c
+++ /dev/null
@@ -1,355 +0,0 @@
-#include <assert.h>
-#include <X11/Xlib.h>
-#include <X11/extensions/XvMC.h>
-#include <vl_display.h>
-#include <vl_screen.h>
-#include <vl_context.h>
-#include <vl_surface.h>
-#include <vl_types.h>
-
-static enum vlMacroBlockType TypeToVL(int xvmc_mb_type)
-{
-	if (xvmc_mb_type & XVMC_MB_TYPE_INTRA)
-		return vlMacroBlockTypeIntra;
-	if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_FORWARD)
-		return vlMacroBlockTypeFwdPredicted;
-	if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_BACKWARD)
-		return vlMacroBlockTypeBkwdPredicted;
-	if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD))
-		return vlMacroBlockTypeBiPredicted;
-
-	assert(0);
-
-	return -1;
-}
-
-static enum vlPictureType PictureToVL(int xvmc_pic)
-{
-	switch (xvmc_pic)
-	{
-		case XVMC_TOP_FIELD:
-			return vlPictureTypeTopField;
-		case XVMC_BOTTOM_FIELD:
-			return vlPictureTypeBottomField;
-		case XVMC_FRAME_PICTURE:
-			return vlPictureTypeFrame;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
-static enum vlMotionType MotionToVL(int xvmc_motion_type, int xvmc_dct_type)
-{
-	switch (xvmc_motion_type)
-	{
-		case XVMC_PREDICTION_FRAME:
-			return xvmc_dct_type == XVMC_DCT_TYPE_FIELD ? vlMotionType16x8 : vlMotionTypeFrame;
-		case XVMC_PREDICTION_FIELD:
-			return vlMotionTypeField;
-		case XVMC_PREDICTION_DUAL_PRIME:
-			return vlMotionTypeDualPrime;
-		default:
-			assert(0);
-	}
-
-	return -1;
-}
-
-Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *surface)
-{
-	struct vlContext *vl_ctx;
-	struct vlSurface *vl_sfc;
-
-	assert(display);
-
-	if (!context)
-		return XvMCBadContext;
-	if (!surface)
-		return XvMCBadSurface;
-
-	vl_ctx = context->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
-
-	if (vlCreateSurface(vlContextGetScreen(vl_ctx),
-	                    context->width, context->height,
-	                    vlGetPictureFormat(vl_ctx),
-	                    &vl_sfc))
-	{
-		return BadAlloc;
-	}
-
-	vlBindToContext(vl_sfc, vl_ctx);
-
-	surface->surface_id = XAllocID(display);
-	surface->context_id = context->context_id;
-	surface->surface_type_id = context->surface_type_id;
-	surface->width = context->width;
-	surface->height = context->height;
-	surface->privData = vl_sfc;
-
-	return Success;
-}
-
-Status XvMCRenderSurface
-(
-	Display *display,
-	XvMCContext *context,
-	unsigned int picture_structure,
-	XvMCSurface *target_surface,
-	XvMCSurface *past_surface,
-	XvMCSurface *future_surface,
-	unsigned int flags,
-	unsigned int num_macroblocks,
-	unsigned int first_macroblock,
-	XvMCMacroBlockArray *macroblocks,
-	XvMCBlockArray *blocks
-)
-{
-	struct vlContext		*vl_ctx;
-	struct vlSurface		*target_vl_surface;
-	struct vlSurface		*past_vl_surface;
-	struct vlSurface		*future_vl_surface;
-	struct vlMpeg2MacroBlockBatch	batch;
-	struct vlMpeg2MacroBlock	vl_macroblocks[num_macroblocks];
-	unsigned int			i;
-
-	assert(display);
-
-	if (!context)
-		return XvMCBadContext;
-	if (!target_surface)
-		return XvMCBadSurface;
-
-	if
-	(
-		picture_structure != XVMC_TOP_FIELD &&
-		picture_structure != XVMC_BOTTOM_FIELD &&
-		picture_structure != XVMC_FRAME_PICTURE
-	)
-		return BadValue;
-	if (future_surface && !past_surface)
-		return BadMatch;
-
-	vl_ctx = context->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
-
-	target_vl_surface = target_surface->privData;
-	past_vl_surface = past_surface ? past_surface->privData : NULL;
-	future_vl_surface = future_surface ? future_surface->privData : NULL;
-
-	assert(context->context_id == target_surface->context_id);
-	assert(!past_surface || context->context_id == past_surface->context_id);
-	assert(!future_surface || context->context_id == future_surface->context_id);
-
-	assert(macroblocks);
-	assert(blocks);
-
-	assert(macroblocks->context_id == context->context_id);
-	assert(blocks->context_id == context->context_id);
-
-	assert(flags == 0 || flags == XVMC_SECOND_FIELD);
-
-	batch.past_surface = past_vl_surface;
-	batch.future_surface = future_vl_surface;
-	batch.picture_type = PictureToVL(picture_structure);
-	batch.field_order = flags & XVMC_SECOND_FIELD ? vlFieldOrderSecond : vlFieldOrderFirst;
-	batch.num_macroblocks = num_macroblocks;
-	batch.macroblocks = vl_macroblocks;
-
-	for (i = 0; i < num_macroblocks; ++i)
-	{
-		unsigned int j = first_macroblock + i;
-
-		unsigned int k, l, m;
-
-		batch.macroblocks[i].mbx = macroblocks->macro_blocks[j].x;
-		batch.macroblocks[i].mby = macroblocks->macro_blocks[j].y;
-		batch.macroblocks[i].mb_type = TypeToVL(macroblocks->macro_blocks[j].macroblock_type);
-		if (batch.macroblocks[i].mb_type != vlMacroBlockTypeIntra)
-			batch.macroblocks[i].mo_type = MotionToVL(macroblocks->macro_blocks[j].motion_type, macroblocks->macro_blocks[j].dct_type);
-		batch.macroblocks[i].dct_type = macroblocks->macro_blocks[j].dct_type == XVMC_DCT_TYPE_FIELD ? vlDCTTypeFieldCoded : vlDCTTypeFrameCoded;
-
-		for (k = 0; k < 2; ++k)
-			for (l = 0; l < 2; ++l)
-				for (m = 0; m < 2; ++m)
-					batch.macroblocks[i].PMV[k][l][m] = macroblocks->macro_blocks[j].PMV[k][l][m];
-
-		batch.macroblocks[i].cbp = macroblocks->macro_blocks[j].coded_block_pattern;
-		batch.macroblocks[i].blocks = blocks->blocks + (macroblocks->macro_blocks[j].index * 64);
-	}
-
-	vlRenderMacroBlocksMpeg2(&batch, target_vl_surface);
-
-	return Success;
-}
-
-Status XvMCFlushSurface(Display *display, XvMCSurface *surface)
-{
-	struct vlSurface *vl_sfc;
-
-	assert(display);
-
-	if (!surface)
-		return XvMCBadSurface;
-
-	vl_sfc = surface->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
-
-	vlSurfaceFlush(vl_sfc);
-
-	return Success;
-}
-
-Status XvMCSyncSurface(Display *display, XvMCSurface *surface)
-{
-	struct vlSurface *vl_sfc;
-
-	assert(display);
-
-	if (!surface)
-		return XvMCBadSurface;
-
-	vl_sfc = surface->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
-
-	vlSurfaceSync(vl_sfc);
-
-	return Success;
-}
-
-Status XvMCPutSurface
-(
-	Display *display,
-	XvMCSurface *surface,
-	Drawable drawable,
-	short srcx,
-	short srcy,
-	unsigned short srcw,
-	unsigned short srch,
-	short destx,
-	short desty,
-	unsigned short destw,
-	unsigned short desth,
-	int flags
-)
-{
-	Window			root;
-	int			x, y;
-	unsigned int		width, height;
-	unsigned int		border_width;
-	unsigned int		depth;
-	struct vlSurface	*vl_sfc;
-
-	assert(display);
-
-	if (!surface)
-		return XvMCBadSurface;
-
-	if (XGetGeometry(display, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable)
-		return BadDrawable;
-
-	assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE);
-
-	/* TODO: Correct for negative srcx,srcy & destx,desty by clipping */
-
-	assert(srcx + srcw - 1 < surface->width);
-	assert(srcy + srch - 1 < surface->height);
-	/* XXX: Some apps (mplayer) hit these asserts because they call
-	 * this function after the window has been resized by the WM
-	 * but before they've handled the corresponding XEvent and
-	 * know about the new dimensions. The output will be clipped
-	 * for a few frames until the app updates destw and desth.
-	 */
-	/*assert(destx + destw - 1 < width);
-	assert(desty + desth - 1 < height);*/
-
-	vl_sfc = surface->privData;
-
-	vlPutPicture(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, width, height, PictureToVL(flags));
-
-	return Success;
-}
-
-Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status)
-{
-	struct vlSurface	*vl_sfc;
-	enum vlResourceStatus	res_status;
-
-	assert(display);
-
-	if (!surface)
-		return XvMCBadSurface;
-
-	assert(status);
-
-	vl_sfc = surface->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
-
-	vlSurfaceGetStatus(vl_sfc, &res_status);
-
-	switch (res_status)
-	{
-		case vlResourceStatusFree:
-		{
-			*status = 0;
-			break;
-		}
-		case vlResourceStatusRendering:
-		{
-			*status = XVMC_RENDERING;
-			break;
-		}
-		case vlResourceStatusDisplaying:
-		{
-			*status = XVMC_DISPLAYING;
-			break;
-		}
-		default:
-			assert(0);
-	}
-
-	return Success;
-}
-
-Status XvMCDestroySurface(Display *display, XvMCSurface *surface)
-{
-	struct vlSurface *vl_sfc;
-
-	assert(display);
-
-	if (!surface)
-		return XvMCBadSurface;
-
-	vl_sfc = surface->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
-
-	vlDestroySurface(vl_sfc);
-
-	return Success;
-}
-
-Status XvMCHideSurface(Display *display, XvMCSurface *surface)
-{
-	struct vlSurface *vl_sfc;
-
-	assert(display);
-
-	if (!surface)
-		return XvMCBadSurface;
-
-	vl_sfc = surface->privData;
-
-	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
-
-	/* No op, only for overlaid rendering */
-
-	return Success;
-}
diff --git a/src/xvmc/tests/Makefile b/src/xvmc/tests/Makefile
deleted file mode 100644
index de095161d2..0000000000
--- a/src/xvmc/tests/Makefile
+++ /dev/null
@@ -1,27 +0,0 @@
-CFLAGS	+= -g -Wall
-LDFLAGS	+=
-LIBS	+= -lXvMCW -lXvMC -lXv
-
-#############################################
-
-.PHONY	= all clean
-
-all: test_context test_surface test_blocks test_rendering xvmc_bench
-
-test_context: test_context.o testlib.o
-	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
-
-test_surface: test_surface.o testlib.o
-	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
-
-test_blocks: test_blocks.o testlib.o
-	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
-
-test_rendering: test_rendering.o testlib.o
-	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
-
-xvmc_bench: xvmc_bench.o testlib.o
-	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
-
-clean:
-	rm -rf *.o test_context test_surface test_blocks test_rendering xvmc_bench
diff --git a/src/xvmc/tests/testlib.h b/src/xvmc/tests/testlib.h
deleted file mode 100644
index af71ad74e1..0000000000
--- a/src/xvmc/tests/testlib.h
+++ /dev/null
@@ -1,42 +0,0 @@
-#ifndef testlib_h
-#define testlib_h
-
-/*
-#define TEST(pred, doc)	test(pred, #pred, doc, __FILE__, __LINE__)
-
-void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line);
-*/
-
-#include <sys/time.h>
-#include <X11/Xlib.h>
-#include <X11/extensions/XvMClib.h>
-
-/*
- * display: IN			A valid X display
- * width, height: IN		Surface size that the port must display
- * chroma_format: IN		Chroma format that the port must display
- * mc_types, num_mc_types: IN	List of MC types that the port must support, first port that matches the first mc_type will be returned
- * port_id: OUT			Your port's ID
- * surface_type_id: OUT		Your port's surface ID
- * is_overlay: OUT		If 1, port uses overlay surfaces, you need to set a colorkey
- * intra_unsigned: OUT		If 1, port uses unsigned values for intra-coded blocks
- */
-int GetPort
-(
-	Display *display,
-	unsigned int width,
-	unsigned int height,
-	unsigned int chroma_format,
-	const unsigned int *mc_types,
-	unsigned int num_mc_types,
-	XvPortID *port_id,
-	int *surface_type_id,
-	unsigned int *is_overlay,
-	unsigned int *intra_unsigned
-);
-
-unsigned int align(unsigned int value, unsigned int alignment);
-
-int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y);
-
-#endif